Movatterモバイル変換


[0]ホーム

URL:


Jump to content
Wikisource
Search

A Welsh Grammar, Historical and Comparative/Phonology

Download
From Wikisource
<A Welsh Grammar, Historical and Comparative
188859A Welsh Grammar, Historical and Comparative — Orthography and PronunciationJohn Morris Jones

Phonology

Orthography and pronunciation

The Alphabet.

§ 7. i. Welsh, in all its periods, has been written in the Latin alphabet.

The ogam inscriptions are Irish. The letters of the ogam alphabet consist of scores and notches on the edge of the stone; one to five scores, cut at right angles to the edge on either side, or obliquely across it, form 15 consonants; one to five notches on the edge form 5 vowels.

The "alphabet of Nemnivus", contained inox., dated 812, and reproduced by Ab Ithel in Dosp. Ed. 10, 11, is stated in thems. to have been formed byNemnivus "ex machinatione mentis suae" in answer to a Saxon's taunt that the Britons had no letters. Most of the signs are forms of Latin characters made to imitate runes; two (ᚾn and ᚢu) are runes, while others seem to be arbitrary inventions. There is no evidence of the use of this alphabet. The "winged alphabet" given by Ab Ithel ibid. 12 consists of two classifications of Scandinavian tree-runes, the top line representing the two schemes of classification. The reason given for supposing the scribe to be a Welshman is too ridiculous to need refutation.

Among the "traditions" invented by the Glamorgan bards in support of their claim to be the successors of the druids was the "wooden book"; though all the accounts of it are inIolo Morgannwg's handwriting, contemporary evidence of its existence in the early 17th cent. is afforded by Rhys Cain's satiricalenglyn (Ab Iolo, Coel. y B. 50); but it cannot be traced further back. The 'bardic alphabet' calledcoelbren y beirdd was a conventional simplification of ordinary characters adapted for cutting on wood; its letters are derived from the handwriting of the period, asb, ∂, (=e),h,n,r, except where it was easier to adapt the Latin capitals, asA,G. With one or two exceptions, such as Ỻ, the "derived characters" denoting consonant mutations, so far from proving the coelbren's antiquity, are its very latest development, Pughe acknowledging himself to be the author of five of them (l.g.c. 260 footnote). Iolo's memoranda (Coel. y B. 27) refer to an old form given by Gwilym Tew in his grammar; but this work is preserved in G.T.'s own hand inp 51, which does not mention the coelbren. The famous transcriber ofmss. John Jones of Gelli Lyfdy compiled two collections of the alphabets known to himp 307, 144, but neither contains anything like the coelbren. Noms. is written in it, for the simple reason that it was easier to write ordinary characters than the coelbren caricature of them. The writing inp 54 pp. 359 ff., stated in ther. to be in "'bardic' characters, which are widely different from Roman characters", bears no resemblance to the coelbren, and is no more "widely different from Roman characters" than the coelbren itself is; it is the hand of an illiterate person; the letters are written separately, but all are clumsy copies of the script characters of the period, mostly formed with awkward curves, the antithesis of the coelbren angles. There is a somewhat similar scribble written upside down on the bottom margin ofb.ch. =p 29, p. 19.—The wooden book consisted of squared inscribed sticks in a frame; it was calledpeithynen from its resemblance to a weaver's reed, and not the reverse, as Iolo asserted, forpeithyn(en) comes regularly from Lat. acc.pectin-em 'comb, weaver's reed'. The absurdity of the supposition that such a device ever served any serious purpose of literature is manifest when one considers what a cartload of wooden books would be required to carry the contents of a small manuscript volume.

ii. The earliest Welsh alphabet given as such is that found in ther.g. col. 1117:a,b,c,d,e,f,g,h,i,k,l,m,n,o,p,q,r,s,t,v,x,y,w,. It containsq, which is not used in Welsh, and omits all the digraphs exceptll; they could not be included in the traditional number, 24.

SirJ. Price's alphabet iny.l.h. (1546) is as follows:a,b,c,d,ď,e,ff,f,g,h,i,k,l,lh,m,n,o,p,r,rh,s,t,v =u,v,y,w.

W. Salesbury gives the following alphabet in his Playne and Familiar Introductiõ, 1567 (written in 1550):A,b,c,ch,d,dd,e,f,ff,g,h,i,k,l,ll,m,n,o,p,r,s,t,th,v,u,w,y. He distinguishes betweenu andv, using the latter for Eng.v, Welshf.

G.R., (1567), who uses,, fordd,ll,w, gives the following alphabet:a,b,c,ch,d,,e,f,g,i,h,l,,m,n,o,p,r,s,t,th,u,,y, omittingng andph (both of which he uses, the latter to the exclusion offf), to make the number 24.

S.V., (1568), gives the following alphabet of 24 letters:a,b,d,dd,e,f,ff,g,i,k,l,ll,m,n,o,p,r,s,t,v,w,y,ch,th, adding thath is the sign of a breathing,j 9/3.

J.D.R., (1592), usedh to form all his digraphs, thusbh =f,dh =dd,gh =ng. His alphabet is as follows:a,b,bh,c,ch,d,dh,e,g,gh,ghh,h,i,lh,l,m,mh,n,nh,o,p,ph,rh,r,s,t,th,u,,y,ỿ. It contains a character for each simple sound in the language, including the two sounds ofy; but it was too cumbrous to win general adoption.

The alphabet of the present day is first met with in D. (1621), with the single difference that D. has two forms of the lettery; thus,a,b,c,ch,d,dd,e,f,ff,g,ng,h,i,l,ll,m,n,o,p,ph,r,s,t,th,u,w,y/ỿ. It omitsmh,nh,ngh,rh. The names now given to the letters are, in the above order, in Welsh spelling (all vowels not marked long to be read short):ā,,ec,ech,,edd,ē,ef,eff,eg,eng,āets,ī,el,ell,em,en,ō,,yff orffī,er,es,,eth,ū,,ȳ. The namesha,he,hi given to the letterh by some writers on Welsh grammar and orthography[1] are figments. The name isāets, borrowed from Eng. or Fr. (Eng.aitch, Fr.ache, Span.atche):

H.arall it sy—Harri
Wyth yw 'r dyn a 'th eura di.—T.A.,c. i 340.

'Thou hast another H.—Henry the Eighth is the man who will ennoble thee.' The first line is to be readAetsarall it sy Harri, as shown by the cynghanedd:t s r—t s r.

Lhuyd, (1707). used χ forch, λ forll, and ꝺ fordd. The last has survived in the form ẟ in ordinary handwriting, but manuscript ẟ is printeddd.

§ 8. The orthography of Mn. W. is almost purely phonetic: each letter of the alphabet has one standard sound, excepty which has two. It will therefore be convenient to give the values of the letters in the modern alphabet, and then, rising the modern characters to represent the sounds of the language, to show in detail how each sound was written in earlier periods, noting any changes which have taken place in the sounds themselves.

The Vowels.

§ 9. The lettersa,e,i,o,u,w,y represent vowel sounds. The following diagram shows the approximate relative positions of the vowels at the present day.ɥ andỿ denote the two sounds ofy. Vowels pronounced with rounded lips are enclosed in brackets. The more open the sound the less the rounding.

The vowel soundsi,e,a,o,w, except in certain diphthongal combinations, have probably undergone no material change from the O. W. period to the present day; the soundsa,e,o, have always been represented by the charactersa,e,o,[2] and the soundi always byi, with some exceptions in Early Ml. W.,§ 16 ii (2).

§ 10. The sound ofa is that of the Englisha infather. It occurs long as intad 'father', medium as in|dol 'fatherly', and short as inmam 'mother'.

The sound does not occur short in English, thea of Eng.man being a more forward sound, which may be denoted byæ. This soundæ is heard in Welsh in a narrow strip stretching from the English border to Harlech, and in Glamorganshire.

§ 11. The sound ofe, when long or medium, is the middlee, as in the Eng.men,let; thusgwên 'smile',gwé|nu 'to smile'; when short it is generally more open, tending towards the Eng.e inthere; thusgwenn 'white'. For its sound in diphthongs, see§§ 29,79.

§ 12. The sound ofi is the closei of the Frenchfini,si, or the North Eng.i inking,machine. The Southern Eng.i is more open. It occurs long as ingwîn 'wine', medium as ingwí|noedd 'wines', short as inprin 'scarcely'.

§ 13. The sound ofo, when long or medium, is the middleo, midway between the closeo in Eng.note and the openo innot; thustôn 'tune',|nau 'tunes'; when short it is more open, tending towards theo ofnot, astonn 'wave',tŏˊnnau 'waves'.

§ 14. i. The sound ofw is that of the Frenchou insou, or the North Eng.oo infood,book. The Southern Eng. sound is more open. It occurs long as ingŵr 'man', medium as ingẃ|rol 'manly', short as intrwm 'heavy'.

ii. (1) The soundw was writtenu in O.W., and thus could not be distinguished (except by the context) from the soundu,§ 15 i, which was also writtenu (though sometimesi,15 ii).

(2) In Early Ml. W., the soundw, both vocalic and consonantal was writtenu (orv) andw, and as the former also represents the soundu, and both represent the soundf, the spelling is often ambiguous. In Late Ml. W. the uncertainty is partly removed by the restriction ofw and the use of (a peculiar shape ofv) to represent thew sounds. The charactersw and represent bothw and almost indifferently. Theoretically perhapsw stood for, and ther.b. scribe wishing to distinguish betweengw̯ŷr 'men' andgŵɥr 'knows' writes themgwyr,gỽyr respectively,r.g. 1118; there seems to be a slight predominance of the value forw, but no systematic distinction is made between the sounds, whole pages frequently occurring, e.g., inw.m., where is used exclusively for both.

☞ In this work Late Ml. W. is transcribedw, as nothing is gained by reproducing a distinction which would often be misleading if taken to have a phonetic significance.

(3) In Mn. W. the sound is represented byw.

G.R. uses; and J.D.R. a peculiar character based onɓ, a late script form of;§ 7 ii.

§ 15. i. (1) In Late Mn. W. the sound ofu, long, medium, and short, is the same as the clear sound ofy,§ 16 i; thus the wordshûn 'sleep' andhŷn 'older' have now absolutely the same sound. But in O. and Ml. W.u had the sound of the Frenchu, that is, ani pronounced with rounded lips. In accented syllables it retained this sound down to the end of the 16th cent., as is shown by the fact that J.D.R. (pp. 33, 34) describes bothu andɥ, and distinguishes between them with a phonetic truth which could only be derived from actual acquaintance with both as living sounds.

(2) In the final unaccented syllable the originalu sound becameɥ as early as the 14th cent.; see ZfCP. iv 118. Hence we findu andy confused from the 14th century on.Kymry 'the Welsh, Wales' often appeared asCymru; seey Cymru 'the Welsh', G.R. p. [v] ; M.Ỻ. (3 Ader.—Title). Later, the misspellingCymru came to be used for 'Wales', the true formCymry being retained as the pl. ofCymro. In the 3rd pl. of prepositions,arnunt 'on them', etc., indywedud 'to say',anoddun 'deep',credadun 'believer',arofun 'intend',munud 'minute',y is in Late Mn. W. wrongly written foru; fortestun 'text',ysgrythur 'scripture', see§ 82 iii (3). The converse error was frequent in the 16th cent., Dr. M. writingfellu,i fynu,gorthrymmudd, etc.

The view that the distinction survived in monosyllables down to a late date is corroborated by the fact that out of about 140 monos. in use containing eitheru orɥ only one,crud 'cradle' (crutỺ.A. 72,r.p. 1418), is now commonly misspelt; and even this misspelling is due to Pughe's bringing the word under the same head ascryd 'quaking, fever' obviously on a false etymological theory. D.D. and Richards havecrud 'cradle',cryd 'fever'.

ii. The O. and Ml. W. sound above described was writtenu. It was therefore not distinguished in writing in the O. and Early Ml. period from the soundw which was also writtenu. We may call O. W.u the frontu, orü, when it corresponds to Mn. W.u, and the backu when it represents Mn. W.w. It is certain that the two sounds were as distinct then as they were later, for in O.W. we find theü sound writteni, as inscipaurjuv. 'barn', Mn. W.ysgubor. Still earlier evidence ofü is furnished by Bede's spellingDinoot of a name which was laterDunawd.

§ 16. i.y has two sounds, the clear and the obscure.

The clear sound ofy is a peculiari-sound very difficult to acquire. It is a dulli produced further back than ordinaryi. The sound is very similar to Frenchu in its effect upon the ear, and has the same absolute pitch; but it is produced quite differently. The Frenchu is ani pronounced with rounded lips, but the Welshɥ is ani pronounced further back, but with open lips; see the diagram,§ 9. Ml. W. had both sounds, writtenu andy respectively; but gradually the rounded sound, which was writtenu, was replaced by the unrounded sound, though still continuing to be writtenu, the result being that Welsh has now the unrounded sound only, writtenu andy.

The soundɥ is long as indŷn 'man' or short as inbrɥn 'hill'. It cannot be medium except when written asu, as inúno 'to unite', and in the wordgɥda forgɥd a,§ 82 ii (2).

In S. W. dialects bothu andɥ are sounded asi or nearly so.

The obscure sound ofy is the sound of the Eng.o inivory. It is medium or short in the penult, or short in an unaccented syllable. It is long in the penult before a vowel orh ascý-oedd,cý-hoedd, and in the name of the lettery.

☞ In this grammar the charactery is used as in ordinary written Welsh to represent both the clear and the obscure sound; but when it is required to distinguish between them, the characterɥ is used to denote the clear, andỿ to denote the obscure sound.—Note thaty is the clearɥ in the diphthongw͡y, and when circumflexed,ŷ.

A special character for the soundỿ was used by some 16th century scribes, and is regularly employed by J.D.R. and Dr. Davies in their grammars. A distinctive character is also needed for the clear sound; andɥ is convenient because it suggestsu which has now the same sound.

Note. The idea thaty has borrowed its clear sound fromu, which, as we have seen, is the exact reverse of the truth, has led some writers to callỿ the primary, andɥ the secondary sound ofy. The former is of course secondary, being the obscured form ofɥ and other sounds.

¶ On the use of the two sounds ofy see§ 82.

ii. (1) In O. W. the sounds ofy are denoted byi, and are therefore not distinguished in writing from the soundi. Thatɥ andi were then distinct requires no further proof than that they are different in origin, and if the difference had been lost it could not have been recovered.

(2) In Early Ml. W.mss., as in theb.b.,y andi are used indifferently to express thei sound and the sounds ofy. Inb.ch. (=a.l. ms. a.)y is used in some parts almost to the exclusion ofi, asbrenyn,tyr forbrenin 'king',tir 'land';yx p. 9 forix 'nine' (printednau ina.l. i 18!) shows that the scribe treatedy andi as identical. In some earlymss. the sounds ofy were represented bye; see the passage in ancient orthography ina.l. ii 36–8, whereỿ lle,ỿ dɥn appear aselle,eden 'the place', 'the man'.

(3) In Late Ml.mss., as in Mn. W., the soundsɥ,ỿ are writteny, and are not confused withi which is writteni (except thaty also represents,§ 25 iii).

In a few monosyllables of frequent occurrence,ɥ by constant repetition advanced to the easier front position ofi towards the end of the Ml. period. These arey 'to',y 'his' or 'her',ny,nyt 'not'. The latter often appears asni,nit inw.m., see 46, 48, showing the thinning of the vowel to be so early. That the sound was onceɥ is shown by the fact thatny̆d, writtennɥdd (dd ≡ doubled, not ẟ) by J.D.R. in 1592, may still be heard in Anglesey.

☞ In this grammar the Ml. W.y 'to' andy 'his' or 'her' are dotted thus,, to distinguish them from the articleyỿ. As the was probably soundedi some time before it came to be so written, it may be readi. [There can be no confusion with, which never stands by itself,§ 25 iii.]

iii. Though not indicated in writing, the difference betweenɥ and ỿ goes back to the O. W. period. That O. W.i represented not only the clearɥ but also the obscureỿ is shown by such forms ascimadas (=cỿfaddas)m.c. Herecyf- comes from *kom‑; theỿ results from the indistinct pronunciation ofo,§ 65 iv (2), and was never soundedɥ; hence the writteni must have meantỿ. See also§ 40 iii (2). In Ml.mss. generally, as in Mn. W., no distinction is made betweenɥ andỿ. But in some parts ofb.ch.,e stands forỿ, andy forɥ regularly; thusYlety yu ety muyhaf ene tref akemeruedaf ac y kyd ac ef erey auenno or teylu,a.l. i 12 ≡ɥ letɥ ɥw ỿ tɥ mwɥhaf ỿnỿ dref a chỿmherfeaf, ag ɥ gɥd ag ef ỿ r͑ei a vỿnno o'r teilu, 'His lodging is the largest and most central house in the town, and with him such as he may please of the household.' The scribe's observance of the rule is remarkable; and though there are many slips due to mechanical copying, his spelling in some cases helps to decide the sound in obsolete forms.

iv. (i) In Early Ml. W.ɥ andỿ were probably nearere than at present. If we assume the lineaɥ more inclined towards the lineai in the diagramp. 11 above, it will be seen at a glance not only why both were writtene at that time, but why theb.ch. scribe usesy to represent bothi andɥ, ande to represent bothe andỿ.

(2) The soundsɥ andỿ in these forward positions were less stable, being not merely felt to be near enough toe to be represented bye in writing, but also liable to be confused withe in speech. Some examples of this confusion survived, and are met with in the later language: (α) Interchange ofỿ ande:Mỿrddin,Merddin D.G. 471;tỿmestl,temestl G. 153;ỿstɥnf. 24,estɥn;cỿbỿddiaeth,a chebỿẟẏaeth.a. 144;ẏ bellỿnnic.a. 126, 146,pellennig;ketỿmdeith,cỿdỿmaith;ỿnnill,ennill;cỿnfigen,cenfigen;Tâl-ỿ-boli̯onm.a. i 315a, explained astâl ebolẏonw.m. 45;Pen-e´-goes for *Pen-ỿ´-goes, see§ 46 ii (3).—(β) Interchange ofɥ ande:velle.a. 148 forfellɥ; Late Mn. W.wele 'behold' for (a)welɥ 'dost thou see?' § 173 iii (3);Mercher forMerchɥrb.a. 17, b.b. 48, see§ 69 v;hwdɥc.m. 31,hwder.m. 173;mɥwn,mewn;Llɥɥn,Lleɥn. Dial.edrech foredrych,‑ech for‑ɥch 2nd sg. pres. subj.§ 176 iv.—(γ) In Ml. W.ɥ hun 'himself, herself' is writtene hun, thee modification being preferred owing to the difficulty of sounding unroundedɥ and roundedü in consecutive syllables, cf.§ 77 viii. Dissimilation also occurs ine Iwerẟonw.m. 59 forɥ Iwerẟon. Similarlyte|ɥrn for *tɥ|ɥrn§ 103 ii (1);diell fordi-hyll§ 146 ii (2).

In Breton *y has generally becomee; thusnevez = W.newydd;pemp = W.pɥmp;kevrann = W.kyvran;ened = W.ỿnɥd.

(3)ỿ before a nasal tended to be lowered towardsa, and is sometimes writtena in theb.ch., ascantafa.l. i 84 forcỿntaf;kannal, do. 154 forkỿnnal;kafreiht do. 130 forkỿṽreith. Henceỿ anda interchange before a nasal:Yngharad,Angharad;ymherawdr,amherawdr;ỿmddifad,amddifad;canhorthwy,cỿnhorthwy;mỿnach,manach, etc.

Unaccenteda is sometimes weakened toỿ in the dialects, but examples are rare in lit. W.:rhỿglỿddu ‘to merit’, forrhaglỿddu, seer͑aclỿẟeiw.m. 428.

(4) In Mn. W. sinceỿ has become quite neutral, it is apt in some cases to be coloured by neighbouring sounds: after or followed byw in the ultima, it becomesw,§ 66 ii. When immediately followed by another vowel it is assimilated to it,§ 82 ii (3).

v. (1) In Ml. W. an inorganicy is written between two consonants at the end of a word in the following groups: 1. cons. +r,l orn; 2.rm,rf,lm,lf; 3.ẟf; 4. rarelyrch,lch; thuspobyl forpobl ‘people’,vy manryfw.m. 59 forfy marf ‘my beard’. In O. W. it appears asi, as inreatirjuv., Mn. W.rhaeadr ‘cataract’, but is of rare occurrence, being usually omitted as in Mn. W., thuscruitr,discljuv.datl,scriblox. It occurs medially asi incenthiliatjuv. ‘singer’ forcenthliat, aso incenitolaidouox., Mn. W.cenedlaethau ‘generations’. In Early Ml. W. it appears asi,y, ande, asperygilb.b. 31 ‘danger’,cathil do. 16 ‘song’,autyl do. 15 ‘ode’,colovena.l. i 10 ‘column’. It occurs sometimes in initial groups:o gynaudb.b. 84 ‘of flesh’;keleuueta.l. i 40 ≡clỿwed ‘to hear’.

(2) The sound intended to be represented was the glide between the consonants, which was becoming perceptible as a dull sound resemblingỿ. It was naturally writteni in O. W.,e inb.ch., these being the signs forỿ, see iii above. It was not written where no audible glide developed, as innt,rth,rẟ, and was rare where the glide was voiceless, as beforech. It did not form a full syllable in Ml. W., at least in the standard pronunciation, for (α) it is occasionally written in groups where it is generally omitted, and which seem never to have been syllabic, as inmeirychw.m. 41 ≡meirchr.m. 28 ‘horses’; (β) it is sometimes found medially where it could not be syllabic, as inkenedyloeẟ.a. 11 ≡kenedloeẟ.a. 169 ‘nations’,dadeleua.l. i 20 ≡dadleu ‘lawsuit’; (γ) it does not affect the accentuation ; thus in

|lofyn gweẟ ḗ|ofyn y gwe|ẟī́|eu,—r.p. 1239

‘Upholder in fearless manner of prayers’, thee ofḗofyn is accented to correspond to thei ofgweẟī́eu; (δ) it does not count as a syllable in Ml. verse; the above is a line of nine syllables; in the following cywydd couplet the cynghanedd requireschalych to be read as an absolute monosyllable, as it is pronounced at the present day:

Pwy a allei, pei pennsaer,
peintẏaw achalychpwynt vy chwaer?—I.G.,r.p. 1408.

‘Who could, though he were a master, paint with chalk my sister’s mien?’

☞ In the quotations in the present work this non-syllabicy is represented thus,ɏ.

(3) In Mn. lit. W. the epentheticɏ is simply dropped; thuspobl,ffenestr,ofn. The non-syllabic pronunciation continued to be the only one admissible in cynghanedd, and so remained the standard literary form; and the mutey came to be dropped in writing to prevent ambiguity. [In one form of cynghanedd, however, exemplified by—

Da osódiad hyd i sawdl.—D.N.,g. 158,

‑l answers a syllable‑iad in the cynghanedd, though it does not count as a syllable in the metre, an inconsistency which shows that such a word as this, treated as a monosyllable in verse generally, sounded like a disyllabic when it ended a sentence.]

In the spoken language, when the word was disyllabic the final liquid was lost, thusperig,ffenest forperigl ‘danger’,ffenestr ‘window’, or metathesized as inewyrth forewythr ‘uncle’. In monosyllables the glide was assimilated to the vowel of the syllable or the second element of its diphthong and became syllabic; thuspobol,cefen,llwɥbɥr,sowdwl,bara’ forpobl ‘people’,cefn ‘back’,llwybr ‘path’,sawdl ‘heel’,barf ‘beard’. Some examples of this assimilation already appear in Late Ml. W., asbudur.a. 18 ‘dirty’,kwbwlc.m. 87 ‘all’,vy marafr.m. 42 ‘my beard’.—The colloquial syllabic pronunciation is the one generally implied in recent verse in the free metres; thus Anne Griffiths’sLlwybr cwbl groes i natur, though so printed in all hymn-books, is intended to be sungLlẃybyr |cẃbwl |gróes i |nátur. But in N. W. dialects the parasitic vowel did not arise in groups containingf; thus in the greater part of N. W.ofn, ‘fear’,cefn ‘back’,llyfr ‘book’,barf ‘beard’ are purely monosyllabic to this day. Forms likemarch,calch are everywhere monosyllabic.

¶ For prostheticỿ- see§ 21 iii,§ 23 ii,§ 26 vi (4).

The Consonants.

§ 17. The values of the letters representing consonants in the Mn. alphabet are as follows:

i. Voiceless explosives (tenues):p ≡ Englishp;t, normally more dental than Eng.t, but varying to Eng.t;c ≡ Eng.k, having two sounds, frontc () beforei,e, likek in Eng.king, backc (q) beforea,o,w,uỿ, likec in Eng.coal.

ii. Voiced explosives (mediae):b ≡ Eng.b;d corresponding to W.t as above;g front and back (,ɡ], like Eng.give,go.

iii. Voiceless spirants:ff orph ≡ Eng.f, labiodental;th ≡ Eng.th inthick (which may be denoted byþ);ch ≡ Scotchch inloch, Germanch innach (χ), butnot Germanch inich (χ̑). Even aftere andi, as inllēch ‘slate’,gwīch ‘squeak’, the ch is the back sound χ.

i + back χ is an awkward combination, and becomes difficult in the short time available when thei is the second element of a diphthong; hencebaich,braich are generally pronouncedbăɥχ,brăɥχ (with the shorta of the original diphthong). This pronunciation is condemned by D., p. 10; but the spellingay is common earlier, e.g. J.D.R. 271. Butbeichiau,breichiau are so sounded, with back χ (not χ̑).

iv. Voiced spirants:f ≡ Eng.v, labiodental;dd ≡ Eng.th inthis (ẟ). O.W. had also the guttural voiced spirant, which may be represented by ᵹ, corresponding toch; see§ 19 i.

v. Voiceless nasals:mh;nh;ngh. The nasals can only be made voiceless by a strong emission of breath, which causes a distinct aspirate to be heard as a glide after the consonant. Thusnh is somewhat similar to Eng.nh ininhale.

vi. Voiced nasals:m;n;ng. The last has two positions corresponding to those ofg, namely frontŋ̑, backŋ.

vii. Voiceless liquids:ll;rh. The former is a voicelessl pronounced on one side. It is produced by placing the tongue in thel position, raising it so as to close the passage on one side, and blowing between it and the teeth on the other. The common imitationthl conveys the effect of the “hiss” (voiceless spirant) in theth, and gives the side effect in thel. Butll is of course a simple sound, which may be described shortly as a “unilateral hiss”. The sound ofrh is the Welsh trilledr made voiceless by a strong emission of breath, causing an audible aspirate glide after it. Briefly, it isr andh sounded together.

viii. Voiced liquids :l;r. The latter is trilled like the strong Scotchr, or the Italianr. The trilledr is a difficult sound to acquire; young children usually substitutel for it. A few never acquire it, but substitute for it a gutturalr (≡ ᵹ). This is almost the only defect of speech to be found among speakers of Welsh; it is calledtafod tew ‘thick tongue’.

ix. Sibilant:s. Welsh has noz; such a pronunciation aszêl ‘zeal’ is pure affectation; unsophisticated persons saysêl,selog. Before as ineisi̯au,s now tends to become Eng.sh, and in some S.W. dialects afteri. But many old speakers cannot pronounceshibboleth at all. Standard Welshs is thess inhiss.

x. Aspirate:h. The aspirate is distinctly sounded, and is never misused except in Gwent and Glamorgan. It is really the voiceless form of the vowel which follows it, or the glide between a voiceless nasal or liquid and a vowel.

xi. Semi-vowels:i;w. As these letters also represent vowel sounds, they will be marked, in this work where it is necessary to point out that they are consonantal. is the sound of the Eng.y inyard; is the Eng.w inwill.

¶ Welsh is the same sound as that which is written in the hypothetic forms of Ar., Kelt., Brit., etc. Thus Mn. W.w̯ir ‘indeed’ is identical with the first syll. of Kelt. *u̯īr-os ‘true’ < Ar. *u̯ēr-os.

§ 18. i. The charactersp,t,c had the values in O.W. of modernp,t,c. They also represented the mutated soundsb,d,g, see§ 103 iii; as inscipaurjuv.scubawr, Mn. W.ysgubor ‘barn’,creaticaulox. ≡creadigawl, Mn. W.creadigol ‘created’. When they have this value they are sometimes doubled; thus inm.c. we findcatteiraul, Mn. W.cadeiriol ‘cathedral’ adj.,carrecc, Mn. W.carreg ‘stone’,hepp, Ml. W.heb ‘says’. Possibly this is due to the influence of Irish spelling. [In Old Ir. original *nt > *d-d writtent and sometimestt.]

ii. In Ml. W.p,t,c no longer representb,d,g medially, but finally after a vowel they continued to do so even down to the Mn. period. The facts are briefly as follows: In theb.b., late 12th cent., the final labial is writtenp, but oftenb (mab 27, 28, 29); the dental is alwaysd, becauset is used for the soft spirant ẟ; the guttural is alwaysc. In the 14th cent. the labial very generally appears asb, though often asp; the dental is alwayst, the guttural alwaysc. In the 15th cent. (e.g. 28) we haveb,d,c. In the 1620 Bibleb,d,g, butc in many forms,unic,lluddedic, etc. The finalc is still written inac andnac, which should beag,nag,§ 222 i (1),ii (3). On the sound of the consonant in these cases see§ 111 v (4).

Finally after a consonantp,t,c have always represented the voiceless sounds.

iii. In Ml. W. and Early Mn. W., initialc is generally writtenk. The chief exceptions are the combinationscl,cr. Medially we findc,k,cc,ck. Finally after a consonant, though we generally havec, we also findk (or evenck); asgrafangk,oerdrangkr.p. 1321,diag̃k etc. do. 1314,digelk do. 1364,Iorkr.b.b. 397,carbunck,.a. 170. In these words the sound was, and is, voiceless. Note that after a vowel, where the sound is nowg, it is never writtenk in Ml. W. Thusk, which represents the tenuis only, is clearly distinguished fromc, which also finally represents the media.

Note. In O. W. and the earliest Ml. W., as inl.l. (about 1150),c alone is used;k appears inb.b. and was general in Ml. and Early Mn. W. G.R. discardedk on the principle of “one sound one letter”, p. 20. But the decisive factor in its banishment from the Welsh alphabet was its replacement byc in Salesbury’s N.T., published the same year (1567). This being one of the many innovations “quarrelled withall” in his orthography, Salesbury, in the Prayer Book of 1586 gave his reason for the substitution: “C for K, because the printers haue not so many as the Welsh requireth,” Llyfryddiaeth 34. It is curious to note that a letter which was thus superseded because of its greater prevalence in Welsh than in English was classed 160 years later among “intruders and strangers to the Welsh language”,Gormesiaid a dieithriaid i’r Iaith Gymraeg, S.R. (1728) p. 1.

§ 19. i. The charactersb,d,g, in O.W. represented initially the modern soundsb,d,g; but medially and finally they stood for the mutated soundsf, ẟ, ᵹ, as ingilbinjuv., Mn. W.gylfin ‘beak’,guirdglasm.c.gw̯ỿrẟᵹlas, Mn. W.gwyrddlas ‘greenish blue’. Medially and finallyf was also represented bym, though in this case the spirant was doubtless nasalized then, as it is still normally in Breton; thusnimerox.niṽer, Mn. W.nifer ‘number’,heithamox., Mn. W.eithaf ‘extreme’.

ii. (1) In Ml. W.,b represented the soundb, but no longer the soundf.

(2) The soundf was written in Early Ml. W.u orv,w andf; thus inb.b.,niuer 7 ≡nifer;vaur 21 ≡fawr ‘great’;sew 45 ≡sef ‘that is’;dihafal 20 ≡dihafal ‘unequalled’. We also findff, asaffv 21 ≡a fu ‘who has been’,bariffvin 53 ≡barfwyn ‘white-bearded,’tiff 50 ≡tyf ‘grows’.

Asu andv also represented the vowelü, and asu,v, andw representedw as well, the orthography of this period is most confusing.

(3) In Late Ml. W. the soundf was written mediallyu orv andfu; finally it was represented byf regularly (the few exceptions which occur, e.g. inw.m., being due to mechanical copying). Thus,.a.,vy 2 ≡ fy ‘my’;llauur 3 ≡llafur ‘labour’;kyfuoethawc 55, Mn. W.cyfoethog ‘rich’;gyntaf 3 ‘first’,dywedaf 3 ‘I say’,ef 3 ‘he’, etc. u andv continued to be used medially forf during the Early Mn. period; but G.R. hasf everywhere, and was followed by Dr. M. in the 1588 Bible, which fixed the Late Mn. orthography.

Asu andv also represented the vowelü, the wordfu may be found writtenvv,vu,uv,uu. But there is much less confusion than in the earlier period, for (1)w is distinguished fromü; (2) finallyu andf are distinguished; thusnev meansneu ‘or’, notnef ‘heaven’.

The distinction between the charactersu andv is a modern one; doublev (i.e.w) is still called “double u” in English.

☞ In the quotations in this grammar the letteru orv (for it wasone letter with two forms) is transcribedu when it stands for the vowel, andv when it represents the consonantf, irrespective of the form in theMS., which depended chiefly on the scribe’s fancy at the moment.

(4) The sound which is now the labiodentalf (≡ Eng.v) was in O. W. and probably also in Ml. W. a bilabialƀ, like the South Germanw. It was the soft mutation ofb orm, and resulted from these bilabial sounds being pronounced loosely so that the breath was allowed to escape, instead of being stopped, at the lips. It was sometimes confused with,§ 26 v; and was so soft that it might, like, be passed over in cynghanedd, e.g.pwyntvy chwaerp. 17 above; see Tr. Cym. 1908–9, p. 34.

iii. (i) The letterd in Ml. W. stands for bothd anddd (ẟ).

(2) In some Early Ml.mss., of which the most important is theb.b., the sound ẟ when it is an initial mutation is generally represented byd, but medially and finally is represented rather illogically byt; thusb.b.,dy divet 19 ≡dy ẟiweẟ ‘thy end’;imtuin 32 ≡ymẟwyn ‘to behave’;guirt 33 ≡gw̯yrẟ ‘green’;betev 63 ≡beẟeu ‘graves’. Medially, however, we also haved, asadaw 41 ≡Aẟaf ‘Adam’; and occasionally, by a slip, finally, asoed 1 ≡oeẟ ‘was’ (conversely, by a rare slip, finaltd, asimbit 70 ≡ym myd ‘in the world’). Inb.ch. usage is still looser.

(3) In the Late Ml. period the sound ẟ is represented byd, rarely bydd, see.a. p. xxii. Initially and mediallyd and ẟ cannot be distinguished at this period, but finally they can, since finald is writtent,§ 18 ii, so that finald must mean the sound ẟ. But it often happens that‑d for‑d and‑t for ‑ẟ are copied from an earlierms.

Whilew. is distinctly Late Ml. W. in the representation ofw,i,y, it has‑d for‑d and medial and finalt for ẟ; also occasionallydd, asar dderchet 120a ≡arẟercheẟ.

(4)dd came generally into use in the 15th cent. In the 16th Sir J. Price, 1546, used; G.R., 1567, used; Salesbury, 1567, useddd and; Dr. M. in the Bible, 1588, useddd, which in spite of J.D.R.’sdh, 1592, has prevailed.

☞ In this grammar Ml. W.d when it stands fordd (ẟ) is transcribed ẟ.

iv. (1) In Ml. W. the letterg stands initially and medially for the soundg. The voiced spirant ᵹ had then disappeared.

(2) Butg is also used as well asng for the soundng (ŋ) (as in Eng.song). When final,g must mean the nasal, for the explosive is writtenc,§ 18 ii; thusllogb.b. 90,w.m. 180,r.m. 87 must be readllong ‘ship’.

☞ In this work Ml.g when it represents the nasalng (ŋ) is transcribed.

(3) Mediallyng sometimes stands forn|g (pronouncedŋg like theng in the Eng.finger); thusBangor, pronouncedBaŋgor. The simple sound represents originalŋg as in angel (≡aŋŋel§ 54 i (2)) < Lat.angelus (≡aŋgelus); the composite sound occurs where the nasal and explosive came together later, and theg is the soft mutation ofc, as inBan-gor, radicalcor;un-glust ‘one-eared’,clust ‘ear’. In O. W. the composite sound appears asnc, asuncenetticionm.c.un-genedigion, gloss on ‘solicanae’. Cf. Bede’sBancor, doubtless the Early W. spelling.

§ 20. i. (i) The soundff is represented in O.W. byf, asfinn,fionoum.c. =ffɥnn ‘sticks’,ffionou ‘roses’; sometimes medially byph as inciphillionm.c. ‘sprouts’,grephioum.c. ‘pencils’,Griphiuda.c. 814,§ 36 ii, andp orpp asGripiudb.s.ch. 1,Grippi(ud)gen. xxx.

(2) In Ml. W. the soundff is represented initially byf, both when it is radical and when it is a mutation ofp, though in the latter caseph is perhaps more usual; rarely we haveff; thusban foherb.b. 5 ‘when they are put to flight’,fort do. 33 ≡fforẟ ‘way’,ny forthint do. 34 ‘they did not cherish’,ny phercheiste do. 21 ‘thou hast not respected’;A fa le e maynta.l. i 160,ms. a.,a phy …ms. d., ‘and where they are’;heb ẟant yn ẏ fennw.m. 453 … yn ẏ phennr.m. 101 ‘without a tooth in her head’;ffoesb.b. 44 ‘fled’. Medially and finally it is generallyff, asdiffuisb.b. 35 ≡diffwys ‘steep’,proffuid do. 85 ‘prophet’,grofftr.m. 52 ‘croft’,anffurvaw do. 29 ‘to disfigure’,gorffen do. 5 ‘to finish’,sarff do. 186 ‘serpent’,hoffw.m. 72 ‘desirable’. It also appears asph, ascorphb.b. 20 ‘body’, (g)orphen do. 76 ‘end’; and often asf, asdeu gorfr.m. 5 ‘two bodies’,anfurɏf do. 29 (≡anffurf) ‘disfigurement’,yn brafw.m. 53 (≡yn braff) ‘strong’,groft do. 73 ‘croft’.

(3) In Mn. W.ff andph are used, the latter generally as a mutation ofp only; but G.R. and J.D.R. useph exclusively.

Many modern writers useph in all positions where they perceive that it is derived fromp, as incorph < Lat.corpus, writingff where it does not appear to them to be so derived, as incyff ‘stem, trunk’,ffon ‘stick’. It is mostly a distinction without a difference:cyff comes from Lat.cippus, andffon is from Pr. Kelt. *spond‑,§ 96 iv (1). The attempted differentiation is a useless one; and as the etymology of too many words is still uncertain, it cannot be carried out. It is better, therefore, to writeff always where the sound is immutable, andph only as a conscious mutation of initialp; thuscorff,cyff,ffon;chwe phunt,chwephunt ‘£6’,gwragedd a phlant ‘women and children’,blith draphlith ‘higgledy-piggledy’.

ii. (1) The soundth (þ) is represented in O. W. byth, asbrithjuv. ‘variegated’; byd, aspapedpinnacm.c.pa beth bynnag ‘whatsoever’; byt afterr, asgurtox.gwrth, Mn. W.wrth ‘against’; and byþ, aspapeþjuv.pa beth ‘what’.

(2) In Ml. W. the sound is generally writtenth, though in some earlymss., asb.ch., sometimest (afterr) askemyrta.l. i 4 ≡kỿmɥrth ‘took’. In Mn. W. it is always writtenth.

Such a form asperffeidẏaw.a. 19 is no exception to the rule. Theth had been voiced todd, and the word wasperffeiddi̯aw. It is so written in Early Mn. W., and the Late Mn. W.perffeithio is a re-formation. See§ 108 iv (2).

iii. (1) The soundch (χ) is writtench in O. W., asbichanox. ≡bỿchan ‘little’. Once we havegch, iniurgchellm.c. ‘fawn’, Mn. W.i̯ỿrchell.

(2) The sound is writtench consistently in Ml. and Mn. W., and there seem to be no variations to note.

§ 21. i. The soundsmh,nh, andngh were writtenmp,nt, andnc in O. W.; andmp,nt andnc,ngk, orgk in Ml. W. These combinations continued to be written throughout the Ml. period, though the modern signs appear as early asw.m. or earlier; see§ 107.

In Early Ml. W. we also findm formh,n fornh, andg forngh; see§ 24 i.

ii. The lettersm,n,ng have always represented the soundsm,n,ŋ; butm also represented in O. W.,§ 19 i;ng may representŋg in Ml. and Mn. W.; andŋ was also writteng in Ml. W.;§ 19 iv.

iii. Initialn has sometimes a prostheticỿ‑; asyrwng eyniver ef acyniver y llys … yryniveroeẟw.m. 40 ‘between his host and the host of the court … the hosts’. It is also writtena asanadreẟc.m. 21 ‘snakes’,anniverw.m. 65.

§ 22. i. In O. W. the soundll was writtenl initially, andll medially and finally; asleillox. ‘others’,lennm.c. ‘cloak’guollungjuv.gw̯ollwng ‘release’. Indluithruimjuv., if rightly analysed intollwyth ‘weight’ andrhwyf ‘oar’, we havedl- forþl‑, the usual imitation of thell sound,§ 17 vii, proving the sound to be as old as the 9th cent., though then usually writtenl- initially. The imitationthl is common in the earliest Norman records, but has not been used by Welsh writers.

ii. In Ml. W. thell sound is represented byll; in somemss., e.g. ther.b., it is ligatured thus, enabling it to be distinguished from doublel as incallonr.m. 106 ‘heart’,Iollor.p. 1369, 1407,kollynr.b. 1073 ‘pivot’, which we now writecalon,Iolo,colyn,§ 54 ii. The ligatured capital has been used from the Ml. period to the present day in lettering done by hand.

iii. In Mn. W.ll is used.

Several attempts have been made from time to time to find substitutes: G.R. used, Sir J. Price and J.D.R. usedlh; Ed. Lhuyd usedlh andλ; butll has held the field.

iv. The soundrh was writtenr in O. and Ml. W. The scribes user forrh even when theh has a different origin, and sometimes even when it belongs to another word, as iny gwanwyn arafr.b.b. 194 fory gwanwyn a’r haf ‘the spring and summer’.

☞ Ml. W.r forrh is transcribed in our quotations.

v. In the late 15th and early 16th cent. the soundrh was represented byrr andR; it was not until the middle of the 16th cent. that the present digraphrh, which seems to us so obvious and natural a representation of the sound, came into general use.

vi. The soundsl andr have always been represented by the lettersl andr.

§ 23. i. The sounds has always been writtens. In O. W. it is sometimes doubled as indrissijuv.drỿssi ‘thorns’,issm.c., Ml. W.ys ‘is’. In Ml. W. it is usually doubled medially between vowels, as inIessub.b. 25, 50,.a. 1, 19, etc.,Saessonb.b. 48,messurb.b. 3 ‘measure’, etc., but sometimes written single as inSaesonb.b. 60. Initialss also occurs, asssillitb.b. 99 ≡syllyẟ, Mn. W.sylli ‘thou gazest’.z fors is rare:tryzor.a. 17 ‘treasure’.

ii. Initials followed by a consonant has developed a prostheticỿ- (writteny,e,i, etc.§ 16), as inỿsgol ‘school’.

It is not derived from the late Lat. prosthetici- as iniscola, since Corn., Bret., Ir.scol do not show it, and it appears in native words in W., asystrad. It arose in W. for the same reason as in late Lat., a syllabic pronunciation ofs- after a consonant. The earliest recorded examples areIstrat,Estrat, besideStrat inl.l. see its index s.v.Istrat. In the spoken language it is not heard except in words in which it is accented, asỿ́sgol,ỿ́strad,ỿ́sbrɥd, etc., and sometimes in derivatives of these, asỿsgóli̯on; butsgúbor,stródur,sgrífen,strɥ̄́d. In O. W. it is not written:scipaurjuv.,stroturm.c.,scribennm.c. In Early Ml. W. we havegwastavela.l. i 4 ≡gwas-stavell for the latergwas ystavellw.m. 183,r.m. 85. In the oldest verse it does not count as a syllable:

StavellGynẟylan ys tywyll heno (10 syll.)r.p. 1045.

‘The hall of Cynddylan is dark to-night.’ In later verse it usually counts after a consonant and not after a vowel:

Mi Iscolan yscolheic (≡Mi ’Scolan ỿscolhe|ïc, 7 syll.)b.b. 81.

‘I am Yscolan the clerk.’ But inb.b. 91 we seem to havescolheic afterwyd, see§ 41 iii (2).

Maesgrifenuwchben y bedd.—L.G.C. 20.

‘There is a legend above the tomb.’

Damasg a roed am isgrîn.—T.A.,a 31101/115.

‘Damask was spread over his coffin.’

Acysgrîni geisio gras.—D.G. 60.

‘And a coffin to seek grace.’

Theỿ- was general in late Ml.mss., but it is possible that when unaccented the actual spoken sound consisted of a gradual beginning of thes, which like a vowel preserved ther of the article, etc. G.R., 1567, says thatyr is used beforest,sc,sp, asyr stalwyn, though some writeyr ystalwyn, p. 68. He himself also writesag scrifennu, p. 69, etc. In the 1620 Bible we findsceler,sclyfaeth,scrifennedic, butyscubor,yspeilio,yscrifen, each word generally written in the same way whether it follows a vowel or a consonant. Ther of the article is retained before forms withoutỿ‑, asyr scrifenyddion Barn. v 14, Matt. vii 29. Theỿ- is introduced more freely in the 1690 edition; but its insertion everywhere is late, and of course artificial, since it never became general in natural speech.

§ 24. i. The letterh has always been employed to denote the aspirate; but it was not used to represent the aspirate glide after until the modern period,§ 22 iv; and in some Early Ml.mss.mh,nh andngh were writtenm,n andg, asemen (≡ymhen)a.l. i 84,eurenynes (≡y vrenhines) do. 4;vy g̃erenhytw. 3a (≡vy ngherennhyẟ);yg̃ g̃adellig̃ do. 9a (≡yng Nghadelling).

ii. In O. and Ml. W.h seems also to have been used to denote a voiced breathing ; see§ 112.

§ 25. i. Consonantal is represented in O. W. byi, asiarjuv.i̯âr ‘hen’,hestoriouox., pl. ofhestawr,cloriouox., Mn. W.clori̯au ‘boards’,mellhionoum.c., Mn. W.meilli̯on ‘clover’. Before‑oü it is also found asu (onceiu), asenmeituouox., Mn. W.amneidiau ‘beckonings’,damcirchinnuoujuv. ‘circuits’;dificiuoujuv. ‘defects’; here it was probably rounded intoü in anticipation of the finalü; cf.§ 76 iii (3). Where it is the soft mutation of front it appears asg in O. W., as inUrbgen in Nennius ≡Urfi̯en, Mn. W.Uri̯en;Morgengen. xxv ≡Mori̯en. Here the was doubtless heard with more friction of the breath being the spirant ᵹ̑ corresponding to front g̑; see§ 110 ii.

ii. In Early Ml. W. is represented byi, except inmss. wherey is used fori,§ 16 ii (2); thustirionb.b. 26, pl. oftir ‘land’,dinion do. 45 (≡dỿni̯on) ‘men’.

iii. In late Ml. W. it is represented initially byi, rarely byy; asIessu,b.b. 25, 50,.a. 1, 19, etc.,Ieuan.a. 78,iarll,iarllesw.m. 136 ‘earl, countess’,iawnr.m. 16 ‘right’,ẏawnhaf do. 24 ‘most proper’,Yessu,Yiessu,.a. 100. Medially it is writteny, asdynnẏonw.m. 32 ‘men’,bedyẟẏaw do. 32 ‘to baptize’,meẟylẏaw do. 34 ‘to think’, etc., etc., rarely asi, asymbilior.m. 3 ‘he may entreat.’

☞ Wheny represents it will be dotted as above in the quotations in this book.

iv. In Mn. W. is writteni; but oftenj in the 18th cent., see e.g. Llyfryddiaeth 1713, 4; 1748, 4, 8; 1749, 2.

v. Voiceless occurs where the word or syllable preceding causes aspiration, and is writtenhi (alsohy in Ml. W.), asẏ hiarllaeth R.M. 178 ‘her earldom’,kennhẏaduỺ.A. 79 ‘to consent’.

If pronounced tenselyhi̯ becomes the palatal spirant χ̑ as in the Germanich, but this does not occur in Welsh:hi̯ remains a voiceless semi-vowel. Cf.§ 17 iii.

§ 26. i. Consonantal is writtengu in O. W. as inpetguarox. ≡pedw̯ar ‘four’. See§ 112 ii (1).

ii. In Early Ml. W. is represented byu,v, andw; in Late Ml. W. byw and. Its representation is the same as that of the vowelw; see§ 14 ii (2). In Mn. W. it is writtenw.

The letterw sometimes appears in the formuu, as inkeleuueta.l. i 40 (≡cɏlỿwed) ‘to hear’.

iii. Initialw̯- had becomegw̯- in the Early Welsh period; see§ 112 ii (1); but it isw̯- under the soft mutation, thusgw̯allt ‘hair,’dỿ wallt ‘thy hair’.

Initialgw̯ may come beforel,r orn, as ingw̯lad ‘country’,gw̯raig ‘wife’,gw̯nâf ‘I do’, each one syllable. The initial combinations are practicallygl,gr orgn pronounced with rounded lips, the rounding taking place simultaneously with the formation of theg, so that the off-glide of theg is heard as. When theg is mutated away the initial isl,r orn with as an on-glide; thusdỿ w̯lad ‘thy country’ sounds likedỿw̯ lā́d, except that the syllabic division isdỿ | w̯lā́d.

iv. In Ml. and Early Mn. W. finalw after a consonant was consonantal; see§ 42. Now thew is made syllabic.

The exceptions to the rule were forms in which‑w represents earlier‑w͡y, ashwnnw; Mn. W.acw, Early Ml. W.raccw, Ml. W.racko;assw,gwrw,banw§ 78 i (2). It may have been made consonantal in the last three by analogy, coming afters,r, singlen.

v. Medial is liable to interchange withf; thuscawod,cafod ‘shower’;cyfoeth,cywaeth§ 34 iv;diawl ‘devil’ for *diafl. The old verbal noun fromlliw ‘colour’ isllifo ‘to dye’, a newer formation islliwio ‘to colour’. The reason for the interchange is thatf was once a bilabial,ƀ,§ 19 ii (4), and so, very similar to, being in effect with friction of the breath at the lips instead of at the back.

vi. (1) Voiceless, by being pronounced tensely, has become a roundedch, writtenchw. It is the result of pronouncing voiceless with the mouth-passage narrowed at the back so as to produce audible friction, which is heard asch (χ) accompanying the. In S.W. dialects the loose voiceless (writtenwh orhw) prevails initially. In O.W., injuv. andm.c.,chw̯i ‘you’ appears ashui; later this word was everywherechw̯i, thech being still heard even in S.W. (though now unrounded in this word, thuschi). Initialchw prevails in Ml. W. and later, aschuervb.b. 83, 84 ≡chw̯erw̯ ‘bitter’,chuec do. 84 ‘sweet’,chuant do. 34 ‘lust’;chwythuw.m. 47 ‘to blow’,chwaer do. 41 ‘sister’,chwedɏl do. 42,r.m. 29 ‘tale’,chwythatỺ.A. 9 ‘breath’,chwant do. 11 ‘lust’, and so generally in Mn. W.; butwh frequently occurs in Ml.mss. and sometimes in Early Mn. poets, aswhechetỺ.A. 147 ‘sixth’,whennychu do. 149 ‘to desire’,whaerr.m. 28,whedlg. 147.

(2) Initial roundedch is heard with as an off-glide, as inchw̯aer; final roundedch has as an on-glide, as iniwch ‘to you’,ewch ‘go ye’. In the latter case the sound isch in all the dialects, noth.


(3) Initialchw sometimes interchanges withgw̯; asGware dy chwarer.m. 154 ‘play thy game’,chwith,gwith do. 301 ‘sinister’; this is due to the variability of original initials‑,§ 101 ii (1); *su̯- >chw̯‑; *u̯- >gw‑.ŋghw̯ forchw̯ is due to a precedingn (nhw̯ >ŋhw̯), aschw̯aneg,anghw̯aneg ‘more’;yn chwaethachr.m. 7yg̃hwaethach do. 85, 108 ‘rather’.

(4) Initialchw̯ has often a prostheticỿ‑, asỿchwaneg ‘more’,ychwaneguw.m. 44 ‘to add’.

(5) Final rounded‑ch, of whatever origin, becomes unrounded if the syllable is unaccented; thuswelewchw.m. 50 ‘ye saw’ iswelech. But‑ɥw̯ch gave‑wch, as incerwch ‘ye love’ for *cerɥw̯ch, see§ 173 viii; sopeswch for *pesɥwch:pas,§ 201 iii (2). The formydych is due to the analogy ofydym; so Late Mn. W.gennych after the 1st pl. for Ml. and Early Mn.gennwch.

Note.

Transcription.—By means of the devices mentioned in the above sections (the use of ẟ,,, etc.) the forms of Late Ml. W. can generally be transcribed so as to indicate the approximate sound while preserving the exact spelling of thems. But, as we have seen, the orthography of O. and Early Ml. W. is so irregular that no such plan is possible. Accordingly, for these periods, the form in thems. is given, followed, where necessary, by a transcription introduced by the sign ≡, giving the probable sound in modern characters.

The works of Early Mn. poets are often found in latemss. and printed books containing not only dialectal forms inconsistent with the forms implied by the rhymes of the bards, but also late inventions, such asei,eich, etc. In these cases the spelling has been standardized in the quotations in this work. The spelling of thems. is here of no importance, as the cynghanedd, rhyme or metre is in every case relied on as showing the exact form used by the author.

All quotations are given with modern punctuation, including the insertion of the apostrophe, and the use of capital letters.

Sounds in Combination.

Syllabic Division.

§ 27. i. In Welsh a single consonant between two vowels belongs normally to the second syllable; thusca|nu ‘to sing’,gw̯e|le|dig ‘visible’; when there are two or more consonants the first belongs to the first syllable, ascan|tor ‘singer’,can|i̯ad ‘song’,tan|w̯ɥdd ‘fire-wood’,can|tref ‘hundred (district)’. A double consonant belongs to both; thus incan|nu ‘to whiten’, the first syllable ends after the stoppage of the mouth-passage for the formation of then, and the second begins before the opening of the passage which completes the formation of the consonant. Thus a double consonant implies not two independent consonants, but a consonant in which the closing of the passage takes place in one syllable and the opening in the next, and both count. This is seen most clearly in a word likedrỿcin ‘storm’, where thec closes as a velarq and opens as a palatal (drỿ́q|k̑in), and yet is not two complete consonants. The consonantsp,t,c,m,s,ng,ll, are double after accented vowels, though written single; thusateb, canasantat|teb,ca|nas|sant. See§ 54.

ii. A consonant which is etymologically double is simplified after an unaccented syllable; ascy|né|finr.m. 183 ‘familiar’ (cyn-nef-in < *kon-dom-īno‑: Lat.domus);whe|ný|chur.b.b. 89 (fromchwant) ‘to desire’;ym|gy|núll|aw, do. 49 (fromcynnull) ‘to gather together’. But this phonetic rule is not regularly observed in writing, except in the final unaccented syllable,cắlonn ‘heart’ (pl.calónnau),Cálann (from vulg. Lat.Kaland‑) etc., being generally writtencalon,Calan, etc.

iii. In modern writing the division of syllables where required, as at the end of a line, is made to follow the etymology rather than the sound; thus it is usual to dividecan-u ‘to sing’ so,can being the stem andu the ending, instead ofca-nu, which is the true syllabic division. In the case of more than one written consonant the division is usually made to follow the sound; thus,can-nu ‘to whiten’,plen-tyn ‘a child’, the etymological division beingcann-u,plent-yn. Ml. scribes divided a word anywhere, even in the middle of a digraph.

In this grammar syllabic division is indicated when required by | as above; and the hyphen is used to mark off the formative elements of words, which do not necessarily form separate syllables.

Diphthongs.

§ 28. A diphthong consists of the combination in the same syllable of a sonantal with a consonantal vowel. When the sonantal element comes first the combination is a falling diphthong. When the consonantal element comes first it is a rising diphthong. “Diphthong” without modification will be understood to mean falling diphthong.

Falling Diphthongs.

§ 29. i. In O. W. falling diphthongs had for their second element eitheri, frontu, or backu. The O. W. diphthongs with their Ml. and Mn. developments are as follows:

O. W.Ml. W.Mn. W.
i{     {\displaystyle \scriptstyle {\left\{{\begin{matrix}\ \\\\\ \\\ \\\ \ \end{matrix}}\right.}}aiaɥ, ae(),ae
oioɥ, oe(),oe
uiw͡y
eieiei, ai
Frontüou (au)eueu, au
Backu{       {\displaystyle \scriptstyle {\left\{{\begin{matrix}\ \\\\\ \\\ \\\ \\\ \\\ \ \end{matrix}}\right.}}
auawaw
euewew
iuiwiw
iuɥw, ỿwɥw, ỿw
ouỿw, ewỿw, ew

ii. (1) Asi in O. W. represented bothi andɥ the exact value of the second element in O. W.ai,oi,ui cannot be fixed; but it was probably receding in the direction ofɥ. Inw͡y it has remainedɥ. The former diphthongs are generally writtenae andoe; but the spellingsay,oy are commonly met with in Early Ml. W., and sometimes inmss. of the Mn. period; asguayt ‘blood’,coyt ‘timber’,mays ‘field’l.l. 120;croyn ‘skin’a.l. i. 24,mays do. 144;Yspayn ‘Spain’,teyrnassoyẟ ‘kingdoms’p 9 r. Inr.m. 118 we havehaearn, in 119hayarn ‘iron’. Though now always writtenae,oe, the sound in N. W. is still distinctlyaɥ, oɥ; thusmaes,coed are readmā́ɥs, cṓɥd. In Mid and S. Wales the sound approaches the spellingae,oe. In parts of S. W. the diphthongs are simplified intoā, ō in the dialects:mā́s, cṓd. In Pembrokeshireoe becomesw̄́-ë and evenw̯ḗ.

(2) Ml. W.ae andoe are derived not only from O. W.ai andoi, but also from O. W. disyllabica|e ando|e; thussaeth <sa|eth < Lat.sagitta;maes <ma|es (rhyming withgormes,b.t. 25) < *maᵹes;troed pl.traed (rhyming withvrithret /bryssẏetr.p. 1042) from *troget‑, *traget‑,§ 65 ii (1). They may also represent a contraction ofa|ɥ, o|ɥ as indā́ed§ 212 iv,trṓent,§ 185 i (1).

iii. Ml. W.ei had an open and a closee according to position; these developed into Mn. W.ai andei; see§ 79 i. The present sound of the formei isəi̯, whereə is an obscure vowel which is hardly, if at all, distinct fromỿ.

iv. O. W.ou (≡) occurs once asau, inanutonaujuv. ‘perjuria’, which inox. isanutonou. Theo was unrounded in Ml. W., becoming an indistinct vowel, open and close, writtene; the two forms became Mn. W.au andeu; see§ 79 ii.

v. O. W.au andeu (backu) have remained the same phonetically, the backu being writtenw in the later language. O. W.iu represented three distinct diphthongs according asi representedi or either sound ofy. The diphthongsɥw andỿw are even now of course both represented by a single groupyw in ordinary writing. The rules for distinguishing between them are those that apply toɥ andỿ generally;§ 82 ii (4).

vi. O. W.ou (backu) represents the diphthongỿw, writtenyw and alsoow at a later period,§ 33 iii (2). Thusdiguolouicheticox.; Ml. W.llỿwỿchedicr.m. 84 ‘shining’,llỿwɥchr.p. 1153, which appear besidellewɥchr.p. 1154, Mn. W.llewɥch corr. intollewɥrch ‘light’;§ 76 vi,viii.

§ 30. The diphthongsae or andoe or followed by form the falling triphthongsaew,oew oraɥw,oɥw, ingw̯aɥw ‘spear’,gloɥw ‘bright’,hoɥw ‘sprightly’,croɥw ‘clear’, which remain strictly monosyllabic in the cynghanedd of the Early Mn. bards. In late pronunciation thew is made syllabic, except when a syllable is added, as in the pl.gloɥw̯on which is still disyllabic. Indāɥwch, contracted fromdā ɥw̯ch, theāɥw has now been simplified intoā́w; see§ 212 iv.

§ 31. i. Unaccentedae in the final syllable was often reduced toe in the Ml. period, especially in verbal forms and proper names; as inadwen foradwaen ‘I know’,chware forchwarae ‘to play’,Ithel forIthael, O. W.Iudhail (≡i̯üẟ-hail).

Pan aeth pawb allan ẏchwarer.m. 116 ‘When everybody went out to play’; see alsor.m. 15, 38, 84, 87, 153, etc.

Lloches adar ichwarae,
Llwyn mwyn, llyna’r llun ymae.—D.G. 37.

‘A retreat for birds to play, a pleasant grove, that is the manner [of place] it is.’ See also D.G. 40, 58, 465 (misprinted‑au in 169).

Nid gŵr heb newidgware:
Nid llong heb fyned o’i lle.—G.Gl.c. i 197.

‘He is not a man, who does not change his pastime; it is not a ship, that does not move from its place.”

For examples ofadwen, see§ 191 ii (2).

ii. (1) The simplification of final unaccentedai andau toe are dialectal and late. Such forms asllefen forllefain,gwele forgwelai are avoided by the Early Mn. bards in their rhymes, but they begin to appear inmss. in the late 15th cent., and were common in the 16th and 17th cent. But the literary forms never fell out of use, and ultimately supplanted the dialectal forms in the written language, though some of the latter have crept in, ascyfer forcyfair, Ml. W.kyveir§ 215 iii (9),ystyried forystyriaid§ 203 iii (2).

(2) The levelling in the dialects of the sounds mentioned gave rise to uncertainty as to the correct forms of some words. The wordbore ‘morning’ began to be wrongly writtenboreu orborau in the 15th cent.[3]; seeg. 190. The formscamrau,godreu,tylau are later blunders for the literary formscamre ‘journey’,godre ‘bottom edge’, pl.godreon,r.m. 147, andtyle ‘hill; couch’. The newychain forychen ‘oxen’§ 121 iii is due to the idea that‑en is dialectal. In Gwyneddỿchain is heard, but is a dialectal perversion likemerchaid formerched.

Tesogforegwna’r lle ’n llon,
Ac annerch y tai gwynion.—D.G. 524.

‘On a warm morning make the place merry, and greet the white houses.’ Seeboreb.b. 31, 55, 82, 92, 108,w.m. 56, 73, etc.

Ni adewais lednaiale
Ynghymry ar fynghamre.—I.G. 201.

‘I left no noble place in Wales on my journey.’ Seekamre,r.p. 1269.

Lluwch ar fre agodregallt,
A brig yn dwyn barúg-wallt.—D.G. 508.

‘Snowdrift on hill and foot of slope, and branch bearing hair of hoar-frost.’ See alsor.p. 1036.

A phan edrychwyt ydyler.m. 146 ‘And when the couch was examined.’

§ 32. The diphthongai is wrongly writtenae by most recent writers (under the influence of Pughe) in the wordsafi̯aith ‘delight’,araith ‘speech’,cyffaith ‘confection’,disglair ‘bright’,goddaith ‘conflagration’,gweniaith (orgweiniaith) ‘flattery’,rhyddiaith ‘prose’,talaith ‘crown; realm’. See§ 202 iv (1). The worddiffaith, Ml. W.diffeith, ‘waste, wild, evil’ (from Lat.defect-us) is generally written so in the good periods (e.g.diffeithb.b. 106,r.m. 183); but some early examples occur of a new formation fromffaeth ‘cultivated’ (from Lat.factus),r.p. 1047, l. 2.

Yn y nef mae ’n unafiaith
Yn sôn archangylionsaith.—Gr.H.g. 101.

‘In heaven in pure rapture there speak archangels seven.’ See D.G. 358, whereafiaith is printedafiaeth in spite of its rhyming withgobaith. See alsog. 122.

Ef a gâr awdl acaraith,
Ef a ŵyr synnwyr ysaith.—H.D.p 99/469.

‘He loves song and speech, he knows the meaning of the seven [sciences].’ Seeg. 118;areithb.b. 9, 15.

Disgleirẟiweir Veir vorwyn.—Ca.,r.p. 1247.

‘Bright chaste virgin Mary.’

Coed osglog, caeaudisglair,
Wyth ryw ɥ̄́d, a thri owair.—D.G. 524.

‘Branching trees, bright fields, eight kinds of corn and three of hay.’ See D.G. 54, 120, 209, 404. Seeb.cw. 8, early editions of Bible, etc.

Falgoddaithyn ymdaith nos.—D.G. 13.

‘Like a bonfire on a night’s march.’ Seegoẟeithr.p. 1042,b.b. 73.

Gwenwyn ydiw eugweiniaith,
Gwynt i gyd gennyt eugwaith.—I.F.m 148/721.

‘Their flattery is poison, to thee their work is all wind.’

Twysog yw, enwog iwaith,
Teilwng i wisgotalaith.—E.U.

‘He is a prince whose work is famous, worthy to wear a crown.’

Troes dilyw tros ydalaith,
Torri ar rif tyrau’riaith.—Gu.O.a 14967/62.

‘A deluge has overflowed the realm, thinning the number of the nation’s towers.’ Seeg. 80, 87, 199, 218, 257.

Tro ’n d’ôl at yr hendalaith;
Digon yw digon odaith.—E.P. 124/283r.

‘Turn back to the old country; enough is enough of travel.’

§ 33. Late Contractions. i. (1) We have seen thata-e ando-e were contracted early intoae andoe;§ 29 ii (2). This contraction also took place later, as inCymrā́eg ‘Welsh’,Groeg ‘Greek’, and in verbal forms such asaed ‘let him go’,rhoed ‘let him give’, rhoes ‘he gave’; see§ 185.

Inr.p. 1189Gro-ec is a disyllable rhyming withchwec, ostec, Cym|ra|ec, tec; in ther.g. 1119 it is stated to be a monosyllable; D.G. uses it as a monosyllable, 53, as well asrhoes 6 ‘gave’,troes 68 ‘turned’, gwnaed 149 ‘let her do’,doed 145, 228 ‘let him come’,ffoes 191 ‘fled’, butffó|es 61. He usesCym|rá|eg as a trisyllable rhyming withteg, 2, 179; so G.Gr.,d.g. 243. This form persisted in the 15th cent.; as

Cymro da iGym|rā́|eg,
Cymered air Cymrudeg.—G.Gl.,m 146/281.

‘A Welshman of good Welsh, let him take the praise of fair Wales.’

In the 15th cent., however, we meet with the contracted form; see T.A.g. 251. Later, this was usual:

Da i̯Gym|rā́eg,di-gymar ṓedd,
Di-dláwd ym mhob dadl ỿ́doedd.—W.Ỻ. 120 (m.S.B.).

‘Good [in] his Welsh, incomparable was he, resourceful in all debate.’

(2) The contraction of the accented penult with the ultima results in an accented ultima§ 41 iii. But in newly-formed compounds, contracted forms such asmaes,troed are treated like other monosyllables, and the accent falls on the penult; thusglỿ́n-faes D.G. 135 ‘vale’,méin-droed do. 262 ‘slender foot’,déu-droed ‘two feet’.

ii. Ther.g., 1119, states thatey is always a disyllable. This is not necessarily the case in the penult, for in such forms askeɥrɥẟ,§ 122 ii (3), pl. ofkaer ‘fort’,treɥthɥchr.p. 1153 fromtraethaf ‘I treat’, etc., it is an old affection ofae. In other cases, however, the diphthong is late, and the disyllabic form is used in poetry down to the 16th cent. Thus:

Lloer yvi a dawn llawerdŷn,
Lleuad rhïaneddLlë|ɥn.—G.Gl.,m 148/191.

‘She is the moon and the grace of many women, the moon of the ladies of Lleyn.’ See also I.G. 388, 405.

Salbri ieuanc sêlbrë|ɥr
Sydd i gael swyddau agw̯ŷr.—Gu.O.a 14967/94.

‘Young Salesbury of the stamp of a chieftain [is he] who is to have offices and men.’

Nid âi na chawr na dynchw̯ɥrn,
Heb haint Dmv, a’n pentë|ɥrn.—T.A.c. ii 81.

‘Neither a giant nor a violent man, without the scourge of God, could take our liege lord.’ Seeg. 176,f. 14, 33. Seehë|ɥrn / të|ɥrn /kedɥrnr.p. 1226.

The contracted form sometimes occurs; as

Penfarheɥrnpan fo’r hirnos.—D.G. 267.

‘A head-dress of iron spikes when the night is long.’—To the holly.

The nameLleyn is now pronouncedLlɥ̄́n, and regarded as an exception to the rule that Welsh is written phonetically.Llŷn, as the name should be spelt, is a contraction ofLlyyn, which also occurs,r.b.b. 307, 342; and has been written in the contracted form from the 16th cent. The contraction is as old as the 14th, for we findllyyn inr.p. 1360, where the metre proves the sound to bellɥ̄́n.

OLŷni Dywyn, yn dau,
O Dywyn i dir Deau.—W.Ỻ.,g. 297.

‘From Llŷn to Towyn, we two, from Towyn to the kind of the south.’

iii. (1) The Mn. W. diphthongsoi, ou andow are always late contractions; as inrhoi forrho|ï fromrhoddi ‘to give’;ymarhóusc.c. 330 forymarho|us ‘dilatory’;rhowch forrho|wch ‘give ye’;rhoist forrho|eist ‘thou gavest’;rhôi forrho|ei ‘he gave’.

These contractions occur in common words in the 14th cent.; seeroir.p. 1206, 1210,rhoi D.G. 206, 521, 524,rhois do. 206,rhoist do. 2,r.p. 1211;rout (printedroit) D.G. 206,rôi, rown do. 243. But uncontracted forms occur even later;tró|ais D.G. 307,tró|ï I.G.,cyffró|ï L.G.C.,d. 16.

(2) The diphthongow is pronounced with theo unrounded, thusəw, where theə is closer than the first element in the Eng.ow, and is scarcely distinguishable from the obscureỿ; in fact theỿw incỿwydd and theow inrhowch are identical. Hence in the 15th, 16th and 17th cent. the old diphthongỿw was often writtenow; as incowydd orkowydd forcỿwɥdd, see Mostynr. pp. 2, 3, etc., 26, 27, etc. etc.

iv. A late contraction may take the form of one of the old diphthongs, or even of a simple vowel; asgla|nháu forgla|nhá|u ‘to clean’;plaur.p. 1222 forplá|eu ‘plagues’;di|léu fordi|lé|u ‘to delete’;aw̯n forá|wn ‘we go’;gla|nhā́d forgla|nhá|ad ‘cleansing’; (g)wnaiw.m. 54, 250 forgwna|ei ‘did’, cf.b.b. 64;cỿ|tûn forcỿ|tú|un ‘united’;bûm forbú|um ‘I have been’;gwy|bū́m forgwy|bú|um ‘I knew’;cau forcáe|u ‘to shut’. These forms occur uncontracted in Ml. W.:gunaunb.b. 81 (≡gw̯na|wn rhyming withwn) ‘I would do’;yn gyttuunr.b.b. 238;cayuỺ.A. 167 (≡cáy|u),kaeuw.m. 24 (≡káe|u). Uncontracted forms are met with as late as the 16th cent.

Dy garu awybū́|um;
Darllain dy bylgain y bûm.—H.S. 5. ‘I have known [what it is] to love thee; I have been reading thy vigil.’ See D.G. 38.

v. A late contraction usually takes place when a word ending in a vowel is followed byi ‘his’ or ‘her’, Ml.y, and often when it is followed by the prepositioni ‘to’, Ml.y.Ac, ag lose their final consonant and form a diphthong with the former, asa’i̯ Ml. W.ae,ay ‘and his, with his’, but not with the latter:ac i ‘and to’.

Ancr wyf fi’n cyweirio i̯fedd.—7 syll.§ 44 vi.
Da i̯Gymraeg, di-gymar oedd.—7 syll.,i above.
Nos da i̯walch onest y Waun.—7 syll.g. 177.

‘Good night to the honest fellow of Chirk.’

Rising Diphthongs.

§ 34. i. The rising diphthongs in the Mn. language are as follows:—

i̯a as incani̯ad, i̯âr;w̯a as inanw̯ar;
i̯e as ini̯echɥd;w̯e as inadw̯en;
i̯o as inrhodi̯o, i̯ôr;w̯i as incedw̯ir;
i̯w as ini̯wrch, rhodi̯wn;w̯o as ingwatw̯or;
i̯ỿ as ini̯ỿrchell;w̯u as ingalw̯ut;
w̯ɥ as inedw̯ɥn;
w̯ỿ as inpenw̯ỿnni.

In Ml. W. is generally writteny,§ 17. The combinationsi̯i,i̯ɥ,i̯u,w̯w do not occur in Mn. W. They occur in verbal forms in Ml. W. but are generally simplified; see§ 36 i,ii.

ii. When or comes before a falling diphthong the combination becomes a mixed triphthong; asi̯ai ini̯aith ‘language’;i̯au inteithi̯au ‘journeys’;w̯aw ingw̯awd ‘song, mockery’;iw͡y inmeddyli̯w͡yd ‘it was thought’,neithi̯w͡yr, D.G. 424 (now generallyneithi̯wr§ 78 i (2)) ‘last night’. We have a tetraphthong in the old pronunciation ofgw̯aɥw (orgw̯aew)§ 30.

iii. When an unaccentedi comes before any other vowel the two are frequently contracted into a rising diphthong; thusdi|ó|ddef ‘to suffer’ becomes a disyllabledi̯ó|ddef D.G. 137. Some early examples occur, aser|i̯ṓed ‘ever’ for *er | i | ṓed ‘since his time’.di̯ṓer ‘by heaven’§ 224 iv (2) is a monosyllable, as the metre shows inr.p. 1206, D.G. 46, 51.di̯awl ‘devil’ must have been contracted into a monosyllable in O. W. when the accent fell regularly on the ultima; otherwise it would have become *dī́|awl.

iv. The rising diphthongsw̯a and w̯o are frequently interchanged; asgw̯atw̯arw.m. 185,gw̯atw̯or D.G. 136 ‘to mock’;marw̯arỺ.A. 39,marw̯or ‘embers’ (cf.maroryn§ 36 iii);caw̯ad, caw̯od ‘shower’;pedw̯ar, pedw̯or ‘four’.

Pedw̯ortrysor tir Iesu.—H.R.,c 7/114.

‘The four treasures of the land of Jesus.’

The change takes place both ways;w̯a becomesw̯o incaw̯adr.m. 180,r.p. 1223, D.G. 57 (rhyming withbrad) andcaw̯odydd orcafodydd D.G. 305 (penult rhyming withbod);w̯o becomesw̯a incỿnaw̯anc.m. 21 forcỿnaw̯on pl. ofceneu§ 125 iii;dyw̯ad fordyw̯od fromdyfod§ 193 ix (3).

v. (1) The rising diphthongsw̯ɥ andw̯ỿ are of course not distinguished in ordinary writing, both being represented bywy; see§ 82 ii (5). Note then thatwy represents three distinct diphthongs, the fallingw͡y as inmŵyn ‘gentle’,sw͡yno ‘to charm’; the risingw̯ɥ, short ingw̯ɥnn ‘white’, long ingw̯ŷr ‘men’; the risingw̯ỿ as intỿw̯ỿnnu ‘to shine’. See§ 38.

(2) In ordinary writing the fallingiw̯ and the risingi̯w are also not distinguished. See§ 37.

§35. i. Many stems end in, which appears before all inflexional endings beginning with a vowel (with the exceptions mentioned in§ 36), but is dropped when the stem has no ending; thusmỿfỿri̯af ‘I meditate’,mỿfỿri̯ant ‘they meditate’,mỿfỿri̯o ‘to meditate’,mỿfỿri̯ol ‘meditating’, butmỿfɥr ‘meditation’.

In words borrowed from Lat. the can be traced to its source in shortĭ; thusmỿfɥr <memoria;sɥnn, sỿni̯af <sentio;ỿstɥr, ỿstỿri̯af <historia. In native words it represents original, as indŷn ‘man’ pl.dỿni̯on from Kelt. *doni̯os: Ir.duine§ 100 iv; cf. also§ 201 iii (6). In a few new formations the is ignored as indi-ỿstỿru ‘to ignore’,dỿnol ‘human’ a new formation which has replaced Ml. W.dỿnẏawlỺ.A. 12, 24, 38, etc.

ii. (1) In Mn. lit. W. generally appears after syllables havingei, as inysbeili̯af ‘I rob’ (ysbail ‘spoil’ < Lat.spolium);teithi̯af ‘I journey’ (taith ‘journey’),geiri̯au ‘words’ (gair ‘word’),neithi̯w(y)r ‘last night’, Ml. W.neithwyr§ 98 i (3). In these cases the is omitted in S. W. dialects and most Ml.mss., as keinhaucb.b. 54 ≡keinhawcb.t. 28; but the oldest Ml. prosemss. (the earlymss. of the laws) and Mn. lit. W. follow the practice of the N. W. dialects and insert the i̯, askeynẏauca.l. i 24ms. a., cf. 22mss. b., d., Mn. W.ceini̯og ‘penny’.

(2) There are, however, several exceptions to this rule besides those mentioned or implied in§ 36. The is omitted before the substantival terminations‑en, ‑es, ‑edd; asdeilen (M.Ỻ. i 155 has the unusualdeili̯en) ‘leaf’,bugeiles ‘shepherdess’,cyfeilles (printedcyfeillies ind.g. 75) ‘amie’,meithedd ‘lengthiness’; before endings of comparison, asmeithed, meithach, meithaf (maith ‘long’),meined, meinach, meinaf (main ‘slender’), exceptrheit-i̯ed, ‑i̯ach, ‑i̯af§ 149 i, stems in‑eidd- asmanweiẟẏachỺ.A. 8 ‘finer’,pereiddi̯af ‘sweetest’, and some stems in‑eith- asperffeithi̯af ‘most perfect’; before the pl. endings‑edd, ‑oedd, asieithoedd ‘languages’; in a few isolated words asteilo ‘to manure’ (butteẏlẏaw inb.ch. 102),adeilad ‘building’ (butadeilẏat inr.p. 1220),cymdeithas ‘society’,eiddo ‘property’.

(3) Medialei before a consonant originally simple must be due to affection by after the consonant; and the inysbeili̯af etc. is the affecting preserved.‑eith- generally represents *‑ekt- a verbal noun and adj. formation, as inperffeith ‘perfect’, and the inperffeithi̯o is probably analogical,§ 201 iii (6). From these the has tended to spread. But there is necessarily no original reason for it whenei comes from‑ek- or‑eg‑; hence the exceptionsmeithach, cymdeithas, teilo (tail < *tegl-§ 104 ii (1)), etc.

iii. is also added to many stems havingi oru; ascil ‘back’, pl.cili̯au cili̯af ‘I retreat’;tir ‘land’, old poetic pl.tiri̯onb.b. 26,r.p. 1144,tiri̯o ‘to land’,tiri̯og ‘landed’ (but pl.tiredd, tiroedd);grudd ‘cheek’, pl.gruddi̯au;llun ‘form’, pl.lluni̯au, lluni̯o ‘to form’,lluni̯aidd ‘shapely’;ystudẏaw, llavurẏawỺ.A. 11 ‘to study’, ‘to labour’. In some of these cases also the is lost in S. W. dialects.

iv. Many stems end in which forms rising diphthongs with the vowels of all endings, except withw§ 36 i; thusgalw̯ ‘to call’,galw̯af ‘I call’,gelw̯aist ‘thou calledst’,gelw̯ynt ‘they called’, etc.

§ 36. i. drops beforew, and drops beforei. The semivowel is sometimes written (asw ory) in Ml. W., but is often omitted. Thus whileb.m. 51 hasmi a gadwwn, mi ae kadwwn, the olderw.m. 71 has in the same passagemi a gadwn, mi ay cadwn. Similarly we havevedyẟẏit inỺ.A. 48 butbedyẟir earlier, p. 42.

The syllable closed by the or remains closed after its loss; thuscad|w̯wn, be|dydd|i̯ir becamecăd|wn, be|dy̆dd|ir (notcá|dwn, be|dý|ddir). By re-formation the is sometimes restored in the spoken lang. in forms likeber|w̯wch ‘boil ye’ impve., on account of the strength of the analogy ofber|w̯i, ber|w̯af, ber|w̯oẟ, etc. But the lit. and ordinary form isbĕr|wch, and the absence ofw̯w in the traditional pronunciation accounts for the well-known W. pronunciation of E.wood as’ood, etc.

ii. drops beforeɥ andu in monosyllables and final syllables; asɥrcha.l. i 20,Ỻ.A. 67 for *i̯ɥrch pl. ofi̯wrch ‘roebuck’;udd ‘lord’ < O. W.I̯ud- (‘*warrior’);peidɥntr.m. 90 (frompeidẏaw ‘to cease’, cf.peidẏw͡ysr.m. 98);Mareduẟr.p. 1194 for *Maredi̯uẟ, O. W.Morgetiudgen. xiii (≡Morᵹeti̯üẟ),Gruffudd < O. W.Griphiud (≡Griffi̯üẟ). It is often found written in Ml. W., asystyrẏychr.p. 1153 ‘thou mayst consider’,hilẏyntỺ.A. 11 ‘they would breed’,llafvurẏus do. 28 ‘laborious’,meẟylẏutw.m. 103 ‘thou wouldst think’; but the spelling is perhaps theoretical; see below.

Initiali̯u in polysyllables has giveni, as inIddew ‘Jew’ for *i̯uẟew;Ithel < *i̯uẟ-hael, O. W.Iudhail.

SeeIẟewp 14/1r. (13th cent.);itewon (t)b.b. 102; so inỺ.A. see its index, and inr.b., seer.b.b. index. Salesbury wroteIuddew, which he inferred from the derivation. The Bible (1588 and 1620) hasIddew; but late editors have adopted Salesbury’s unphonetic spelling. D. includesi̯u among rising diphthongs; but his only example is the artificialIuddew.

It is seen thati̯u becameu in the syllables which were accented in O. W., andi in syllables unaccented at that period,§ 40. The simplification must therefore have taken place before the shifting of the accent; and Ml. W. forms withẏu (≡i̯u) are analogical formations, and perhaps artificial.

iii. sometimes drops beforeo; as in the prefixesgo‑, gor- forgw̯o‑, gw̯or‑; thus Ml. and Mn. W.goleuni ‘light’, O. W.guolleunijuv. But analogy has tended to restore it; thus while we findathraonm.a. i 256, ii 319 forathrawonỺ.A. 112,r.m. 19,r.p. 1234 ‘teachers’,canaonb.a. 38,m.a. i 261, 315 forkanawonr.b.b. 147 ‘whelps’,lleoth.m. ii 234, 235 forllewotỺ.A. 10 ‘lions’,marorynỺ.A. 25 formarworyn D.G. 363 ‘ember’, it generally remained in these words. Late examples of its loss: Ml. W.etwo (varying withetwa by§ 34 iv) givesettor.p. 1357, Mn. W.eto (≡etto) ‘again’. Sopenwag became *penwog whencepennog ‘herring’, the pl. retaining the:penw̯aig L.G.C. 158, Ml. W.penw̯eica.l. i 66.

*gwolchi ‘to wash’ gavegolchi, whencegylch ‘washes’; but in Ml. W. the latter wasgwylch, asy dwfvyr awylchpob pethỺ.A. 18 ‘water washes everything.’

Môr awylchmwyn amgylch Môn.—Ca.,r.p. 1244.

‘The sea washes the sweet coast of Môn.’

iv. drops before owing to the extreme difficulty of pronouncing the combination, but it remains before vocalicw; thusgweithi̯wr ‘worker’,gweithi̯w͡yd ‘was worked’, butgweithw̯ɥr ‘workers’ (not *gweithi̯w̯ɥr).—Of course vocalici remains in all cases:ysbī́-wr ‘spy’, pl.ysbī́-w̯ɥr.

v. drops after following a consonant, or following a diphthong; thusceidw̯ad for *ceidw̯i̯ad ‘keeper, saviour’,geirw̯on for *geirw̯i̯on, pl. ofgarw̯ ‘rough’,hoyw̯on for *hoyw̯i̯on, pl. ofhoyw̯ ‘sprightly’. But when follows a simple vowel the remains, as inglew̯i̯on, pl. ofglew ‘bold’,glaw̯i̯o ‘to rain’.

It is kept ingw̯i̯alen when contracted (as in D.G. 60) forgw̯i|á|len,§ 75 vi (2).

vi. drops afteru, as induon for *dui̯on, pl. ofdu ‘black’,goreuon for *goreui̯on pl. ofgoreu ‘best’.

vii. drops afterr orl following a consonant, asmeidrol formeidri̯ol ‘finite’ (veidrẏawlr.p. 1233,veidrawl do. 1234),budron for *budri̯on, pl. ofbudr ‘dirty’,crwydrad forcrwydri̯ad ‘wanderer’,meistraid formeistri̯aid ‘masters’,teimlo for *teimli̯o ‘to feel’,treiglo fortreigli̯o ‘to roll’.

This rule is not always observed. In some late Biblescrwydrad has been altered intocrwydriad. We also findmeistriaid in Mn. W.;dinistri̯o always retains, andmentri̯o occurs formentro.

Ambiguous Groups.

§ 37. i. As above notediw in ordinary writing represents both the rising diphthongi̯w and the falling diphthongiw̯.

ii.iw in the ultima followed by a consonant isi̯w, asi̯wrch ‘stag’,rhodi̯wch ‘walk ye’,cofi̯wn ‘we remember’,wỿrddi̯wn ‘a myriad’. The only exceptions are the Mn. formsiw̯ch for Ml.ɥw̯ch ‘to you’, andniw̯l for Ml.nɥwl§ 77 v,§ 90.

The Demetian disyllabicni|wl (D.D. s.v., D.G. 150nī́-wl /nā́-wyr) is < *niw̯wl < *niw̯ɏl <nɥwl with irregular epenthetic vowel§ 16 v (3) (ỿ >w after§ 66 ii (2)).Nifwl existed beside *niw̯wl. But the standard form appears to be a monosyllable (D.G. 70níw̯l /nṓs); and all the derivatives are fromniw̯l‑, asniwliog orniwlog ‘misty’,niwlen ‘a veil of mist’.

Initiali̯ŵ became *ü̯ŵ and thenüw̯ inuwd ‘porridge’ < Ml. W.iwt (≡i̯ŵd)r.b. 1061, Bret.iot; buti̯wrch remained because it is easier so than if another consonant were added to the group at the end of the syllable.

iii. In all other casesiw isiw̯; thus (1) finally, as ini’w̯, Ml.yw̯ ‘to his’,rhiw̯ ‘hill’,briw̯ ‘wound’,edliw̯ ‘to reproach’,heddiw̯ ‘to-day’.

There is no exception to the rule in lit. W. In the Powys dialectheddiw is soundedheddi̯w, and in Gwyneddheiddi̯w; but the Demetianheddi’ impliesheddiw̯. The bards always rhymed it asheddiw̯, till it came to be writtenheddyw in the 15th cent. (one example inr.p. 1286), an artificial restoration, see§ 77 v.

Nid oes fyd na rhyd na rhiw̯
Na lle rhydd na llawrheddiw̯.—D.G. (to the snow), 408.

‘There is no world or ford or hill or any free place or ground to-day.’ See also D.G. 16, 26, 82, 86, 126, 153, 194, etc.

Ni fu hawdd nofioheddiw̯
I un a ffrwd yn i ffriw̯.—T.A.,f. 22.

‘It has not been easy to swim to-day for one with the stream in his face.’

(2) In the penult or ante-penult, asdiw̯edd ‘end’,ni|w̯eidio ‘to harm’,ciw̯dod ‘race, people’. Exceptions are the borrowed wordssi̯wrnai ‘journey’,si̯ŵr ‘sure’, anddi̯wrnod ‘day’ when contracted, as in Gr.O. 88, fordi|ẃrnod for Ml. W.diw̯ỿrnawd,w. 1a (generally in Ml. W.diw̯arnawt, a S. W. form).

iv.iw is disyllabic when it is formed by adding a syllable beginning withw to a syllable ending ini; thusgweddi ‘prayer’,gweddī́-wn ‘let us pray’,gweddī́-wr ‘suppliant’. In such words thei is generally written in Mn. W. with a diaeresis—gweddïwr.

v. The combinationiwy has four sounds: (1) the mixed triphthongi̯w͡y, as inneithi̯w͡yr,§ 34 ii. It occurs in verbal forms when the terminations‑w͡yf, ‑w͡yd, ‑w͡ys are added to stems in,§ 35; asrhodi̯w͡yf ‘I may walk’,tybi̯w͡yd ‘it was thought’.

(2)ī́w͡y disyllabic. It occurs when the above endings are added to stems in vocalici, asgweddī́w͡yf (3 syll., see example in§ 201 ii (2)); and in compounds ofdi- with stems havingw͡y, as indi-w͡yr ‘not bent’ (gŵyr ‘bent’).

(3)iw̯ɥ or (4)iw̯ỿ according to position, as inlliw̯ɥddg. 164 ‘painter’, pl.lliw̯ỿddi̯on;diw̯ɥd ‘diligent’ spv.diw̯ỿtaf. These sounds may occur either wheniw̯ is followed byɥ orỿ or wheni is followed byw̯ɥ orw̯ỿ in word-formation.

§ 38. i. The distinction between the falling diphthongw͡y and the rising diphthongw̯ɥ, both writtenwy, is an important one. The difference between them is seen most clearly in monosyllables such asgŵɥr ‘he knows’,gw̯ɥ̂r ‘men’. In other positions they are liable to be confused in the dialects, and in a few cases we find confusion even in lit. W.

In ordinary written W. the falling diphthong when long is denoted byŵy (only used initially and afterg,ch), but when short or unaccented there is no method in ordinary use by which it can be distinguished; in that case it is printedw͡y, where necessary, in this book. The rising diphthong is indicated by marking the a consonant.

ii. In monosyllableswy represents the falling diphthong except when preceded byg orch; thusdŵɥn ‘to bring’,brŵɥn ‘rushes’,cŵɥn ‘complaint’,clŵɥd ‘hurdle’,llw͡ybr ‘path’,hŵɥnt ‘they, them’,cw͡ymp ‘fall’. Words beginning withg orch have usually the rising diphthong, asgw̯ɥn ‘white’,gw̯ɥrdd ‘green’,gw̯ŷdd ‘trees’,chw̯ɥrn ‘roaring’,chw̯ŷth ‘blows’; the exceptions areGŵɥ ‘the Wye’,gŵɥdd ‘goose’,gŵɥdd ‘presence’,gŵɥl ‘vigil, holiday’,gŵɥl ‘modest’,gŵɥll ‘goblin’,gŵɥr ‘knows’, ''gŵɥr ‘a bend’,gw͡ystl ‘pledge’,gŵɥth ‘anger’,chŵɥdd ‘swelling’.

Note the following words which conform to the rule, though spelt like some of the above-mentioned exceptions:gw̯ŷdd ‘trees’,gw̯ŷl ‘sees’§ 173 iv (1),gw̯ɥll ‘darkness’.

iii. When a word has the falling diphthongŵɥ in its simple form, the diphthong remains so in all derivatives; thusmŵɥn ‘gentle’,mw͡ynach ‘gentler’,mw͡ynhau ‘to enjoy’;cŵɥn ‘complaint’, pl.cw͡yni̯on, v.n.cw͡yno ‘to complain’. Similarly the rising diphthong remains rising, theɥ becomingỿ according to rule,§ 82 ii (5); thusgw̯ɥn ‘white’,gw̯ỿnnach ‘whiter’,gw̯ỿnnu ‘to whiten’.

In N. W. dialectsw͡y has come to be soundedw̯ɥ in the penult afterc,g orch, ascw̯ɥno forcw͡yno ‘to complain’;gw̯ɥddau forgw͡yddau ‘geese’;chw̯ɥddo forchw͡yddo ‘to swell’. But originalw̯ɥ, which in the penult is properlyw̯ỿ, has becomew in all dialects, aschwthu forchw̯ythu ‘to blow’,chwrnu forchw̯ỿrnu ‘to roar’,gwnnu forgwỿnnu ‘to whiten’; see§ 66 ii.

iv. When a word in its radical form begins withwy the diphthong is the falling one; thusŵɥ ‘egg’,ŵɥth ‘eight’,w͡ythnos ‘week’,w͡ybr ‘sky’,w͡ylo ‘to weep’,ŵɥl ‘weeps’,w͡yneb ‘face’.

w͡ybr, w͡ylo andw͡yneb are frequently mispronounced; and in N. W. dialects thew ofw͡yneb having been made consonantal ag has been prefixed to it givinggw̯ɥneb. This vulgarism hardly occurs before the 19th cent.

Rhaid im ddŵɥn pridd ar f’w͡yneb[4]
Rhag bod i’m adnabod neb.—D.G. 307.

‘I must bear earth upon my face, so that no one shall know me.’ Seewrth f’w͡yneb D.G. 23,yn f’w͡yneb do. 442.

Amlwg fydd trŵɥn a’rw͡yneb;[4]
Afraid i ni nodi neb.—E.P. 212.

‘Plain is the nose on a face; we need mention no one.’

A’r anadl oll a’rw͡yneb[5]
Fal aroglau si̯opau Si̯êb.—D.G., 330.

‘And all the breath and face like the perfume of the shops of Cheapside.’ See alsog. 49.

Osw͡yneb[5]i̯arll sy ’n y bedd,
I̯arll a aned erllynedd.—D.N.,c. i 161.

‘If an earl’s face is in the grave, an earl was born last year.’

So always in the Bible; seefy w͡yneb[6]Gen. xliii 3,Ex. xxxiii 20,Lev. xvii 10, etc.;eu hw͡ynebau,[6]Gen. xlii 6, etc. An early indication of the mispronunciation is found iny wynebeu,b.cw. (1703), p. 7, which should beyr w͡ynebeu, but has not yet becomey gw̯ynebeu.

v. Finalwy is always the falling diphthong; aspwɥ ‘who?’Conw͡y, Myfanw͡y, arlw͡y ‘a spread’,dirw͡y ‘fine’,llỿw͡y ‘beautiful’, Taw͡y; also medialwy followed by a vowel, asmw͡yar ‘blackberries’,gw͡yar ‘gore’.

Tlawd a ŵyr taludirw͡y:
Ni thelir math Lowri mŵy.—T.A.,a 14879/20.

‘The poor are accustomed to pay forfeit; they will never more forfeit such a one as Lowri.’

But in the Ml. 2nd sg. pres. ind. of verbs with stems, as ingelw̯ɥ ‘thou callest’,kedw̯ɥ ‘thou keepest’§ 173 iii (1), Mn. W.gelw̯i, cedw̯i, the diphthong is of course the rising one.

vi. When a word haswy in the last syllable anda in the penult, thewy is the falling diphthong; thusarw͡ydd ‘sign’,arglw͡ydd ‘lord’,annw͡yd ‘cold’,addw͡yn D.G. 355 ‘gentle’,cannw͡yll ‘candle’,gwanw͡yn ‘spring’,cadw͡yn ‘chain’,annw͡yl ‘dear’;aw͡yr ‘air’,aw͡ydd ‘desire’, seex below. Except in compounds, such astanw̯ɥdd ‘firewood’, etc.; see§ 83 iii.

Rhaid i’r gwan ddal ygannw͡yll
I’r dewr i wneuthur idŵɥll.—E.P. 235.

‘The weak must hold the candle for the bold to do his deceit.’

Oer gennych eiragwanw͡yn:
Oerach yw ’myd er ỿchmŵɥn.—T.A.,c. i 342.

‘Cold you deem the snow of spring: colder is my plight because of you.’ See D.G. 321, 408, 525.

Aur a gâd yn ddwɥgadw͡yn,
A’i roddi’n faich i’r ddynfŵɥn.—D.G. 64.

‘Gold was brought in two chains, and laid as a burden on the gentle maiden.’ See alsog. 250.

Dyfynnodd i’w dai f’annw͡yl
Da o le mae ’n dala iŵɥl.—H.D.,p 99/430

‘He has summoned to His mansions my dear one—it is a good place where he is keeping his holiday.’ See§ 54 i (3).

vii.wy is the falling diphthong when it is derived from Kelt.ei corresponding to Irishīa orē, aspŵɥll ‘thought’, Ir.cīall,gŵɥdd ‘goose’, Ir.gēd, gw͡ystl ‘pledge’, Ir.gīall, etc.; or when it is derived from Latinē, ig orī, as inrhŵɥd ‘net’ fromrēte,cŵɥr ‘wax’ fromcēra,eglw͡ys ‘church’ fromecclēsia,egw͡yddor ‘alphabet’ fromābēcēdārium,gwenw͡yn ‘poison’ fromvenēnum,dŵɥs ‘intense’ fromdēnsus,sŵɥn ‘charm’ fromsignum;sỿ́nnw͡yr ‘sense’ fromsentīre. Rule vi may be verified in many words by applying the test of derivation; e. g.cannw͡yll fromcandēla,cadw͡yn fromcatēna,[7]parádw͡ys fromparadīsus.

Geiriau da a gwŷr i’wdŵɥn
A ddinistr y ddauwenw͡yn.—D.I.D.,f. 11.

‘Good words and men to bring them will destroy the two poisons.’

Y doeth ni ddywaid aŵɥr;
Nid o sôn’ y dawsynnw͡yr.—G.I.H.,g. 144.

‘The wise does not say what he knows; it is not from talk that sense comes.’ See alsog. 111, 175, 234, 296.

viii.wy is the falling diphthong in the substantival terminations‑rw͡ydd ‘‑ness’,‑w͡ys ‘‑ians’, and in the verbal terminations‑w͡yf,‑w͡ys,‑w͡yd, but is the rising one in‑w̯ɥr pl. of‑wr ‘‑er’.

The ending‑w͡ys ‘‑ians’ added to names of places is probably derived from the Latin‑ēnses.

Hyd Iork y bu hydrefdŵɥs,
A’r gwanwɥn ar yGwennw͡ys.—L.G.C. 421.

‘As far as York it has been a very autumn, while it was spring to the men of Gwent.’

ix. The following words may be mentioned as those most commonly mispronounced:wy is the falling diphthong incerw͡yn ‘vat’,disgw͡yl ‘look, expect’,Gw͡ynedd ‘Venedotia’,Gw͡yndɥd, id.,morw͡yn ‘maiden’,terw͡yn ‘fervent’; it is the rising diphthong inoherw̯ɥdd ‘because of’,cychw̯ɥn ‘rise, start’,erchw̯ɥn ‘protector, [bed]-side’,dedw̯ɥdd ‘happy’. Seeterw͡yn /gŵyn /brŵynr.p. 1206;cerw͡yn /coll-lw͡yn D.G. 347.

Y ferch addfw͡yn oW͡ynedd,
Sy ymysg osai a medd.—D.G. 314.

‘The gentle maid of Gwynedd, who lives in the midst of wine and mead.’ See also L.G.C. 219.

Mi a euraf bobmorw͡yn
O eiriau maiol er imŵɥn.—D.G. 281.

‘I will gild every maiden with words of praise for her sake.’ See also D.G. 126, 236, 297, 298, 356, andg. 119, 229, 243.

Ar i farch yr âi f’erchw̯ɥn
Yn y llu ddoe’n llew o dn.—T.A.g. 234.

‘On his steed went my protector in the host yesterday, a man like a lion.’ See also L.G.C. 143, D.G. 510.

The wordkyfrw̯ɥs ‘shrewd’ (rhyming withhenwerɥs andynɥs inb.t. 78, and withpriscprys andchuischw̯ŷs inb.b. 57) is now soundedkyfrw͡ys on account of the difficulty of the consonantal groupfrw̯. The wordcelw̯ɥdd has undoubtedly the rising diphthong; seekelw̯ɥẟ /kynnɥẟr.p. 1223, cf. 1251, and D.G. 338; probablygŵɥdd /gelwydd, D.G. 256, is a misreading, but this form occurs in the 16th cent., seef. 36.

x.w͡y after a vowel has generally been changed tow̯ɥ, except in verbal terminations. Thusaw͡yr /hŵɥr /llŵɥrr.p. 1029, and generally so rhymed, see D.G. 395, 416, is now pronouncedaw̯ɥr, and the rhyme withɥr occurs already in the 13th cent.:aw̯ɥr /sɥrb.t. 23, G.Y.C.r.p. 1418. Similarlyaw͡yẟ /rŵɥẟ /arw͡yẟr.p. 1180 is latera-w̯ɥdd.Pow͡ys L.G.C. 381 is pronouncedPow̯ɥs§ 192 ii (2);tyw͡yll as intywill /canvillb.b. 30,tyw͡yll /gannw͡yll /pw͡yllr.p. 1045,tyw͡yll /amw͡yll D.G. 267,tŵɥll /tyw͡yll do. 117, 283 is nowtỿw̯ɥll, and already in D.G. rhymes withhyll 71, 285, 421, and withcyll 173, 185;ew͡ynr.p. 1036, laterew̯ɥn ‘foam’. On the other handglanhā́-w͡yd ‘was cleansed’ and all similar inflected forms are still so pronounced.

Lat.ăvĭdus would have given *ew̯ydd in Welsh;aw̯ɥdd cannot be derived from it, see§ 76 iii,iv.

Accentuation.

§ 39. i. In a polysyllabic word, one syllable is always pronounced with more emphasis than the others; this is called the syllable bearing the principal accent, or, simply, the accented syllable. In Welsh the accent is a stress accent.

A syllable may be emphasized either by raising the tone of voice or by a more forcible utterance. The two things may go together; but speakers of various languages unconsciously adopt one or the other as their principle of accentuation. The first produces musical or pitch accent, the second produces expiratory or stress accent. In Pr. Aryan the accent before the dispersion is believed to have been predominantly pitch, though vowel gradation,§ 63, points to the working of a strong stress accent. In Keltic, as in Italic and Germanic, the accent became predominantly stress, and has remained so, though its position has varied greatly.

☞ The syllable bearing the principal accent is denoted by an acute accent ´ placed above its vowel.

ii. The remaining syllables of the word are also pronounced with varying emphasis, but this may generally be disregarded, and they may all be considered as unaccented syllables. In some cases, however, one of them may attain a decided prominence in comparison with the others; such a syllable may be said to bear a secondary accent.

☞ The vowel of the syllable bearing the secondary accent is denoted where necessary by the grave accent `.

iii. Most monosyllables are stressed, but many frequently-recurring monosyllables bear no stress, but are pronounced in conjunction with another word. These areproclitics, which precede the accented word, andenclitics, which follow it.

The Welsh proclitics are the articley, yr, the prefixed pronounsfy, dy, etc., which are always unstressed. Usually also the relativesa, yẟ, yr, y, the negative, interrogative and affirmative particles, most conjunctions as thea inbara a chaws ‘bread and cheese’, and often prepositions as therhag inrhag ofn ‘for fear’.

The Welsh enclitics are the auxiliary pronounsi, di, etc. They are often written inmss. where they do not count in the metre, as inArduireaue trib.b. 36 (Arẟwyrëaf-i drf) forArddwyreaf dri (5 syll.) ‘I will exalt Three’. These may however be accented for emphasis.

§ 40. i. In Mn. W. all polysyllables, with a few exceptions named in§ 41, are accented on the penult; as|naf ‘I sing.,cán|i̯ad ‘a song’,can|i̯á|dau ‘songs’.

ii. The position of the accent was certainly the same in the Late Ml. period. This is proved by the fact that in the 14th cent. the cynghanedd was fully developed in its modern form in which the penultimate accent plays an important part, ZfCP. iv 123 ff.

iii. (1) But certain vowel values point to a period when the accent fell generally on the ultima. The evidence seems to show that this was the case in O. W., and that the transition took place in the Early Ml. W. period.

(2) The clear soundɥ occurs in the ultima only; the obscure soundỿ, which must have been the sound when unaccented, occurs in all other syllables. Hence the ultima must at one time have borne the accent. In monosyllables which have always been unaccented such as the articleyr, y, the sound isỿ; but in those which have always been accented, such asdyẟ ‘day’, it isɥ. There has been no shifting of the accent inỿ dɥ̄́ẟ ‘the day’, which therefore preserves the accentuation that resulted in the vowel sequenceỿ…​ɥ. Hence a word likemỿ́nɥẟ, which contains this sequence, must once have been accented *mỿnɥ̄́ẟ.

Similarly Brit.ŭ remains (writtenw) in the ultima; but appears asỿ in other syllables,§ 66 i;—ȩi remained and becameai in the ult., but becameẹi givingei (≡əi) in the penult,§ 79;—Brit.ā isaw in the ult.,o in the penult,§ 71 i;—uw in the ult. isu in the penult,§ 77 x; fromi̯ü we findü in the ult. and monosyllables, the easieri in the penult,§ 36 ii.

(3) In one or two words the vowel of the old penult has dropped since the separation of W. and Bret.; thus W.crȳ́ẟ ‘shoemaker’ < *cerȳ́ẟ < Brit. *kar(p)íi̯ō: Bret.kere,§ 86 i (5);—W.ysbryd < *spryd < *spyrýd < Lat.spiritus: Bret.spered.

On the other hand in some words an intrusive vowel developed before the accented syllable; Ml. W.dyly ‘deserves, owes’ comes through *dylý < *dlyᵹ,§ 199 ii (2); theỿ spread from this to other forms of the verb.—Ml. W.taraw ‘to strike’,tereu ‘strikes’ < *taráw, *tereu < *traw, *treu. The vowel did not spread from these totrawaf; the late Mn.tarawaf is an artificial lit. form,§ 202 i (3).

(4) The accent inýsgol, ýstrad, etc., now falls on a syllable that at one time had no existence. It is obvious that the shifting took place after the introduction of the prosthetic vowel. There is no evidence of that vowel in O. W. In the earliest Ml. W. we findIstrat andStrat,§ 23 ii. The latter may be an archaic spelling, but it seems to show that the accent was on thea. We may therefore infer that the transition took place in the Early Ml. period. In some words the prosthetic vowel was never firmly established; and the accent remains in its original position in these,§ 41 i.

iv. In Brit. the accent was apparently free as in Pr. Ar. As unaccentedā was shortened, it is seen that in *brā́teres (>broder) the accent was on the ante-penult; asā which remained accented givesaw, the accent to giveo must have shifted to theer in O. W., according to the general rule at that period. By the second shifting it went back to its original position, the new penult. Two shiftings must be assumed to explain such a form asýsbryd, which involves a shifting from *(y)sprýd, which in turn implies a shifting fromspírit-us.—It will be seen in the following pages that British cannot have shared the fixed initial accentuation of Goidelic.

§ 41. In some words in Mn. W. the accent falls on the ultima. These are

i. A few disyllables in which the first syllable is (1)ỿs- or (2)ỿm‑; as (1)ysgrī́n ‘shrine, coffin’,§ 23 ii,ystrŷd ‘street’,ysgrḗch ‘screech’,ystṓr ‘store’; (2)ymwḗl ‘do thou visit’,ymā́d ‘do thou leave’. But most words with these initial syllables are accented regularly, asýsgol ‘school’,ýsbryd ‘spirit’,ýsgwyd ‘to shake’,ýmdaith ‘journey’,ýmgudd D.G. 374 ‘hides’. In some cases we have both accentuations, seeýmwel below; occasionally with different meanings, asýmladd ‘to fight’,ymlā́dd ‘to tire one’s self’;ýmddwyn ‘to behave’,ymddŵyn ‘to bear’.

Y dydd a’r awr, ni’m dawr, dod;
ýmwelâ mi dan ámod.—G.I.H.,tr. 91.

‘Fix the day and hour, I care not [when]; visit me under [that] condition.’

Arthur o’i ddolur oedd wan,
Ac oýmladdcad Gámlan.—L.G.C. 450.

‘Arthur was weak from his wound, and from fighting the battle of Camlan.’ See also T.A.,c. ii 78.

Y ferch wéddw̯ ddifrychéuddeddf
Wedi’rymlā́dda’r drem léddf.—D.E.,p 112/840.

‘The widowed woman of spotless life after the prostration and disconsolate aspect.’

ii. The reduplicated pronounsmỿfī́, tydī́, etc. Rarely these are accented regularly; see§ 159 ii (2).

iii. (1) Words in which the last syllable has a late contraction,§ 33, such aspa|ra|tói for Ml. W.pa|ra|tṓ|i ‘to prepare’,cy|tū́n for Ml. W.cy|tū́|un ‘united’,Gwr|théɥrn forGwr|thḗ|ɥrn,Cỿm|rā́eg forCỿm|rā́|eg,pa|rhā́d forpa|rhā́|ad ‘continuance’. It is seen that in these words the accent in Ml. W. was regular, and kept its position after the ultima was merged in the penult.

(2) In the wordysgolhái̯g, Ml. W.yscolheic ‘scholar’, the contraction in the last syllable seems to have taken place early in the Ml. period, asNid vid iscolheic nid vid eleic unbenb.b. 91 (10 syll.; readscol|heic,§ 23 ii), but it was necessarily subsequent to the fixing of the present accentuation; inb.b. 81 the uncontracted form occurs, rh. withguledic. A similar form ispen-áig ‘chief’. The wordffelaig seems to have been accented regularly; thus inr.p. 1221 we haveffeleic/ffilij, the latter being the Lat.filii.

Tudur waed Tewdwr ydoedd,
Aphenáigcyff Ieuan oedd.—Gu.O.,g. 196.

‘He was Tudor of the blood of Tudor, and chief of the stock of Ieuan.’

iv. A few words recently borrowed from English; asapêl, ‘appeal’.

v. Disyllables in whichh stands between two vowels are accented regularly; thuscýhyd as inCýhyd a rhai og háearn D.G. 386 ‘[spikes] as long as those of an iron harrow’; andhyd gýhydc.c. 312 ‘full length’;cỿ́hoedd ‘public’, as ingýhoedd/gáeat,r.p. 1283;gwéheirdd D.G. 20 ‘forbids’. Contraction has taken place in some of these, thuscỿ́hoedd > *cóhoedd >coedd, D.G. 524; sogwáhan >gwân, which gave rise togwahân. This appears to be the reason forgwahân, cyhŷd, gwahárdd, etc. in recent W.

§ 42. In Ml. and early Mn. W. finalw afterd, ẟ, n, l, r, s was consonantal,§ 26 iv; thusmeddw̯ ‘drunk’,marw̯ ‘dead’,delw̯ ‘image’, were monosyllables, sounded almost likemeddf,marf,delf. Hence when a syllable is added thew is non-syllabic for the purposes of accentuation; thusméddw̯on ‘drunkards’,márw̯ol ‘mortal’,márw̯nad ‘elegy’,delw̯au ‘images’,árddelw̯ ‘to represent, to claim’. The is usually elided between two consonants, asmédd-dod ‘drunkenness’, forméddw̯dod. Inb.b. 84 we haveuetudaud (≡feẟw̯dawd), but in Ml. W. generally such words were written without the, asmeẟdawt,r.p. 1217, 1245, 1250, 1269,Ỻ.A. 147;gweẟdawtb.t. 31,r.p. 1261 ‘widowhood’. Thew inserted in these words in recent orthography is artificial, and is commonly misread as syllabicw, thusmedd|w|dod, the accent being thrown on the ante-penult, a position which it never occupies in Welsh. The correct formmédd-dod is still the form used in natural speech. When final, in polysyllables, the is now dropped, and is not written in late W., so there is not even an apparent exception to the rule of accentuation; thusárddelw̯ ‘to claim’,sýberw̯ ‘proud’ are writtenárddel,sýber. Ingwárchadw̯ ‘to guard’,ymóralw̯ ‘to attend (to)’, metathesis took place about the end of the Ml. period, givinggwárchawd, ymórawl, which becamegwárchod, ymórol in Mn. W.

In all standard cynghanedd the in these words is purely nonsyllabic:

Daarẟelw̯ kýnnelw̯ Kýnẟelw̯kéinẟawn.—r.p. 1229 (9 syll.)

‘A good representation of the exemplar of Cynddelw exquisitely gifted.’ The accentuation ofKýnẟelw̯ corresponds to that ofkéinẟawn. Cf.kývarch /kýfenw̯, 1230.

I llórf a’m pair ynllẃyrfarw̯
O hud gwir ac o hoedgarw̯.—D.G. 208.

‘Its [the harp’s] body makes me faint away from real enchantment and sore grief.’

Dynmarw̯a allai f’árw̯ain
Weithian drwy eithin a drain.—D.I.D.,g. 182.

‘A dead man might lead me now through furze and thorns.’

F’enaid hoengeirw̯afonydd,
Fy nghaniad dy 'fárw̯nadfɥ̄́dd.—Ỻ.G.,f.n. 30.

‘My beloved of the hue of the foam of rivers, my song thy dirge shall be.’ Cf.i fárw̯nad efṓ D.I.D.,g. 184.

Marw̯nadym yw awr yn d’ôl.—T.A.,a 14894/35.

‘It is a lament to me [to live] an hour after thee.’

Pwy a’theilw̯pe â’th wayw onn?—T.A.,a 14975/102.

‘Who will challenge thee if with thy ashen spear?’

The last example shows thateilw̯ could still be a pure monosyllable at the end of the 15th cent., for the present disyllabic pronunciation mars the cynghanedd. Even stronger evidence is afforded by the accentuationdéu-darw̯ / dódi B.Ph.B., Stowe 959/98b. Although final was non-syllabic,yn oryr following it was generally reduced to’n or’r, being combined with the to formw̯n orw̯r,§ 26 iii.

A’ch gwaed, rhyw ywch gadw̯’rhëol.—T.A.,a 14965/46.

‘With your blood it is natural to you to guard the road.’

Murnio da, marw̯’ny diwedd.—D.Ỻ.,f. 31.

‘Stowing away wealth, [and] dying in the end.’

In a compound likemarw̯nad the was not difficult, forw̯n (roundedn) is common in Welsh,§ 26 iii. But the colloquial pronunciation is nowmaw̯rnad, with metathesis of. In 16th and 17th cent.mss. we also findmarnad andbarnad. The combination is more difficult in such compounds asdérw̯goed ‘oak-trees’,márw̯ddwr ‘stagnant water’,chw̯érw̯-der ‘bitterness’; and though the etymological spelling persisted in these, the pronunciationdér-goed, már-ddwr, chw̯ér-der is doubtless old.

Lle dírgel gerllawdérw̯goed.—D.G. 321.

‘A secret place near oak-trees.’ Cf.dérw̯gist, T.A.,g. 232.

Tro fychwer’deryn felysdra.—Wms. 657.

‘Turn my bitterness into sweetness.’

Gyrchwérw̯dero garchárdai;
Newyn y lleidr a wna’n llai.—D.W. 112.

‘[Charity] drives bitterness from prisons; it makes less the hunger of the thief.’

Note 1. The rule that such words asmarw, delw are monosyllabic was handed down by the teachers of cynghanedd, but the bards of the 19th cent. hardly knew what to make of it. Thus R.G.D. 97 usesmarw anddelw, and E.F. 185 usesenw andgarw as monosyllables, while at the same time rhyming them. They no more rhyme as monosyllables than if they weremarf,delf, orenf,garf. In standard cynghanedd,marw̯ rhymes withgarw̯, tarw̯ only, anddelw̯ withelw̯, gwelw̯ only; see below. The disyllabic pronunciation may be traced as far back as the 15th cent. In a couplet attributed to D.G. (seed.g. 322)bw rhymes withgalw, a rhyme condemned by S.V. becausegalw̯ is a monosyllable whose vowel isa,P.Ỻ. xcii.

Some old rhymes aresyberw̯/hirerw̯/derw̯/chw̯erw̯,b.b. 69;agerw̯/chw̯erw̯/syberw̯/gochw̯erw̯,b.a. 19;helw̯/delw̯, ib.;dyveinw̯/dyleinw̯,b.t. 21;divanw̯/llanw̯,m.a. i 475;ymordlw̯/salw̯, do. 466;cadw̯/achadw̯/bradw̯, I.G. 422;enw̯/senw̯, do. 407;geirw̯/teirw̯, D.G. 500;syberw̯/ferw̯, E.P. 203.

Note 2. Inhwnnw,acw (earlierraccw) thew was vocalic; also probably in other forms in which it is a reduction of‑w͡y, see§ 78 i (2).

§ 43. i. No Welsh word or word fully naturalized in Welsh is accented on the ante-penult. Such forms asSáesoneg, Sáesones are misspellings ofSáesneg, Sáesnes.

A’r gyfreith honno a droes Alvryt vrenhin o Gymraec ynSaesnecr.b.b. 79 ‘And that law did king Alfred turn from Welsh into English.’ See ib. 64, 95, 96, etc.

The following words for different reasons are now sometimes wrongly accented:cathólig,oméga,[8]penígamp ‘masterly’,períglor ‘parson’,lladmérydd ‘interpreter’,ysgelérder ‘atrocity’,oléw̯ydd ‘olives’.

A thálu’r ffingathólig.—S.C.

‘And to pay the catholic fine.’ Cf.c.c. 25; I.G. 491; L.M.,d.t. 196.

Cyngorperígloréglwys.—M.R.,f. 12.

‘The counsel of a church parson’.

Penáig y glod,penígamp
Pennod i chompod a’i champ.—M.B. (m. D.G.), A 14967/183.

‘Master of the [song of] praise, supreme the height of its compass and achievement.’

Alpha acOmégamáwr.—A.R. (1818),e.g. p. xiii.

‘Great Alpha and Omega.’ Cf.Ỻ.M. 2. See Wms. 259, 426, 869.

ii. A few words recently borrowed from English are accented on the ante-penult, asmélodi, philósophi; but derivative forms of even these are accented regularly, e.g.melódaidd, philosóphydd.

§ 44. i. In a regularly accented word of three syllables the first syllable is the least stressed; thus incan|i̯a|dau the stress oncan is lighter than that ondau, both being unaccented as compared withi̯a. Hence the vowel of the first syllable is liable to drop when the resulting combination of consonants is easy to pronounce initially; as in Mn. W.pladur ‘scythe’, for Ml. W.paladur,c.m. 95 (paladurwyrw.m. 425, 426); Mn. W.gw̯rando ‘to listen’, for Ml. W.gw̯arandaw,r.m. 16,c.m. 29; Mn. W.Clynnog for Ml. W.Kelynnawc,[W 1]Ỻ.A. 124.

Some shortened forms are found, though rarely, in Ml. prose and verse:gw̯randaw,c.m. 27;kweirẏwyt forkyweirẏwyt ‘was equipped’,r.p. 1276 (they was written, and then deleted as the metre requires);pinẏwnr.p. 1225 from E.opinion;grennyẟ do. 1055 forgarennyẟ.

Fordywedud ‘to say’ we generally havedwedud in Early Mn. poetry (writtendoedyd in the 16th cent.); sotwysog, E.U. § 32,b.cw. 71, fortywysog ‘prince’;cledionc.c. 334, 390, pl. ofcaled ‘hard’;clonnau forcalonnau ‘hearts’, inTyrd, Ysbryd Glân, i’n clonnau ni, R.V.

ii. In words of four or more syllables, when pronounced deliberately, the first syllable has a secondary accent, asbèn|di||dig ‘blessed’, pl.bèn|di|ge|díg|ion. This also applies to trisyllables with the accent on the ultima, ascỳf|i̯aw|nhā́d ‘justification’. The least stressed syllable is the second; and this is often elided, in which case the secondary accent disappears; as in Mn. W.gorchfýgu forgòrchyfýguỺ.A. 15, and in Mn. W. versetragẃyddol fortrà|gy|ẃy|ddol ‘eternal’,partói for|ra|tói ‘to prepare’,llythrénnau forllỳthyrénnau ‘letters’,perthnásau ‘relations’ forpèrthynásau, etc.

Gwaeddwn, feirdd, yndragẃyddol;
Gwae ni nad gwiw yn i ôl.—Gu.O.,a 14967/120.

‘Bards, let us cry for ever; woe to us that it is useless [to live] after him.’ Seeg. 160, 255.

Yn ddyfal beunydd ibartói.—Wms. 259.

‘Assiduously every day to prepare.’

iii. When a vowel is elided, as in i, ii, or v, the same vowel disappears in the derivatives of the word; thuspladurwyr ‘mowers’;twysogesb.cw. 11 ‘princess’ fromtwysog, fortywysog;tragwyddoldeb ‘eternity’,ymbartói ‘to prepare one’s self’,’wyllysgar ‘willing’ (ewyllys, ’wyllys ‘will’).

Wedi ’mrawd yma’r ydwyf;
Ato, Dduw,ymbartói’dd wyf.—L.Mor. (m. I.F.).

‘After my brother I tarry here; to him, Oh God, I am preparing [to go].’ (The metre proves the elision, but not its position.)

Intragwyddoldeb the lost syllable is the second, so that there is no departure from the general principle laid down in ii; but inpladurwyr the first is lost because the word is formed from the reducedpladur. Ifpaladurwyr had been reduced directly it would have given *paldurwyr; similarlytwysoges, etc.

iv. Occasionally in Mn. W. haplology takes place, that is, a consonant, if repeated in the following syllable, is lost with the unaccented vowel; aserledigaeth forerlidedigaeth ‘persecution’,crediniol forcredaduniol,§ 132 (8), ‘believing’. (Cf. Eng.singly forsingle-ly,Bister forBicester, Lat.stipendium forstipipendium, etc.)

v. An unaccented initial vowel sometimes disappears, as in Late Ml. W.pinẏwnr.p. 1225 ‘opinion’, borrowed from Eng.;’wyllys forewyllys in verse; and in Late Mn. W.machlud ‘to set’ (of the sun) for Ml. and Early Mn. W.ym-achludd, D.G. 121,§ 111 vii (3). As a rule, however, this elision only takes place after a vowel:

Tebig yw ’r galennig lân
I’dafeddo wlad Ifan.—I.D.,tr. 142.

‘The fair new year’s gift is like threads from the land of [Prester] John.’ Another reading isI edafedd gwlad Ifan, I.D. 22.

Ac ef gyda’i ogyfoed
Yw gŵr y wraig oreu’rioed.—L.G.C. 318.

‘And he with his mate is the husband of the best wife [that] ever [was].’

In the dialects it is very common:morol ‘attend (to)’ forymorol, molchi forymolchi ‘to wash’,deryn foraderyn ‘bird’,menyn forymenyn ‘butter’,mennyẟ forymennyẟ ‘brain’, etc.

vi. In a few disyllables the vowel of the final unaccented syllable is sometimes elided; thusónid ‘but’ appears generally asond in Mn. W. Other examples met with in Mn. (rarely in Late Ml.) verse aremɥnd formỿ́ned ‘to go’,tɥrd fortỿ́red ‘come!’gweld forgwéled ‘to see’,llond forllónaid ‘full (capacity)’,cans forcánys ‘because’,namn fornámyn ‘but’, all except the last two in common use in the dialects. Similarlyér ỿs becomesers,§ 214 vii.

Ancr wyf fi’n cyweirio i fedd,
Ondaroamɥndi orwedd.—D.G. 295.

‘I am an anchorite making ready his grave, only waiting to go to rest.’

Cansar ddiwedd pob gweddi,
Cof cywir, yr henwir hi.—D.G. 235.

‘For at the end of every prayer, unforgotten she is named.’

Maẟeu,kanysti yw’r meẟic.r.p. 1298 (7 syll.).

‘Forgive, for Thou art the Healer.’ The length of the line shows thatkanys is to be readkans. It occurs writtencans inw.m. 487.

Ni edrychodd Duw ’r achwyn;
Ni mynnodd aur,namni ddwyn.—G.Gl.,m 148/256.

‘God did not regard the lamentation; He desired not [to have] gold, but to take him away.’ See also I.G. 380.

See examples oftyrd, dyrd in§ 193 viii (2).

vii. The vowel of a proclitic is often elided

(1) After a final vowel,ỿ is elided in the articleỿr,§ 114; the pronounsỿn ‘our’,ỿch ‘your’ (now writtenein,eich),§ 160 ii (1); the oblique relativeỿ orỿr,§ 82 ii (1),§ 162 ii (2); the prepositionỿn,§ 210 iv.

(2) Before an initial vowel,ỿ is elided infỿ ‘my’,dỿ ‘thy’,§ 160 i (1).

(3) The relativea tends to disappear even between consonants,§ 162 i.

(4) The vowel ofpa orpỿ ‘what?’ sometimes disappears even before a consonant, as inp’le ‘where?’§ 163 ii (2).

(5) Afterpa,rɥw tends to becomerỿ andr’,§ 163 ii (6).

§ 45. i. (1) Compound nouns and adjectives are accented regularly; thusgwī́n-llan ‘vineyard’,cadéir-fardd ‘chaired bard’,gwág-law orlláw-w̯ag ‘empty-handed’.

Gw̯áwd-laismwyalch argóed-lwyn,
Ac ëos ar lïos lwyn.—D.G. 503.

‘The musical voice of a thrush in a grove, and a nightingale in many a bush.’

Yn i dydd ni adai wan
Acw ’nllaw-w̯ag,Gwenllian.—L.G.C. 232.

‘In her day she, Gwenllian, left not the weak empty-handed there.’

(2) Even a compound of an adjective and a proper name may be so accented; as

Dágrau amurddedíg-Rys
Yw’r môr hallt, os gwir marw̯ Rhys.—G.Gl.,m 146/171.

‘The salt sea is tears for noble Rhys, if it is true that Rhys is dead.’

SeeUchél-Grist, D.G. 259. The nameBendigéid-fran ‘Bran the Blessed’, was so accented, and thef was lost,§ 110 iii (3), givingBendigéidran (corrupted intoBenegridran in Emerson’s English Traits, xi).

Bondo gw̯ýdrBendigéidran.—T.A.,a 14976/166;c. ii 83.

‘The glass eaves of Bendigeidran.’

(3) When the first element has one of the mutable soundsai, au, w, ɥ it is mutated in the compound, becomingei, eu, ỿ, ỿ respectively, because it is no longer ultimate when the compound is treated as a single word; thusgwéith-dy ‘workshop’ (gwaith ‘work’),héul-des ‘heat of the sun’ (haul ‘sun’),drỿ́g-waith ‘evil deed’ (drwg ‘evil’),melỿ́n-wallt ‘yellow hair’ (melɥn ‘yellow’). In old compoundsaw also is mutated, as inllófrudd,§ 110 iii (1).

☞ A compound accented as above may be called astrict compound.

ii. (1) But the two elements of a compound may be separately accented; thuscṓel gréfydd ‘false religion’,gáu bróffwyd ‘false prophet’,hḗn w̄́r ‘old man’ (sometimes accented regularly,hénwr,b.cw. 64).

(2) The difference between a secondary accent and a separate accent should be noted. A secondary accent is always subordinate to the principal accent; but when the first element of a compound has aseparate accent it is independent of the accent of the second element and may even be stronger if the emphasis requires it. Again, the first element when separately accented has the unmutatedai, au, w, orɥ in its final syllable; thus incỿ̀d-nabỿ́ddiaeth ‘acquaintance’ there may be a secondary accent oncỿ̆d (shortỿ) but incɥ̄́d gỿnúlli̯ad there is an independent accent oncɥ̄d (longɥ). In fact, when there is a separate accent, the first element is treated as an independent word for all purposes of pronunciation (accentuation, vowel quantity, and vowel mutation).

☞ A compound accented as above may be called aloose compound.

(3) Sometimes the elements of a loose compound are now hyphened, thuscoel-grefydd; but as any positive adjective put before a noun forms with it a loose compound, in the vast majority of such compounds the elements are written as separate words. See§ 155 iii.

iii. An adjective or noun compounded with a verb or verbal noun forms a loose compound, ascỿ́nffon lónni ‘to wag the tail’,prýsur rédant ‘they swiftly run’.

Fel y niwl o afael nant
Ydíson ymadáwsant.—R.G.D. 149.

‘Like the mist from the grasp of the valley have they silently passed away.’

iv. (1) Prefixes form strict compounds with nouns, adjectives, and verbs; asáthrist ‘very sad’ (trist ‘sad’),ám-gylch ‘circumference’,cýn-nal ‘to hold’, etc., etc.

(2) But compounds with the prefixesan‑, di‑, cyd‑, go‑, gor‑, gwrth‑, rhy‑, tra- may be either strict or loose; asán-awdd orán háwdd ‘difficult’,§ 148 i (6);án-aml/ýnysg. 103,án áml,§ 164 i (1);dí-wair, dí wáir ‘chaste’;rhý-wyr ‘high time’ andrhɥ̄́ hw̄́yr ‘too late’;trá-mawr Gr.O. 51,trā́ máwr ‘very great’;trá-doeth do. 52,trā́ dṓeth ‘very wise’.

Dí-dad,amddifad ýdwyf,
Adī́ fráwdwedi i farw̯ ẃyf.—L.Mor. (m. I.F.).

‘Fatherless, destitute, am I, and without a brother after his death.’

Y mae’r ddẃyais morddíw̯air.—D.G. 148.

‘The bosom is so chaste.’

Fwyn adī́ wáirf’enaid yw.—D.G. 321.

‘Gentle and chaste—she is my soul.’ Cf. D.G. 306.

Trā́ dā́im y trȳ́ déu-air.—I.F.,c 18/11.

‘Very good for me will two words turn out.’

In late Mn. W. new compounds are freely formed with these elements separately accented; thustra, go andrhy are placed before any adjectives, and treated as separate words;§ 220 viii (1).

When both elements are accented, the second has generally the stronger accent, unless the prefix is emphatic; ingor-úw̯ch ‘above’,gor-ī́s ‘below’, the first element has lost its accent, though these are also found as strict compounds, thusgóruwch, O.G.,g. 257, Gr.O. 34.

§ 46. i. Expressions consisting of two words in syntactical relation, such as a noun and a qualifying adjective or a noun and a dependent genitive, are in some cases accented as single words. ☞ These may be calledimproper compounds. Mutable vowels are mutated (ɥ >ỿ, etc.) as in single words.

They differ from proper compounds in two respects: (1) the initial of the second element is not softened except where the ordinary rules of mutation require it; (2) the words are arranged in the usual syntactic order, the subordinate word coming last, except in the case of numerals,ii (5) below.

Cf. in Latin the improper compoundspater-familias,juris-dictio, in which the first element is an intact word, by the side of the proper compoundspatri-cida juri-dicus in which the first element contains the stem only.

ii. Improper compounds accented on the penult consist of—

(1) Some nouns qualified byda, asgẃr-da ‘goodman’,gw̯réig-dda ‘good wife’,hín-dda ‘fair weather’,géir-da ‘good report’. Names of relatives withmaeth, astád-maeth ‘foster father’,mámaeth (formám-faeth,§ 110 iii (1)) ‘foster mother’,máb-maeth, bráwd-maeth, chw̯áer-faeth. A few other combinations, such ashéul-wen ‘bright sun’[9] (haul fem.,§ 142 iii),cóel-certh ‘bonfire’ (lit. ‘certain sign’). See also (3) below.

A bryno tir â braint da
Yn i árdal â’nẃr-da.—L.G.C. 249.

‘He who buys land with good title in his neighbourhood will become a goodman.’

(2) Nouns with dependent genitives:tréf-tad ‘heritage’,dỿ́dd-brawd ordỿ́dd-barn (alsodɥ̄́dd bráwd, dɥ̄́dd bárn) ‘judgement day’,pén-tref ‘village’,pén-cerdd ‘chief of song’,pén-tan ‘hob’. See also (3) and (4) below.

(3) Nouns with adjectives or genitives forming names of places; asTré-for orTré-fawr, Brỿ́n-gwyn, Mỿnỿ́dd-mawr, Abér-maw, Mín-ffordd, Pén-tir, Pén-mon, Pén-mon Máwr.[10]

Even when the article comes before the genitive, the whole name is sometimes thus treated, the accent falling upon the article; asPen-ỿ́-berth near Pwllheli,Tal-ỿ́-bryn in Llannefydd,Clust-ỿ́-blaiẟ near Cerrig y Drudion,Moel-ỿ́-ci (pron.Moɥ|lỿ́c|i), a hill near Bangor,Llan-é-cil near y Bala,Pen-é-goes near Machynlleth,Pen-é-berth near Aberystwyth (e forỿ,§ 16 iv (2)). Cf.(7) below.

Mi af i ganu i’m oes
I benáig oBen-é-goes.—L.G.C. 429.

‘I will go to sing while I live to a chieftain of Penégoes.’

(4) The wordduw (ordɥw) followed by the name of the day in the genitive; asDúw-sul as well asDúw Sū́l orDȳ́dd Sū́l ‘Sunday’; soDúw-llun ‘Monday’,Dúw-mawrth ‘Tuesday’, andDíf-i̯au forDúw I̯áu ‘Thursday’. Similarly(w)-gwyl ‘the day of the feast (of)’.

Echrỿ́s-haint, och, wir Iesu!
Ddyfod i IâlDdíf-i̯audu.—T.A.,g. 235.

‘A dreadful plague, Oh true Jesus! that black Thursday should have visited Yale.’ See§ 214 vii, ex. 2.

Both accentuations are exemplified in—

Bûm i’r gog swyddogDduw Sū́l;
Wy’ ddí-swydd, a hynDdúw-sul.—T.A.,a 14976/108.

‘I was an officer of the cuckoo on Sunday; I am without office, and this on Sunday.’ (Gwas y gog ‘the cuckoo’s servant’ is the hedge-sparrow.)

(5) A numeral and its noun, asdéu-bwys ‘2 lbs.’,dẃy-bunt ‘£2’,cán-punt ‘£100’, etc. Cf. E.twopence, etc. Though the order is the same here as in proper compounds, and the mutation is no criterion, it is certain that most of these are improper compounds. In the case ofun, proper and improper compounds can be distinguished:ún-ben ‘monarch’ is a proper compound, the second element having the soft initial, butún-peth is precisely the combinationū́n pḗth ‘one thing’ under a single accent.

(6) The demonstrative adjective after nouns of time. See§ 164 iii.

(7) Very rarely the article with its noun, as inÈ-fenéchtyd forỿ Fenechtyd ‘the monastery’, in which the article, taken as part of the word, acquired a secondary accent.

iii. Improper compounds accented on the ultima consist of—

(1) A few combinations of two monosyllabic nouns, of which the second is a dependent genitive and the first has lost its accent; aspen-rháith ‘autocrat’,pen-llā́d ‘summum bonum’,prỿ-nháwn forpryt nawn.

Yr eog, rhýwi̯ogben-rháith,
At Wén dos eto ún-w̯aith.—D.G. 148.

‘Thou salmon, gentle master, go to Gwen once more.’

A’m cérydd mawr i’m cári̯ad,
Ac na’th gawn yn lláwnben-llā́d.—D.G. 513.

‘And my great punishment for my love, and that I might not have thee as my whole delight.’

(2) A number of place-names of similar formation, asPen-tɥ́rch.

Note.—(1) From this and the preceding section it is seen that accentuation does not always accord with the formation of words. A loose compound is etymologically a compound, but its elements are accented as separate words. An improper compound is etymologically a combination of separate words accented as one word. The accentuation of improper compounds is to be accounted for thus: in O. W. we may assume thatgwr da,Aber Maw,Pen ỿ berth were originally accented as they would be if they were formed now, with the main stress in each case on the last word. When each combination came to be regarded as a unit, the main stress became the only accent; thus, *gwr-dā́, *Aber-máw, *Pen-ỿ-bérth. This was at that time the accentuation of ordinary words, such as *pechadū́r,§ 40 iii. When the accent shifted, and *pechadū́r becamepechádur, *gwr-dā́ becamegẃr-da, *Aber-máw becameAbér-maw and *Pen-ỿ-bérth becamePen-ỿ́-berth. In most cases of a combination like the last, each noun retained its individuality, and the original accentuation remained; hencePèn-ỿ-bérth, which is a common place-name, is usually so accented, and the accentuationPen-ỿ́-berth is exceptional. In such a phrase asprỿ́t náwn ‘time of noon’, each noun retained its meaning to the Ml. W. period; then, when the combination came to be regarded as a unit, the first element became unstressed, resulting inprỿt-náwn, whenceprỿ-nháwn,§ 111 v (5).

(2) Improper compounds having thus become units could be treated as units for all purposes; thus some of them have derivatives, such asgwr-dā́-aeth, ‘nobility’,tref-tád-aeth ‘heritage’,di-dref-tád-us.g. 306 ‘to disinherit’,prỿnháwn-ol ‘evening’ adj.

(3) On the other hand, in some proper compounds each element was doubtless felt to preserve its significance; and the persistence of this feeling into the Ml. period resulted in loose compounds.

§ 47. i. In compound prepositions the elements may be accented separately, asóddi ár. But the second element has usually the stronger accent; and in some cases the first element becomes unaccented, as in Ml. W.y gánn, which becamegan ‘by’ in Late Ml. and Mn. W. by the loss of the unaccented syllable.

On the analogy ofy gánn,y ẃrth, etc., derivative and other old prepositional and adverbial formations retained the O. W. accentuation, asodán, yr͑ẃng, yrháwg.

The separate accent often persists in Mn. W., as inóddi ẃrth (Ml. W.y wrth), and in adverbial phrases likeóddi ýno (in the dialectsṓdd ýno as in Ml. W.). In the latter the first element may become predominant, thusṓdd yno ‘from there’ in the spoken language (often contracted toṓẟno and evenṓno).

ii. In prepositional and adverbial expressions formed of a preposition and a noun (whether written separately or not), the last element only is accented; thusuwch-bén ‘above’,dra-chéfn ‘again’,ger-brón ‘before’,uwch-láw ‘above’,ymlā́en ‘forward’,ynghȳ́d ‘together’,i gȳ́d ‘together’,eri̯ṓed ‘ever’.

These expressions thus form improper compounds accented on the ultima. The adverbachlā́n (achlân) ‘wholly’ is similarly accented.

Hḗais fal orohī́an
I chlṓd yng Ngwyneddachlā́n.—D.G. 235.

‘I have sown her praises like a paean through the whole of Gwynedd.’

iii. Many adverbial expressions of three syllables, consisting of a monosyllabic noun repeated after a preposition, form improper compounds accented on the penult; asol-ỿ́n-ol ‘track in track’, i.e. ‘in succession’,[11]ben-drá-phen ‘head over head’,law-ỿ́n-llaw ‘hand in hand’, etc. The first noun may have a secondary or separate accent, asblìth drá-phlith ‘helter-skelter’. The first noun being in an adverbial case has a soft initial.

A daufrawd ieuaf ar ôl
Eli énw̯ogol-ỿ́n-ol.—G.Gl.,c. i 201.

‘And two younger brothers in succession after the famous Eli.’

Oes hwy no thri, Siôn, y’th roer,
Law-ỿ́n-llawâ’th law̯én-lloer.—T.A.,a 14866/74b.

‘For a life longer than three, Siôn, mayst thou be spared, hand in hand with thy bright moon.’ See also E.P. 240.

Ael-ỿ́n-aelâ’i elỿ́ni̯on.—D.N.,c. i 160.

‘Brow to brow with his enemies.’

Dal-ỿ́n-nalrhwng dwy lánnerch.—D.N.,m 136/147.

‘Face to face between two glades’;ýnnal forýn-nhal,§ 48 ii.

Daw o déidi̯audad-í-dad,[12]
Gollwyn hen,—nid gwell un had.—W.Ỻ.

‘He comes from forebears, father to father, like an ancient hazel-grove—there is no better seed.’

Arglwyddilī́n ó-linynt.[13]—L.G.C. 460.

‘They are lords from line to line.’

Seewers dragwers.a. 164 ‘reciprocally’,gylch ogylch do. 166 ‘round about’,ddẃrn trá-dwrn,láw drá-llaw, L.G.C. 18. In many cases the first noun also is preceded by a preposition, as

Marchog olī́n ó-linoedd. L.Mor.,i.mss. 292.

‘He was a knight from line to line.’

Seeo lwyn í-lwyn D.G. 141,o law í-law do. 145. Cf. Late Mn. W.í-gam ó-gam ‘zig-zag’.

The ordinary accentuation is also met with in the bards:

O lẃyn i lẃyn,ail Énid.—D.G. 84.

‘From bush to bush, [maiden] second to Enid.’

iv. Whenpa orpỿ is followed by a preposition governing it, the latter only is accented:pa-hám (forpa am,§ 112 i (2)) ‘what for? why?’ often contracted intopam by the loss of the unaccented syllable,§ 44 vii. So were doubtless accented the Ml. W.pahára.l. i 108, 134,pa hár do. 118 (forpa ar) ‘what on?’pa rácb.b. 50,pyrácr.m. 126 ‘what for?’

§ 48. i. When the syllable bearing the principal accent begins with a vowel, a nasal, orr, it is aspirated under certain conditions,§ 112 i (4); thusce|nhéd|loedd ‘nations’, fromcenedl;bo|nhé|ddig (vonheẟicr.p. 1331) frombonedd ‘gentry’,§ 104 iv (1);cy|nháli̯wyd, fromcynnal ‘to support’ fromcyn +dal (d normally becomesn, notnh,§ 106 ii);di|háng|ol fromdi-anc ‘to escape’;a phlannhédeur.p. 1303 ‘and planets’, usuallyplanedau;kenhadeuw.m. 184, oftener in Ml. W.kennadeu do. 42 ‘messengers’.

A’i aur a’i fedd y gŵyr fo,
Fonhéddig,[14]fy nyhuddo.—L.G.C. 188.

‘With his gold and mead doth he use, as a gentleman, to comfort me.’

ii. On the other hand, anh required by the derivation is regularly dropped after the accent; ascýnnes ‘warm’, forcýn-nhes fromcyn +tes (t givesnh,§ 106 iii (1));bré|nin ‘king’, forbrḗn|nhin frombre|en|nhin from *breentin, Cornishbrentyn;tán|nau ‘strings’, fortán|nheu from O. W.tantoum.c.;ḗang ‘wide’, foréh-ang from *eks-ang‑;ánawdd.a. 109 forán-hawdd ‘difficult’;áraul ‘bright’, forár-haul, which appears asarheul inr.p. 1168. Theh is, however, retained between vowels in a few words, asḗhud ‘foolish’,dḗhau anddḗau ‘right (hand), south’; and innrh,nhr,[15]nghr, andlrh, asánrhaith ‘spoil’,ánhrefn ‘disorder’,ánghred ‘infidelity’,ólrhain ‘to trace’.

Theh is also dropped after a secondary accent, as in brènini̯áethau ‘kingdoms’. So we havecènedláethau ‘generations’,bòneddígaidd ‘gentlemanly’ (voneẟigeiẟr.g. 1129).

iii. Note therefore the shifting of theh in such a word asdiháreb ‘proverb’, Ml. W.dihaerebr.p. 1326, pl.dìarhébi̯on, Ml. W.diaerhebẏonr.b. 974, 975, 1083. The word has etymologically twoh’s:di-haer-heb, but only that is preserved which precedes the principal accent.

iv. The above rules may be briefly stated thus: an intrusiveh sometimes appears before the accent, and an organich regularly disappears after the accent. It is obvious that the rule cannot be older than the present system of accentuation; it is indeed the direct result of that system, and is probably not much later in origin. The first change was the weakening and subsequent loss ofh after the accent, giving such pairs asbrenin,brenhinoedd;angen,anghenus (< *n̥ken‑, Ir.ēcen);cymar, cymharu (< Lat.compar‑): hereh vanishes in the first word of each pair. Later, on the analogy of these, other pairs were formed, such asbonedd, bonheddig;cenedl, cenhedloedd; where an intrusiveh appears in the second word of each pair.

In O. W., when the accent fell on the ultima, it was easy to saybre|en|nhī́n; but when the accent settled on the penult, it required an effort to sound the aspirate after the breath had been expended on the stressed syllable. Hence we find, at the very beginning of the Ml. period,breenhineẟ andbreeninl.l. 120. But the traditional spelling, withh, persisted, and is general inb.b., asminheu 12;synhuir (≡synnhwyr) 17;ag̃hen ag̃heu 23;breenhin 62; though we also find a few exceptions, askag̃ell 35. Inr.m. it still survives in many words, asbrenhin 2;ag̃heu 5 (butangeu ib.);mwyhaf 11;minheu 12; but more usuallyvwyaf 13;minneu 3;gennyf 8;synnwyr 13;amarch 36;llinat (forllin-had) ‘linseed’ 121. In ther.p. we findánawẟ 1227, 1264, 1270, 1299;áneirdd, ánoew̯ 1226;diagɏr (fordí-hagr) 1289;lláwir (forllaw-hir ‘long-handed’) 1207, 1226;láwhir 1214, withh inserted above the line—an etymological correction;áwrḥonn 1271, withh deleted by the underdot—a phonetic correction.

Intrusiveh makes its first appearance later, and is rarer in Ml. W. than losth. Ina.l., ms. a., we findboneẟyc ii 6, 14, but in thisms.n may be fornh; in latermss.bonheẟyc i 176–8,ms. e.;bonheẟic in Ml. W. generally. In other cases it is less usual; thuskennadeu is the form inr.m., though the olderw.m. has sometimeskenhadeu 184, 249;kenedloeẟr.b.b. 259,.a. 169, so generally.

The orthography of the 1620 Bible generally observes the phonetic rule; thusbrenin, brenhinoedd Ps. ii 6, 2;cenedl, cenhedloedd do. xxxiii 12, ii 1;angeu, anghefol do. vi 5, vii 13;aros, arhosodd Jos. x 12, 13;bonheddig, boneddigion Es. ii 9, 1 Cor. i 26;ammarch, ammherchi Act. v 41, Rhuf. i 24; etc. There are some irregularities and inconsistencies; e.g.diharebion Diar., title, i 1, andanghall Diar. i 4 beside the phoneticangall do. viii 5. The Bible spelling was generally followed, and the use ofh medially was fairly settled on phonetic lines, when Pughe introduced confusion by discarding it wherever his mad etymology failed to account for it. His wildest innovations, such asglanâu,parâu forglanháu,parháu, were rejected by universal consent; but his principle was adopted by the “new school” including T. Charles, Tegid and G. Mechain, who disregard the accent, and insert or omit h in all forms of the same vocable according to their idea of its etymology.[16] Silvan Evans (Llythyraeth, 68) writes as if the cogency of this principle were self-evident, and imagines that to point out the old school’s spelling ofcyngor without, andcynghorion with, anh, is to demonstrate its absurdity. In his dictionary he writesbrenines, boneddig, etc., misquoting all modern examples to suit his spelling; underammeuthun (his misspelling ofamheuthun) he suppressesh in every quotation.

In spite of the determined efforts of the “new school” in the thirties, present-day editions of the Bible follow the 1620 edn. with the exception of a few insertions of etymologicalh, as inbrenin, ammarch, which appear asbrenhin, ammharch.

Quantity.

§ 49. In Mn. W. all vowels in unaccented syllables are short.

Unaccented syllables here include those bearing a secondary accent, in which the vowel is also short, as incĕ̀nedláethau, though before a vowel it may be long in deliberate pronunciation, as indḕalltẃri̯aeth.

In Late Ml. W. the same rule probably held good, but not necessarily earlier. In O. W. it was clearly possible to distinguish in the unaccented penult the quantities preserved later when the syllable became accented,§ 56 iv.

§ 50. Vowels in accented syllables in Mn. W. are either (1)long, as thea incân ‘song’; (2)medium as thea incanu; or (3)short, as thea incann ‘white’,cannu ‘to whiten’.

In monosyllables a long vowel (excepti oru) is generally circumflexed beforen,r orl,§ 51 iv, and in any other case where it is desired to mark the quantity. Short vowels are marked by ` which is sometimes used instead of doubling the consonant, as in D.D. s.v.càn =gan ‘with’, and beforel which cannot be doubled in writing;dàlb.cw. 91,hèl do. 95,càlon Hyff. Gynnwys (1749) pp. 3, 20, 319 bis.

☞In this grammar the circumflex has been retained in most cases where it is, or might be, used in ordinary writing. But where the position of the accent has to be indicated, ˉ́ is used; where there is no need to point out the accent, and the word is not usually circumflexed, ˉ is used. As every long vowel must be accented in Mn. W., it will be understood that ˉ, ˉ́ and ˆ in Mn. W. words mean the same thing. In Brit. and earlier a vowel marked ˉ is not necessarily accented. As ` is required to denote a secondary accent it would be confusing to use it to mark a short accented vowel; hence ˘́ is used here for the latter purpose, where necessary. The accent mark ´ denotes accent without reference to quantity. A medium vowel can only be indicated by showing the syllabic division; thus|nu.

Note. The medium vowel, or short vowel with open stress, which occurs in the penult, is not heard in English where a penultimate accented vowel, if not short as infathom, is long as infather. Silvan Evans calls the medium vowel “long”, and J.D.R. often circumflexes it. But thea of|nu is not long, except in comparison with thea ofcán|nu; beside thea ofcân it is short. It is a short vowel slightly prolonged past the point of fullest stress, so as to complete the syllable, and the following consonant is taken over to the ultima.

§ 51. i. If a vowel in a monosyllable is simple its quantity is determined by the final consonant or consonants, the main principle being that it is long before one consonant, short before two, or before a consonant originally double; see§ 56 ii.

ii. The vowel is short before two or more consonants, or beforep,t,c,m,ng; ascănt ‘hundred’,tŏrf ‘crowd’,pŏrth ‘portal’,bărdd ‘bard’,ăt ‘to’,llăc ‘slack’,căm ‘crooked’,llŏng ‘ship’.

Nearly all monosyllables ending inp,t orc are borrowed; some from Irish, asbrăt ‘apron’, most from E. ashăp,tŏp,hĕt,pŏt,cnŏc, which simply preserve the original quantity. E. tenuis after a long vowel becomes a media, as W.clôg < E.cloak, W.grôdg. 157 < E.groat, re-borrowed asgrôt; so the late borrowingscôt, grât (but in S. W.cǒt}.

W.ăt is an analogical formation,§ 209 vii (2);ac, nac should beag, nag in Mn. orthography§ 222 i (1),ii (3).

Exceptions to the above rule are the following:

(1) In N. W. words ending ins orll followed by another consonant have the vowel long; astrīst ‘sad’,cōsb ‘punishment’,hāllt ‘salt’ adj., etc., except in borrowed words, ascăst ‘trick’. In S. W., however, all such words as the above conform to the rule.

(2) The vowel is long when it is a late contraction,§ 33 iv; asânt ‘they go’, fora-ant;bûm ‘I have been’, forbu-um;bônt ‘they may be’, forbo-ont;rhônt ‘they give’, forrho-ant. Inɥ̂m ‘we are’,ɥ̂nt ‘they are’, the vowel is pronounced long; it is marked long by J.D.R. 94; but E.P.,ps. lxxv 1, rhymesynt withhynt, and in Ml. W. it is writtenynt (not *yynt); hence the lengthening is probably due to false analogy.

Cânt ‘they shall have’ is forca-ant and has longa; butcant ‘sang’ is forcan‑t; and is therefore short. Evengwĕld,§ 44 vi, fromgwêl, has thee shortened by the two consonants; a fortiori, incant ‘sang’ where the final double consonant is older, thea must be short. Silvan Evans (s. v.canu) adopts the error of some recent writers, and circumflexes thea incant, even where it rhymes withchwant, and in quoting Gr.O. 82, where no circumflex is used. The word never rhymes withânt,gwnânt, etc.

☞The vowel is circumflexed when long before two consonants, except where the length is dialectal.

(3) The mutated formdēng ofdeg ‘ten’ preserves the long vowel of the latter in N. W.

iii. The vowel is long if it is final, or followed byb,d,g,f,dd,ff,th,ch,s; as ‘house’,llē ‘place’,māb ‘son’,tād ‘father’,gw̯āg ‘empty’,dōf ‘tame’,rhōdd ‘gift’,clōff ‘lame’,crōth ‘womb’,cōch ‘red’,glās ‘blue’.

Exceptions: (1) Words which are sometimes unaccented,vi below.

(2) Words borrowed from English, assăd ‘steady’,tw̆b,fflăch (fromflash),lăch (fromlash).Sŭd, also writtensŭt, ‘kind, sort’ fromsuit (cf. Chaucer, Cant. Tales 3241) is now short; but in D.G. 448 it is long, rhyming withhud.

(3) Some interjectional words, such aschwăff, pĭff, ăch. The interjectionoch is now short, but is long in the bards; seeOch/Gōch D.G. 464.Cȳff is now sometimes incorrectly shortened.

☞A long vowel need not be circumflexed before any of the above consonants. In the case of a contraction, however, the vowel is usually marked; thusrhôdd ‘he gave’ forrhoodd forrhoddodd. In such forms the circumflex is unconsciously regarded as a sign of contraction, and may be taken to indicate that the vowel is long independently of the character of the consonant.

The circumflex is also used innâd ‘cry’ to distinguish it fromnăd ‘that not’.

iv. If the vowel be followed byl,n orr, it may be long or short:tâl ‘pay’,dăl ‘hold’,cân ‘song’,căn ‘white’;câr ‘relative’,căr ‘car’.

Each of these consonants may be etymologically single or double.Dăl is from *dalg-§ 110 ii (2), so that the finall represents two root consonants. In O. and Ml. W. finaln andr when double in origin were doubled in writing, as inpenn, ‘head’, Irishcenn, in other cases of course remaining single as inhēn ‘old’, Irishsen; thus the principle that the vowel is short before two consonants, long before one, applied. The final consonant is now written single even in words likepen, and only doubled when a syllable is added, as inpennaf, cf. Eng.sin (O. E.sinn) butsinner (though even medial‑nn- is now sounded‑n- in Eng.). It is therefore necessary now to distinguish between long and short vowels in these words by marking the vowels themselves.

☞In a monosyllable, a long vowel followed byl,n orr is circumflexed; thus,tâl ‘pay’,cân, ‘song’,dôr ‘door’,dêl ‘may come’,hŷn ‘older’. Buti andu need not be circumflexed, since they are always long before these consonants, except inprin, andin (= Ml. W.ynn ‘to us’), and a few words from English aspĭn,bĭl. The common wordsdȳn,hēn,ōl are seldom circumflexed.

Ml. W.‑nn is still written in some words, e.g. inonn ‘ash’ pl.ɥnn, as in the namesLlwyn Onn,Llwyn Ynn. Doubling the consonant is preferable to marking the vowel when it is desired to avoid ambiguity, as incann ‘white’,a yrr ‘drives’. It is not sounded double now when final; but the consonant is distinctly longer e.g. inpĕn than inhēn. In Corn.,penn becamepedn.

Note. Thea is long intâl ‘forehead, front, end’, and was circumflexed down to the latter part of the 18th cent.; see D.D. s.v.,g. 68. Thel is etymologically single, as is seen in the Gaulish nameCassitalos. In the spoken language the word survives only in place-names, and is sounded short in such a name asTàl-y-bónt because this has become an improper compound accented on the ultima,§ 46 iii, so that its first element has only a secondary accent,§ 49. When the principal accent falls on it, it is long, as inTrwyn-y-tâl near the Rivals.Tegig̃il otâl,Edeirnaun, Iâlb.b. 74 ‘Tegeingl to its end, Edeirnawn, [and] Yale.’ The rhyme withIâl shows the quantity oftâl.

Y fun araf, fain, eirian,
A’rtâlfal yr aur mâl mân.—D.G. 330.

‘The calm, slender, bright girl, with the head like finely milled gold.’

v. When the word ends inll the quantity varies. In N. W. it is short in all such words exceptōll, hōll; in S. W. it is long, except ingăll ‘can’,dŭll ‘manner’,mw̆ll ‘sultry’,cy̆ll ‘loses’, and possibly some others.

vi. Many prepositions, adverbs and conjunctions, which are long by the above rules, by being often used as proclitics have become short even when accented, more especially in N. W.; asrhăg ‘against’,hĕb ‘without’,nĭd, năd ‘not’,dăn ‘under’ (originally onen),măl, făl, fĕl ‘like’,ăg (written ac) ‘and’,năg (writtennac) ‘nor’; butā́g ‘with’.

The long vowel is preserved in some of these in S. W. The wordnes ‘until’,§ 215 i (2), was circumflexed even by N. W. writers as late as the 18th cent., seenêsg. 237; it is now soundednĕs (alreadynes inb.cw. 83, 115 besidenês ‘nearer’ 13, 109, 110). In D.G.dan ‘under’ has longa:

Serchog y cândāny dail.—D.G. 225.

‘Lovingly it sings under the leaves.’

§ 52. i. If the vowel in a monosyllable is the first element of a diphthong, its quantity depends chiefly upon the form of the diphthong.

ii. The vowel is long inae, oe, w͡y; thustrāed ‘feet’,ōen ‘lamb’,hŵyr ‘late’,cāe ‘field’,cāem ‘we might have’,dōe ‘yesterday’,mw̄y ‘more’,cŵyn ‘complaint’,hw̄ynt ‘they’,blōesg ‘blaesus’,rhw̄ysg ‘pomp’,māent ‘they are’,trôent ‘they might turn’.

But except before‑sg,w͡y is short before two or more consonants orm; astw̆ym, tw̆ymn, ‘hot’,rhw̆ym ‘bound’ (alsorhw̄ym),cw̆ymp ‘fall’ (now pron.cw̯ɥ̆mp in N. W.),llw̆ybr ‘path’,rhw̆ystr ‘hindrance’,brw̆ydr ‘battle’,pw̆ynt ‘point’;—hw̄ynt is influenced byhw̄y ‘they’. Similarlymāent formed from, and influenced bymāe. The other cases are examples of contraction:cāem <cā-em,trôent <tro-ynt.

iii. The vowel is short in all other falling diphthongs; asbăi ‘fault’,by̆w ‘alive’,trŏi ‘to turn’,llăid ‘mud’,brĭw ‘wound’,dŭw ‘god’,bŭwch ‘cow’,hăul ‘sun’,ăur ‘gold’,dĕwr ‘brave’,băwd ‘thumb’,măwl ‘praise’, etc.

Exceptions: (1) In N. W.aw, ew are long when final only; astāw! ‘be silent’,bāw ‘dirt’,llēw ‘lion’,tēw ‘fat’; otherwise short as above. In S. W. the diphthongs are short in both cases.

(2)au is long intrāul ‘wear, expense’,pāun ‘peacock’,gwāudd ‘daughter-in-law’,ffāu ‘den’,gwāun ‘meadow’,cāul ‘rennet’,pāu ‘country’. The formgwaen is a recent misspelling ofgwāun. In West Gwynedd the word is pronouncedgwĕun (eə), Ml. W.gweun, O. W.guoun.

(3) The vowel is long inâu when contracted fora-au, as inplâu ‘plagues’; but incău forcáe-u,§ 202 iii, it is short. It is long inâi fora-ai, andôi foro-ai when final, asgwnâi, trôi 3rd sg. impf.; but ŏi foro-ai not final, as intrŏis fortró-ais. On account of the long vowelgwnâi, trôi, etc. are generally sounded and often writtengwnae, troe, etc.; but in the bards‑âi rhymes withai, seewnâi / ehedaig. 242. Both forms are seen in Ml. W.gwnai W.M. 25, 54,gwnayr.m. 237 (ae=ay,§ 29 ii (1)).

(4) The vowel is long ino’i, a’i, da i̯, etc.,§ 33 v, of course only when accented. In Ml. W.o’i, a’i are writtenoe, ae oroy, ay.

§ 53. When the accent in a polysyllable falls on the ultima, the above rules apply as if the ultima were a monosyllable; thus, short,pahắm ‘why?’,penắig,§ 41 iii (2),parhắu ‘to continue’,gw͡yrdrŏ́i ‘to distort’; long,Cymrā́eg, parhā́nt (forparhá-ant),gw͡yrdrṓi (forgw͡yrdro-ai) ‘he distorted’,penllā́d ‘summum bonum’.

Inparhau, caniatau, etc., some recent writers circumflex thea, possibly a practice first intended to indicate the long vowel in the uncontracted form‑ha-u,§ 54 iii. When contracted thea is short. In D.D. and Bible (1620) it is not circumflexed. J.D.R. 144 writescadarnháu. But see§ 55 ii.

§ 54. In the accented penult—

i. (1) The vowel is short, if followed by two or more consonants, or byp, t, c, m, ng, ll, s; ashărddwch ‘beauty’,plĕntyn ‘child’,cănnoedd ‘hundreds’,by̆rrach ‘shorter’,ĕstron ‘stranger’,ĕpil ‘progeny’,ăteb ‘answer’,ămeu, ‘to doubt’,ăngen ‘need’,ăllan ‘out’,Iĕsu ‘Jesus’,glăndeg ‘fair’,glănw̯aith ‘cleanly’,tăni̯o ‘to fire’,ty̆bi̯af ‘I suppose’. There is no exception to this rule, though beforem the vowel is sometimes wrongly lengthened in words learnt from books, such astrămor ‘foreign’,ămwys ‘ambiguous’.

Silvan Evans marks many obsolete words, such asamwg, amug with longā, for which there is no evidence whatever; it merely represents his own misreading of Ml. W.‑m‑, which always stands for‑mm‑.

(2) The consonants above named are each double in origin. In Ml. W.t, c, s were usually doubled in this position, asatteb,racco orracko,messur; but‑m- is generally written single, owing to the clumsiness of‑mm- and its frequency; possibly‑p‑, which is not very common, followed the analogy of‑m‑;ll andng being digraphs can hardly be doubled in writing. In early Biblesm andp are doubled; and G.R. wrotegaḷḷu, doubling (his =). As however each is etymologically double (except in borrowed words), the double origin is sufficiently indicated by writing the letter; thusateb is necessarily the same asatteb;mesur is necessarilymessur. So every medial or finalm,ng orll meansmm,ŋŋ, orỻỻ etymologically, and is so pronounced in the accented penult.

☞ But in the case ofn andr the consonant is not necessarily douhle; hence a distinction must be made between single and doublen andr. Thea incannu ‘to whiten’ is short because it is followed bynn, representing originalnd (cf. Lat.candeo); thea incanu ‘to sing’ is medium because it is followed by a singlen (cf. Lat.cano). The distinction is made in nearly all Ml.mss., and generally in Mn.mss. and printed books down to Pughe’s time.

(3) The accented syllable is “closed” (stopped, blocked) by the first of the two consonants, thusglắn|deg,plĕ́n|tyn,cắn|nu. Even and cause the preceding consonant to close the penult; thusglắn|w̯aith fromglân ‘clean’. Ml. scribes, knowing that the syllable was closed by two consonants, and not knowing that the second in this case was or, sometimes doubled the first consonant, as indynnẏonw.m. 32, (g)lannweithr.m. 52; but as a rule, perhaps, it is written single, as indynẏonr.m. 21, (g)lanweithw.m. 72. A consonant originally double cannot be distinguished from one originally single in this case; thustắn-i̯o ‘to fire’, fromtân ‘fire’, andglắ-i̯o ‘to land’, fromglann ‘shore’, form a perfect double rhyme. It is therefore unusual to double the consonant in the modern language in these forms;glannio andtorriad are writtenglanio andtoriad, which adequately represent the sound (cf.pentref forpenntref, etc.). Thus inysgrifennw͡yd ‘was written’ the doublen indicates that thew is a vowel; inysgrifenw̯yr ‘writers’, the singlen indicates that the is consonantal. Hence some words likeannw͡ylc.m. 70,synnw͡yrr.m. 116 are now written with onen owing to a common, but by no means general, mispronunciation ofw͡y asw̯ɥ; seeP.Ỻ. xcvi, whereLlyr /ssynwyr is condemned as a false rhyme.

ii. The vowel is medium if followed byb,d,g,[W 2]ff,th,ch,l, singlen, or singler; as|baith ‘hope’,á|deg ‘time’,|gur ‘idle’,é|ffaith ‘effect’,é|thol ‘to elect’,|chod ‘sin’,|nu ‘to sing’,|re ‘morning’,|lan ‘new year’s day’.

In this case the accented syllable is “open” (free), that is, it ends with the vowel, and the consonant is carried on to the next syllable. See§ 50, Note;27 i.

In a few forms we have a short vowel beforel, as inIŏ́l|o (often mis-read|lo);cắl|on ‘heart’;cŏ́l|yn ‘sting’, O. W.colginnjuv.;bŏ́l|wst ‘colic’ < *bolg‑;dĕ́l|ir ‘is held’ fordĕ́l|i̯ir§ 36 i <*dĕ́lᵹ̑ir. In Ml. W. such forms are written with doublel,§ 22 ii.

Doublel cannot be from originalll, which gives the voiceless Welshll (). It occurs only in a new hypocoristic doubling as inIol-lo, or where a consonant now lost closed the syllable before disappearing: incắlon the lost consonant is; incŏ́lyn it is < ᵹ̑; drops beforeo, and beforey§ 36 iii,ii;—cắlon (Corn.colon, Bret.kalon, kaloun) < *kalu̯ond‑: W.colweẟb.a. 6 ‘heart’,coludd ‘entrail’: Skr.kroḍá‑ḥ ‘breast, interior’: Gk.χολάδες, O. Bulg.želąd‑ŭkŭ ‘maw’ withɡh- (q/ɡh alternation).—For Early Mn. W.cắlyn ‘to follow’ the Ml.canlyn has been restored in writing.

A short vowel also occurs incădwn, ty̆bir, etc.§ 36 i.

iii. The vowel is long if followed by a vowel orh; asḗog (‘salmon’,dḗ-hau ‘right, south’,Gwen|llī́|an.

iv. It is short in all falling diphthongs; ascắe|ad ‘lid’,mw̆́y|af ‘most’,llĕ́i|af ‘least’,rhw̆́y|dau ‘nets’,llw̆́y|brau ‘paths’,hĕ́u|log ‘sunny’,tĕ́w|dwr ‘thickness’,bỿ̆́w|yd ‘life’,cnắw|dol ‘carnal’.

But in N. W. the vowel is medium inaw, ew, iw before a vowel, that is thew is heterosyllabic; thus|w̯el ‘silent’,|w̯i ‘to be silent’,llé|w̯od ‘lions’,|w̯ed ‘harm’. In S. W., however, these are soundedtắw|el, tĕ́w|i, llĕ́w|od, nĭ́w|ed.

§ 55. i. The above are the quantities of the vowels in the Mn. language. They were probably the same in Ml. W. where the vowel is simple. Thusmap ormab, tat, gwac had a longā like their modern equivalentsmāb, tād, gwāg; for where the vowel was short and the final consonant voiceless (= Mn.p, t, c), the latter was doubled, as inbrattr.g. 1117, Mn. W.bratt D.D., orbrat (≡brăt) ‘rag, apron’. In the case of Ml. single‑t, both the long vowel and the voiced consonant are attested in the spelling of foreigners; thus the place-name which is nowBōd Feirig, which in Ml. W. spelling would be *Bot veuruc, appears in Norman spelling in the Extent of Anglesey, dated 1294, asBode-ueuryk (Seebohm, Trib. Sys.¹ App. 6), wherebode doubtless meansbōd, the Mn. W. sound. Again in the Extent of Denbigh, dated 1335, the Mn. W.Rhōs appears asRoos (op. cit. 72), showing the vowel to be long befores then as now. The N. W. long vowel beforest is attested in 1296 in the Ruthin Court Rolls p. 15, l. 10 in the spellingNeeste of the nameNest. The distinction between medium and short in the penult is everywhere implied in Ml. spelling; and we are told inr.g. 1120 that the vowel is long when followed by another, as thei inGwenlliant, Mn. W.Gwen-llī́-an. Thus the quantity of a simple vowel was generally the same in all positions in Ml. and Mn. W., even local usage agreeing; except in shortened words§ 51 vi.

ii. But in diphthongs many changes must have taken place. As a “vowel before a vowel” was long then as now,trṓ-ï must have had a longō, so that, when first contracted, it was still long; it remains long in Montgomeryshire; thus the shorto intrŏi̯ is probably late. Similarly shortĕi fore-i,ău fora-u,ŏu foro-u. Other diphthongs also probably differ, and we can infer nothing as to Ml. W. quantity in diphthongs from the Mn. W. pronunciation.

§ 56. i. The quantity of a vowel in British determines its quality in Welsh; but its quantity in Welsh depends, as we have seen, on the consonantal elements which follow it in the syllable.

ii. A short accented vowel in Brit. or Latin ollowed by a single consonant was lengthened in Welsh; thus Brit. *tălos gavetâl,§ 51 iv Note, *rŏtā (cognate with Lat.rŏta) gaverhōd, Lat.sŏnus gavesôn, etc. This took place after the change in the quality of long vowels, for while originalā givesaw§ 71, longā lengthened fromă remainsâ. It also took place after the reduction ofpp, tt, cc intoff, th, ch, for the latter are treated as single consonants for this purpose; thus Lat.saccus became *saχos with singleχ, which gives sach (≡sāχ) in Welsh. Long vowels remained long, as inpûr from Lat.pūrus. On the other hand, a vowel originally long was shortened before two consonants; thus theō of Lat.fōrma becameǖ, which was shortened in the Welshffŭrf. Hence the general rule§ 51 i, which probably goes back to Early Welsh and beyond; for the lengthening of short vowels originated at the time of the loss of the ending, and is due to compensation for that loss.

iii. There is no reason to suppose that this lengthening took place only in monosyllables. Thus O. W.litan ‘wide’ (: Gaul.litanos in Κογκο-λιτανος,Smertu-litanus, etc., Ir.lethan) was probably sounded *llỿ-dā́n, whileguinlann was doubtless *gwinl(l)ắnn. In Ml. W. when the ultima became unaccented this distinction was lost, thea ofllýdān being shortened,§ 49, and thenn ofgwín-llann being simplified,§ 27 ii. The rule forbidding the rhyming of such a pair was handed down from the older period, and is given inr.g. 1136; such a rhyme is calledtrwm ac ysgawn ‘heavy [with 2 consonants] and light [with one]’. But the bard’s ear no longer detected any difference in the unaccented ultima; he is therefore instructed to add a syllable to find out whether the syllable is “heavy” or “light”:kallonneu (lll‑l) is given as an example to show that theon(n) ofkallon [sic] is “heavy”, andamkaneu to show that thean ofamkan is “light”. The Early Ml. bards avoidtrwm ac ysgawn; but in the first poem inb.b., where the rhyme is‑ann, several forms in‑an occur, asimuan 1 (:gwanaf ‘I wound’),darogan 7 (:canaf ‘I sing’), which shows that the distinction was beginning to disappear. The Late Ml. poets frankly give it up; e.g. Ca.bychan / glan / kyvan(n) /diflan(n) /darogan / …kalan(n) /kan /Ieuan(n),r.p. 1233–4. Yet in O. W. the distinction was a real one, for it is reflected in the ordinary spelling of words; asbichanox. ‘little’ (cf.vychanetw.m. 44,r.m. 31),atarox. ‘birds’ (cf.adarenb.b. 107),scribennm.c. ‘writing’ (cf.yscrivennuỺ.A. 2),corsennox.,guinlannjuv., etc. The dimin. endings‑yn, ‑en appear as‑inn,‑enn; the pl. ending‑i̯on is always‑ion.

iv. In the unaccented penult in O. W. the distinction between an open and a closed syllable was preserved; the vowel must have been shorter in the latter, as it was later when the penult became accented.

v. The diversity in the present quantity of vowels beforell ands, and the fixing of the present quantities of diphthongs, are due to complicated actions of analogy, which it would take too much space here to attempt to trace.


The Aryan vowels in Keltic

§ 57. Parent Aryan had the following vowel-system:

Short vowelsaeiouə
Long vowelsāēīōū
Short diphthongs aieioiaueuou
Long diphthongsāiēiōiāuēuōu
Short vocalic
Long vocalicl̥̄m̥̄n̥̄r̥̄

e ando were probably pronounced open;u has of course its Latin value ≡ Welshw (not Welshu);ə was an obscure vowel whose exact quality is uncertain, but which was probably not unlike W.ỿ; vocalic,,, arose from reducedel, em, en, er; when long they represent the contracted reductions of two syllables§ 63 vii (2).

§ 58. i. The Aryan short vowels remained unchanged in Primitive Keltic, exceptə, which becamea as in all the other branches except Indo-Iranian, in which it becamei, seevii below.

ii. Ar.a (Lat.a, Gk.α). Lat.dacruma (lacruma), Gk.δάκρυ, Goth.tagr: W. pl.dagrau ‘tears’ < Pr. Kelt. *dakruu̯a.—Ar. *ag̑ō > Lat.agō, Gk.ἄγω: Ir.agaim ‘I drive’, W.af fora-af for *aᵹaf ‘I go’ < Pr. Kelt. *ag‑.—Lat.sāl, sălis, Gk.ἅλς, Goth.salt: Ir.salann, W.halen ‘salt’ < Pr. Kelt. *sal‑.

iii. Ar.e (Lat.e, Gk.ε). Ar. *bher- > Lat.ferō, Gk.φέρω, O. E.beran ‘to bear’: Ir.berimm ‘I bear’, W.ad-feraf ‘I restore’ < Pr. Kelt. *ber‑.—Ar. *medhu- > Gk.μέθυ ‘wine’, O. H. G.metu ‘mead’, O. Bulg.medŭ ‘honey’, Skr.mádhu ‘honey’: W.medd ‘mead’,meddw̯ ‘drunk’ < Pr. Kelt. *medu- *medu̯‑.—Ar. *ek̑u̯os > Lat.equus, Skr.ás̑va‑ḥ: Ir.ech ‘horse’, Gaul.Epo- (inEpo-redia, etc.), W.eb-ol ‘colt’ < Pr. Kelt. *eku̯‑.

iv. Ar.i (Lat.i, Gk.ι). Ar. *u̯id- (√u̯eid- ‘see, know’) > Lat.video ‘I see’, Gk. Hom.ϝίδμεν, Goth.witum ‘we know’: Ir.fiss ‘knowledge’, W.gw̯ŷs ‘summons’ < Pr. Kelt. *u̯iss‑,§ 87 ii.—Ar. *u̯liq- (√u̯eleiq- ‘wet’) > Lat.liqueo: Ir.fliuch ‘wet’, W.gw̯lyb ‘wet’ < Pr. Kelt. *u̯liq.

v. Ar.o (Lat.o, Gk.ο). Ar. *ok̑tṓ(u) > Lat.octō, Gk.ὀκτώ: Ir.ocht, W.ŵyth ‘eight’ < Pr. Kelt. *oktō§ 69 iv (2). Ar. *loɡh- (√leɡh- ‘lie’) > Gk.λόχος ‘bed, couch, ambush’, O. Bulg.są-logŭ ‘consors tori’: W.go-lo-i,r.p. 1040, ‘to lay, bury’ < Pr. Kelt. *log‑.—Ar. *toɡ- (√(s)theɡ- ‘cover’) > Lat.toga: W.to ‘roof’,§ 104 ii (2).

vi. Ar.u (Lat.u, Gk.υ). Ar. weak stem *k̑un- > Gk. gen. sg.κυνός, Goth.hunds, Skr. gen. sg.s̑únaḥ: W. pl.cŵn ‘dogs’ < Pr. Kelt. *kun-es.—Ar. *sru‑t- (√sreu- ‘flow’) > Gk.ῥυτός ‘flowing’, Skr.srutáḥ ‘flowing’, Lith.srutà ‘dung-water’: Ir.sruth ‘stream’, W.rhwd ‘dung-water’ < Pr. Kelt. *srut-.

vii. Ar.ə (seei). Ar. *pətḗr *pətér- > Lat. pater, Gk.πατήρ, Goth.fadar, Arm.hair, Skr.pitár‑: Ir.athir ‘father’ < Pr. Kelt. *(p)atīr. Ar. *sət- (√sē- ‘sow’) > Lat.satus: W.had ‘seed’ < Pr. Kelt. *sat‑,§ 63 vi (1).

§ 59. i. The Aryan long vowelsā,ī,ū remained; butē becameī; andō in stem syllables becameā, in final syllablesū.

ii. Ar.ā (Lat.ā, Gk. Dor.ᾱ, Att. Ion. η). Ar. *bhrāt-ēr, ‑er‑, ‑ōr, ‑or- > Lat.frāter, Gk. Dor.φρᾱ́τηρ ‘member of a clan’, Goth.brōþar, Skr.bhrā́tar‑: Ir.brāthir, W.brawd ‘brother’, pl.broder, brodorion§ 124 i < Pr. Kelt. *brāt-īr, ‑er‑, ‑or‑.—Ar. *māt-ēr, ‑er‑, ‑r- > Lat.māter, Gk. Dor.μᾱ́τηρ, Skr.mātár‑: Ir.māthir ‘mother’, W.modr‑yb ‘aunt’ < Pr. Kelt. *māt-ēr,‑r‑.

iii. Ar.ē (Lat.ē, Gk.η). Lat.vērus, O. Bulg.věra ‘faith’: Ir.fir, W.gwīr ‘true’ < Pr. Kelt. *u̯īros.—Lat.rēx, Skr.rā́j- ‘king’: Ir., Gaul.rīx, W.rhī < Pr. Kelt. *rīks, *rīg‑.

iv. Ar.ī (Lat.ī, Gk.). Ar. *qrīt- (√qrei̯ā- ‘buy’) > Skr.krītaḥ ‘bought’: Ir.crīthid ‘inclined to buy’, W.prīd ‘precious’ < *Pr. Kelt.qrīt‑. O. H. G.rīm, O. E.rīm ‘number’: Ir.rīm, W.rhīf ‘number’ < Pr. Kelt. *rīm‑. Ar. suffix *‑īno‑, as in Lat.su-īnus: W.‑in§ 153 (10) < Pr. Kelt. *‑īno‑.

v. Ar.ō (Lat.ō, Gk.ω). Lat.ōcior, Gk.ὠκύς, Skr.ās̑úḥ ‘quick’: Ml. W.di-awc, Mn. W.di-og ‘idle’ < Pr. Kelt. *āk-us.—Lat.ignōtus, nōtus, Gk.γνωτός: Ir.gnāth ‘known, accustomed’, W.gnawd ‘customary’ < Pr. Kelt. *gnātos.—Lat.flōs, O. H. G.bluot ‘bloom’: Ir.blāth, Ml. W.blawt ‘blossom’ < Pr. Kelt. *blāt‑.

In final syllables Ar.ō > Kelt.ū; this becameǖ, laterī in Brit., and affected a preceding vowel,§ 69 i; it remains as‑ī in W. ‘dog’§ 132 (1). But when followed by a final nasalō becameo in Pr. Kelt.; thus Ir. gen. pl.fer ‘of men’ implies *u̯irŏn from *u̯irōm *‑ōm: Gk.‑ων).

vi. Ar.ū (Lat.ū, Gk.). Lat., Gk.τῡ́-νη), O. Icel.þū, Avest.: Ir., W. ‘thou’ < Pr. Kelt. *.—O. H. G.rūna, O. Icel.rūn ‘secret, rune’: Ir.rūn, W.rhīn ‘secret’ < Pr. Kelt. *rūn‑.—Lat.cūlus: Ir.cūl, W.cīl ‘back’ < Pr. Kelt. *kūl‑.

§ 60. The Aryan short diphthongs remained in Pr. Kelt.; see examples in§§ 75,76. In the long diphthongs the long vowels developed as elsewhere; thusāi,āu remained;ēi >īi̯;ēu >īu; in syllables not finalōi,ōu becameāi,āu respectively; in final syllablesōi >ūi, later doubtlessū, but seemingly still written‑ουι in Gaulish, Rhys CIG. 5;ōu >ūu̯;§§ 75,76.

§ 61. i. (1) Aryan, (Lat.ul, or; Gk.αλ, λα, αρ, ρα; Germ.ul,ur; Skr.r̥, r̥) probably remained in Pr. Kelt., but developed in all the groups asli,ri. Thus Ar. *ml̥k̑‑t- (√melg̑- ‘milk’) > Lat.mulctus: Ir.mlicht, blicht, W.blith ‘milch’ < *mlikt- < Pr. Kelt. *ml̥kt- (W.ar-mel ‘the second milk’,mel-foch ‘suckling pigs’ < F-grade *melg̑‑).—Ar. *k̑l̥‑t- (√k̑el- ‘hide’) > Lat.oc-cult-us: Ir.clethi ‘celandum’, W.clyd ‘sheltered’ < Pr. Kelt. *kl̥t‑.—Ar. *pr̥t- (√per-) > Lat.portus, O.H.G.furt: Gaul.‑ritum, O. W.rit, Mn. W.rhyd ‘ford’ < Pr. Kelt. *(p)r̥t‑.—Ar. *qr̥m-is ‘worm’ > Skr.kŕ̥mi‑ḥ, Lith.kirmis: Ir.cruim, W.pryf 'worm' < Pr. Kelt. *qr̥mis.—Ar. *dr̥k̑- (√derk̑‑) > Gk.ἔδρακον ‘I saw’, Skr.dŕ̥s̑- ‘look’: Ir.drech ‘aspect’, W.drych ‘appearance’,e-drychaf ‘I look’ < Pr. Kelt. *dṛk‑.

Ir.cru comes fromqri beforei, e oru, as shown bycruimther ‘priest’ which appears in ogam asqrimitir < Early W.primter, Thurneysen Gr. 135; therefore this proves nothing as to Kelt.. But Kelt. *r̥k gave Brit. *r̥kk > W.rych as indrych above,rhych < *pr̥k-§ 101 iii (1), Zupitza KZ. xxxv 256, while Kelt.rik gives W.ryg as in cryg§ 101 ii (2).

(2) Before vowels and and, Ar. preserved an older form of these sounds, which we may writeₑl,ₑr, where represents an indistinct or murmured vowel. These give Kelt,ar,al, see§ 63 iii.

ii. Ar.l̥̄,r̥̄ (Lat.,; Skr.īr, ūr for both) appear in Pr. Kelt. aslā, rā. Thus Ar. *pl̥̄-no- ‘full’ (√pelē‑) > Skr.pūrṇá‑ḥ: Ir.lān, W.llawn ‘full’ < Pr. Kelt. *(p)lānos.—Ar. *ml̥̄‑t- (√melā- ‘grind’) > W.blawd ‘flour’ < Pr. Kelt. *mlāt‑.—Ar. *g̑r̥̄n- (√gerāˣ- ‘rub, grind’) > Lat.grānum, Skr.jīrṇá‑ḥ ‘worn out’: Ir.grān, W.grawn ‘grain’ < Pr. Kelt. *grān‑. See§ 63 vii (2).

62. i. (1) Ar., (Lat.em,en; Gk.α; Germ.um, un; Skr.a) remained in Pr. Kelt., and appear asam,an in Brit. and Gaul., and *em, *en in Ir. (becomingē beforec, t, andim, in beforeb, d, g). Thus Ar.k̑m̥tóm ‘hundred’ > Lat.centum, Gk.ἑ-κατόν, Goth. hund, Lith.szim̃tas, Skr.s̑atá‑m: Ir.cēt, W.cant.—Ar. *dn̥t- ‘tooth’ > Lat.dent‑, Goth.tunþus, Skr.dat‑: Ir.dēt, W.dant.—Ar. *n̥- negative prefix > Lat.in‑, Gk.ὰ‑, Germ,un‑: Ir.in-gnath ‘unwonted’,ē-trōcar ‘unmerciful’, W.an-§ 156 i (5).

(2) Before vowels and and, the forms wereₑm,ₑn, see§ 61 i (2); these gaveam,an in Kelt., and appear so in Ir. and W.; thus W.adanedd ‘wings’ < *pₑtₑníi̯ās; O. W.‑ham, W. ‑(h)af spv. suffix < *‑isₑmos. But whenₑn followed the accent it seems to have becomeann in Kelt. (throughn̥n?); thus Ir.anmann ‘names’ < *án’mₑna < *ánəmₑnə§ 121 iv,§ 63 v (2); Ir.Ērenn ‘of Ireland’ < *ēriann < *īu̯erii̯ₑn-os beside W.Iwerẟon ‘Ireland’ < *īu̯éri̯on‑;—Brit.Britann- < *qrítₑn-§ 3 iii; with the same suffix W.pell-enn-ig ‘stranger’;—W.griddfan ‘groan’ pl.griddfannau§ 203 ii (4);—W.Gofannon, Gaul.Gobannicnos, Ir.goba ‘smith’, gen.gobann; etc.—Final‑ann either remains as‑an, or is reduced to‑a§ 110 v (2), or tended to become‑ant (through‑and?)§ 121 iv,203 ii (4).

This development is precisely parallel to that of the R-grade ofei̯ after the accent in Brit., which gaveai̯i̯ > W.‑oeẟ, the second becoming. Similarlyou̯ after the accent gives W.‑eü̯, prob. from´‑ou̯u̯-§ 76 iii (2).

ii. Ar.m̥̄, n̥̄ were doubtlessmā, nā in Pr. Kelt. Thus Ar. *sn̥̄- (R² of √senē‑, see§ 63 vii (2)), > Ir.snā-that, W.no-dwydd ‘needle’.—Ir.gnāth, W.gnawd ‘known, accustomed’ might be from *g̑n̥̄- like Lat.gnā-rus, but is more probably from *g̑nō- like Lat.nōtus, √g̑enē‑. The Gaul,‑gnatus ‘born’ is assumed to have ā, in which case it may be from *g̑n̥̄‑; but it may haveă fromə, like W.ynad ‘judge’, Early Ml. W. pl.hyg̃neid B.B. 10, 84 < *hyn-ᵹnat < *seno-gnat- ‘elder’ <*g̑nə‑t‑, √g̑enē- ‘give birth’.

Aryan vowel gradation

§ 63. i. In Parent Aryan, while the consonants of any morphological element were comparatively stable, its vocalism varied according to circumstances; this variation is called “vowel gradation” or “ablaut”. The system is similar to, but less highly developed than, that of the Semitic languages, in which the only fixed elements of a word are its consonantal skeleton. In Aryan what may be regarded as the standard vowel wase; this is the full grade, and may be denoted by F. It interchanged witho; this grade may be denoted by F°. In either case the vowel might be lengthened, becomingē orō; the lengthened grades may be denoted by L and L°. The vowel might become more or less indistinct; in this case we write it below the line thus; this is the reduced grade, R. Lastly it might vanish altogether; this is the vanishing grade, V. The same syllable in different combinations may occur in any or all of these grades.

ii. Taking the root *sed- ‘sit’ as an example, the system is as follows (forz in V-grade see§ 97) :

VRFL
zdsₑdsedsodsēdsōd

Examples: V *‑zd‑: W.nyth, Lat.nīdus, E.nest, etc. < Ar. *ni-zd-os§ 97 ii, W.syth < *si‑zd‑, ibid.—R *sₑd‑: W.hadl < *sₑd-lo-§ 111 vii (1).—F *sed‑: W.gorsedd ‘high seat’ < Kelt. *u̯er-en-sed‑;eistedd ‘to sit’ met. for *eitsedd < *ati-en-sed‑; Gaul.esseda ‘war-chariot’ < *en-sed‑; W.annedd ‘dwelling’ forann-hedd < *n̥do-sed‑,cyntedd ‘porch’ < *kintu-sed‑;heddwch ‘peace’ < *sed‑; Lat.sedeo, etc.—F° *sod‑: W.hudd-ygl, Ir.suide ‘soot’§ 100 v; W.aros ‘to stay’ < *pₑri-sod‑t-§ 187 iii.—L *sēd‑: Lat.sēdēs, whence W.swydd ‘office’.—L° *sōd‑: W.soddi ‘to sink’,sawdd ‘subsidence’ < *sōd‑, O.E.sōt, E.soot.

ed- ‘eat’:—V *d‑: W.dant, Lat.dent‑, etc. < *d‑n̥t- (participial stem) ‘*eater’. F *ed‑: W.ŷs ‘eats’ < *etˢti < *ed-ti, Lat.edo, est.—L *ēd‑: Lat.in-ēdia, Skr.ādyáḥ ‘eatable’.

ret- ‘run’:—F *ret‑: W.rhedaf ‘I run’,gwa-redaf ‘I succour’, Gaul.Vo-reto‑.—F° *rot‑: Ir.roth, W.rhod ‘wheel’, Lat.rota.—L° *rōt‑: W.rhawd ‘troop’, Ml. W.gwarawt ‘he succoured’ < *u̯o-rāt- < *upo-(re)rōte.

In Kelt. becomesa before explosives, as well as beforel, r, m, n, see iii below. Thus W.adar ‘birds’ < *pₑtₑr‑;adanedd ‘wings’ < *pₑtₑníi̯ās; besideedn ‘bird’ < *petn‑, √pet- ‘fly’. In Italic also we seem to havea for it, as in Lat.quattuor < *qₑtu̯ores; in Gk.ι inπίσαρες; Hirt, Abl. 15, Meillet, Intr.² 73.

iii. When the vowel is followed by one of the sonantsl, r, m, n, the scheme is as follows,er being taken as the example:

VRFL
rr̥ ₑrerorērōr

Examples: suffix *‑ter‑:—V *‑tr‑: W.modryb ‘aunt’ <mā́-tr-əq, Lat. gen.mā-tr-is.—R *‑tr̥‑: Skr.mā-tr̥-kā ‘grandmother’.—F *‑ter‑: W.bro-der ‘brothers’, Gk. acc.πα-τέρ-α.—F° *‑tor‑: W.bro-dor-ion ‘brothers, clansmen’, Gk. acc.φρᾱ́-τορ-α.—L *‑tēr‑: Gk.πατήρ.—L°‑tōr‑: Gk.φρᾱ́-τωρ.

bher- ‘bear’: R *bhr̥‑: W.cymryd ‘to take’ < *kom-bhr̥‑t-—F *bher‑: W.cymeraf ‘I take’ < *kom-bher‑; Lat.fero, Gk.φέρω, etc.

k̑el- ‘hide’:—R *k̑l‑: W.clyd ‘sheltered’ < *k̑l̥‑t‑, Lat.occultus§ 61 i (i).—F *k̑el‑: W.celaf ‘I conceal’.—L *k̑ēl‑: Lat.cēl-o.

Before these sonants appears asa in Kelt., givingal, ar, am, an. In other branches thus : Ar.ₑl, ₑr give Gk.αλ, αρ, Lat.al, ar, Germ.ul, ur, Skr.ir ur (for both), Lith.il ul, ir ur; Ar.ₑm, ₑn give Gk.αμ, αν, Lat.am, an orem, en (venio§ 100 i (4),tenuis below), Germ.um, un, Skr.am, an, Lith.im um, in un.

The V-grade occurs only before vowels. The formr̥, n̥, etc. of the R-grade occurs only before consonants; the formₑr, ₑn, etc. before vowels, and before and. Where in the derived languages the latter appears before other consonants, a vowel following it has been elided since the Ar. period. I use ’ to mark this elision.

Examples: V-grade ofel in W.glas ‘green’ seevii (3); ofer inrhannvii (2); ofen inglinvii (4).

R-grade before consonants,l̥, r̥, m̥, n̥, see examples in§§ 61,62.

R-grade before vowels: W.malaf ‘I grind’ < *mₑl‑, √melāˣ- ‘grind’;—araith ‘speech’, Ir.airecht < *ₑreq‑t‑, √ereq- ‘speak’: O. Bulg.reką ‘I speak’ (with V-grade of 1st syll.);—archaf ‘I ask’, Ir.arco < Kelt. *ar’k- < *pₑrₑk̑‑, √perek̑‑: Lat.precor (with V-grade of 1st syll.);—carr ‘car’, Ir.carr, Gaul. (‑Lat.)carr(‑us) < Pr. Kelt. *kar’sos: Lat.currus < *qr̥s-os;—darn ‘fragment’ < *dₑr’n- < *dₑrə‑n‑: Skr.dīrṇáḥ ‘split, divided’ < *dr̥̄n- < *dₑrə‑n‑, √derā- ‘split’;—sosarn ‘causeway’: Skr.stīrṇáḥ ‘strewn’, √sterō‑;—carn ‘hoof’, Galat. κάρνον ‘trumpet’: √k̑erāˣ()‑;—teneu ‘thin’, Corn.tanow, Ir.tana: Gk.τανυ‑, Lat.tenuis, Skr.tanú‑ḥ, all < Ar. *tₑnu‑;hafal ‘like, equal’, Ir.samail ‘likeness’ < *sₑmₑl‑: Lat.similis;—ganed ‘was born’ < *g̑ₑn‑, *√g̑enē‑.

R-grade before: W.carw̯ deer ' < *k̑ₑru̯-os: Lat.cervus < *k̑eru̯-os;—marw̯ ‘dead’: Lat.mortuus§ 204 ii (5);—before: W.myned§ 100 iv.

The formsl̥, r̥, m̥, n̥ are generally classed as V-grade; but the vowel of the syllable cannot be said to have vanished when it has converted the consonantr into the vowel. In fact is the form thatₑr takes before a consonant, and must therefore be the same grade.

iv. The treatment of the diphthongsei, eu (properlyei̯, eu̯) is parallel, and corresponding tol, r, m, n, and vocalici, u to vocalicl̥, r̥, m̥, n̥. Thus:

VRFL
i, (ₑi̯ >)ii̯ei̯oi̯ei̯oi̯
u, (ₑu̯ > )uu̯eu̯ou̯eu̯ou̯

The R-grade formsi, u occur before consonants only; the formsₑi̯, ₑu̯, which becameii̯, uu̯, occur before vowels.

Examples: V-grade: W.berw̯i ‘to boil’, Lat.ferveo < *bheru̯‑, √bhereu̯‑;—W.duw ‘god’ < *dw͡yw̯, Lat.deus both < *dei̯u̯-os, √dei̯eu̯‑,vii (4).

u̯ei̯d- ‘see, know’:—R:gwedd ‘aspect’ < *u̯id-ā;gw̯ŷs ‘summons’,gw̯ŷs ‘it is known’, both < *u̯itˢt- < *u̯id‑t‑; Lat.vid-eo;—F:gŵydd ‘presence’ < *u̯ei̯d‑,arw͡ydd ‘sign’ < *pₑri-u̯eid‑; Gk.εἴδομαι;—F°: Gk.οἶδα < *u̯oi̯d-a.

k̑leu̯- ‘hear’:—R:clod ‘praise’ < *k̑lu-tó‑m§ 66 v; Gk.κέ-κλυ-θι;—F:clust ‘ear’ < *k̑leu‑t‑st-§ 96 ii (3).

deu̯k- ‘lead’:—R:dyg-af ‘I bring’,dwg ‘brings’ < *duk‑;dwyn ‘to bring’ < *duk‑n‑;—F: Lat.dūco, O. Lat.douc-o, Goth.tiuh-an < *deuk‑;—L°:dug 'brought' < *(du)-dōuk‑,§ 182 ii (2).

The V-grade disappears between consonants; see √qonei̯d-vii (4)geneu̯- ib.; seeviii (2) and§ 100 ii (2).

v. (1) As seen above, Ar. had the vowele interchanging witho; the vowelsi andu are secondary, being vocalized forms of and.

(2)a occurred in Ar. only in special cases, which Meillet, Intr.² 139 gives thus: 1. in child-language, as Skr.tata, Gk.τάτα, Lat.tata, W.tada; 2. in certain isolated words, possibly borrowed, as Lat.faba; 3. in a few endings, as 3rd sg. mid. *‑tai, Gk.‑ται, Skr.‑te; and 4. initially, interchanging with zero, as Gk.ἀστήρ: Lat.stella, W.seren, E.star.

As shown by Meillet (ib. 140) initiala- may coexist with the F- or L-grade of the following syll., as in Gk.ἀ(ϝ)έξω with F *u̯eg- beside αὔξω, Lat.augeo with V *u̯g‑; cf. ἀστήρ. This seems to imply thata- might be a movable preformative, but it does not prove that it was outside the ablaut system; in fact, the common gradationā : ə necessarily implies the ablaut ofa, asē : ə does that ofe; seevi.

Many indications point toa being an Ar. survival of a pre-Aryan sole vowela, which ordinarily split up in Ar. intoe ando. It is preserved in child-language because this is conservative; thus while Ar. *tata gives W. tad ‘father’, in W. child-speech it remains astáda. In the ordinary languagea stands side by side withe/o, or occurs where we should expecte/o, in the following cases: 1. initially; 2. before *ə̯ or; 3. before gutturals. Thus 1.at‑,ati‑:et‑,eti- pref. and adv. ‘beyond, and, but’§ 222 i (3); O. W.anu, Ir.ainm ‘name’ < *án(ə)mn̥, Armen.anun ‘name’: Gk.ὄνομα < *ónəmn̥, √onō‑/​anō‑.—2. The ending of the neut. pl. nom.-acc. is *‑ə; now the neut. pl. ofo/e-stems is‑ā from *‑aə, where *‑a- represents the stem vowel instead of‑o- (or‑e‑); similarly the fem. ofo/e-stems is formed with‑ā- for *‑aə‑; buti̯o/​i̯e-stems have beside‑iā- < *‑iaə- the fem. form‑i̯ē- < *‑i̯eə‑. Cf. alsoā : ōix below. In the dat. sg. of cons. stems both‑ai and‑ei occur, as Gk. infin. suff.‑μεναι: Osc.diúveí, patereí, Solmsen KZ. xliv 161 ff.

In the positions indicated,a has R- and L-grades. Thus, 1. Initially: F *am- in Gk.ἀμφί, Lat.ambi‑: R *m̥- in Ir.imb, imm, W.am, ym‑, Skr.abhí-taḥ (a- < *m̥‑) ‘on both sides’; F *ar- in W.arth, Gk.ἄρκτος: R *r̥- in Lat.ursus, Skr.ŕ̥kṣah§ 98 i (2); F *ag̑- in Lat.ago, Gk.ἄγω: L *āg- in Lat.amb-āges.—2. Beforeə̯ or: F *ā (< *): R *ə, see vi; F *ai- in Gk.αἴθω, Ir.aed ‘fire’, W.aelwyd: R *i- in Skr.idh-má‑s ‘firewood’. For the fem. ofi̯o/​i̯e-stems there is beside‑i̯ā- and‑i̯ē- a form‑ī‑; this may be explained thus: RF *ii̯aə, *ii̯eə giveii̯ā, ii̯ē: RR *ii̯ə >ī,vii (2). Cf.vii (5).

3. Before gutturals : √ak̑‑:oq- ‘sharp, rugged’, as Gk.ὄκρις, ὀξύς, Lat.ocris, W.ochr: Gk.ἄκρος, Lat.acus, W. (h)agr ‘ugly’;—√dek̑/g̑h- ‘to seem good, acceptable; to apprehend, teach’;e in Lat.decus, decet, Ir.dech, deg, ‘best’:o in Lat.doceo, Gk.δοκέω, δόγμα:a in Gk.διδάσκω (< *διδαδσκω), διδάχη, W. da ‘good’ < *dag‑, Gaul.Dago‑, Ir.dag- ‘good’.

vi. (1) The long vowelsē, ō, ā had R- and V-grades;ē had also the F°-gradeō. The R-grade of each isə. Before a vowelə regularly disappears, giving the V-grade, as in Skr.dá-d-ati ‘they give’, where‑d- is the V-grade of √dō‑. It also occurs before consonants, as in Skr.da-d-máḥ ‘we give’ beside Gk.δί-δο-μεν; but the disappearance ofə between consonants is believed to be due to analogy or elision after the Ar. period. It is however lost in syllables not initial or final in Germ., Balt.-Slav., Armenian, Iranian; Meillet, Dial. 63.

ə appears to come from a guttural spirant resembling ᵹ (§ 110 ii (2)), which played the same part as the sonants, so that the ablaut series ofē is parallel to that ofei̯ orer, the F-gradeē being for *eə̯; thus V (ə non-syllabic, lost) ; Rə (syllabic); Fē for *eə̯; F°ō for *oə̯; corresponding to V (non-syllabic); Ri (syllabic); Fei̯, F°oi̯. This explains whyə is the R-grade of all the long vowels.

In cases where the F-grade has not survived, or has survived only in Indo-Iranian, whereā̆, ē̆, ō̆ all appear asā̆, so that the quality of the vowel is unknown, it is usual to write itāˣ.

Examples: √dō- ‘give’: F:dawn ‘gift’ < *dō‑n‑; Lat.dō-num; Gk.δί-δω-μι.—V:rho-ẟ-ant ‘they give’ < *pro-d-n̥ti; Skr.dá-d-ati < *dé-d-n̥ti. √dhē- ‘put’:—F: Gk.τί-θη-μι;—V:rho-ẟ-ant ‘they put’. See§ 179 ii.

sthā- ‘stand’: R:gwa-sta‑d ‘level’ < *upo-sthə‑t‑; Lat.stā-tus; W.sa‑f ‘stand’ < *sthə‑m-§ 203 vii (5);—F:saw‑dl ‘heel’ < *sthā‑tl‑.

sē- ‘sow’:—R:had ‘seed’ < Ar. *sə‑t‑; Lat.să-tus.—F:hīl ‘progeny’, Ir.sīl < *sē‑l‑; Lat.sē-vi,sē-men.

(2)ə generally appears asa in the European languages, as in the above examples. (Cf.§ 110 ii (2).) But in Gk. if the F-grade isē orō, the R-grade often appears as ε or ο. Elsewheree besidea is probably to be explained as due to a variant of the root, with short vowel; thus W.tref, O. W.treb ‘homestead’ < *treb‑; Lat.trabs < *trəb‑; Gk.τέρεμνον, τέραμνον both < *terəb‑; √terē̆b‑.

vii. (1) As a rule the same morphological element could not contain two F-grade syllables, though, of course, a word, made up of more than one element, might. The diversity in different languages of words of the same origin is largely due to the preservation of various groupings of grades; see for example *qetu̯er- in (4) below.

(2) A large number of roots were disyllabic. A characteristic form of Ar. root had a short vowel in the first syllable and a long in the second. A very common form of reduction was RR, i.e. R-grade of both syllables. When the consonant between the vowels was one of the sonantsi̯, u̯, RR wasₑi̯ə, ₑu̯ə, which gaveii̯ə, uu̯ə; these were generally contracted toī, ū respectively; we may call this contraction R². On the analogy of these it is assumed that RRₑlə, ₑrə, ₑmə, ₑnə gave respectively R²l̥̄, r̥̄, m̥̄, n̥̄,§ 61 ii,§ 62 ii. The uncontracted RR forms also survived, as in Gk.παλάμη; < *pₑlə-mā, √pelā‑, beside W.llaw(f), Ir.lām < Kelt. *lā-mā < *pl̥̄-mā;—W.taradr ‘auger’, Ir.tarathar < ER *tₑrə‑tr‑, beside Gk.τέρετρον < FR *terə‑tr‑, √terē- ‘bore’;—W.rhaeadr ‘cataract’ < RR *rii̯ə‑tr‑, beside Lat.rīvus < R² *rī‑, √rei̯ā- ‘flow’. In many cases theə dropped, seevi (1), as in Lat.palma < *pₑl(ə)-mā; we may denote this by R(R). Beside these we also have VR formslə, rə, mə, nə; thus beside W.gwaladr ‘ruler’ < RR *u̯ₑlə‑tr‑, we have W.gw̯lad ‘country’, Ir.flaith ‘lordship’ < VR *u̯lə‑t‑, √u̯elē(i)- (: Lat.valēre, E.wield);—W.gw̯lân ‘wool’ < VR *u̯lən-ā, beside Lat.lāna, Skr.ū́rṇā < R² *u̯l̥̄n-ā;—W.rhann ‘share’, Ir.rann id. < VR *prə‑t-snā, beside Lat.part- < R(R) *pₑr(ə)‑t‑, beside Skr.pūr-t-ám ‘reward’ < R² *pr̥̄‑t‑, √perō‑;—W.ystrad ‘dale’, Gk.στρατός < VR *strə‑t‑, beside W.sarn ‘causeway’ < R(R) *stₑr(ə)n‑, beside Skr.stīr-ṇá‑ḥ ‘strewn’, Lat.strā-tus < R² *str̥̄‑, √sterō- ‘spread out’.—When the long vowel afterl, r, m, orn wasā orō we cannot distinguish in Kelt. between R² and VF, since in Kelt.l̥̄, lā, lō, all give; we can only infer the probable original from a comparison of cognates; thus O.E.flōr ‘floor’ < VF *plā‑r- suggests that Ir.lār, W.llawr ‘floor’ contain VF *plā‑r‑, √pelā‑; and Gk.παλάμη < RR as above suggests W.llaw < R². But where the vowel wasē as in √pelē- ‘fill’, we know e.g. that W.llawn, Ir.lān ‘full’ come from R² *pl̥̄‑n‑, since VF *plē‑n- as in Lat.plēnus would give W. *llīn, which does not exist, and does give Ir.līn‑, which is seen inlīnaim ‘I fill’.

R(R) is postulated instead of RV because the loss ofə is late; this agrees with the fact that we havear in W., implying *ₑr the form before a vowel, the loss of which is therefore secondary, and notry from * the form before a consonant. Similarlyi may be taken as R(R) ofeiē; thus RRii̯ə > R(R)ii̯(ə) >i. Wherery occurs in W. beside forms implying an original long vowel we may assume that the former comes from a variant with short vowel of the root; thus W.gw̯rysg ‘boughs, twigs’ < *u̯r̥d‑sq‑, RV of √u̯erod‑; Lat.rādīx < *u̯r̥̄d‑, R² of √u̯erōd‑, O.E.wrōt < *u̯rōd‑, VF of √u̯erōd‑.

(3) A few examples are appended:

g̑helē- ‘green, yellow’: VR *g̑hlə- > Kelt. *gla‑st- > Brit.‑glasos ‘tawny’ (Gildas), W.glas ‘green’; FV *g̑hel- > Lat.hel-us.

gelāk/g- ‘milk’: RR *gₑlək- > Gk.γάλα, γάλακτος;—VR *glək- > Lat.lact- (whence W.llaeth); *gləkt‑s > Ir.glass ‘milk’, W.glasdwr ‘milk and water’.

qeu̯ēp- ‘blow’: VF *qu̯ēp- > Lith.kvė̃pti ‘blow’;—RR *qₑu̯əp- > W.cawad ‘shower’, Ir.cūa, gen.cūad;—VR *qu̯əp- > Lat.vapor, Gk.καπνός.

ɡei̯ē- ‘live’: R² *ɡī- > Lat.vī-vu‑s, W.buan ‘quick’ < Brit. *bī-u̯o-no‑s§ 76 ix (2);—R(R) *ɡi- > W.by‑w ‘live’,by‑d ‘world’, Gk.βίος;—VF *ɡi̯ē- > Gk.ζῆν.

bheu̯ā- ‘be’: R(R) *bhu- > Lat.fu-turus, Gk.φύ-σις, Kelt. *bu-tā > W.bod ‘to be’;—L°V *bhōu̯- > W.bu§ 189 iv (3);—VV *bh()- >f- in Lat.fīo,b- in W.byẟ§ 189 iv (1).

(4) When the second syllable has a short vowel, the treatment is similar: RRii̯ₑ > R²ī, etc., as before; RV isĭ. Examples:

dei̯eu̯- ‘god, day’: FV *dei̯u̯-os > Lat.deus, W. *dwyw >duw ‘god’;—R² *dīu̯- > Lat.dīv-us;—RV *dĭu- > W.dyw ‘day’;—RL *dii̯ēu̯- > Lat.diēs, W.dydd ‘day’.

qonei̯d- ‘nit’: FR *qonid- > Gk.κονίς gen. κονίδος ‘nit’; VR *qnid- > O.E.hnitu, E.nit, O.H.G.hniz ‘nit’; *s(q)nid-ā > W.nedd ‘nits’, Ir.sned ‘nit’;—FV *qond- > Lith.kandìs ‘moth’; *sqond- > W.chwann-en ‘flea’.

g̑eneu̯- 'knee': FR *g̑enu > Lat.genu;—F°R *g̑onu > Gk.γόνυ;—with‑en‑, ‑er- forming names of parts of body: base *g̑eneu̯-en‑: VR² *g̑nūn‑, by dissim. > Kelt. *glūn- > Ir.glūn, W.glīn ‘knee’;—base *g̑eneu̯-er‑: RVV *g̑ₑn()r- > *ganr- > W.garr ‘knee’ (afal garr ‘knee cap’).

qorou̯‑: FR *qoru- > Gk.κορυ-φή;—VF *qrou̯- > W.crug ‘heap, hillock’.

bhereu̯ ‘boil’: FV *bheru̯‑, seeiv above;—VR *bhru- > W.brwd ‘hot, fervent’, Lat.dē-frŭ-tum ‘new wine boiled down’.

*qetu̯er- ‘four’: RF° *qₑtu̯or- > Lat.quattuor;—FR *qetuₑr- > W.pedwar, Ir.cethir, Gk. τέτταρες;—RL° *qₑtu̯or- > Skr.catvā́raḥ, Goth.fidwor; FR (before cons.) *qetu̯r̥- > Gk. τετρα‑, becoming byviii (1) *qetru- it gives Gaul.Petru‑, W.pedry- as inpedry-fan.

(5) Long diphthongs must represent radical disyllables, and their reduced grades can only be explained from the disyllabic forms. Thusēi̯ must be FV of *eə̯ei̯ or *eə̯ai̯ (ei/aiv (2)); the R of the first syll. isə which vanishes before a vowel, leavingei orai (properly VF of *eə̯ei̯ or *eə̯ai̯); if the second is reduced we getii̯, before a cons.i (properly VR of *eə̯ei̯ or *eə̯ai̯). We know thatēi̯ interchanges in roots withei̯ē orei̯ā; this implies a metathesis of the sonants, for the latter forms represent *ei̯eə̯ or *ei̯aə̯; the RR of these is *ii̯ə which givesī, seevii (2). Thus we have as reduced grades ofēi̯ the formsei orai,i(),ī; for convenience these may be distinguished thus: R₁ₑei, R₁ₐai, R₂ii̯, i, R₃ī. The same principle applies to the long-diphthongs. [It has been assumed thatai isəi̯ (withə as R ofē), but Skr. hasay for it, whereasə isi in Skr. Besides, we should expectəi̯ likeₑi̯ to give *ii̯, as perhaps it does, forii̯ may also be for *əi̯ RV of *eə̯ei̯.] Examples :

sēi̯- ‘late, long’: F *sēi̯- > Skr.sāyám ‘evening’, before cons. *sē- > Lat.sērus, W.hir ‘long’, Ir.sīr;—R₁ₑ *sei- > W.hŵyr ‘late’ (< *sei-ros),hŵy ‘longer’, Ir.sīa (< *seison < *sei-isōn);—R₂ *si- > W.hyd ‘length’ (< *sit‑);—R₁ₐ *sai- > W.hoedl ‘lifetime’, Lat.saeculum ‘age’, both < *sai-tlo‑m§ 75 i.

u̯erēi̯- ‘laugh (at), shame’: VR₂ *u̯ri- > *u̯ri‑zd- whence Lat.rīdeo, Skr.vrīḍ-ā ‘shame’;—(VF *u̯rē- or else) VR₃ *u̯rī- > Kelt. *u̯rī‑t- > W.gw̯rīd ‘blush’;—RR₂ *u̯ₑri- > *u̯aritā in W.dan-wared ‘to mimic’;—RR₂ *u̯ₑrii̯- > W.gw̯arae ‘play’§ 75 v (4).—From √u̯erē- (without): RR *u̯ₑrə- > *u̯arat- > W.gw̯arad-w͡yẟ ‘shame’ (by dissim. for *gwarad-rw͡yẟ);—R(R) *u̯ₑr’- > *u̯ar‑t- > W.gwarth ‘shame’; *s‑u̯ar‑d- > W.chwarẟ ‘laughs’; *s‑u̯ar-tīn-ī > W.chwerthin ‘laughter’§ 203 vii (3).

viii. (1) Certain combinations produced by the above laws are unstable; thusu̯r̥ is liable to becomeru, as in *qetru-vii (4); andu̯ₑr may becomeur as in *dhur- for *dhu̯ₑr‑: *dhu̯or‑,§ 91 i. Whileu̯ₑ, lₑ, rₑ, etc., may remain and giveu̯a, la, ra, etc., in Kelt., they may be, and oftenest are, reduced tou, l̥, r̥, etc. Hence we are not obliged to postulateeue, ele, ere, etc., where there is no evidence of the firste in surviving forms. Thus:

su̯ep- ‘sleep’: F *su̯ep-no- > Lat.somnus (< *su̯epnos), Skr.svápnaḥ ‘sleep, dream’;—R *sup-no- > Gk. ὔπνος, W.hun ‘sleep’, Ir.sūan.

plethē‑: RR *pl̥thə- > Gk. πλάτα-νος, Gaul. ‑λιτα-νο‑ς, O.W.lita‑n, W.llyda‑n ‘broad’; FV *pleth- > Skr.práth-aḥ ‘breadth’, W.lled ‘breadth’;—RV *pl̯th- > W.llys ‘court’§ 96 ii (5), Gk. πλατ-ύς;—RV *plₑth- > Armen.layn ‘broad’;—(withoutl,§ 101 ii (2)) RF *pₑthē- > Lat.patē-re, etc.

(2) Other combinations are unpronounceable or difficult; thusu̯i̯ cannot be sounded before a cons.; in that case drops. Generallyi̯, u̯ drop between consonants, seeiv above.

ix. Some roots have more than one ungraded form; thus radicalā may stand beside radicalō, as in *arā- or *arō- ‘scratch, bite; plough, dig’: Lat.arā-re has F *arā- of the first, Gk. ἄροτρον has R *arə- of the second. The F of both, with‑d- extension, occurs in Lat.rād-o, rōd-o. Besideā we have sometimes to assumea, as in Skr.rádati ‘scratches, digs’ (notə here, which givesi in Skr.). In many cases all the forms cannot be explained without assuming an alternation of long and short vowel in the root; this may have come about by false analogy. Another common form of root alternation is *tēu̯‑: *teu̯ā- or *g̑hēi̯‑: *g̑hei̯ā- (Lat.hiā-re); seevii (5).

Note.—Ablaut is not to be confused with the changes due to accentuation or other causes in the derived languages, such as the shortening of unacc.ā in Brit.§ 74, or the loss of a vowel in such a word ascawr§ 76 iii (4), which would be *cur if the loss were primitive§ 76 ii (1).

Keltic vowels
in British and Welsh

§ 64. From what has been said in§§ 5762 we arrive at the following vowel system for Pr. Kelt.:

Short vowelsa e i o u
Long vowelsā ī ū
Short diphthongsai ei oi au eu ou
Long diphthongsāi ūi āu, īu
Short vocalicl̥ m̥ n̥ r̥

The Short Vowels.

§ 65. i. The short vowelsa,e,o remain unchanged in W.; see examples in§ 58; so Latina,e,o; unless affected by other vowels§§ 6770. The exceptions are the following:

ii. (1) Before a gutturalo in many cases becamea, apparently when unaccented in Brit.; thus W.Cymro < *kom-brógos, butCymraes ‘Welsh-woman’ < *kom-brogíssā: *brog‑, W.bro ‘border, region’ < *mrog‑, VF of √marog‑, whose FV gave Lat.marg-o;—W.troed ‘foot’ < acc. *tróget-m̞, pl.traed < acc. pl. *troget-áss (< *‑ń̥s: Skr.‑aḥ), or from gen. pl. *troget-ón (< *‑ṓm which was generally accented in Ar.) as ingwŷr traed ‘infantry’; √t/dhregh‑: Gk. τρέχω, τρόχος; the √ had alsoa by Ar.a/e/o altern.§ 63 v (2), as in Ir.traig ‘foot’ < *tragets, but we can hardly suppose Ar.o/a in the same word in Brit.—Similarly in Lat. loanwords, as W.achos ‘cause’ <occā́sio.—W.achub <*occū́p- for Lat.occup-§ 73 ii (4).

(2) On the other handa >o in Pr. Kelt. before Ar.ɡh in W.oen, Ir.ūan ‘lamb’ < *ognos < *hnos: O.E.ēanian ‘yean’§ 101 iii (1).

(3) In Brit.e becamei beforeg followed by a vowel; so partly in Gaul; as W.ty ‘house’ O. W.tig < *tigos < *tegos, Brit.Cato-tigirni, also spelt (in Cornwall)Tegerno-mali besideTigerinomalum Rhys LWPh.² 404, Gaul.Tigernum, Ir.teg ‘house’,tigerne ‘lord’, √(s)theg-§ 92 i.—W.hy ‘bold’ < *segos: Gaul. Σεγο-μαρος, √seg̑h‑: Gk. ἔχω < *seg̑hō, Skr.sáhaḥ ‘might’.—W.gwe-ly ‘bed’ < *u̯o-leg‑: Ir.lige < *legii̯o-legh‑.—Wheree appears it is due toa-affection; as inbre ‘hill’ < *brigā§ 103 ii (1); thuslle ‘place’ < acc. *ligan < *leg-m̥, √legh‑;gre ‘herd’ < *greg-m̥ = Lat.gregem;—godre ‘bottom (edge of garment), foot (of hill)’ < *u̯o-treg-m̥, √treɡh‑, see(1), pl.godryon, godreon, both inr.m. 151.

But before a consonanteg remained: W.gwair m. ‘hay’ < *u̯egr‑: Ir.fēr;—W.tail ‘manure’ < *tegl-§ 104 ii (1);—W.arwain ‘to lead’ < *ari-u̯eg‑n-u̯eg̑h‑: Lat.veho;olr͑ein, etc.§ 203 iv (1);—W.tew ‘thick’§ 76 viii.

iii. (1) The mid vowels e and o were pronounced close in Brit. before nasal + explosive and becamei andu respectively. Examples:

e before nas. + exp. > W.y; thus W.hynt ‘way’ Ir.sēt < *sent‑: O.H.G.sind ‘way’ < *sent‑.—O.W.pimp, Ml. W.pymp ‘five’, Gaul. πεμπε- < Pr. Kelt. *qeŋqe < Ar. *penqe.—W.cy-chwynnuỺ.A. 133 ‘to rise’, later ‘to start’, Ir.scendim < Ar. *sqend-§ 96 iii (2).—They becomese by a-affection, asGwent <Venta; cf. E.Wintchester 'Venta Belgarum'. In Lat. loanwords we havey, astymp <tempus ;tymor <tempora ;cymynn(af) <commend-o;esgynn(af) < ascend-o, etc.; but most nouns have‑enn, Mn. W.‑en, aselfen <elementum;ffurfafen <firmāmentum;ysgrifen <scribenda, all fem., having been treated like native nouns in‑enn§ 143 i;mynwent fem. ‘graveyard’ alone has‑ent < pl.monumenta. (Calan is from Vulg. Lat.Kaland‑, which occurs.)

o before nas. + exp. > W.w; thustrwnc < *tronq-§ 99 v (3);—twng ‘swears’: Ir.tongim ‘I swear’;—hwnt ‘yonder’: Bret.hoñt§ 220 ii (5).—The change took place in Lat. loanwords, aspwnn ‘burden’ <pondus;ysbwng <spongus; except in fem. forms, asllong ‘ship’ <longa (nāvis). W.pont ‘bridge’ < Brit. acc. *pontan (<‑m̥) put for Lat.pontem, became fem. The 3rd pl. subjunct.‑ont instead of *‑wnt is prob. due to the analogy of the other persons, which have‑o‑.

(2) The same change took place before a liquid and explosive, though here with less regularity.

e + liq. + exp. > W.y; thus Ml. W.kymyrth < *kombert-et, witha-affectionkymerth§ 181 vii (1);—gwyllt ‘wild’ < Brit. *gu̯eltis: Ir.geilt§ 92 iv.—But usually it remains ase; thus fornyrthb.b. 68, the ordinary form isnerth m. ‘strength’; soperthyn < Lat.pertin- owing to preference for the sequencee . . y.—merch ‘maid’,perth ‘bush’ are fem.; andmellt pl. ‘lightning’,gwellt pl. ‘grass’ may be neut. pl. in *‑ā or fem. pl. in *‑ās.

o + liq. + exp. > W.w ; thus W.i̯wrch ‘roebuck’, Bret.iourc’h; O. Corn,yorch: Gk. ζόρξ;—W.twrch ‘boar’, Bret.tourc’h: O. Corn.torch, Ir.torc;—W.swllt ‘money, shilling’ < Lat.sol’dus.—torch ‘torque’ is fem.: Ir.torc. But other exceptions occur ascorff ‘body’ < Lat.corpus;porth m. ‘gate’ f. ‘harbour’ has exchanged genders and keepso in both. Formations likegor-ffen etc. are also exceptions.

(3) The same change took place beforern. Thuse: W.chwyrn ‘whirling’ < *spern‑,§ 96 iv (1);—W.Edyrn besideEdern <Eternus.—W.gwern ‘alder’, andcern§ 95 ii (3) are fem.; souffern ‘hell’ < Vulg. Lat.īferna.—o: W.asgwrn ‘bone’ < *ast-korn-§ 96 ii (4);—W.dwrn ‘fist’: Ir.dorn.—But W.corn ‘horn’ < Lat.

e beforerr > W.y; asbyrr ‘short’: Ir.berr;—W.gyrr ‘a drove’ < *gerks-§ 95 iv (2). Buto remains, as incorr ‘dwarf’,torri ‘to break’.

(4) In many Lat. loanwordse oro beforer + cons. becamea (on the analogy of the R-grade insarn etc.?); thussarff <serpens;carrai ‘lace’ <corrigia;parchellb.b. 55 besideporchella.l. i 276 <porcellus;tafarn <taberna;Padarn <Paternus;Garmon <Germānus.

(5)e beforess >y; as inys (ỿs,ɥ̄́s§ 82 ii (1)) < *esti ‘is’;—ŷs ‘eats’ < *essi < *ed-ti: Lat.est. Also before Lat.st as intyst ‘witness’ <testis. But either affection or the sequencee . . y (ore . . u) causes it to bee, as inffenestr ‘window’,testun ‘text’ <testimōnium.

iv. (1) In the present penultỿ appears fore ando before a nasal whether followed by another consonant or not; as incychwỿnnu,tỿmor§ iii (1);ffỿnnhawn, nowffynnon < Lat.fontāna;tỿner < Lat.tenerum;mỿfɥr < Lat.memoria,mỿned ‘to go’: Bret.monet;mỿnwent besidemonwent < Lat.monumenta. But many exceptions occur, ascenedl ‘nation’,Conwy; and derivatives likegwenu ‘to smile’ (:gwên ‘smile’),tonnau ‘waves’ (:tonn ‘wave’) do not show the change (exc.hỿnaf ‘oldest’ assim. to the cpv.hŷn,§ 148 i (11)).

(2)o >ỿ in the prefixes *ko‑, *kom‑, *kon‑, *to‑, *do‑, *ro‑; as W.cỿwir ‘correct’ < Kelt. *ko-u̯īros;rhỿ-fawr ‘very great’ < *(p)ro-māros; see§ 16 iii; except when the vowel of the root is lost, as in W.cosp ‘punishment’, Ir.cosc < *kon-sq-§ 96 iii (5); W.rhodd ‘gift’ < *(p)ro‑d-§ 63 vi (1).—When separately accentedrhỿ has acquired a new strong formrhɥ̄́, asrhɥ̄́ ddā́ ‘too good’; similarly *dỿ, *ẟỿ, writtendi in O. W. (< *do ‘to’), as a preposition became *ẟɥ > Ml. W. > Mn. W.i ‘to’§ 16 ii (3). Socyn before the equative, now soundedcɥn, and dialectallyk̑ĭn.

v. (1)o anda interchange after§ 34 iv. So we havegwa- besidego- forgwo- < *u̯o‑: Gaul.vo- < Ar. *upo; thusgwa-red-wr ‘saviour’ < *u̯o-reto-u̯ir‑: Gaul.Voretovir‑;—W.gwas ‘servant’: Ir.foss < *upo‑st-§ 96 ii (2). The 15th cent.pedwor§ 34 iv (so Salesbury’s Dic. s. v.) has a new, perhaps local,o fora§ 63 vii (4).

We also find the interchange afterü (cons. or voc.), asbreuan for *breuon§ 76 iv (2);bū́an for *büon§ 76 ix (2); (Anglesey dial.neuoẟ forneuaẟ).

(2) Afterm- there is an older change ofa too, as in W.môr ‘sea’, Gaul.Aremorici, Ir.muir: Lat.mare; W.myned < *monet‑, Bret.monet < *mami̯et-§ 100 iv;—W.morwyn < *marein-§ 125 v (1).

(3)e after becomeso/a in the following cases: Ar. *uper > Pr. Kelt. *u̯er > Gaul.ver‑, Bret.war ‘on’, W.ar,gwar‑,gwor‑,gor-§ 36 iii;—W.Cadwallon < Brit.Catu-vellaunos;—W.gosper < Lat.vesper‑. Probably the above show the influence of Brit.u̯o‑; cf. Ir.for- < *u̯er- on the analogy offo- < *u̯o‑. Generallyu̯e remains, as inchwech ‘six’ < *su̯ek̑s.

vi. (1) After post-tonica becamee; thuswyneb ‘face’ < *éni̯-eq,§ 100 v, < *éni̯-aq- < *éni̯-əq, √ōq = Skr.ánīkam ‘face’ < *eni-əqom. But when pre-tonic thea remained, as inwynab‑, in composition, from *eni̯aq;gwyẟi̯ad < *u̯idi̯ətó§ 180 iv (1).

(2) Pre-tonici̯o prob. becamei̯a; thus we haveaea < *‑ii̯a‑´, but no *aeo < *‑iio‑´, so that the latter perhaps became *‑ii̯a‑´§ 75 vi (2). So the rel.a <*i̯a < Ar.i̯os,§ 162 vi (1).

§ 66. i. Pr. Kelt.i andu remained in Brit. Brit.i was open, and is transcribed ε by the Greeks, as in Πρετ(τ)ανικὴ (νῆσος) : W. (ynys)Prydain, buti by the Romans as inBritannia (Gk. ι was close, Lat.i open). Brit.i gave W.y, which isɥ in the ult. and accented monosyllables,ỿ in non-ultimate syllables and proclitics. Brit.u remains, now writtenw, in the ultima and monosyllables, and becomesy (≡ỿ) in all other syllables. See§ 40 iii. Examples: W.drɥch ‘appearance’,edrỿchaf ‘I look’ < Pr. Kelt. *dr̥k‑,§ 61 i;—W.cŵn ‘dogs’,cỿnos ‘little dogs’ < Pr. Kelt. *kun‑;—W.cỿbɥdd ‘miser’ < Lat.cupidus;—W.terfɥn ‘end’ < Lat.terminus.

ɥ andỿ may interchange withe, andỿ witha,§ 16 iv.

u before a labial may develop irregularly,§ 73 ii.

ii. (1)ỿ in the penult, whether fromi oru becomesw in Mn. W. beforew in the ult., as incwmwl ‘cloud’ forcymwl < *cumbul- < Lat.cumulus;swmbwl < *stimbul- < Lat.stimulus;cwmwdcomot’ < Ml. W.kymwt;dwthwn <dythwn <dydd hun§ 164 iii. When a syllable is added, bothw’s becomeỿ, ascỿmỿlau ‘clouds’.

(2) After the obscureỿ becamew; as (g)wrthẏeuỺ.A. 83 ‘miracles’ forgw̯ỿrthi̯eu. In the spoken lang. and frequently inmss. we havegwnnach forgw̯ỿnnach ‘whiter’,’wllɥs forew̯ỿllɥs ‘will’, etc. Theỿ was artificially restored in most of these forms in the lit. lang.—G.R. 31 states that the rising diphthong always becomesw in the penult, the falling diphthong never, citing as examplesgw̯ynn,gwnnach;gwinw̯ydd,gwinwdden;celw̯ydd,celwddog, butcŵyn,cw͡ynaw;gŵydd,gw͡yddau, etc. J.D.R. writeswỿ ingwỿnnach,gwỿrddach 63, but (g)wrthieu [xvii].

iii. (1) Unaccented initialu̯i- before sonants became *u̯u- > *gw̯w- > *gw‑,§ 36 i. Thusgŵr ‘man’ < *u̯ur-os < *u̯ir-ós;—gwrth- ‘contra‑’,wrth ‘against’ < *u̯urt- < *u̯irt- < *u̯ertó: Ir.frith < *u̯r̥t-§ 211 iv (2);—gwnn ‘I know’ < *u̯indṓ,§ 191 iii (1). Thew thus produced is not mutated toỿ in the penult, e.g.gẃrol ‘manly’,gẃraidd id.,ẃrthyf ‘by me’; andgwnn seems to show that it was not liable to affection; in that casegw̯ŷr ‘men’ is analogical.

(2) Before other consonants initial unaccentedu̯i- oru̯e- became *oi- giving W.ü‑, as in Ml. W.ugeint ‘twenty’ < Kelt. *u̯ikn̥tí: Ir.fiche;—W.ucher ‘evening’ < (*u̯isqer- <) *u̯esper-§ 96 iv (2).

(3) Generally, however, initialu̯i- becamegw̯y- regularly: asgw̯ŷs < *u̯id‑t-§ 63 iv;—gw̯ynt < *u̯int- < *u̯ent- < *u̯ēnt‑: Lat.ventus;—gw̯yw ‘withered’,§ 75 vii (3);—gw̯yrth ‘miracle’ < Lat.virtus. Butgw̯ỿ- later becamegw‑,ii (2) above.

iv. Ar.i in the ultima, or ending the first element of a compound gave Gaul. and Brit.e. Thus Gaul.are‑, W.ar- < *are- < *ari- < *pₑri;—W.am < *m̥be < *m̥bhi: Lat.ambi‑, Gk. ἀμφί;—W.môr < *more, Gaul.more < *mori: Lat.mare. The reason that final unaccented shorti does not affect a preceding vowel is probably that it had becomee.

v. Pretonicu becameo, as ini̯ôn ‘lord’ < *i̯ud-nó‑s,i̯ôr ‘lord’ < *i̯ud-ró‑s: W.uẟ§ 100 i (1); see§ 104 iv (3);bôn m. ‘base, stem’ < *bud-nó-§ 104 iv (1);clod ‘praise, fame’ <k̑lutóm: Ir.cloth (gen.cluith) id. <k̑lutóm, Gk. κλυτόν, Skr.s̑rutám ‘what has been heard, tradition’, √k̑leu- ‘hear’.

Affection of Short Vowels.

§ 67. A short vowel (but no long vowel) was liable to be affected by a sound in a succeeding syllable. Affection is of two kinds in Welsh : 1.ultimate, when it takes place in the syllable which is now the last, having been brought about by a sound in a lost termination; 2.non-ultimate, when it takes place in the present penult or antepenult, the affecting sound being generally preserved in the ultima. Ultimate affection is caused bya ori sounds ; non-ultimate by the latter only.

§ 68.Ultimatea-affection.ĭ andŭ became respectivelye ando in the ultima when the lost ending hada; thusgw̯edd ‘aspect’ < *u̯id-ā§ 63 iv;—bod ‘be’ < Kelt. *bu-tā§ 189 iv (6);ciwed ‘rabble’ < Lat.cīvitās;—gramadeg < Lat.grammatica;colofn < Lat.columna.

Hence adjectives havingɥ, (<ĭ) orw (<ŭ) in the ultima change these toe ando in the fem., the affection being due to the lost fern, ending‑ā; thus Brit. *u̯indos, *u̯indā gave respectivelygw̯ynn, gw̯enn ‘white’.

The adj. *briktos had regularly fem. *briktā, which by the rule became *brektā; now *ikt >īth and *ekt >eith, lateraith§ 108 iv (1); hencebrīth ‘speckled’, f.braith, which is thus seen to be quite regular.

The affection is original only in adjectives of the‑os/​‑ā declension; but after the loss of the inflexional endings, it spread by analogy to other stems; e.g.crwnn ‘round’ < Brit. *krundis (: Ir.cruind) has f.cronn on the analogy oftrwm < Brit. *trumbos (: Ir.tromm) f.trom; andgwyrẟ < Lat.vir’dis has f.gwerẟ on the analogy offfyrf, fferf < Lat.firmus, firma. Doubtlessdeilien wyrdd in M.Ỻ. i. 155 represents a local survival of the old fem., as intonn wyrt (‑t‑ẟ)w. 9a ‘green wave’.

§ 69.Ultimatei-affection. i. This was caused by,ī (fromī, ē, ō orū), or by accentedĕ́ orĭ́. Kelt. post-tonices before a vowel became and caused this affection§ 75 vii (1), soe(p) see ib.; also Lat.i, and sometimese, before a vowel.

ii. (1)a becomes Ml. W.ei, Mn. W.ai:eil, ail ‘second’ < *ali̯ós: Lat.alius;—yspeit, ysbaid ‘space’ < Lat.spatium;—rhaib ‘spoil’ < Lat.rapio;—beirdd ‘bards’ < *bardī;meib ‘sons’ < Brit. *mapī;—ugeint, ugain ‘twenty’ < *u̯ikantí < Ar.u̯ī̆k̑m̥tí;—lleidr ‘thief’ < Lat.latrō;—deigr ‘tear’ < *dakrū§ 120 iii (1).

(2)ak orag before a consonant, which becomesae in Ml. and Mn. W.§ 104 ii (1),iii (1),§ 108 iv (1), is affected toek oreg which gives Ml. W.ei, Mn. W.ai, see ib. ThusSaxones >Saeson butSaxō > *Sex >Seis, Sais;—*kaktos ‘serf’ (< *qaptos) >caeth, but pl. *kaktī >ceith, caith ‘serfs’;—*dragnos >draen ‘thorn’§ 104 ii (1), pl. *dragnesa > *dragni̯a >drein,drain.

(3) In disyllables before consonant groups containingr, and beforech, the affection ofa appears asɥ, which alternates withei in Ml. and early Mn. W. Thusheyrnb.t. 29,r.m. 121,r.p. 1362,r.b.b. 47, pl. ofhaearn ‘iron’;—r͑eydɏrr.p. 1301 besider͑yeidɏrr.p. 1222, pl. ofrhaeadr ‘cataract’;—kedyrnw.m. 51 besidekedeirn do. 40, pl. ofcadarn ‘mighty’;—soalarch pl.eleirch,elyrch§ 117 i;—tywarchen pl.tyweirch, tywyrch§ 126 i (2);—paladr, pl.peleidɏrw.m. 179, Mn. W.pelydr;—Mn. W.bustych, menych,§ 117 i. Also in the procliticgeir >gyr ‘near’§ 214 ii.

Theɥ is probably the result of thickening the beforer + cons. and beforeχ in an unaccented syllable. (In accented syllables asbeirẟ, thei is still pure, but it has becomeɥ beforeχ§ 17 iii.) Thusei >ỿɥ >ɥ. Fromr + cons. it spread to cons. +r. Probablygwesgyr (singler) forgwasgar§ 173 iv (1) is due to false analogy.

(4) In polysyllables before a labial also,a is affected toɥ; as inmodryb < *mātr-aq§ 122 iv (2);cyffelyb, ethryb also from *‑aq- < *‑əq-ōq- ‘face’, cf.§ 143 iii (8);Caer-dȳ́f ‘Cardiff’:Taf.—‑am- becomes‑eu or‑yf, except in analogical formations; see§ 76 vii (1).

iii. (1)e becomesɥ:engyl ‘angels’ < Lat.angelī;—cyllyll ‘knives’ < Lat.cultellī;—so,cestyll,gwëyll§ 117 i;—erbyn ‘against’ < Kelt. *ari qu̯ennōi§ 215 ii (4);—gwŷl ‘sees’§ 173 iv (1).

There appears to be no certain example ofe becomingei;dyweit ‘says’ may be from *u̯at-§ 194 i (1).

(2)ek oreg before a consonant when affected becameik orig which givesī regularly; asnith ‘niece’ < *nektí‑s§ 86 ii (1);—llith ‘lesson’ < Lat.lectio.

iv. (1)o becomesei (Mn.ai) orɥ:yspeil, ysbail ‘spoil’ < Lat.spolium;—seil, sail ‘foundation’ < Vulg. Lat.solea for Lat.solum, cf. E.soil;—myfyr ‘thought’ < Lat.memoria;—ystyr ‘meaning’ < Lat.historia;—dŷn ‘man’ < *doni̯os: Ir.duine;—mɥ̂r ‘seas’ < *morī§ 122 ii (4);—esgyb ‘bishops’ < Lat.episcopī;—Selyf <Salomō;—tair Ml. W.teir for *ty-eir ‘three’ fem. < *tisorés§ 75 vi (3);—pair, Ml.peir ‘caldron’: Ir.coire§ 89 iii.

It is seen thatei occurs beforel andr; but in disyllables we haveɥ before the latter.

(2)ok orog before a consonant, which givesoe in W., becomesw͡y when affected; thusoen ‘lamb’ < *ognos, pl.ŵyn < *ognī;ŵyth ‘eight’ < *ok̑tō.

v.u becomesɥ: Merchyr§ 16 iv (2) <Mercurius;—cŷn ‘chisel’ < Lat.cuneus;—asgwrn ‘bone’ pl.esgyrn;—ŷch ‘ox’ < Ar. *uqsō, whence O.H.G.ohso, Skr.ukṣā (Av.uxš- implies‑q‑); the pl.ỿchen (< Ar. *uqsénes, whence Skr.ukṣáṇaḥ, E.oxen) hasỿ fromu unaffected,§ 66 i.

u does not becomeei;deifr as pl. ofdwfr is doubtful (m.a. i 556) except as a late and artificial form; see Silvan Evans s. v.

vi. When any of the above changes takes place in the ultima,a in the penult becomese; seekedyrn, elyrch, pelydr, Selyf, esgyrn above.o also becamee, asgosod ‘to set’gesyd ‘sets’, liable to becomeỿ beforest, as Ml. W.ebestyl, ebystyl <apostolī, sg.abostol <apostolus. In Ml. W. the affection extended, as in the last example, to the ante-penult.

vii. Theei due to affection as above, alsoei fromek oreg, had open, and was thus distinct from originalei which had close. The former (e̦i) givesei, ai; the latter (ẹi) givesw͡y§ 75 iii (1).

On later modifications ofɥ, ei, see§§ 77,79.

§ 70.Non-ultimate affection. i.a and sometimeso in the syllable which is now the penult becamee when the following syllable hadī orĭ (nowi orɥ), except where theĭ was itself affected toe,§ 68. Thuscerydd ‘reprimand’ < *karíi̯o(s) besidecaredd ‘fault’, Ir.caire, < *karíi̯ā;—Ml. W.gwedy ‘after’, O. W.guotig;—Ml. W.pebyll ‘tent’ < *papíli̯o < Lat.pāpilio;—Ebrill <Aprīlis;—cegin <coquīna;melin <molīna; etc. In Ml. W. the affection extends over two syllables, asederyn ‘bird’, Mn. W.aderyn, pl.adar.

o seems to undergo the change chiefly after a labial or before a guttural, where it might have becomea if unaffected.

The restoration ofa in the antepenult in Mn. W. is due to the vowel in that syllable becoming obscure because unaccented, in which case it was natural to re-form etymologically.

ii. (1) Before the same change took place, anda ando appeared ase in O. W.; but thee was further affected by the, and becameei in Ml. and Mn. W.; thusMariānus > O. W.Meriaungen. iii. > Ml. W.Meirẏawnr.b.b. 81, Mn. W.Meiri̯on;—so O. W.Bricheniauca.c. 895, Mn. W.Brycheini̯og;—O. W.mepiongen. xii, Mn. W.meibi̯on ‘sons’. See§ 35 ii.

In the dialect of Powysceili̯og ‘cock’,ceini̯og ‘penny’ are pronouncedcel̯iog, ceni̯og. This is perhaps a simplification ofei,§  78 v, rather than old e retained.

(2) Originale also becameei before; thusEterniānus >Edeirnaunb.b. 74Edeirnonw.m. 50,r.m. 35, Mn. W.Edeirni̯on (now wrongly speltEdeyrnion);—sopencerdd ‘chief of song’ Ml. pl.penkeirẟẏeitr.p. 1230, Mn. W.penceirddi̯aid;—anrheg ‘gift’ pl.anr͑eigẏonr.p. 1221 (generallyanr͑egẏonr.b.b. 394,r.m. 257, nowanrhegi̯on);un-ben ‘mon-arch’,unbeẏnẏaeth,a.l. i. 34, 382, ‘sovereignty’ (nowunbennaeth, new formation);gorwedd ‘to lie’,gorweiddi̯og ‘bed-ridden’;gweini̯aith§ 32 forgweni̯aith ‘flattery’.

(3) In later formations does not affect the vowel ; forms likepersonnẏeit, Albanẏeit etc.§ 123 iv, andcari̯ad, meddi̯ant, etc., are extremely common in Ml. and Mn. W. Also forms likeari̯an ‘silver’ in which is not original, but comes fromg.

iii. The Ml. and Mn. diphthongae, whether fromak- orag- before a consonant, or froma-e, becomesei beforeī or, as in Ml. W.keithiwet < Brit.-Lat. *kaktīu̯itās;saer ‘craftsman’ pl.seiri;gwaedd ‘cry’,gweiddi ‘to cry’;draen ‘thorn’,dreiniog ‘thorny’. Similarlyog..ī orug..ī >ei..i; as ingweini ‘to serve’ < *u̯o-gnīm‑,heini ‘active’ < *su-gnīm‑:gnīm-§ 203 vii (4). Beforeɥ it becomes, as inkeyrydd pl. ofkaer ‘fort’. But, except in a few cases such as the above, this affection is usually ignored in writing, especially in the Mn. period.

iv. The affecting sound has disappeared incenwch ‘ye sing’ for an earlier *cenɥw̯ch§ 26 vi (5); in the Ml. formsEdeirnon etc.§ 35 ii; and in such forms asceidw̯ad forceidw̯i̯ad,§ 36 v.

v. The affection ofa ando by a lost stem-ending‑ī‑, ‑i̯o‑, ‑ū‑, of the first element of a compound is similar to ultimate affection:a >ei inmeitinb.a. 18 ‘morning’ (Mn. W.er’s meitin ‘some hours ago’) < *matū-tī́n- (treated as a compound) < Lat.mātūtī́num;—o >ỿ insỿl-faen:sail,§ 69 iv.

In Ml. W.meinoethb.t. 68,meinẏoeth do. 45 ‘midnight’ <mediā nocte, we seem to have early metathesis of, thusmeinẏoeth < *menẏoeth < *meda-ni̯okte. The formsmeinyẟb.t. 31,meinẟyẟ do. 55 ‘mid-day’ are formed on its analogy.

The Long Vowels.

§ 71. i. (1) Pr. Kelt.ā (from Ar.ā andō) remained in Brit. In Early W. it became an openō like Eng.a incall, which we may writeɔ; in O. W. this becameo in unaccented syllables,au (≡aw) in accented syllables. Latinā also shared this development.

The Early W.ɔ is attested in Bede’sDinoot (≡Dünɔt), Ml. W.Dunawt < Lat.Dōnātus. In all syllables except the ultima it becameo, asbroder ‘brothers’ < Pr. Kelt. *brāteres; in this positionaw fromā occurs only in late formations likemawrion pl. ofmawr ‘great’, and after§ 148 i (6). But in the ultima and in monosyllablesɔ > O.W.au ≡ Ml. W.aw, as O. W.brautox. ‘judgement’ < Pr. Kelt. *brāton, trintautjuv. sk. < Lat.trīnitātem; Ml. W.brawt, trindawt. In Mn. W. aw remains in monosyllables, asbrawd, but in the now unaccented ultima it has becomeo, as intrindod. The following table summarizes the history of Brit. (and Lat.)ā:

Brit. (Lat.)Brit.Early W.O.W.Ml.W.Mn.W.Mn.W.
*brāteres    }{\displaystyle \scriptstyle {\left.{\begin{matrix}\ \\\\\ \\\ \ \end{matrix}}\right\}\,}}ā———ɔ/
\
o———o———obroderpenult.
trīnitātemau———aw/
\
otrindodult.
*brātonawbrawdmonosyll.

(2)ā when unacc. was shortened and givesa§ 74 i (1); this might happen in monosyllables asa ‘of’§ 209 vii (5),a ‘whether’§ 218 iii. When acc. in Brit. and unacc. later, it giveso, as inpob§ 168 i (3),mor§ 151 i,o ‘from, of’§ 209 vii (5),o, ‘if’§ 222 v (1).

ii. (1) Ml. W.aw in the unaccented ultima (whether fromā as above, or fromou§ 76 iii) survives in the spoken language incanllaw ‘hand-rail’,darllaw ‘to brew’,distaw ‘silent’,eirlaw ‘sleet’,ysgaw (alsoysgo) ‘elder-tree’,llỿsfrawd ‘brother-in-law’; in compounds with numerals, asdeunaw ‘18’,dwyawr ‘2 hours’,teirawr ‘3 hours’, etc. (exceptdẃylo fordẃylaw ‘hands’); and in compounds ofmawr, asdĭ́rfawr ‘very great’,trỿstfawr ‘noisy’ (except in place-names,Trefor,Coetmor). In a few book-words which have gained currency it is not a genuine survival: astraethawd ‘treatise’,catrawd ‘regiment’,bydysawd ‘universe’,rhaglaw ‘deputy’; and the formsllïaws ‘multitude’,cyfiawn ‘just’,Ionawr ‘January’,ansawdd ‘quality’, are influenced by the written language, which, however, had alsollios,cyfion,Ionor,ansodd Io.G. 187, formerly; see examples below.Chwefror has o always (generally soundedChwefrol by dissimilation). The recent written language has been influenced by mechanical ideas of etymology in the substitution ofaw for the regularo inffỿddlon ‘faithful’,dwylo ‘hands’,union ‘straight’,cinio ‘dinner’,anodd ‘difficult’, cpv.anos (§ 48 iv,§ 148 i (6)); all these appear witho in early Mn. poetry, and are pronounced witho in the spoken language. On the misspellingathraw forathro see§ 76 v (5).

Ni fyn cariad i wadu,
Na’i ddangos ilioslu.—D.G. 69.

‘Love will not be disavowed, or manifested to many a host.’

Gwahawdd Saeson bobIonor
I’r Deau maent ar hyd môr.—L.G.C. 155.

‘They invite Saxons every January to the South across the sea.’

Anoddrhyngu bodd y byd.—T.A.a 14967/29.

‘It is difficult to please the world.’

(2)awin the ultima began to be reduced too in the Ml. period; thus we findEdeirnonw.m. 50,achosỺ.A. 4,Meirẏonr.b.b. 13. But the bards even in the Mn. period continued to write theaw for the purposes of rhyme. In recent times, owing to ignorance of the older language, they have sometimes writtenaw for originalo, as “esgawb” foresgob ‘bishop’, “dyniawn” fordynion ‘men’. This is not due to a confusion of the sounds ofo andaw (for thea inaw is a purea, quite distinct fromo), but to the blundering notion that as someo’s may be writtenaw, anyo may. The Early Mn. poets generally useaw correctly, guided by a living literary tradition. The distinction is seen in Ml. W.yscol ‘school’,iscolb.b. 81 from Lat.sc(h)ola andyscawlw.m. 189 ‘ladder’ < Lat.scāla, bothysgol in Mn. W.

(3) In a few casesaw comes fromo:praw(f) besideprofi < Lat.prob‑;mawl besidemolaf ‘I praise’, Ir.molim;tymawrr.p. 1244 for the usualtymor < Lat.tempora. In each case theo comes before or after a labial. In Vulg. Lat. there was a tendency to lower a vowel before a labial so thatprob- might become *prɔb- >prawf. But it is more likely that all these are due to false analogy.

Inawr ‘hour’, andnawn ‘noon’ we haveaw < Lat.ō. These have been explained as late borrowings ; but historically this is improbable. Possibly the pronunciation ofhōra varied in Lat., since Gk. ω (≡ō̦) was popularly soundedō̦ (γλῶσσα > Ital.chio̦sa);ō̦ would giveɔ >aw. Fornawn see§ 76 iii (4).

iii.āg > O. W., Ml. W.eu, Mn. W.eu, au; thusbreuant ‘wind-pipe’, O. W.‑brouannou < *brāgn̥t‑: Ir.brāge gl. cervix, O. Bret.brehant;—W.pau ‘country’, O. Bret.pou, Corn.pow < Lat.pāg-us;—soāk orāg before a consonant: W.gwaun, O. W.guounl.l. 156, 196 ‘lowland’, Ir.fān < *u̯ākn- < *u̯o-ak‑n-§ 104 iii (1);—W.ceulo ‘to congeal’ < *cāgl- < Lat.co-āg’l-o. But beforet theā is shortened§ 74 iv.

iv.‑ān- often givesonn in the present penult:cronni:crawn§ 202 v (2);—ffynhonnau ‘fountains’ <fontān‑;—Meirẏonnyẟg.c. 122,r.b.b. 263, besideMeirẏonyẟ do. 303, 306, <Mariān‑.

§ 72. i. Pr. Kelt.ī (< Ar.ē, ēi̯, ī) remained in Brit., and Brit. and Lat.ī remain in W.,§ 59 iii,iv. Further examples: W.hīr ‘long’, Ir.sīr < Pr. Kelt. *sīros < *sē-ro‑s: Lat.sērus;—W.gwīn < Lat.vīnum. It is, of course, shortened in W. before two consonants; asgwĭ́n-llan ‘vineyard’.

ii. Lat.ī is treated asē or Kelt.ẹi in W.paradw͡ys <paradīsus < Gk. παράδεισος; andsynnw͡yr <sentīre. In rustic Lat.ī was often soundedē, but whether only in words with original ei is not clear, Lindsay, p. 29. In Lat.ign thei was often written long, or was writtene; hence it probably differed little from Lat.ẹ̄, and so gives W.w͡y, assŵyn ‘charm’ <signum.

§ 73. i. Pr. Kelt.ū, which remains in Ir., and apparently remained in Gaul., as shown by the spelling ου in the second element of Αὐγουστό-δουνον, advanced in Brit. towardsü, for it appears asī in W., while Lat.ū borrowed into Brit. givesu in W.; thus Pr. Kelt. *dūnom > Ir.dūn, W.dīn ‘fort’,dinas ‘city’: O. E.tūn, E.town;—Pr. Kelt. *glūn- > Ir.glūn, W.glīn ‘knee’§ 63 vii (4);—Pr. Kelt. * > Ir. ‘dog’, W.,§ 89 iii.—But Lat. pūrus gives pūr, mūtus gives mūd, etc.

ii. Some irregularities occur in the development of Lat.ū and Brit. and Lat.ŭ before a labial:

(1) Lat.ū incūpa givesī in W.cib, Bret.kib. This seems to be the only example in W., and may be due to fluctuation betweenü andi before a labial; cf. conversely W.uffern ‘hell’, Bret.ifern < Vulg. Lat.īferna, Lat.inferna.

(2) Brit. and Lat.ŭ beforeb followed by a vowel gave W.u; asdu ‘black’ for *duv < *dub‑: Ir.dub ‘black’; W.cuẟygl for *cufygl < Lat.cŭbic’lum. But beforen, r, l, ub giveswf regularly, as indwfn,dwfr,§ 90.

(3)ŭ beforem is regular, as shown by W.twf ‘growth’,tỿfu ‘to grow’ < *tum-§ 201 i (8). But Lat.ŭ innumerus givesi innifer. This may be due to a dial. pronunciation of Lat.u asü; cf. Osc.Niumsieís ‘Numerii’, and the Oscanized Lat.Niumeriis ‘Numerius’. Lat. itself hadü beforem in an unacc. syll., asmaximus, maxumusmaxümus. The soundü would be identified with Brit.ü, and prob. lengthened, giving the same result. W.ufyll ‘humble’ < Lat.hŭmilis may perhaps be similarly explained, but withu fori as inuffern.

(4)ŭ beforep is regular, as seen incybyẟ ‘miser’ < Lat.cŭpidus,syberw̯ ‘proud’ < Lat.superbus. In W.achub < Lat.occŭpo theu may be due to the lengthening of theŭ when it came to be accented, as it did in Brit.§ 65 ii (1).

For Lat.ē see§ 75 iii (1); for Lat.ō see§ 76 ii (1).

§ 74. i. (1) In Brit.ā was shortened when unaccented. Thus W.pechadur ‘sinner’ < *peccătṓr- < Lat. acc.peccātōr-em besidepechod ‘sin’ <peccā́tum;—W.meitin ‘morning’ < *meid-din < *matū-tī́n- < Lat.mātūtī́num;—W.agw͡yẟawr for *afgw͡yẟawr < Lat.ābēcēdā́rium;—W.Madrun < Lat.mātrṓna besidemodryẟ ‘aunt’ < Kelt. *mā́traqī;—W.ceiliagw͡ydd ‘gander’, Ml. W.keẏlẏacuyta.l. i 280 < *kali̯ako-géidos besideceiliog ‘cock’ < *kali̯ā́kos;—W.paratói ‘to prepare’,§ 201 iii (5), besideparod ‘ready’ < Lat.parā́tus, etc., etc.

Naw mwy i frag nacheiliagwydd,
Naw gwell i synnwyr na gŵydd.—S.T.,c 16/93.

‘Nine times more boastful than a gander, nine times more sensible than a goose.’ (The recent spellingsparotoi, ceiliogwydd are false; the words are pronounced as spelt above.)

For the apparent exception inIonawr a sufficient explanation is the secondary accent which was required to distinguishJầnuā́rius fromFèbruā́rius, and which for emphasis might even become primary.

(2) Words likeswyddogol ‘official’ are formed in W., and mostly late, by adding‑ol to‑og, and are not derived in full from Brit., for Brit.‑āk-ā́l- would give‑ag-ol. The wordlluosog is an old formation, but it is not formed from the original ofllïaws; the latter hasā́ fromō, the formation being *‑ōs-tāts, while the former hasŏs, the formation being *‑os-tos, extended to *‑os-tākos,§ 75 iii (3).

ii. It is seen in the above examples that other long vowels remained long when unaccented; and thatī andū need not have been accented to cause affection of a preceding vowel.

iii. In Ir. the shortening of long vowels is carried further and is independent of the Brit. shortening ofā. The latter had not set in in Pr. Kelt, as is shown by the development ofāu, which when unaccented in Brit. gaveau, while Kelt.au gaveou§ 76 v (5).

iv. All long vowels were shortened before groups of sonant + explosive, as ingwynt ‘wind’ < *u̯entos < *u̯ēntos; so Lat.ventus. W.dyall < *dii̯ált- < *dii̯ā́lt-§ 75 vi (4). Also before two explosives; *‑o-akt- *‑ākt- > *‑akt- >‑aeth§ 203 i (4).

The Diphthongs.

§ 75. i. (1) Ar.ai remained in Kelt. It appears in Ir. asāi, āe, in Gaul. asai ore. Before a consonant it appears in O. W. asoi, and in Ml. and Mn. W. asoe ()§ 29. Thus W.coeg ‘empty’ (as a nut without a kernel),coeg-ddall ‘purblind’, Ir. caech ‘one-eyed’: Lat.caecus ‘blind’, Goth,haihs ‘one-eyed’, Skr.kekara‑ḥ ‘squinting’ < Ar. *qaiq‑;—W.hoedl ‘lifetime, life’, Gaul.Setlo-ceni-(ae Deae): Lat.saeculum < *sai-tlo‑m§ 111 vii (1); W.coed ‘wood, forest’, Gaul.Ceto-briga < *kaito‑: Goth.haiþi, O. H. G.heida, E.heath, Lat.bu-cētum (ē forae owing to confusion with the suffix‑ētum).

(2) Before a vowelai fell together withii̯, seeiv below. Butas in the penult, followed bye (ori), gave a newai which gives W.oe >o§ 78 i (1); thus Brit. *karaset > *karoe,caro ‘may love’. Followed byī́ it falls together withii̯ and gives‑ei, as *u̯órnasīm >arnei; when theī was unacc. it gives‑i as *u̯órnasīm >erni§ 209 vii (1).

But in the ante-penult a vowel befores drops§ 113 i (2); hence *kara-se-re >kar-her ‘may be loved’.

(3) Kelt.āi > W.w͡y, as inmwy ‘greater’ < *mā́-i̯ōs or *mā́ison: Ir.mao for *mau < *mā́i̯ōs. When unaccented it was shortened and so givesoe, as prob. in Ml. W.moeỺ.A. 142 ‘more’.

A newāi was produced before a vowel in Brit. whenās was followed byī ore; thus *karā́s-īt > *karāi̯īt >karwy§ 183 ii (1).

A newāi might be produced before a cons. by metath. of§ 100 v; thus Lat.occā́sio > W.achos > but Brit. pl. *accā́si̯ones > *accā́i̯sones > Ml. W.achwysson.

(4) W.oe >ae after orm, etc.;oe >w̯ae afterg§ 78 ii (2).

ii. (1) Ar.oi remained in Pr. Kelt., and appears in Ir. asōi, ōe. In W. it becameu before a consonant. Thus Ar. *oinos ‘one’ > Gk.οἰνός, οἰνή ‘ace’, O. Lat.oinos, Lat.ūnus: Ir.oen, W.un ‘one’.—W.ud inanudon ‘perjury’, Ir.oeth ‘oath’: Goth.aiþ‑s ‘oath’.—W.grug ‘heather’ for *gw̯rug (Pemb. dial.gw̯rīg), Ir.froech < *u̯roiko‑s: Gk.ἐρείκη < *u̯ereikā.

Before or after in Brit.,oi becameai which gives W.oe (oy); as ingloyw ‘shiny, glossy’ < *ɡloi-u̯o‑s: Gk.γλοιός < γλοιϝος,§ 92i;—ky(h)oeẟ ‘public’ < *ky-w̯oeẟ < *ko-u̯oid‑: W.gŵyẟ ‘presence’ < *u̯eid‑, √u̯eid- ‘see’; here‑w̯- dropped; where it remained,woe again gavewae§ 78 ii (2); thusgwaethaf for *gwoethaf < *u̯aidisamos < *u̯oidisamos < *u̯o-ed-isamos < *upo-ped-isₑmo‑s§ 148 i (5).

(2) Before a back voweloi gave W.w͡y; as *‑oi̯an > *‑wy‑n >‑wn§ 180 iii (1); cf.§ 76 v (4). But beforei ore the dropped§ 100 vi, ando before the vowel developed likeu before a vowel, that is, asou̯; thus *dó esō > *dói̯ǖ > *dóī > *dóu̯ī > *deu§ 76 v (1), whencedeuaf§ 193 x (5); and *do eset > *doi̯et > *doet > *dou̯et >daw, or without diphthongization *do-et >do, see ib.; so *moi estō > *mo estī > *mo ys >moes§ 200 ii. Followed byī after the accent it gives‑i, as in‑ẟi f. sing. ‘to her’ < *´‑doi̯ī < *´‑do-sī§ 210 x (1).

(3) Ar.ōi gave Kelt.āi and developed accordingly.

iii. (1) Ar.ei remained in Pr. Kelt. In Gaul. it is writtene or ει, asDevo-gnata, Δειουονα. In Ir. it appears asē orīa. In W. before a consonant it becamew͡y. Thus W.gŵyẟ ‘presence’ for *gw̯ŵyẟ < Ar. *u̯eid-§ 63 iv;—mor-dw͡y ‘sea voyage’ < *mori-teig-§ 103 ii (1), etc.

In Brit. and Gaul. it was probably sounded asẹi̯. Latinē which was soundedẹ̄, was identified with this sound in Brit., and shared its development in W., thusrēte >rhwyd,rēmus >rhwyf,plēbem >plwyf,cēra >cwyr, etc. Lat.oe which seems to have varied fromö to appears in W. asi, oe orw͡y, asciniaw ‘dinner’,poen ‘pain’,cwyn ‘supper’.

(2) Before a vowelei fell together withii̯, see below.

(3) Ar.ēi before a vowel > Kelt.ī > W.i. Thus W.dī́od, Ml.diawt ‘drink’ < *dhēi-āti‑s, √dhēi- ‘suck’.—W.llī́aws ‘multitude’, Bret.liez < Brit. *līā́ssās < *līāstāts < *(p)līōs-tāts, a noun in *‑tāt- from the cpv. *plē-i̯ōs: Lat. insc.pleores, Gk. πλείων. Before Kelt.o it becomesu, as inllüosog, Ml.lluossauc < Brit. *līu̯ossā́ko‑s an extension *osso‑s < *(p)līos-to‑s an adj. formed from *plē-i̯ōs like Lat.honestus fromhonōs; see§ 76 ix (2),§ 74 i (2),§ 169 iii (3). Before a consonantēi >ē giving Kelt.ī, W.i.

iv.ai andei fell together withii̯ before vowels.After the accent the became ẟ, in other positions it remained as. Thus:

(1) Accentedíi̯ (orái̯ oréi̯), which is generally in the penult, but may be ante-penultimate, gives W.‑yẟ; thus W.rhyẟ ‘free’ < *príi̯os: Goth.freis, Eng.free;—trefyẟ ‘towns’ < *trebíi̯es;—trydyẟ ‘third’ m. < *tritíi̯os; with‑a in the ult. it gives‑eẟ, astrydeẟ ‘third’ f. < *tritíi̯ā. In the ante-penult‑yẟ‑, as W.ysbyẟad ‘thorn’: Ir.scē, gen. pl.sciad.

(2) Post-tonic´‑ii̯ gave *ai̯i̯, which becameoeẟ,§ 62 i (2); thusmoroeẟ ‘seas’ < *mórii̯a: Lat.maria;—Ml. W.gw̯ladoeẟ ‘countries’ < *u̯látii̯es;—dannoeẟ fem.m. m. 8 ‘toothache’ < *dántii̯ā;—oeẟ ‘would be, was’ < *´sii̯ēt,§ 180 ii (3).

v.Before the accent, in the penult the result varies according to the quality of the accented vowel in the (now lost) ultima; thus:

(1)ii̯é > W.‑i, as intri ‘three’ m. < *trii̯és (accented like the f.) < Ar. *tréi̯es (f. *tisorés) see§ 103 i (3);—W.trefi ‘towns’ < *trebii̯és.

(2)ii̯ī́ > Ml. W.‑ei, Mn. W.‑ai, as in W.r͑ei, rhai ‘some’§ 165 vi,carai ‘would love’§ 180 ii (2); cf.nei, naivii (2).

(3)ii̯ó > W.‑yw as inrhyw ‘some’§ 165 vi; cf. gwywvii (3).

(4)iiá > O. W.‑ai, Ml. and Mn. W.‑ae,‑e, also Ml. W.w͡y; as in O. W.guarai, latergwarae, gware, chwarae, chware ‘to play’, Bret.c’hoari, Corn.hwary < Brit. *(s‑)u̯arii̯ā́ < *u̯ₑrii̯- √*u̯erēi§ 63 vii (3); a variant isguarvyb.b. 50 =gw̯arw͡y.

vi.Before the accent in the ante-penult the result varies according as the accent fell on the lost ultima, or on the penult.

(1) In the former case the penult had generally a reduced vowela (<ə or) ; the combination‑ii̯a‑´ gave W.‑aea- (also writtenaya), O. Bret,‑oia‑, Bret.‑oua‑, ‑oa‑. Thus W.claear ‘lukewarm’, Bret.klouar: Gk. χλιαρός (Ar. alternationk /gh);—W.gaeaf, gayaf, Bret.goañv ‘winter’ < *g̑hii̯əmó‑s: Skr.himáḥ, Lith.žëmà, Gk. χειμών, χεῖμα > Lat.hiems: Gaul.Giamon.., Ir.gem-red (e foria);—W.traean ‘third part’: Ir.trian;—W.rhaeadr ‘cataract’ < *rii̯ə-tró‑: Ir.riathor: Lat.rīvus, √rei̯ā- ‘flow’;—W.daear ‘earth’, Bret.douar < *g̑hðii̯ₑrā§ 98 iii.—‑isa- or‑esa- gives the same result: Pr. Kelt. *isarno- (*is R-grade of *ais: Lat.aes): Gaul.Ysarno- Iserno‑: W.haearn, hayarn ‘iron’, O. W.Gur-haiernngen. xxiii, O. Bret.hoiarn.

Before the loss of the accented ending the accent must have shifted to the present penult, which had the next highest stress. In Gwent and part of Dyfed the unaccenteda was generally lost; thusdaer ‘earth’ nowdâr§ 29 ii (da’r a nen Wms. 785). The O. W.dair, dayrl.l. 120,gaemb.s.ch. 3 represent this dialect. The reduction is general inclaer ‘bright‘ besideclaear ‘lukewarm’, with differentiation of meaning. Fromdaer comesdaerawlỺ.A. 130, 164.

(2) After a labial the above group takes the form‑w͡ya- interchanging with‑ïa‑; thus W.mw͡yar ‘blackberries’,mïaren ‘bramble’, Ir.smēr ‘blackberry’ < *smii̯ar- < *smii̯oró- (§ 65 vi (2)) : Gk. μόρον (μῶρον, Hes.) (with μ < *smi̯‑?), Lat.mōrum prob. < Gk.;—W.gw͡yal (for *gw̯w͡yal) inmorw͡yal ‘laminaria’,gw̯ïal ‘twigs, osiers’ < *u̯ii̯ə‑l‑, √u̯ei̯āˣ- ‘weave’: Lat.vieo, etc.;—W.gw͡yar ‘blood, gore’ < *u̯ii̯ₑr‑: Lat.vīrus;—W.bw͡yall ‘axe’ < *bii̯ald‑, met. for *bii̯adl‑: Ir.biāil, Ml. Bret.bouhazl < *bii̯adl‑: O. N.bilda, O. H. G.bīal < *bīþl *bītl: Lat.findo, √bheied‑?—W.hw͡yad ‘duck’, Gwyn. dial.chw̯ïadan < *s‑u̯ii̯at‑, √a-u̯ei‑: Skr.váya‑ḥ ‘bird’;—W.mw͡yalch ‘ousel’ < *mesₑl‑: Lat.merula < *mesula, O.H.G.amsala, Ger.Amsel.—éso developed similarly (since post-tonics did not give); thus *su̯ésores >chwïor-eẟ ‘sisters’, with‑eẟ added, Bret.choarezed with two additions.

The labial changed the diphthong (Early W. *oi) towy. Under the new accentwy remained, but becamei in the present ante-penult; thusmẃyar:mïáren—*gw̯ẃyal:gw̯ïálen, a new pl.gw̯ḯal being then formed from the latter. Where the sound comes in the present ante-penult in old formations, the form is undecided ; thus O. W.guiannuinox. ‘Spring’, Early Ml. W.guaiannu(i)n, guayanuhina.l. i 142, alsogwahanwyn do. 308, Ml. and Mn. W.gwannw͡yn, gwanw͡yn < Brit. *u̯esant‑´: Skr.vasantá‑ḥ ‘Spring’, Lat.vēr < *u̯ēsr.

(3) When the followinga oro was affected, the diphthong becamey ore, liable to be assimilated and lost; thusr͑yeidɏr, r͑eydɏr ‘cataracts’,heyrn ‘irons’§ 69 ii (3), Gwyn. dial.hɥ̂rn forhɥɥrn;—W.tair ‘three’ f., Ml. and O. W.teir for *tyeir (cf.breint,Seint§ 103 ii (1)), Ir.teoir < *tisorés: Skr.tisráḥ; so W.pedair ‘four’ f., Ir.cetheoir < *qetesorés: Skr.cátasraḥ.

Such forms asheiyrn,rheieidr are quite late and artificial. But some old re-formations occur when the diphthong stood in the present ante-penult, asdeyeryn (‑yn‑in)b.a. 12 ‘earthen’,heyernin ib. ‘of iron’,daeerinr.p. 1281,mïéri, pl. ofmïáren.

(4) Secondly, the vowel following the diphthong is accented. In that case the diphthong becamee ory liable to be assimilated and lost, as in (3) above. Thus W.ëog ‘salmon’ < *esā́k‑: Ir.ēo, gen.iach;—W.deall, dyall, dallt§ 82 ii (3) ‘understanding’,deallt-wriaeth id., N.W. dial.dāllt < *dii̯ált- < *dii̯ā́lt-§ 74 iv, met. for *dii̯ā́-tlo‑, √dhei̯ā- ‘appear, perceive’: Skr.dhyā-yati ‘thinks’,dhīraḥ ‘intelligent’; Ml. W.dyat ‘thought’ < *dii̯á‑t- with analog. accentuation for original *dhii̯ə-tó‑; Ml. W.gorffywys, later, withy lost,gorffwys ‘rest’ < *u̯er-qii̯éi‑st‑, √qei̯ēi̯‑: Lat.quiēsco.

(5) Latin pretonici ore before a vowel is treated as, thusdiáb(o)lus givesdi̯awl§ 100 ii (1).

vii. Except as above,‑es‑,‑is- before a vowel developed differently from‑ii̯‑, chiefly because post-tonics did not, like, become.

(1) In the penult after the accent‑es- >‑i̯‑ ; thus Ml. W.tei ‘houses’ < *tigi̯a < *tigesa§ 104 ii 2;—W.clyw ‘hearing’ < *klou̯i̯-§ 76 v (2) < *kléu̯es‑, nom. *kléu̯os: Ir.clū, Gk. κλέος < *kléu̯os, neut.s-stem.—So‑ep‑: W.ceifn ‘distant cousin’ < *kóm-ni̯ōs < *kóm-nepōt‑s, see§ 123 v.

(2) In the penult and ante-penult, whenés came before‑e‑, contraction took place, andése >ei > W.w͡y; thus W.wy‑t ‘art’ < *ése tū < Ar. *esi ‘art’;—W.neithi̯wyr < *nokti di̯éser-§ 98 i (3).—Soépe: W.twymn < *tepesm(e)n-§ 86 i (3).

In the penult‑és- before‑ī- gaveoe; thus W.chwaer for *chwoeri (4), Corn.hoer < *su̯ésīr < *suésōr;—W.doe ‘yesterday’ < *désī < *ghði̯esei: Lat.heri, Gk. χθές, Skr.hyaḥ.—Corn.noi ‘nephew’ < *népōts.—‑es- beforeī́ prob. gaveei (like‑ii̯- before‑ī́‑, seev), and Ml. W.nei, Mn.nai ‘nephew’ may represent *nepṓts (accented like the f. *neptís: Skrnaptī́ḥ).—‑es- before‑i̯´- gavey, as in Ml. W.y ‘his’ < *esi̯ó,y ‘her’ (fore?) < *esi̯ā́s,§ 160 iv.

Lat.‑ai̯i̯- > Ml. W.ei, Mn. W.ai as inMei,Mai ‘May’ <Mai̯i̯us (Sommer 225); Ml. W.Kei <Caius.

(3) Before lostu oro,‑es- or‑is- givesyw (ew); as Ml. W.Ywein,Ewein, laterOwein < *Esu-gani̯os: Gaul.Esugen(ios), Ir.Eogan: Gk. Εὐγένιος. So perhaps in the (pretonic) penult: W.gwyw ‘withered’ < *u̯isú‑: Ir.feugud gl. marcor, Icel.visenn: Lith.výstu ‘I wither’, Lat.viēsco.

Soīs before lostu oro givesiw, andais givesoew: W.gwiw ‘good’ < *u̯īsus < *u̯ēsu‑s: Gaul.Visu-rix: Skr.vásu‑ḥ, Gk. εὖ, √eu̯eseu̯‑;—W.gwaew ‘spear’ for *goew§ 78 ii (2) < *gaison: Gaul.gaison whence Lat.gaesum: Ir.gae.

Lat.e in the penult givesew before losto oru: W.llew ‘lion’ <leo;pydew ‘pit’ <puteus;olew ‘oil’ <oleum. But Lat.i in the same position gave which affected the vowel: W.yspeil <spolium; so sometimese: W.cŷn ‘chisel’ < Lat.cuneus.

viii. (1) In final syllables, lost in W., Ar.ai,oi,ei becameī in Brit. and Gaul.; thus the nom. pl. ending of nouno-stems, which in Pr. Kelt., as in Lat. and Gk., was *‑oi (instead of Ar. *‑ōs), became‑ī (though‑oi also survives in a North Italian Kelt. insc.:Tanotaliknoi, Rhys, CIFI. 60); thus Brit. *bardos pl. *bardī > W.bardd, pl.beirdd.

‑āi unaccented >ai >ī, thus Gaul. Βηλησαμι dat. of a name whose nom. occurs asBelisama;ōi >ūi, in Pr. Kelt, laterū,§ 60, cf. Ir. dat.fiur ‘to a man’ < dat. *u̯irōi;‑ēi doubtless gave‑ī.

(2) But in monosyllables Ar.‑ai, ‑oi, ‑ei remained in Kelt., and developed as follows in W.:

‑ai >‑oe, thus Ar. *u̯ai > *gwoe >gwae§ 78 ii (2).

‑ei >wy; W.wy ‘they’ < *ei: Ir.ē.

‑oi >wy; W.pwy ‘who?’ < *qo-i = Lat.quī§ 163 vi; when unaccented it becameeu (O. W.ou, oi)§ 78 iii, thus Ar. *moi, *toi > W.meu, teu§ 161 iv.

§ 76. i. The Ar. diphthongsau, eu, ou were distinct in Pr. Kelt., but tended later to become one sound, which is writtenou. In Gaul.eu was still written as well asou in forms having originaleu, as inteuto- beside τοουτιους andNeviod… besideNoviodunum; we also find αυ,iii (4). In Brit. we may assumeou for all three. In W. it takes a variety of forms according to its position. The same development is shared byuu̯ whether from Ar.uu̯§ 63 iv or from Lat.u before a vowel.

ii. (1) Before a consonant, excepts, the diphthong becameu (≡ü) in W.,ūa in Ir. Thus W.tud ‘people, country’, Ir.tūath < *teutā, Gaul.teuto‑: Goth.þiuda, etc.;—W.rhudd ‘red’, Ir.rūad < *roudh-os, Gaul.Roud-ius: Goth.rauþs;—W.cudd ‘hidden’,cuddio ‘to hide’ < *qeudh‑: Gk. κεύθω, O. E.hȳde, E.hide;—W.bugail ‘shepherd’ < *bou-koli̯ós < *ɡou-qoli̯ós: Gk. βουκόλος.

In Brit. it was probably sounded; and Latō (≡ọ̄) andū shared its development; thus W.ffurf < Lat.fōrma;mur < Lat.mūrus, etc.

(2) But originaleus gives W.ew, asrhew ‘ice’ < *preus‑: Lat.pruīna < *prusuīna;—W.trew ‘sneeze’ < *(s)treu‑s‑, √pstereu-§ 96 ii (4);—W.blew ‘hair’ < *bleus-§ 101 iii (2).

The reason seems to be that *eus became *eh before the degradation of the first element of the diphthong.

(3) The diphthong was liable to be simplified by dissimilation when the following syllable contained oru; thus Ar. *tauros ‘bull’ became Kelt. *tauru̯os (in imitation of *u̯eru̯ā > Ml. Ir.ferb ‘cow’, Vendryes MSL. xii. 40), whence Kelt. *taru̯os > Ir.tarb, W.tarw ‘bull’. Later, whenau had becomeou in Brit., *ou-tūt- > *o-tūt- > W.odid ‘rarity’, beside Ir.ōthad, uathad < *au-tāt‑, both from *pau‑: Lat.pau-cu‑s, O.H.G.fōh, E.few.

iii. (1) Before a vowel the diphthong becameaw when unaffected. Thus W.naw ‘nine’ < Brit. *nou̯an < Ar. *neu̯n̥;—W.baw ‘dirt’ < *bou̯‑, besidebudr ‘dirty’ < *bou-tro-peu̯(āˣ)‑: Lat.pūs, etc.§ 101 iii (2);—W.awydd ‘desire’ for *aw̯w͡yẟ (rh. withrhŵyẟ§ 38 x) < *au̯eid‑: Lat.avidus < *au̯id‑, √au̯ei̯‑.—So Brit.au for unacc.āu as in Ml. W.andaw ‘listen’, met. for *adnaw < *áti-gnā-u- ‘attend to’ < *g̑n̥̄‑u̯-g̑enē‑: Lat.nāvus, Ir.aithgne ‘cognitio’. So alsouu̯ for Lat.u before a vowel, as W.cystrawen ‘syntax’ < Lat.construenda.

(2) But in the penult (the present ult.) post-tonic´‑ou- gives Ml. W.‑eu, Mn. W.‑au; thus the pl. endings *´‑ou̯es, *´‑ou̯a give W.‑eu, ‑au, as incadau ‘armies’ < *kátou̯es,dagrau ‘tears’ < Ar. *dák̑ruu̯ə; similarlyangau ‘death’ < *ánkou̯‑;cigleu ‘I have heard’ < *k̑ū́k̑lou̯a§ 182 i.

In this case‑eu does not affect a precedinga as it does when it is itself the result of affection§ 69 vi, as inteneu < *tanouís.

The above change may be due to a doubling of, see§ 62 i (2), thus *auu̯ > *üu̯ > *üü̯ > *öü > O. W., Ml. W..

(3)i̯ou̯- gives W.ieu (≡i̯eü). Thus W.ieuanc ‘young’ < Brit. *i̯ou̯ankos < Ar.i̯uu̯n̥k̑os: Lat.juvencus§ 100 i (1);—W.Ieuan < *I̯ou̯ánnes forlōánnes;—Mn. W.Iau, Ml. W.Ieu ‘Jove’ < Brit. gen. *I̯ou̯-os for Lat.Jovis;—Mn. W.iau, Ml. W.ieu ‘yoke’ < *i̯ou̯-ón < *juɡ-óm, seevi (1). Here we have the assimilation of to by which it becomesü; cf. the assim. of toü in‑i̯oü in O. W.,§ 25 i.

An alternative formi̯ef‑, if- appears in the penult:iefanc, ifanc;Iefan, Ifan. The latter is attested in the 14th cent:ivanghetc.m. 84. Later it is common:Pawb yn eu rhifynifanc S.C.,c. i 114 ‘all in their [full] number young’.

Ifanc, ifanca ofyn:
Henaint, at henaint y tyn.—S.Ph.br. iv 391.

‘The young seeks the young: old age is drawn to old age.’ The formi̯ef- is probably older, but cannot be verified; Ml. W.ieu- is ambiguous, but doubtless generally meanti̯eü‑. The latter form is seen in

Paham, a minneu ’nieuanc,
Yr wyf yn rhwym ar fy nhranc?—B.A. 133/77.

‘Why, when I am young, am I bound at death’s door?’ The dialects now haveif‑, asIfan,ifanc, buti̯enctid fori̯euenctid ‘youth’.

(4) The ante-vocalic formaw may occur before a consonant where the vowel after it has dropped, as in W.cawr ‘giant’ < Brit. *kou̯arós: Gaul. Καυαρος, Ir.caur (< W.?) √k̑eu̯ā‑. We also haveaw regularly for Lat.au, as inawdur < Lat.au(c)tṓrem;llawẟ ‘praise’ <laudem; Ml. W.Pawl <Paulus (the biblicalPaul is merely the Eng. form, and is pronouncedPôl).

W.nawn ‘noon’ < *nouna possibly dial. Lat. fornōna ( < *nou̯enā), cf. PelignianNounis ‘Nonius’, and Lat. old spellingnoundinum. Sir John Rhys suggests the influence of Brit. *nou̯an. See§ 81 iii (2).

(5) Except when affected as iniv (4),v (3)(5), Brit.āu̯ gaveu in W., as inbu ‘has been’ < *(be)bā́u̯e < Ar. *bhebhōu̯e§ 189 iv (3);caru ‘to love’ < *karā́-u-§ 202 ii. When unaccentedā was shortened,iii (1),§ 74.

iv. The penultimate affection of the diphthong has the formsew, yw, andeu; thus

(1) Beforeĭ orī remaining asy ori, it appears asew, as innewyẟ ‘new’, Bret.nevez < *nouíi̯os < *neu̯ii̯os;—W.cnewyll ‘kernels’ < *kneu̯‑: E.nut < *knu‑d‑;—W.ewythr ‘uncle’ < *au̯on-tēr: Lat.avun-culus < *au̯on- (nom. *au̯ō seev (5)); Bret.eontr (eo foreu;i lost), Corn.euitor (‑tortr̥?); W.ewyllys ‘will’ < *ou̯i‑, √au̯ei‑. Similarlyrhewin ‘ruin’ derived from the Lat.ruīna.

(2) Before when pretonic it iseu (≡), the being lost ; thus W.breuan ‘handmill’ for *breuon < *broui̯ón- (: Corn.brou, Bret.breo, Ir.brāu all from nom. *broui̯ō, Ir. gen.broon): Goth.quairnus, E.quern, √ɡerā‑.

But when accented it isyw as in ultimate aff. ; thuscyw ‘young of an animal’ < *kóui̯ō pl.cywi̯on < *kóui̯ones, seev (6);—distrywi̯af < *dī-stróui̯a-mi,v (2);—llywiaf ‘I steer’:llyw ‘rudder’ ib.

In late formations has no effect:gwrandawi̯ad ‘hearing’ fromgwrandaw ‘to listen’.

(3) Where it remained a diphthong before a consonantiii (4), its affected form iseu; thusceurip 94/179r. ‘giants’ nowceiri (inTre’r Ceiri, etc.) by§ 77 ix, pl. ofcawr; the usual pl.cewriw. m. 441,.a. 44 is a re-formation;—W.beudy < Brit. *bou̯i-tigos, a later formation than *boukoli̯osii (1) (cf. Lat.nāufragus, laternāvi-fragus);—Ml. W.Meuruc§ 77 viii <Mauricius, Ml. W.cyngheussaeth < *con-caus‑i̯‑act‑:cyngaws ‘lawsuit’ < Lat.causa.

(4)ā́u̯i̯ becameā́i̯, givingw͡y§ 75 i (3); asandwyo ‘to mar, spoil’ met. for *ad-nwy-o < *ati-nāu̯‑i̯‑,niwed ‘injury’ for *nwyet§ 78 iv <nā́u̯-i̯at- < *nōu̯‑i̯‑: Lith.novýti ‘to afflict’ < *nōu̯ī‑.

v. In the presentultima the diphthong, when affected, takes various forms, as follows:

(1) The ordinary affection is Ml. W.eu, Mn. W.au; this occurs:

1 . Before unaccented‑ī; asdau ‘two’ m., Ml.deu, O. W.dou < *dóu̯ī < Ar. *duu̯ō(): Gk. δύω, δύο, Lat.duo, Skr.duvā́(u);—W.tau ‘is silent’ < *tou̯īt < * (s)tup-ēit, besidetaw ‘be silent!’ < *tóu̯e;—W.cenau ‘whelp’ < *kanóuī < *kanóu̯ō: Ir.cana: from *k()ₑn‑: Lat.canis.

2. Before accentedĭ; as W.teneu ‘thin’ (Corn.tanow, Bret.tanao) < *tanouís < *tₑnuu̯ís: Lat.tenuis, Skr.tanúḥ f.tanvī́.

3. Before a consonant; as W.haul ‘sun’ < Brit. *sau’li̯ós < *sāu̯eli̯ós: Gk. ἠέλιος, ἥλιος, Dor. ᾱ̓έλιος Lith.sáulė, Skr.sū́rya‑ḥ, Lat.sōl < *sāuol < *sāuel. (´‑li̯- would have given W.II; hence we assume Brit.‑li´‑; see also§ 113 i (5).)

Ml. W.eur, Mn. W.aur ‘gold’ cannot be fromaurum which gave Ml.aur (≡awr), and Bret.aour. The Mn. W.aur, Ml.eur represents the adj. *auri̯os foraureus, which spread from expressions likemodrwy aur ‘gold(en) ring’, etc. The noun is seen inef guisgus aur (uw)b.a. 38 ‘he wore gold’.

¶ The above is the ordinary affected form, which is used e.g. in the formation of the 3rd sg. pres. ind. of verbs; thustereu ‘strikes’:taraw =saif:saf§ 173 iv (1). It is seen that when‑eu is the result of affection as above, ana before it is affected toe; seeiii (2).

(2)‑óu̯i̯- gives‑yw. Thus W.dilyw ‘flood’ (now generally misspeltdiluw) < *dīlóu̯i̯o- < Lat.dīluvium;—distryw ‘destruction’ < *dī-stróu̯i̯‑: Goth.straujan, Lat.destruo; the vb. isdistrywi̯afiv (2);—W.llyw ‘rudder’, < *lóu̯i̯o‑: Ir.lue < *lu-ii̯o‑: Gk. πλόος, √pleu‑;—W.clyw ‘hearing’ < *klou̯i̯- <k̑léu̯es-§ 75 vii (1).

There is no reason to suppose thatuu̯ becameiu̯ in Brit., as stated by Pedersen, Gr. i 61;yw is fromóu̯i̯ as above.Clywaf ‘I hear’ is a denominative fromclyw, cf.clywẏafc.m. 32 (the pres. stem of √k̑leu̯- meant ‘to be named’, andclywaf cannot come directly from it; cf. Meillet, MSL. xv 337).

(3)‑ā́u̯i̯- became‑ā́i- which gives‑w͡y§ 75 i (3); thus W.wy ‘egg’ < Brit. *ā́u̯i̯on < Ar. *ṓu̯i̯om: Gk. ὤιον, ᾦον, Lat.ōvum;—Cornwy <Cornā́u̯i̯‑(a);—Aethwyr.p. 1419 < *Oethwy§ 78 ii (3) <Octā́vius.

Pedersen Gr. i. 66 suggests that Ir.og is borrowed from W., but this is improbable, and does not help to explain the‑g. Thurneysen IA. xxvi 26 insists upon a Kelt. *ugos, *uges. The fact, however, seems to be thatu̯i̯ under certain conditions became in Ir. a spirant writteng; thus Ir.ugaire ‘shepherd’ < *ou̯i̯-ārius:ōi,ui ‘sheep’, Lat.ovis; Mn. Ir.ughachd ‘will’ < *ou̯i̯-akt‑, √au̯ei̯‑,iv (1). Eng.egg is from Icel.egg < Pr. Germ. *ajja- < *ōu̯i̯o‑.

(4)‑ou̯ī́ or‑ou̯i̯´- was similarly simplified to‑o-ī́,‑oi̯´‑, which gives‑w͡y; thus W.dwy ‘two’ f. < *dou̯ī́ < *duu̯ái: Lat.duae, Skr.duvé < *duu̯ái;—W.aswy ‘left (hand)’ < *at-sou̯i̯-ā́: Skr.savyá‑ḥ ‘left’.

‑wy as in (3) and (4) may be weakened to‑eu; asCerneu, asseu; these are not direct affections, as shown by the unaffecteda‑; also tow, assw, see§ 78 iii, i.

(5)‑āu̯ī́, ‑āuí- or‑āu̯i̯´- by the shortening of unacc.ā became‑au̯ī́, ‑au̯í- or‑au̯i̯´- simplified to‑a-ī́,‑a-í- or‑ai̯´‑, which gives‑oe. The simplification here was late, so that‑au̯ī́ did not, like‑asī́, give‑ei. It did not take place in Bret. and Corn., in which the groups appear as‑ou (‑ow). In W.‑oe generally becomes‑o,78 i (1). Examples: ‑(g)no in proper names;Iud-noel.l. 176, 187,Balch-noe D. G. 43;Gueithgnol.l. 144,Guiþno (wrongly wr.guipno)gen. v, Mn. W.Gwyddno;Mochnob.b. 61,BeunoỺ.A. 119, Mn. W.Tudno, Machno, etc., all < *‑gnāu̯i̯ó‑s: Lat.Gnaeus < *gnā-u̯i̯os < *g̑n̥̄‑, √g̑enē- ‘be born’. (With the accent on theā́ it gave‑nwy by (3), asMochnwyb.b. 47,Gronwy§ 78 i (2), weakened to‑neu, see (4), asGuitneub.b. 98, 106,Iudnoul.l. 73, 77, etc.);—W.clo ‘lock’ < *qlāu̯-ís: Gk. κληίς, Lat.clāvis, √(s)qlāu̯‑;—W.noe ‘large bowl’ < *nāu̯i̯ā́: Lat.nāvia, nāvis, Gk. ναῦς, Ion. νηῦς, Skr.nā́uḥ;—W.athro ‘guardian, teacher’, < *altrāu̯ī́ < *altrāu̯ṓ < *altro-au̯ṓ,§ 155 ii (1): *au̯on-iv (1); pl.athrawon, alltrawon < *altrāu̯ónes; f.elltrewyn ‘stepmother’ < *altrāu̯ónī; Bret.aoutrou ‘seigneur’, Corn.altrou ‘fosterfather’.

The mas. sg. isathro in all Ml. W. texts:b.b. 86;a.l. i 338;w.m. 128, 452–3;r.m. 100–1, 202;Ỻ.A. 3, 6, 49, 107, 113;r.p. 1225, 1241, 1255, 1345, 1348;r.b. 975; Io. G. 640, etc.; and in the early edns. of the Bible. The lateathraw (Salesbury, Dic.) is an artificial form deduced from the pl.Cae Athro (near Carnarvon) is so named locally;Cae-athraw is a misspelling which came through the Sunday school from late edns. of the Bible. The sg.alltraw is also artificial. (So in late W.cenaw is written forcenau in defiance of the pronunciation in all the dialects, which iscene orcena implyingcenau§ 6 iii.) Pughe's fem.elltrewen is his own invention;‑en would not affect the‑aw- to‑ew‑.

Other examples of the same development, though the orig. formation is not so clear in these, are—W.glo ‘coal’ for *gw̯loe < Brit. *gu̯lāu̯ís, √g̑u̯el(āˣ)‑: E.coal, Skr.jválati ‘blazes’;—W.gro ‘gravel’ < *grāu̯ís < *ɡhrōu‑, √ghreu̯‑: Lat.rūdus, E.grit;—W.tyno ‘plain, meadow’ for *tno, O. W.tnoul.l. 32, 44, 74, Bret.tnou (: W.teneu, √ten- ‘stretch’?).

(6) Doublets occur for several reasons.—1. Difference of accentuation in Brit.; thus W.gwryw ‘male’,benyw ‘female’ < Brit. *uiróu̯i̯os, *banóu̯i̯os, besideguru,banua.l. i 272 = Gwyn. dial.gwrw,banw for *gwrwy *banwy < *u̯irou̯i̯ós, *banou̯i̯ós.—2. Difference of ending, as inceneu ‘whelp’ < *kanóuō, see (1) above, besidecnyw ‘young of an animal’ < *k(a)nóu̯i̯ō, whence, by§ 101 ii (2),cyw ‘young of an animal’ pl.cywion < *k(n)óu̯i̯ones.—3. Difference of strong and weak forms; asasswy besideasseu andassw,Gronwy besideGronw,Cornwy besideCorneu, see§ 78.

Note. It is to be observed that‑ō does not producei-affection in Bret.; hence W. aff.dau, but Bret. unaff.daou, < *dúu̯ō. On the other hand W. aff.haul, Bret. aff.heol both from *sau’li̯os. The assumption usually made thatāu gives W.‑eu, ‑au based entirely upon these two words (taken as *d()āu, *sāu’l‑) does not explain the difference in Bret.

vi. (1)ug before a vowel >uu̯ in Brit. and developed like ordinaryuu̯ orou̯. Thus W.traw-af ‘I strike’ < *trug-ami < *prug- for *pu̯r̥g‑, √(s)phu̯ereɡ-§ 97 v (3);—after‑,iii (3), W.iau ‘yoke’: Lat.jugum, Gk. ζυγόν, Skr.yugám all < Ar. *jugóm;—beforei retained asy, W.llewych ‘light’ < *lug-isk‑;—before losti, O. W.poullor-aur, Ml. W.peullawrb.t. 25 ‘writing tablet’ < Lat.pugillāres;—before‑ū, W.go-lev ‘light’ < Brit. *u̯o-lugū.—Forug before see§ 104 ii (2).

(2) Butoug has the regular development ofou̯ before a consonant, and gives *üᵹ >ü, as W.llu ‘host’, Ir.slūag < *sloug‑,§ 95 i;—W.tru, tru-an ‘wretched’, Ir.trūag < *troug-os;—W.bu-arth ‘farmyard’ < *bou-gart‑: Lat.hortus§ 99 vi.

vii. In Brit.m between vowels or sonants was already loosened to nasalizedv or; after a vowel it is therefore treated partly as a consonant and partly as the second element of a diphthong.

(1)am generally givesaf as in the spv. ending‑haf§ 147 iv (2),hafal ‘like, equal’: Ir.samail§ 94 i; affected it gives mediallyef before a vowel,eif before,ef oreu beforen, as indefnydd ordeunyddf. 37 ‘material’ < *dam-níi̯o‑: Ir.damnae id., √demā- ‘build’;cyntefig ‘primitive’:cyntaf ‘first’;—finally,eu, as W.edau, edeu ‘thread’ < *etamī, O. W.etem (≡edɥṽ?), pl.edafeẟ < *etamíi̯ās < *petə‑, √petē‑; so Ml. W.gwelleu ‘shears’, Mn. W.gwellau pl.gwelleifiau; Mn. lit.gwellaif is deduced from the pl.;hynaif is doubtless analogical; sodrycheif, dyrchaif,§ 188 iii. The variant of‑eu is‑yf:crog-edyf ‘dropwort’.

(2)em givesef finally, as innef§ 100 v; mediallyef as ingefell ‘twin’ < Lat.gemellus; orỿf as inDyfed <Demeta; or (beforew͡y)y(w) as intywyll§ 111 i (2),tywyẟ§ 86 i (5); affected, finally,‑yf or‑eu; ascleẟyf orcleẟeu ‘sword’ < *klad-emō (cf. Gk. ἀκρεμών), √qolād‑; pl.cleddyfau a new formation; soneẟyf orneẟeu ‘adze’,§ 130 i.

(3)om givesof as indof ‘tame’: Lat.domāre; affected,yf as inSelyf§ 69 iv (1); before‑n- it givesaf as insafn ‘mouth’ < *stom‑n‑: Gk. στόμα; affected,eif as inceifn§ 75 vii (1), simplified toef,§ 78 v, in the improper compoundcefnderw, O. W. pl.ceintiru§ 137 ii.

(4)um before a vowel gives‑w(f),‑ỿf‑, as intw(f) ‘growth’,tyfu ‘to grow’: Lat.tumeo; beforen it givesaw,af orw, as in Ml. W.ysgawn > S.W.ysgon, alsoysgafn, Mn. W. and N. W.ysgafn, Ml. W.ysgwn§ 101 iv (3).

viii. (1) After the prefixes *ko‑, *to‑, *do‑, *ro- an initialu̯- was heterosyllabic, and theo of the prefix becomesỿ regularly§ 65 iv (2), as incỿ-wir ib.,tỿ-wysog ‘prince’,dỿ-wedaf§ 194 ‘I say’,rhỿ-wynt ‘hurricane’.

(2) The vowel also develops regularly beforegu̯, as intew ‘thick’, Ir.tiug < *tegu̯‑: E.thick; cf.ii (2).

ix. (1)iu̯ andīu̯ occurred as V- and R-grades ofei̯eu̯,ei̯ēu̯, etc.; Kelt.īu̯ also < Ar.ēu̯. Thei orī appears regularly in W. asy ori. Thusbyw ‘live’ <ɡiu̯-§ 63 vii (3);—W.lliw < *līu̯‑: Lat.līvor.

(2)īu̯o gives W.üo orüa (§ 65 v (1)); thus W.buan ‘quick’ < Brit. *bīu̯o-no‑s§ 63 vii (3);—W.hual ‘shackle’ for *ffual < *fīu̯ol < Lat.fībula;—originalīo >īu̯o >üo as inlluossawc§ 75 iii (3).

Later Modifications of Vowels.

§ 77. i. In Late Ml. W.ɥ, when short, becamei beforeg (then writtenc) orng. Thus in the unaccented ultima in Late Ml. W. we generally findic, sometimesing, astebicw.m. 122, 129, 142,r.m. 164, 213, etc.;meddicw.m. 141,r.m. 113, 212, 306,r.p. 1298;kyvingr.m. 110 (butkyvyngw.m. 46, 465,r.m. 32). Such words are rhymed by the bards with monosyllables havingi (notɥ):

Ond dychmygion dyniondig,
A cham oedd pobdychymig.—D.G. 22; see 246.

‘[They were] but jealous men’s fancies, and every fancy was false.’

Rhinwedd mab Ieuanfeddig
Ar dy rudd fal aur adrig.—L.G.C. 348.

‘The virtue of Ab Ieuan the physician will dwell as gold on thy cheek.’

A’ifrig yndebigi dân.—D.E.,g. 125.

‘And its tips like fire.’—To a woman’s hair. See 133, and D.G., 27, 285. Seetebig / diwig / cerrig / llewig / rhyfig, etc. E.P. 283.

In a monosyllable beforeg the vowel is long,§ 51 iii, and therefore remainsɥ, as inplŷg ‘fold’,crŷg ‘hoarse’ seeplycw.m. 89,r.m. 65; but beforeng it is short,§ 51 ii, henceing ‘anguish’, which is forɥngr.p. 1286, 1407; cf. the derivativeỿg̃derr.m. 119.

The only words in which the vowel is soundedɥ in the unaccented ult. are compounds ofplyg, cryg etc., asdyblyg D.G. 258, (g)w͡yrblyg 255,ogryg 244,deuẟyblycỺ.A. 68; also the 3rd sing. pres. ind. of verbs with stems ending in‑og, asysgyg D.G. 370 ‘shakes’ (though we haveennicc.m. 13 fromannoc ‘incite’). In other cases the sound isig. The late Mn. spellingsmeddyg, tebyg etc., are purely artificial, deduced frommeddỿgon, tebỿgu etc. A few words of this class are still written phonetically, ascerrig.

The sound wasɥ in Early Ml. W. as shown by the rhymecerryg / plyg C.m.a. i 241, and the assonancemetic / bidb.b. 76 (≡meẟɥg / bɥd); andỿ the mutation ofɥ remains in the penult. Hence we have two forms: (1)‑ig for‑ɥg <‑ĭc‑, which becomes‑ỿg- in the penult; (2)‑ig for‑ig <‑īc‑, which is‑ig- in the penult. Thus (1)meddig < Lat.medĭcus, pl.meddỿgon, (2)lleithig < Lat.lectīca, pl.lleithigeu. In Mn. W. one or two words of the second class have passed over to the first:perigl ‘danger’ < Lat.perīc’lum;cynnig ‘to offer’ < Lat.condīco, though still soundedperigl,cynnig are writtenperygl,cynnyg because, by false analogy, derived forms have come to be sounded withỿ asperỿglus, cỿnỿgi̯af. In Ml. W. the penult hadi in these, asperiglwysr.b.b. 44–5,periglusỺ.A. 146,berigleur.b.b. 121,gynigwytw.m. 168,gynnigẏwytr.M. 234,kynnigẏwys do. 144.

ii.ɥ becomesi in the unaccented ult. in some cases afterg orng; thusergytw.m. 110, 111 ‘shot’;ergitr.m. 80, 81,r.b.b. 42, nowergid (writtenergyd);efengil R.Ỻ.,f. 5, E.P. 278 ‘gospel’; so sounded now though writtenefengyl;megis / dis, D.G. 315;cregin for *cregyn. But as a ruleɥ remain ; egyr ‘opens’,diogyn ‘idler’,negydd ‘denier’,dengys ‘shows’, are so pronounced, owing to the influence of analogical forms withoutg orng.

iii. In the same positionɥ frequently becomesi after penultimatei orei; thusllinɥnw.m. 75 ‘string’, butllinin four lines earlier, also 78,llininr.m. 54, 56 (each time),dibinc.m. 91 ‘hang’,amẟiffinp 21/1r. ‘to defend’,gwlithinw.m. 455,r.m. 102 ‘dewdrop’,giliẟw.m. 9, 134 ‘other’,originỺ.A. 122 ‘a moment’, (double dim. ofawr ‘hour’),dilind.g. 343 ‘to follow’.

Derfel wrth ryfel a thrin
Dewr oedd, a da iwreiddin.—D.I.D.,g. 178.

‘He was a brave Derfel in war and encounter, and of good stock.’

Herwydd nas gwnaiddyhirin
Fentri̯o i oes o fewn trin.—S.T.,g.r. 369.

‘Because a dastard would not risk his life in battle.’

But analogy has always tended to preserve the termination‑yn:

Ysbys y dengys y dŷn
O ba radd y bo iwreiddɥn.—T.A.,f. 33.

‘Plainly does a man show of what degree his origin is.’

O chyrch dyrfa, deca’ dŷn,
Daw i’w haraildihirɥn.—D.G., 173.

‘If she hies to a gathering, fairest maid, a knave comes to watch her.’

The sound is nowi, asllinin,dibin,amddiffin,giliẟ,gwreiẟin, etc.; they written is an etymological spelling. Sometimes it is wrongly written, as iners meityn forers meitin§ 70 v. This may also occur in Ml. W. as inyr meitynw.m. 17,r.m. 11 beside the correcter meitinw.m. 128, 138,yr meitinr.m. 280, cf.meitin/ffinb.a. 18;dilyt besidedilitw.m. 41.

iv. In Mn. W.ɥ followed byi in some common groupings becamei; thuscerɥ di ‘thou lovest’ becameceri di, andceri supplantedcerɥ as the regular form. Sowrthɥf i, wrthɥt ti becamewrthif i, wrthit ti, and the 1620 Bible haswrthif, wrthit; sogennif, gennit; but later the Ml. forms withy were restored in writing. [The dialects developed new formations.]

v. The diphthongɥw is now soundediw after front consonants: afterc (≡) incywk̑iw (but pl.cỿwi̯onqəui̯on), afterr inrhyw and its compoundsamryw, cyfryw, etc., indryw, ystryw, gwryw, aftern inbenyw, and initially inyw ‘is’,yw ‘to his’ now writteni’w. (Gwryw, benyw, yw ‘is’ are not dialectal forms in N. W., but are sounded with‑iw in reading or quoting.) Ml. W.nywlr.m. 46,w.m. 64 ‘fog’ is now writtenniwl,§ 37 ii. Indistryw, dilyw the‑iw sound is earlier, on account of the precedingi; both are often spelt with‑iw in Ml. W. Afterd and the soundiw is still earlier; thusydiw,heẟiw are so spelt in Ml. W. in MSS. wherei andy are distinguished.

The only words remaining now withɥw arebɥw, clɥw, llɥw ‘prince’ andllɥw ‘rudder’ (also soundedlliw),gwyw ‘withered’ in addition toDuw which is soundedDɥw in Late Mn. W.; and compounds of theselledfyw, hyglyw, etc.

vi. In the Mn. languageɥ in the unaccented ult. is soundedi beforell in some words; ascyllyll ‘knives’,gwyntyll ‘fan’; in some, ascandryll ‘shattered’ (lit. ‘100 bits’), bothɥ andi are heard; others haveɥ always, assefyll. This modification sometimes appears in late MSS.; but is not recognized in the rhymes of the bards.

vii. (1) In Ml. W.u (≡ü) was unrounded toi after the labial ingovut ‘pain’; the usual Ml. form isgovutw.m. 138 l. 15; 231; butgofitw.m. 138 l. 4; 131, 141, etc.; Mn. W.gofid.

(2) In a few casesɥ came to be rounded after a labial; thuspump ‘five’ for an earlierpɥmp, O.W.pimp,bustl ‘gall’ for *bɥstl: Bret.bestl (Bret.e = W.y§ 16 iv (2)).

viii. As it was difficult to pronounce unroundedi orɥ and roundedü in consecutive syllables, assimilation took place: *iẟunt ‘to them’ (cf.iẟaw ‘to him’) becameuẟunt and always appears so in Ml. W. seea.l. i 2;p 17/1r.;Ỻ.A. 7, 8, 11, 21, etc.;w.m. 6, 26;r.m. 4, 7, etc. The natural sound in Gwyn. isuẟun, though the artificial Mn. lit.iddynt and the analogy ofiddo may have influenced the pronunciation of some speakers. Similarlyei becameeu, as inr͑eudusw.m. 21,r.m. 13,r.p. 1238 forrheidus ‘needy’;teulu ‘household troops’ forteilu, the form implied in the spellingteylu ofa.l. i 2, 12, etc.;eulun often later foreilun, and now soundedeulun. In the reverse order we have Ml. W.Meuruc forMeuric.

ix. In Mn. W.u having come to be soundedɥ, it becomesi in those positions whereɥ would be so treated: thusbarrug, esgus, cynnull are soundedbarrig, esgis, cynnill.—D.G. rhymesmenig / sarrug 8.—Before ori it is soundedi. Dr. M. writesiniawn Job i 1; we now sayini̯on ‘straight’ foruni̯on,inig forunig,tostirio fortosturio, etc. Hencecarut ti becamecarit ti, and‑it in Late Mn. W. replaced‑ut as the 2nd sg. impf. ending.

x.u being rounded in O. and Ml. W., finalch after it retained its rounding; thusuch ‘higher’ ≡üch, sometimes writtenuwch in Late Ml. W.; when theu was unrounded the glide remained, and the sound becameɥw̯ch as implied inywchr.p. 1295; this is the present sound; it is writtenuwch in Mn. W. But in the penult we haveuch, as inuchel ‘high’. Hence the mutation,uw: u,§ 81.

xi. The modern pronunciation cited in this section is that of Gwynedd, where the soundɥ oru is quite distinct from the soundi.

§ 78. i. (1) The diphthongoe oroy, O. W.oi, remains finally in only two words:noe ‘basin’,doe ‘yesterday’; Ml. W. hadmoe ‘more’ also. Elsewhere it is regularly reduced to‑o, as increto ‘may believe’ for *cred-hoe appearing ascred-doeb.b. 53, a stray survival,§ 183 ii; and in‑no in personal names for‑noe,clo for *cloe, etc.§ 76 v (5); inam-do ‘shroud’ for *am-doe§ 104 ii (2);th or may be lost after it as inheno ‘to-night‘ < O. W.henoidjuv. sk.henoethr.p. 1040; it becameui by assim. inhunnoidox. >hunnuidm.c. > Ml. and Mn. W.hwnnw; andhinnoid gavehynny by analogical assimilation (‑d‑ẟ in O.W.). A late example isy ddannodd ‘toothache’ < Ml. W.y ẟannoeẟ§ 75 iv (2), in which however the final‑ẟ remains.

Final‑aeth >‑a in the same way inyna, etwa forynaeth, etwaeth.

(2) Similarlyw͡y, O. W.ui, may be reduced tow; cf.hwnnw above. Thusllw ‘oath’§ 104 ii (2);Gronwyw.m. 110, 111 >Gronw do. 101, 104, 105;Gronwy, Goronwy for *gw̯ronwy < *(i)ro-gnā́u̯i̯os§ 76 v (5); both forms survived:Pont Ronw (Llanedwen) is calledPont Ronwy by some, but whether the latter is of lit. origin is difficult to decide. Soassua.l. i 144 (≡assw) <asswy ‘left’;—guru, banu < *gwrw͡y, *banw͡y§ 76 v (6);—raccw § 210 x (3)[W 3].—Before a consonant: aor. 3rd sg.‑w͡ys >‑ws§ 175 i (5);tyngwtb.a. 4 fortyngw͡yt;adeilwt, rannwtg.c. 106, 108; and doubtless impf. 1st sg.‑wn is for an earlier *‑w͡yn§ 180 iii (1);‑wn for *‑w͡yn§ 215 iii (1). Somwrthwlw.m. 46,r.b. 968, D.G. 430,myrthwlr.m. 32 besidemortuylb.ch. 77,morthwyl, mwrthwyl D.D.,morthwyl Bible, spoken lang.mwrthwl pl.mỿrthw͡ylion. Late Mn. W.neithi̯wr ‘last night’ <neithi̯w͡yr§ 34 ii, Ml. W.neithẏwyrs.g. 43.

Some cases occur of the late substitution ofwy forw:madws ‘high time’w.m. 22,b.m. 14 (: Sequ.matu.., Lat.mātūrus) is given by Wm.S. and D.D. s.v. asmadwys, which is not attested;—cyfarws,w.m. 454, 459–60, latercyfarwys, see Silvan Evans s.v.

ii. (1) In some wordsoe in the ultima was reduced toe, andw͡y toy; thus *nammoen ‘not more [than]’ becamenamenb.a. 15, 16 ‘only’, andnamwynr.p. 1056 gave Ml. and Mn. W.namyn ‘but, except’,§ 222 iii (3);—*mahar-oin (variantmaharuin,b.s.ch. 3), Early Ml. W.maharaena.l. i 278, Ml. and Mn. W.maharen ‘ram’, pl. *meheruin >meheryn;mahar- < *mas-ₑro- ‘male’: Lat.mās, suff.§ 153(5), +oen§ 65 ii (2);—*adwoen (writtenadwaen but rh. withhoen, poenb.b. 70) >adwaen, adwen ‘I know’;—brenhinoetb.b. 53 >brenhineẟ, but‑oeẟ remains in N. W. and Mn. Lit. W.;—socefnderweẟ, ewythreẟ. The change seems to be due to unrounding by dissimilation with a labial in the word (teyrneẟ followed the synonymousbrenhineẟ). Later examples areCawlw͡yd, Mawddw͡y now soundedCowlɥd, Mowddɥ; cf. alsoa(w)w͡yr >aw̯ɥr, etc.§ 38 x.

(2) After a labial O. W.oi > Ml. W.ae; as O. W.guoilautb.s.ch. 6 > Ml.gw̯aelawt, Mn.gw̯aelod;—W.gw̯ae ‘woe’ for *gw̯oe < *u̯ai: Lat.vae, Goth.wai;—W.gwaeẟ ‘cry’ for *gw̯oeẟ, Ir.fāed < *u̯aid‑: Lith.waidi ‘lamentation’;—baeẟ ‘boar’ < *boeẟ (writtenbaeẟ but rh. withoeẟb.t. 26, l. 17).

Afterg‑,oi (oe, oy) becamew̯ay,w̯ae as ingw̯ayw̯ ‘spear’ for *goyw̯§ 75 vii (3) writtengvaev but rhyming withgloev (gloyw̯)b.b. 72;—gwaed ‘blood’ for *goed = Bret.goad, Leonhe c’hoad ‘his blood’ (c’h <g); seegwaet rh. withcoet, eirẏoetr.p. 1046.

(3) In the penultoi (oe) becameae beforew͡y inaelwyd ‘hearth’: Corn.oilet, Bret.oaled§ 104 iv (3);—Aethw͡y< *Oethwy§ 76 v (3).

iii.‑w͡y, or rather Early W.‑ui, was liable when unaccented to be weakened to > Ml. W.eu; thuseu ‘their’ for *wy from *eisṓm§ 160 iv;—meu, teu§ 75 viii (2),§ 161 iv;—pi-eu ‘whose is?’ witheu for *wy < *eset§ 179 ix (3),§ 192;—asseu, Corneu, Guitneu, Iudnou76 v (4), (5);neu§ 219 i (2).

iv. (1)ui (w͡y) finally or before a vowel was liable to be metathesized toyw; as inyw ‘is’ for *wy§ 179 ix (3);—yw ‘to his, to her’ for *w͡y§ 160 iv (2);—nyw ‘who…​not…​him’ for earliernuy§ 160 ii (2).—After a dental it becameiw,§ 77 v, as in Ml. W.ydiw ‘is’ for *yd-wy;—W.niwed ‘harm’ for *nw͡yet§ 76 iv (4).—In Bret. and Corn. this metathesis was carried further: Bret.piou, Corn.pyw, pew: W.pwy ‘who’, etc.

(2) This might happen before a consonant also; but in that case *yw becameü; thus *dw͡yw̯ ‘god’ > *dyw‑w̯ >duw; the form *dwyw is attested inb.t. 10, where, though speltduw, it rhymes withplwyw (=plwyf?); and it remained in all derivatives, as O. W.duiutit ‘divinity’, Ml. W.dwywes ‘goddess’,dwywawl, Mn. W.dwyfol ‘divine’; the formsduwies ‘goddess’,duwiol ‘pious’ etc. are late deductions fromduw;—similarly Early Ml.W. verbal noundeweduyta.l. i 146, 152, etc.gwedy dywedwydw. 15a ‘after saying’ > Ml. W.dywedut ‘to say’; thew͡y remains indywedwydatw.m. 63,r.m. 45 ‘saying’,dywedwydẏats.g. 171 ‘babbler’.

v. In the penultoe, ae, ei tend to becomeo, a, e respectively before two consonants, more especially in Mn. W.; thusotvar.p. 1208,s.g. 303, Mn. W.odfa foroedfa ‘appointment, meeting’; Mn. W.addfed ‘ripe’ foraeddfed, Ml. W.aeẟvetw.m. 73,Ỻ.A. 166,r.b.b. 175; Mn. W.glendid for Ml. W.gleindit ‘cleanliness, beauty’. (Dial.gwergloẟ forgweirgloẟ,cosnoth sgernoth forcoesnoeth esgeirnoeth.)

ae >a inaeth-§ 108 iv (2).

vi. In the ult.ae sometimes becamee§ 31.

§ 79. i. (1) Old and Ml. W.ei appears asai andei in Mn. W. With some exceptions,§ 81 iii (1),ai appears in the ultima and in monosyllables, andei (pronouncedəi§ 29 iii) in other syllables. Thus Mn. W.ai stands in the syllable generally accented in O. W., andei in the syllable then unaccented. The natural inference is that the Mn. mutationei/ai is an exaggeration of a difference in the pronunciation ofei going back to O. W.

(2) O. W.ei was originallye̦i with open,§ 69 vii. But in unaccented syllables it came to be soundedẹi to avoid lowering the tongue to and raising it again toi in the short time available. The same thing took place in accented syllables ending in a group of consonants, asbeirẟ, since the time required to pronounce the consonants left less time to sound the diphthong. But in accented syllables with a simple or no consonantal ending thee̦i remained. Ml. W.ei therefore representedẹi ande̦i; the former gave Mn. W.ei, soundedəi; the latter gaveai. The old distinction is reflected in the Gwynedd pronunciation of a preceding guttural:ceiniog, ceirch are soundedk̑əini̯og k̑əirch; butcaib, cais areqaib, qais; the velar and palatal alternate in the same word:qaib, k̑əibio; it may be added that before ordinaryỿ (≡ə) the consonant is the velar, thuscybyẟ, cynnar areqəbɥẟ, qənnar. It is seen therefore that the first element ofəi must be from close, for it differed from that ofai which comes from open, and also from the oldỿ (≡ə). The present soundəi seems to be as old as the 16th cent., forrhəir contracted forrhỿ-hir (rhə-hir) is writtenrheir ing.r. 101. The present soundai is at least as old as the 14th cent.:gwnai (<gwnaei) is rhymed withdelei inr.p. 1271 by M.D., and withdiveir.p. 1293 by G.V. The oldest appearance of thespellingai seems to occur in the Red Book:benn r͑aithr.p. 1194,diwair do. 1200,kain 1205,arynaic 1227,kain, main 1318; but Norman scribes heard thee̦i asai much earlier, to judge by such a form asTrefwalkemay in the Extent of Anglesey dated 1294 (Seebohm, Trib. Sys. ¹App. 10), Ml. W.Gwalchmei, Mn. W.Gwalchmai.

ii. O. W.ou (≡) has a somewhat similar history. Theo was probably close in unaccented and open in accented syllables. In Ml. W. it was unrounded in both cases, giving a closeə and an openə, both written e, so that the two sounds of the diphthong were writteneu. The closeə remains in Mn. W.eu, soundedəu; the openə gavea in Mn. W.au. That the former was a closeə and not a close is shown by the fact that in Gwyneddceunant, ceulo are soundedqəunant qəulo. The two soundseu andau occur in the same positions in the word asei andai respectively; see§ 81.

Vowel variation in Modern Welsh

§ 80. The above are the changes that have taken place in vowel sounds. Many of them depend upon accentuation or the influence of neighbouring sounds; hence in the Mn. language a vowel may have its original sound in one form of a word, and a changed sound in another, or two different changes of an original vowel may appear in two different forms of a word. It will be convenient now to bring together the more important variations of the same originals that occur in Mn. W.

Vowel Mutation.

§ 81. i.Vowel mutation is the regular alternation of vowels and diphthongs according to their position in a word. Certain sounds occurring in the ultima and in monosyllables are regularly modified in other positions.

The following is a table of the vowel mutations (numbered for reference). The numbers in the last column indicate the sections where the changes resulting in the mutation are dealt with.

No.In final,
and mono-,
syllables.
In
other
syllables.
Examples.§
1aieiadail,adeilad;caib,ceibio79 i.
2aueuhaul,heulog;aur euraid79 ii.
3awotlawd,tlodion,tlodi,tlotaf71 i.
4wỿtrwm,trỿmnion,trỿmach66 i.
5ɥỿbɥr,bỿrion,bỿrder66 i.
6uwubuwch,buchod,buches77 x.

As a general rule the respective forms appear only in the positions indicated. The exceptions are noted below.

ii. There is no exception to the rule thatai andau appear asei andeu in the penult. Such forms asdaiar, graian, haiarn, rhaiadr, traian, cauad, cauodd, gauaf, cynhauaf are not exceptions but misspellings ofdaear, graean, haearn, rhaeadr, traean, caead t caeodd, gaeaf, cynhaeaf, the diphthongae (also writtenay§ 29 ii) being one which does not undergo mutation in Lit. W., but remains the same in all positions (unless affected§ 70 iii). Seedayarr.m. 4, 5, 73, 78, etc.,w.m. 100, 456, 459,daearb.b. 70,w.m. 107,r.m. 97,gaeafr.b.b. 277,r.p. 1269,kynhaeafw.m. 73,r.m. 53,r.b.b. 271,p 14/11r.,kynhayafb.t. 8,haearnr.m. 118,hayarn 119,r͑aeadɏrr.p. 1255. The sound is attested in cynghanedd lusg:

Cyfled i chaeâ daear.—D.G. 205.

‘Her demesne is as wide as the earth.’

Ba le mae’r gorsied gaead?—L.G.C. 372; cf. 28, l. 1.

‘Where is the closed gorget?’

The spellingai, as indaiar, used by Salesbury and in the early Bibles, is a mistranscription of Ml. W.ay, due to the fact that Ml. W. y sometimes represents,§ 25 iii. (Salesbury hasdayar also, andgayaf always.)gauaf is phonetically correct now thatu has come to be soundedɥ, so that the error is only an orthographic one exactly similar to writingdun fordɥn ‘man’. Incauodd etc. the error was suggested by the fact that the verbal noun iscau ‘to shut’, a contraction ofcay|u orcae|u§ 33 iv. Such spellings as the latter-daytraithawd for the usual and correcttraethawd are due to bungling etymological theories. Pedersen, Gr. i 67, imagines from these false spellings that the difference between andai is small in diphthongs and vanishes where the second element is heterosyllabic. It is not heterosyllabic in these diphthongs, see§ 54 iv; andɥ andi are perfectly distinct wherever the dialect distinguishes betweenɥ andi as vowels. The possible forms in the penult areae, eu, ei, now sounded in Powysaɥ, əɥ, əi, and in Gwyneddəɥ, əɥ, əi. No one in Powys or Gwynedd sounds ani indaear.

iii. The exceptions to the general rule are the following (‘ultima’ being understood to include ‘monosyllable’):

(1)ei occurs in the ultima when followed by two consonants, or byl forlᵹ,r forrr, thusbeirdd ‘bards’,teifl ‘throws’,eithr ‘except’,gweheirdd D.G. 20 ‘forbids’,meirw̯ pl. ofmarw ‘dead’,deil ‘holds’ for *deilᵹ,ceir ‘cars’, pl. ofcarr. Beforel usage varies:lleill ‘others’,y naill ‘the one’,ereill oreraill ‘others’. In polysyllables it sometimes occurs beforem orch;dychleim Gr.O. 90 ‘leaps up’,myneich ‘monks’. Butai appears beforenc, nt, sg, ascainc ‘branch’,maint ‘size’,henaint ‘old age’,braisg ‘thick’; also inAifft, enghraifft, aillt.

As a contraction ofe-i the diphthong is now written and spokenei (that isəi̯), asceir, gwneir; butai was common formerly, ascair, gwnair.

eu is now commonly written, when absolutely final, in polysyllables, except when it is a plural or pronominal ending; asgoreu, goleu, dechreu forgorau, golau, dechrau. It survived from Ml. W. under the influence of dialectal‑e, and its use was extended in the 19th cent. because of an idea that‑au suggested the pl. ending.

In Ml. W.ei andeu appear in all positions, so that the mutation is not represented in writing,§ 79. But‑é-u,‑á-u were distinct, as are contracted‑éu, ‑áu now:diléu, parháu,§ 33 iv.

(2) The mutationaw: o is not of general application. The penultimateo does not come from the ultimateaw, but both come fromɔ; see§ 71 i. Hence whenaw is an original diphthong < Brit. or Lat.ou orau, it remainsaw in the penult, as inawdur < Lat.au(c)tōrem; socawgiau pl. ofcawg < late Lat.caucus;awydd, etc.§ 76 iii;canawon, athrawon,§ 36 iii. This showsnawn which givesprynhawnol, prynhawngweith, etc., to be from *nouna§ 76 iii as opposed toawr which givesoriau, oriog, etc., and is from *(h)ō̦ra§ 71 ii (3). In late formationsaw <ā is unmutated as inmawrion§ 144 iii (1),ardderchawgrwydd besideardderchogrwydd. Before a consonant, penultimateaw is soundedəw, and sometimes writtenow, ascowgiau D. 40,ardderchowgrwydd.

Where Ml. W. aw in the unaccented ult. has becomeo,§ 71, the mutation of course disappears; thus it appears in Ml. W.pechawt, pechodeu, but is lost in Mn. W.pechod, pechodau. Where at the same time theaw represents a Brit. diphthong, as ingwrando, gwrandawaf, the rule of mutation is reversed. So in final‑o for affectedau, inathro, athrawon§ 76 v (5).

(3)w appears in the penult in some words; see§ 66 ii, iii. For other exceptions to mutations 4 and 5 see§ 82.

(4) The mutationuw: u occurs only beforech,§ 77 x. In late formations it is neglected; thus besidelluwch ‘(snow)drift’, we have the oldlluchio ‘to hurl’, and the newlluwchio ‘to drive (dust or snow)’. For the derivatives ofduw see§ 78 iv (2).

(5) On unmutated forms in loose compounds see§ 45 ii (2).

82. i. From the table in the above section it is seen that the use of the two sounds ofy is regulated by the law of vowel mutation. The general rule in its special application to these sounds may be stated as follows:

y has theɥ sound in monosyllables and final syllables, and theỿ sound in all syllables not final; asedrɥch, edrỿchwch, brɥn, brỿniau, mỿnɥdd, wỿnỿddoedd, bỿrddau, prɥd, prỿdferth, dỿfod, cỿfỿngder.

ii. The exceptions to the rule are—

(1) A few proclitics, which, though monosyllabic, have theỿ sound. These areỿr, ỿ ‘the’,ỿn ‘in’,fỿ ‘my’,dỿ ‘thy’,ỿn ‘our’,ỿch ‘your’,mỿn, ỿm ‘by’ (in oaths).

Pre-verbalyẟ, yr, y (whether the relative,§ 162, the affirmative particle,§ 219 ii, or the conjunction,§ 222 x) is now always sounded withỿ. Inb.ch. it is regularly writtened, e (implyingỿẟ, ỿ,§ 16 iii); seea.l. i 2, 4, 6, 12, etc. But in the 15th cent. and later it was often writtenir, i, asI'r tri oessawlira’r teirswyẟ, L.G.C.r.p. 1412,o Vranideuan do. 1411. J.D.R. and D. regularly write it withy (≡ɥ); but Dr. Davies later in his D.D. (opp. p. 1) says that the sound isỿ. The explanation doubtless is that it was originallyɥ andỿ according to the accent; and both survived, theɥ becomingi (like the preposition,§ 16 ii (3)). It is often non-syllabic after a vowel in poetry; if its vowel is written it must be read asi orɥ forming a diphthong with the preceding vowel,§ 33 v.

Hen(e)iddio ir wy’, hyn oedd reid.—I.G.,p 53/91 r.
Ac yno ytrîc enaid Rrys.—H.D. (auto. ?),p 67/149 r.

But it is most commonly elided, in which case we have to assume that the lost vowel wasỿ,§ 44 vii (1).

Astudio’dd wyf, was didwyll.—An.,p 54/27 r. (15th cent.).
Meddylio’r wyf, mau ddolur.—G.C.,p 64/122 r. (Auto. S.V.).
Thomas ddulas, lle’dd elwyf.—H.D. (auto. ?),p 67/212 r.
Ac yno’trîc enaid R(h)ys.—R.C. (auto.),p 68/19 r.

ys as a proclitic isỿs, asỿs gwir ‘it is true’, often’s gwir§ 221 iii; when accented it isŷs ‘there is, people are’.

Llenwi, dros yr holl ýnɥs,
Dagrau ar ruddiau yrŷs.—Gut.O.,a 14967/120.

‘Over the whole island, there is a shedding of tears on cheeks.’

(2) The old formsɥmɥ, ɥttɥ, ɥmi, ɥtti, etc. ofimi, iti, etc. hadɥ in the penult,§ 212 ii.gɥda also hasɥ; but this is forgɥd â, Ml. W.ẏ gyt a,§ 216 ii (2). G.R. writes itgida and J.D.R.gyda (hisyɥ); both these pronunciations survive.

(3) Non-ultimatey before a vowel is now mostlyɥ; but originally it wasỿ regularly, for it may come fromo as indỿ‑,rhỿ‑, or was followed by so that at first there was no hiatus. In many cases theỿ was assimilated to the following vowel§ 16 iv (4), and contraction took place; thus Early Ml. W.deodreven (≡dỿodrevɏn)a.l. i 80 >doodreven do. 94 > Mn. W.dodrefn ‘furniture’;r͑ỿodresb.a. 5 >r͑ootdresr.b.b. 195 >rhodres ‘pomp’;kỿoeẟr.p. i 206 > *cooeẟ >coeẟ§ 41 v;gwelỿeu >gwelỿau >gwelâu Ps. cxlix 5, Can. vi 2 (1588 and 1620),b.cw. 23 ‘beds’; *cỿd-dỿ-un > Ml. W.cỿt-tu-un > Mn. W.cỿtū́n§ 33 iv;dỿlỿed > *dỿleed >dỿlḗd§ 199 ii (2). But it also remained unassimilated, as inhundyeur.m. 4,dylyet do. 5,camlyeur.p. 1297. In that case it tended to becomee§ 16 iv (2), thusdeunr.p. 1217,deall besidedallt I.D. 12, N.W. dial.dâllt;godreon besidegodryon§ 65 ii (3);darlleaf§ 203 iv (3); or was raised toɥ, which broke up later intoəɥ (writteneu); thusgodreuon J.D.R. [xxi] forgodryon,lletteuodd Gen. xxxii 21,dyleuaf so printed ind.g. 35, besidegwelyeu J.D.R. (whoseyɥ) [xiv, xix],dhỿlyei [xix, xxi],dhỿlyedic [xvi, xix]. Cf.rhɥ <rhỿ§ 65 iv (2). (Buthɥawdl is a misspelling ofhuawdl, Ml.W. huawdɏlr.p. 1301.) It is probable that the misspellingboreu forbore§ 31 ii 2 sprang fromboreuach the debased form ofboryach.

(4)yw follows the rule, asbɥw, bỿwɥd, bỿwi̯og;clɥw, clỿwed;llɥw, llỿwɥdd, llỿwi̯o, etc., except in late formations, especially from forms in whichɥw becameiw§ 77 v, as inamrɥwi̯o (soundedamriwi̯o),distrɥwi̯af (distriwi̯af), etc. J.D.R. writesamrywio (yɥ) [xvi],distriwiaw [xix]. Soniwli̯og, niwloedd§ 37 ii.

(5) The rising diphthongw̯y follows the rule:gw̯ɥrdd, gw̯ỿrddi̯on, etc.; butw̯ỿ generally becomesw§ 66 ii. Such a pronunciation asgw̯ɥntoedd is recent; butgw̯ɥwo ‘to wither’ may be old, as theɥ may have resisted mutation between two’s.

(6) The rule does not apply to the falling diphthongw͡y, in which the y is consonantal. In thisy must necessarily beɥ always, asmŵyn, mw͡ynach, mw͡yni̯on; and theɥ remains when thew͡y is mispronounced asw̯ɥ§ 38 iii, asGw̯ɥ́nedd forGw͡ynedd.

iii. (1) In the wordssylw, gw̯yry, the finalw andy were non-syllabic§ 42,§ 110 ii; hence they isɥ; thussɥlw (butsỿlw̯i),gw̯ɥrỿ.

(2) With the exceptions mentioned inii, the sound which is now common toy andu, if it occurs in the penult, is to be writtenu. Thus we writemunud, munudau, papur, papurau. Following this rule the translators of the Bible were misled by the late disyllabic pronunciation ofsylw to write itsulw; in late editions this error is corrected.

(3) In a few casesu in the ultima has come to be mistaken fory and mutated toỿ in the penult; as inysgrythyrau in the Bible (butysgrythurau correctly in the 1727 edn.) pl. ofysgrythur (< Lat.scriptūra) regarded asysgrythyr on the analogy ofllythyr ‘letter’. Sotestynau fortestunau, Early Mn. W.testunion, pl. oftestun (<Lat.testimōnium) treated astestyn;corynau ‘crowns’,corynfoelb.cw. 33 ‘bald-headed’, fromcorunr.b.b. 171 (< Lat.corōna) treated ascoryn.

Vowel Affection.

§ 83. The following tables show the affected and unaffected forms of vowels as they alternate in Mn. W.

i.Ultimatea-affection,§ 68.

No.Unaffected.Affected.Examples.
1ɥegwɥ̆n, f.gwĕn;crȳf, f.crēf
2wotrw̆m, f.trŏm;tlw̄s, f.tlōs

The affected sound occurs in the ultima. It is occasionally found in the penult in compounds, as incromlech (crwm ‘arched’); and in superlatives, asgwennaf, tromaf§ 147 iii. Inbychan, f.bechan, thee seems to be a variant ofy, chosen for the f. on the analogy of the usual f.e caused by affection.

ii.Ultimatei-affection,§ 69,§ 76 v.

No.Unaffected.Affected.Examples.
1a{   {\displaystyle \scriptstyle {\left\{{\begin{matrix}\ \\\\\ \ \end{matrix}}\right.}}ai orei brân, brain;bardd, beirdd;dal, deil
 ɥ tafarn, tefeirn ortefɥrn
2ae ai  draen, drain;cỿrraedd, cỿrraidd
3e     }{\displaystyle \scriptstyle {\left.{\begin{matrix}\ \\\\\ \\\ \\\ \ \end{matrix}}\right\}\,}}ɥ {     {\displaystyle \scriptstyle {\left\{{\begin{matrix}\ \\\\\ \\\ \\\ \ \end{matrix}}\right.}}angel, engɥl;ateb, etɥb;seren, sŷr
4oagor, egɥr;ffon, ffɥn
5wasgwrn, esgɥrn;swrth, sɥrth
6oe w͡y  oen, ŵɥn;croen, crŵɥn
7aw{   {\displaystyle \scriptstyle {\left\{{\begin{matrix}\ \\\\\ \ \end{matrix}}\right.}}au oreu taw, tau;taraw, Ml. W.tereu
 ɥ taraw, Mn. W.terɥ

The change occurs only in the ultima.a oro in the penult becomese§ 69 vi. Finalw, being originally consonantal, does not count as a syllable for the purposes of affection:marw ‘dead’, pl.meirw.

As to the formsai andei of No. 1, see§ 81 iii (1); the formɥ occurs only in the unaccented ultima,§ 69 ii (3).

The formɥ of No. 7 is not a phonetic development ofeu, but is due to false analogy; whentaraw had becometaro the 3rd sing. pres. ind.terɥ was formed from the latter on the model ofagor: egɥr. See§ 173 iv (3).

iii.Penultimate affection,§ 70. The affecting sound is usually preserved in the ultima, but has in some cases disappeared,§ 70 iv.

No.Unaffectedbeforeis affected toExamples.
1aeimab, meibion;cym-ar, ‑heiri̯aid
2eeigorwedd, gorweiddi̯og
3ai orɥetruan, trueni;plant, plentɥn
4aei oreigwaedd, gweiddi;draen, dreini̯og
5aeɥcaer, ceɥrɥdd;saeth, seɥthɥdd
6aeueuaeth, euthum
7awi orɥewtaw, tewi, tewɥch;cawr, cewri

Note 1.—No. 1 occurs only in old formations;‑i̯ad denoting the agent affects,‑i̯ad abstract does not,§ 143 iii (18),iv (5).—No. 2, though common as a fixed affection, is comparatively rare in inflexion. No. 3 is usual in inflexion, but rare in composition, e.g.rhán-dir ‘allotment’,háf-ddɥdd ‘summer’s day’,cán-ddrɥll ‘shattered’,á-dɥn ‘wretch’,ád-fyd ‘adversity’,tán-llɥd ‘fiery’,hád-ɥd ‘seed’ (had +ɥd, buthéd-ɥn ‘a seed’).—No. 4 is only written in old combinations, asgweiẟir.m. 174,seiri; it is rare before, see§ 144 iii (2).—Nos. 5 and 6 also occur only in set forms, and is now wrongly writteneu, asmeusydd.

Note 2.—In Ml. W.a in preceding syllables had becomee beforeỿ, or before one of the above affections; in Mn. W. thea is generally restored,§ 70 i, asenr͑ydeẟ nowanrhydedd ‘honour’,gwerendewɥch, nowgw̯randewɥch. It occasionally remains as inlleferɥdd ‘speech’ (:llafar id.), and even spreads, as inllefaru forllafaru.

Note 3.ɥ in the falling diphthongw͡y does not affect:arw͡ydd etc.§ 38 vi.

Note 4.u does not affecta:canu, parchu, etc. Butcrededun occursr.p. 1368, 1424, besidecredadun do. 1298, 1235.

The Aryan consonants in Keltic and British

§ 84. The Aryan parent language had the following consonant system:

 Labial.Dental.Palatal.Velar.Labio-
velar.
Explosives:     
Tenuesptqq
Tenues aspirataephthk̑hqhqh
Mediaebdɡɡ
Mediae aspirataebhdhg̑hɡhɡh
Spirants:     
Voiceless s, þ   
Voiced z, ðj  
Sonants:     
Nasalsmn   
Liquids l, r   
Semivowels  [ə̯]

Note 1.—In the aspirated tenues the breath was allowed to escape after the explosion; thusth was probably sounded somewhat like thet in W.tad, or like t before an accented vowel in Eng. or N. German, in all of which breath is heard as an off-glide. Aryant on the other hand was sounded like French or South Germant with no escape of breath between the explosion and the vowel.

The exact pronunciation of the aspirated mediaebh, etc., is not known. The conventional European pronunciation isb +h, etc., as in Eng.abhor, adhere. In India the element represented byh is a voiced throat spirant. But the sounds were undoubtedly simple like the aspirated tenues, and were probably voiced forms of the latter.

Note 2.—It is generally held that there were as above three series of gutturals. The palatals were sounded on the hard palate like W.c inci or E.k inking. The labiovelars were sounded between the root of the tongue and the soft palate, so far back that the lips were naturally rounded, as in the formation of the vowelu, W.w, E.u infull. These two series are established by such equations as Skr. = Lat.c < Ar., and Skr.k, c = Lat.qu < Ar.q. But another equation often occurs: Skr.k, c = Lat.c, which points to Ar.q intermediate between the two others, too far back to give Skr. and too far forward to give the labialized Lat.qu. In the Western languages Kelt., Ital., Germanic, Greek, there is no difference between Ar. andq; both givek which is generally accommodated to the following vowel; thus Ar.k̑m̥tom gives W.cant pronouncedqant, not *k̑ant. Where a guttural occurs in a form only found in Western languages, we can only write itk, g, etc., with no diacritic mark. In the Eastern languages (except Tocharish) the palatals became sibilants, thus > Skr. (ansh sound); but the velars remain, or becametch sounds (as infetch) before front vowels, thusq > Skr.k, orc (atch sound), the latter before an Ar. front vowel.—Meillet, Intr.² 63 ff., admits only two series, andq, and regards Skr.k = Lat.c as a special treatment of Ar. in Skr. and the Eastern group. He points out that the supposedq occurs chiefly beforer, beforea, and afters.

The frequent alternation of andq§ 101 iv (1) makes it probable that originally, at any rate, the two are the same. A recent advance fromq to has taken place in Eng. beforeă, now soundedæ̆; thus old borrowings in W. haveq, as in the Anglesey dial.qap ‘cap’,qaban ‘cabin’,qari̯o ‘to carry’, but later borrowings have ask̑ab ‘cab’,k̑ábinet ‘cabinet’,k̑arej ‘carriage’, thea being the same, but the with a perceptible glide. The example shows howq may become before a forward vowel, and how the, once introduced, may remain before a back vowel. The same processes might have taken place in Ar., and it is quite possible that andq represent an original neutralk.

Note 3.—The “sonants” play a special part in Ar. phonology; they occupy an intermediate position between consonants and vowels, and in R-grades become vocalic; see§ 63.

It is usual to include in the Ar. nasalsŋ̑, occurring only beforek̑, g̑, k̑h, g̑h, andŋ occurring only beforeq ɡ, etc. These are secondary sounds due to the assimilation ofm, n to gutturals; and it is not certain that such assimilation had taken place in Ar. We find e.g.mt in *k̑m̥tóm still remaining in Lith.szim̃tas,§ 62 i.

The Explosives.

§ 85. In Pr. Kelt. the aspirated explosives fell together with the unaspirated, thusth anddh are treated ast andd respectively; there is one exception§ 92 iii. The velars fell together with the palatals, thusq, like, gavek. Hence,g̑, g̑h, ɡ, ɡh all appear asg in Kelt.

§ 86. i. Ar.p (Lat.p; Gk. π; Germ.f; Skr.p) and Ar.ph (Skr.ph; Gk. φ) disappeared in Kelt. (1) initially before a vowel, (2) initially before a sonant, (3) between vowels, (4) between a vowel and a sonant, (5) between a sonant and a vowel, (6) between sonants.

Examples: (1) Ar. *pib- > Skr.píbāmi ‘I drink’, Lat.bibo (< *pibō): Ir.ibim ‘I drink’, O. W.ibenjuv. sk. ‘we drink’, W.yfaf ‘I drink’.—Ar. √pet- ‘fly’ > Lat. penna < *petsnā: O. W.etn, Mn. W.edn ‘bird’ < *petno‑.—Ir.athir ‘father’ < *pətēr, W.edryd ‘parentage, descent’ < *pətr̥‑t‑,edrydd ‘patrimony’ (e.g.m.a. i 247) < *pətrii̯o‑,edryf id. < *pətr̥‑m‑: Lat.pater, Gk. πατήρ, etc., Skr.pitr̥tvá‑m ‘paternity’,pítriyaḥ ‘paternal, ancestral’, Gk. πάτριος, etc.—Ir.air‑, W.ar- ‘fore‑’, Gaul.are- < *pₑri‑: Gk. παρά.—Ar. *pr̥t- > Lat.portus: O. W.rit, W.rhyd ‘ford’,§ 61 i.—Ar. *pl̥̄n- > W.llawn,§ 63 vii (2).

(2) Ar. *pro > Lat.pro‑, Gk. πρό, Skr.prá: Ir.ro‑, O. W.ro‑, W.rhy‑.—Ar. *plār- > O. E.flōr: Ir.lār, W.llawr ‘floor’,§ 63 vii (2).—Ar. *prii̯os > Goth.freis, O. E.frēo, E.free: W.rhydd ‘free’.—W.lliaws§ 75 ii (2).—W.llydan§ 63 viii (1).

(3) Ar. *nepōt- > Skr.nápāt‑, Lat.nepōs: Ir.niæ, Ml. W.nei ‘nephew’,§ 75 vii (2).—Ar. *upo- > Pr. Kelt. *u̯o- > Gaul.u̯o‑, Ir.fo‑, W.gw̯o‑,go‑.—W.twymn ‘hot’,twymyn ‘fever’ < *tepes-men‑,§ 75 vii (2).

(4) Ir.tene ‘fire’ < *tepnet‑, W.tân id. < *tₑp‑n‑: Lat.tepeo, Skr.tápas ‘heat’.—Ir.solam, W.hylaw ‘handy’ < Pr. Kelt. *su-lām-os < Ar. *pl̥̄‑,§ 63 vii (2).—W.dyro ‘give’ < *do-pro‑d‑, *√dō‑,§ 63 vi (1).—*mpl- *mpr- > *aml‑, *amr- > W.af‑l‑,af‑r‑, asaf-les ‘harm’, Ir.am-less thusaf- spread foran- < *n̥- (neg. prefix) beforel andr, see§ 156 i (5).

(5) Ir.col, Bret.col, W.cŵl ‘fault’: Lat.culpa, O. Lat.colpa.—Ir.cilornn gl. urceus, O. W.cilurnn (≡cỿlwrnn) gl. urnam, W.celwrn, Bret.kelorn: Lat.calpar, Calpurnius, Gk. κάλπη, Skr.karpara‑ḥ ‘shell’ (Kelt.ĭ orĕ in first syll. unexplained).—W.crydd ‘shoemaker’ (for *cerydd§ 40 iii (3)), Bret.kere < *karpíi̯ō, Ir.cairem < *karpimo‑: Lat.carpisculum, Gk. κρηπίς: √qerāˣp- ‘shoe’. *mp > *m > W.f orw: W.tywydd ‘weather’ for *tyw̯w͡yẟ < *tempes-edō: Lat.tempestas.

The view thatrp, contrary to every analogy, givesrr is based upon one or two examples in which the group may have beenrps or evenrs, as Ir.serr, W.serr ‘bill-hook’ (: Lat.sarpo), which may be < *serp‑s- or *ser‑s- (cf., withoutp, Skr.sr̥ṇī́ ‘sickle’), and upon such an equation as W.gwarr ‘the back between the shoulders’ and Lith.várpa ‘ear of corn’.

(6) *mpl, *mpr gave *ml, *mr, W.fl, fr, ascyflawn ‘full’ < *kom-pl̥̄n‑,cyfran ‘share’ < *kom-prət-snā§ 63 vii (2).

ii. (1) Beforet, Ar.p becameq > k (§ 89 ii) in Pr. Kelt. Thus Ar. *septm̥ > Pr. Kelt. *sektm̥ > Ir.secht n‑, W.saith: Lat.septem, Gk. ἑπτά, etc.—Ar. *qap-tos > Pr. Kelt. *kaktos > Ir.cacht, W.caeth ‘serf’: Lat.captus.—Ar. *neptís > Pr. Kelt. *nektís > Ir.necht, W.nith ‘niece’: Lat.neptis.—W.llithro ‘to slip’ < *sliktr- < *slip‑tr‑, *√slei‑b- extension of √slei‑: E.slip, etc.,§ 95 i.

Before or afters also,p was liable to becomeq in Kelt.,§ 96 iv; also beforen, seeiv below.

(2) Initially in anticipation of medialq, Ar.p becameq in Italo-Keltic; as Ar. *penqe ‘five’ > Skr.pañca, Gk. πέντε: Lat.quinque, Pr. Kelt. *qeŋqe > O. W.pimp, Ml. W.pymp, pump, Ir.cōic, Gaul. πεμπε‑.—Ar. *peq, *poq- > Gk. πέπων, πόπανον: Lat.coquo (< *quequō),coctus, Bret.pibi, W.pobi ‘to bake’ (< *qoq),poeth ‘hot’ (< *qoq‑t‑).

(3) In anticipation ofk orq, Ar.p- seems in some cases to have becomet‑; thus Ir.torc (besideorc), W.twrch ‘boar’: Lat.porcus; seeturio§ 101 iii (1);—W.tanc ‘peace’: Lat.pax, pango, √pāk̑/g̑‑;—W.teg ‘fair’, Gaul.Tecos: O. E.fægr, E.fair, √pē̆k‑;—W.gwar- theg ‘cattle’: Lat.pecus, Lith.pėkus, Skr.pás̑u ‘cattle’; W.talch ‘flake’: Lat.plancus, E.flag-stone, flake, √pelāq‑;—W.twll ‘hole’ < *tuk-slo‑s,tyllu ‘to pierce’: Lat.pungo, punctum,peuk̑/g̑‑. It seems also as ifp at the end of a root or stem beginning with a guttural sometimes becamet, as W.pryd ‘personal appearance’, Ir.cruth: Lat.corpus, Skr.kr̥p- ‘aspect’ < *qr̥p‑;—W.cawad ‘shower’: Ar. *qeuēp-§ 63 vii (3);—W.caled ‘hard’, as a noun ‘difficulty’b.b. 65: ? Gk. χαλεπός (χ- <qh‑).

iii. Ar.p, before disappearing in Kelt., doubtless first became a bilabialf, thenh. When the stop of thep was beginning to be loosened, any reaction in favour of the explosive articulation would naturally take the form of transferring the stop, that is, of substituting for the loosening labialp, the labiovelarq; or, where the word had a guttural already, the dentalt. Befores, both the substitution and the regular development took place; the former,‑qs‑, attested later as‑x- in Gaul.Crixos, gives W.‑ch‑; the latter,‑fs‑, gives W.‑ff‑. Beforet, I have assumed the former, as the substitution ofq forp, known to occur, seems more likely than that of χ forf, so thatpt >qt > χt is more probable thanpt >ft > χt.

iv. Befores,p >q >k after a rounded vowel; thus *upsel- > *uksel- > *ouksel- > W.uchel, Ir.ūasal; *lopsq- > *loksk- > W.llusgo§ 96 iii (5);—similarly beforen; *supn- > *supn- > W.hun ‘sleep’, Ir.sūan,§ 63 viii (1); *n̥-u̯o-dup‑n- > W.an-o-ẟun ‘bottomless’, cf.annwfn§ 102 iv (2), √dheup/b‑; so possibly beforet; W.tuth ‘trot’ < *tupt‑: O. Bulg.tŭpati ‘palpitare’,tŭpŭtati ‘palpitare, calcare’, Gk. τύπτω. Originalq beforet had becomek earlier (in It.-Kelt.), and develops ask, as inpoeth above.m beforep prevents the diphthongization: W.llost < *lompst-§ 96 ii (3).

§ 87. i. Ar.t (Lat.t; Gk. τ; Germ.þ, ð; Lith.t; Skr.t) and Ar.th (Gk. θ; Skr.th) appear in Pr. Kelt, ast. Thus Ar. *tauros > Lat.taurus, Gk. ταῦρος: Ir.tarb, W.tarw;—Ar. *tep‑: W.tes, twymn, tân§ 86;—Ar. *trei̯es > Skr.tráyas, Gk. τρεῖς, Lat.trēs: W.tri, Ir.trī, ‘three’.—Ar. *arətrom > W.aradr ‘plough’: Gk. ἅροτρον.—Ar. *pl̥thə- > Gk. πλάτανος, Gaul. ‑λιτανος, O. W.litan, W.llydan ‘broad’,§ 63 viii (1).

ii. In Ar. the firstt in the grouptt had become an affricative; this stage is represented thustˢt; in Skr. it went back tott (just astst, with originals, gavett in Skr.), in Gk. it became στ, in Germ.ss, in Lat.ss, in Pr. Kelt.ss, appearing in W. generally ass. Example: basemeleit- ‘honey’: FR *melit-tos ‘honeyed’ > W.melys ‘sweet’, Ir.milis. Asdt becamett,§ 93 i, the same result followed; thus Ar. √u̯eid- ‘see, know’, gave *u̯id‑t- > *u̯itt- > *u̯itˢt- > W.gwŷs ‘it is known’§ 63 iv: Skr.vitta‑ḥ ‘known’;—Ar. √(s)k̑heid‑/​(s)qheid‑: R-grade nasalized > Lat.scindo, F-grade *keid‑t- > *keitˢt- > W.cŵys ‘furrow’, Ir.cēis.—So W.cas ‘hate’ < *k̑əd‑t‑, √k̑ād‑: E.hate; F-grade, W.cawdd ‘insult’: Gk. κῆδος. Similarly Ar.dd >dᶻd >zd,§ 91 ii.

§ 88. Ar. (Lat.c; Gk. κ; Germ.h,‑g‑; Lith.sz; Skr.), Ar.k̑h (Gk. χ), Ar.q (Lat.c; Gk. κ; Germ.h,‑g‑; Lith.k; Skr.k, c), Ar.qh (Gk. χ, Skr.kh) appear in Kelt. ask. Examples: Ar. *k̑m̥tóm ‘100’ > Lat.centum, Gk. ἑ-κατόν, O. E.hund, Lith.szim̃tas. Skr.s̑atá‑m: Ir.cēt, W.cant < Pr. Kelt. *kn̥tom.—Ar. √qā‑, F-grade Lat.cārus, Skr.kā́yamāna‑ḥ ‘fond’, R-grade W.caraf ‘I love’.—Ar. *qap- > Lat.capio: W.cael§ 188 iv—Ar. *qrek̑t > Ir.crecht, W.craith ‘scar’ < Pr. Kelt. *krekt‑: Skr.karjati ‘injures’ < *qerg̑‑, √qereg̑‑.—Ar. *ₑreqt- > Pr. Kelt. *arekt- > W.araith ‘speech’§ 63 iii.

§ 89. i. Ar.q (Lat.qu; Gk. π, but τ before ε or η, and κ before or after υ; Germ.hw, ‑f‑, ‑w‑, ‑g‑; Lith.k; Skr.k, c) and probably Ar.qh (Skr.kh; Gk. φ, θ?) wereq in Pr. Kelt. This remains asq in the ogam inscriptions, but becamec in Ir.; in Gaul. and Brit. it appears asp.—Examples: Ar. *qetu̯er- (in various grades§ 63 vii (4)) > Lat.quattuor, Skr.catvā́raḥ: W.pedwar, Ir.cethir.—Ar. √qelāˣ‑/​qēl- > W.pell ‘far’ (< *qel-s-o‑): Gk. τῆλε.—Ar. √seq: Lat.inquam < *insquām: Ml. W.hep, heb ‘says’.—W.prynaf < I buy§ 201 i (4); Ar. √leiq- > Gk. λείπω: W.llwyb‑r ‘track’: Lat.linquo (n-infix).

ii. (1) Beforet, s and prob.n, Ar.q becamek in Kelt. Thus Ar. *poqt- > *qoqt- > *qokt- > W.poeth,§ 86 ii (2).—Ar. *noqt- (√noɡ) > Kelt. *nokt- > Ir.nocht, W.noeth ‘naked’: Lat.nūdus < *noɡedhos.—W.gw̯lyb, O. W.gulip ‘wet’ < *u̯liq: Lat.liqueo; W.gw̯lith ‘dew’ < *u̯likt- < *u̯liqt‑;gw̯lych ‘liquid’ (such as gravy, etc.) < *u̯lik‑n- oru̯lik‑s‑: *√u̯eleiq.

For Ar.sq, qs, see§ 96 iii.

(2) Afterl orr also (but notl̥, r̥), we havek for Ar.q; thus W.golch ‘slops’,golchi ‘to wash’ (Ir.folcaim) < *u̯olk- < *u̯olq-§ 100 ii (2): *u̯liq, as above.—W.cynnyrch ‘crop, produce’ < *kon-derq, √dereq: Gk. δρέπω ‘I mow, reap’, δρεπάνη ‘sickle’. Except in compounds, where the initial of the second element is treated as an initial, asgorffwys,§ 75 vi (4).

(3) Beforeu it appears ask, as in Ml. W.cw ‘where?’ < the Ar. interrog. stem *qu‑,§ 163 i (7),vi.

iii. Ar.k̑u̯ orqu̯, likeq, givesp in Brit. and Gaul.: W.prys ‘brushwood’ <k̑u̯rist‑: O. H. G.hrīs ‘twig’,hurst, E.hurst, O. Bulg.chvrastŭ ‘brushwood’, √k̑u̯ereis‑.—W.paircaldron’ < *qu̯ori̯ó‑) Ir.coire id.: O. N.hverna (‘pot’ < *qu̯er‑: Lat.scrīnium.—Ar. *ek̑u̯o‑s > Lat.equus, Gk. ἵππος (ι <?), Skr.ás̑vaḥ: Ir.ech, Gaul.Epo‑, W.ebol ‘colt’.—W.penn ‘head’, Gaul. Πεννο‑, Ir.cenn < *qu̯enno- < *qu̯ept-sno‑: Goth.haubiþ, E.head, Germ.Haupt, base *qau̯epet- met. for *qapeu̯et- (Sütterlin IF. xxix 123) whence Lat.caput (< *qapu̯et‑).—In later formations: Ar. √māk̑- ‘grow’: R *mək̑- > Gk. μακρός: W.mag-u ‘to nurture’; *mak-u̯o‑s > W.mab ‘son, youth’, Ir.macc, ogam gen.maqqi.—W.epil ‘offspring’ < *eb-hil < *ek-u̯o-sīl‑, √sē-§ 63 vi (1), cf. W.gwe-hil-i̯on 1 Bren. xiii 33 < *u̯o-sīl‑.

But beforeu it givesk, as in W.ci ‘dog’ < * < *ku̯ū< *k̑u̯ō = Skr.s̑vā́; cf.ii (3) above.

In the Roman period, therefore, there was no Brit.q orqu̯, and Lat.qu givesk; as incarawys, garawys ‘lent’ <quadragēsima; W.cegin ‘kitchen’ <coquīna.

iv. It was clearly possible to distinguish in Ar. betweenq andqu̯; probably the rounding in the latter was much more pronounced. Butqu̯ was also felt as a double consonant, and gives ‑ππ- in Gk., whereasq gives ‑π- only.

§ 90. Ar.bh (Lat.f,‑b‑; Gk. φ; Germ.b; Lith.b; Skr.bh) and the rarer Ar.b (Lat.b; Gk. β; Germ.p; Lith.b; Skr.b) both appear asb in Pr. Kelt. Examples:bh: Ar. √bher- > Lat.fero, Gk. φέρω, E.bear, Skr.bhárati ‘bears’: Ir.berimm ‘I bear’, W.cymeraf ‘I take’ < *kom-ber‑.—Ar. *bhrātēr, *bhrāter- > Lat.frāter, Gk. φρᾱ́τωρ ‘member of a clan’, E.brother, Skr.bhrā́tar‑: Ir.brāthir, W.brawd ‘brother’.—Ar. √bhereu̯- > Lat.ferveo: W.berwi§ 63 vii (4).—Ar. *bhu- > W.bod,§ 63 vii (3).—Ar. √enebh‑: VF *nebh- > Gk. νέφος, νεφέλη, Lat.nebula: Ir.nēl ‘cloud’ < *neblo‑, Ml. W.nywl ‘fog’ < *nebli̯o‑; see§ 37 ii.— —b: Ar. √breg̑h- ‘short’: Lat.brevis, Gk. βραχύς: Ir. berr, W.byrr ‘short’, see§ 101 ii (2).—Ar. √dheub- > Goth.diups ‘deep’: R *dhub- > Gaul.Dubno-, W.dwfn ‘deep’, Gaul.dubron, W.dwfr ‘water’.

§ 91. i. Ar.dh (Lat.f-, -d-, -b-; Gk. θ; Germ.d; Lith.d; Skr.dh) and Ar.d (Lat.d; Gk. δ; Germ.t; Lith.d; Skr.d) appear asd in Kelt. Examples:dh: Ar. *dhu̯or-: *dhur- > Lat.foris, Gk. θύρᾱ, E.door: W.dôr < *dhur-ā,drws ‘door’ < *dhru-st- < *dhu̯r̥-§ 63 viii (1), Ir.dorus (intrusiveo?).—Ar. *dhub- > W.dwfn§ 90.—Ar. √ereudh- > Lat.ruber, Gk. ἐρυθρός: W.rhudd ‘red’ < Kelt. *roud-. — —d: Ar. √demā-: F°R *domə- > Lat.domi-tus: W.dof ‘tame’; RR *dₑmə- > Gk. ἀ-δάματος: W.dafad ‘sheep’.—Ar. *dék̑m̥ > Lat.decem, Gk. δέκα, Goth.taíhun, Lith.dẽszimt, Skr.dás̑a: Ir.deich n-, W.deg ‘ten’.—Ar. √dei̯eu̯- > W.duw, dydd,§ 63 vii (4).—Ar. √u̯erō̆d-: R²u̯r̥̄d- > Lat.rādīx; VR *u̯r̥d- > Gk. ῥάδαμνος, Lat.rădius: W.gw̯raidd ‘roots’; RV *u̯r̥d- > Ir.frēm ‘root’ < *u̯r̥d-mā, W.greddf ‘instinct’ for *gw̯reẟf§ 102 iii (2) < *u̯r̥d-mā,§ 63 vii (3).

ii. Ar.d ort +d(h) becamedᶻd(h), which gavezd in Kelt., and fell together with Ar.zd, giving Ir.t (tt), and W.th§ 97 ii. Thus W.peth ‘some, a certain quantity of, something, thing’,beth ‘what?’, Ir.cuit ‘part, share’ < *qid-dm̥: cf. Lat.quid-dam.—W.rhathu ‘to scrape, smooth’ < *rəd-dh- (or *rəd-zdh-): Lat.rādo,§ 63 ix.—W.meth ‘miss, failure’ < *mit-dh-: Ir.mis-, mith- ‘miss-’, E.miss, √meit-: Lat.mūto.

§ 92. i. Ar. (Lat.g; Gk. γ; Germ.k; Lith.ž; Skr.j), Ar.g̑h (Lat.h; Gk. χ; Germ.g; Lith.ž; Skr.h), Ar.ɡ (Lat.g; Gk. γ; Germ.k; Lith.g; Skr.g, j), Ar.ɡh (Lat.h; Gk. χ; Germ.g; Lith.g; Skr.gh, h) all appear in Kelt. asg. Examples:: Ar.g̑r̯̄n- > Lat.grānum, Goth.kaúrn, Lith.žírnis, Skr.jīrṇá-ḥ: Ir.grān, W.grawn§ 61 ii.—Ar. √g̑enē- > Lat.genitor, Gk. γένεσις, Skr.jánati ‘begets’: W.geni ‘give birth’.—Ar. √areg̑- > Lat.argentum, Gk. ἅργυρος, Skr.rajatá-m ‘silver’: W.ariant, Ir.airget ‘silver’ < Pr. Kelt. *argn̥t-. — —g̑h: Ar. *g̑hei-em- > Lat.hiems, Gk. χεῖμα: W.gaeaf,§ 75 vi (1).—Ar. √seg̑h- > Gk. ἕχω (< *seg̑hō), Skr.sáhate ‘vanquishes’: W.hy ‘bold’ < *seg-os, Gaul.Sego-;hael ‘generous’ < *sag-lo- < *sₑg̑h-lo-;haer ‘importunate’ < *sag-ro‑. – –ɡ: Ar. √ɡlei̯- ‘sticky, liquid’: Lat.glūs *<gloi‑s, Gk. γλοιός < *γλοιϝος: O.W.gloiu gl. liquidum, W.gloyw̯ ‘shiny’,gloyw̯-ẟu ‘glossy black’ < *ɡloi-u̯o‑s§ 75 ii (1), Ir.glē, glae ‘bright’ < *ɡlei-uos, *ɡloi-uos, O. Corn.digluiuhitox. 2 gl. eliqua, W.gloywi ‘to drain (after boiling), to clarify, to polish’; R *ɡli- > Lat.gli‑s, Gk. γλί-νη, Lith.gli-tùs ‘smooth, sticky’, Ir.glenim, W.glynaf ‘I adhere’.—Ar. √(s)theɡ- > Lat.tego, Gk. στέγος, τέγος, Skr.sthágati ‘covers’: Ir.tech, teg, O. W.tig, W.ty ‘house’ < *tegos; F° *(s)thoɡ- > Lat.toga, W.to ‘roof’,§ 104 ii (2). – –ɡh: Ar. √ɡhabh- > Lat.habeo, Lith.gabanà ‘armful’: W.gafael ‘to take hold’, Ir.gabim ‘I take’. Ar. √leɡh- ‘lie’ > Lat.lectus, Gk. λέχος: W.lle ‘place’, Ir.lige ‘bed’, W.gwe-ly ‘bed’; L *lēɡh- > Lith.pãlėgis ‘confinement to bed’; F° *loɡh-§ 58 v.

ii. Ar.ɡ (Lat.v, gu aftern,g before cons. andu; Gk. β, δ before ε or η, γ before or after ν; Germ.kw; Lith.g; Skr.g, j) gave Pr. Kelt.b. Thus Ar. √ɡei̯ē- > Lat.vīvo, Gk. βίος: W.byw, etc.,§ 63 vii (3).—Ar. *ɡōus > Lat.bōs (Umbr.-Samn. form for true Lat. *vōs), Gk. βοῦς: Ir., W.biw, pl.bu.

iii. But Ar.ɡh (Lat.f‑, ‑v‑, ‑b‑, gu aftern; Gk. φ, θ; Germ.w, g; Lith.g; Skr.gh, h) forms an exception to the general rule,§ 85, and does not fall together with the unaspirated consonant. It remained a rounded guttural in Pr. Kelt., and gaveg in Ir. with loss of rounding; but the rounding was retained in Brit., and we have in W. initiallygw̯, mediallyf (≡v) between vowels. Thus Ar. √ɡhen- > Gk. θείνω, φόνος, Lat.dē-fen-do: Ir.gonim ‘I wound’, W.gwanu ‘to stab’ < *gw̯on-§ 65 v,gw̯anaf ‘swathe’ (hay cut at one sweep).—Ar. √gher- > Lat.formus, Gk. θερμός, E.warm: Ir.gorim ‘I warm’, W.gori (< *gw̯ori§ 36 iii), Bret.gori, gwiri ‘to incubate’, W.gori ‘to suppurate’,gôr ‘pus’, W.gw̯rēs ‘heat’,§ 95 iii (1).—Ar. √ɡhelē- ‘green, yellow’ > Lat.flāvus: W.gw̯elw̯ ‘pale’,gw̯ellt ‘straw, grass’, Ir.gelim ‘I graze’,gelt- ‘fodder’; the doublet *g̑helē- > Skr.hári‑ḥ ‘yellow, greenish’, Gk. χλόη ‘verdure, grass’, χλόος ‘green’: W.gledd ‘turf’,glas ‘green’,glas-wellt ‘grass’,§ 101 iv (1).—Ar. √ɡhedh- > Gk. ποθέω, θέσσασθαι: Ir.guidim ‘I pray’, W.gw̯eddi ‘prayer’.—Medially: Ar. √sneiɡh- > Lat.ninguit, nix, nivis, Gk. νίφα: Ir.snigid ‘rains’,snechta ‘snow’, W.nyf ‘snow’.—Ar. √dheɡh- > Lat.foveo, Gk. τέφρα: Ml. Ir.daig ‘fire’, W.deifio ‘to singe’.

iv. Unlikeku̯, which is treated asq in Kelt., Ar.gu̯ (g̑u̯, ɡu̯) does not fall together withɡ. The changeq >p is Gaul.-Brit. but not Goidelic, while the changeɡ >b is Pankeltic, and therefore much earlier. The double consonantgu̯ remained, and gives medially W.w, Ir.g, as in W.tew ‘thick’ < *tegu̯os, Ir.tiug: E.thick§ 76 viii.—Ar.ghu̯ develops likegh, giving initially W.gw̯‑, Ir.g‑; thus Ar. *g̑hu̯el‑t- > W.gw̯yllt, Ir.geilt ‘wild’: Goth.wilþeis, E.wild, parallel to Ar.[W 4]g̑hu̯er- > Lat.ferus, Gk. θήρ.

v. When the guttural follows a nasal we have the following results:

> W.m (formm), Ir.mb; as Ar. *n̥ɡen- > Ir.imb, W.ymen-yn ‘butter’: Lat.unguen.

h > W.ng (≡ŋŋ), Ir.ng; as W.llyngyr ‘lumbrici’: Lat.lumbrīcus.—W.angerdd ‘heat’ < *n̥-ɡher‑d‑;angar ‘heat’ < *n̥-ɡhₑr‑, √ɡher‑, seeiii;ager ‘steam’§ 99 vi (1).

nɡhu̯ > W.w, Ir.ng; as W.ewin ‘nail’, Ir.ingen < *n̥ɡhu̯‑, √onoqh/gh‑: Skr.nakhá‑ḥ ‘nail’, Gk. ὄνυξ, Lat.unguis.

ng̑hu̯ > W.f, Ir.ng; as W.tafod ‘tongue’, Ir.tenge: O. Lat.dingua (Lat.lingua), E.tongue < Ar. *dn̥g̑hu̯ā.—W.llyfu ‘to lick’ < *ling̑h‑u̯‑: Ir.līgim, Gk. λείχω, λιχνεύω Lat.lingo, √leig̑h‑.

The first two groups contain two consonants each;ɡ >b, and consequently the nasal becamem; butɡh remained a guttural so that the nasal becameŋ, and the group becameŋŋ, which was unrounded in W. as in Ir. The other groups contain three consonants; in Ir. the dropped as usual, leavingŋŋ; but in W. the remained,ŋŋ >ŋ before a consonant, andŋ dropped,§ 106 ii (1).

§ 93. i. In Ar., when two explosives came together, a tenuis before a media became a media, and a media before a tenuis became a tenuis; thusp + d > bd, andb + t > pt. Only the second could be aspirated, and the aspiration, if any, of the first was transferred to it; thusbh + d > bdh. In this case if the second was a tenuis it became an aspirated media, thusbh + t > bdh; this however only survives in Indo-Iran.; elsewhere we have two tenues; thus Gk. has κτ fromgh + t, as in εὐκτός: εὔχομαι, Meillet, Intr.² 106. So in Italic and Keltic; thus Lat.lectus, Ir.lecht ‘grave’, √leɡh‑; W.gwaith, Ir.fecht < Pr. Kelt. *u̯ekt‑, √u̯eg̑h-§ 100 i (2).

ii. (1) Ar.‑pt‑, ‑kt‑, ‑qt‑, ‑qt‑, all gave‑kt- in Pr. Kelt.,§§ 86 ii,88,89 ii; this appears in Ir. as‑cht, in W. as‑i̯th, etc.§ 108 iv (1).

(2) In other groups of dissimilar explosives the first was assimilated to the second in Pr. Kelt.; thustk >kk > Ir.cc, W.ch; as W.achas ‘hated’, Ir.accais ‘curse’ < *akkass- <*ad-kəd‑t-§ 87 ii.—W.achar ‘loves’ < *akkar- < *ad-qər‑: Lat.cārus§ 88. Lat.‑pt- was introduced too late to become‑kt- as above, and so becamett, as the habit of assimilation persisted in Brit.; this gives W.th; aspregeth ‘sermon’ <pre̦ceptum,ysgrythur <scriptūra.

(3) When the group consisted of mediae, the double media became a single tenuis in Brit., giving a media in W.; thusdg > gg > Brit.c > W.g; it gives Ir.c orcc soundedgg, Mn. Ir.g. Examples: Ir.acarb, W.agarw̯ ‘rough, rocky, unfertile’w.m. 180 < *aggaru̯- < *ad-g̑hₑr’su̯‑: Ir.garb, W.garw ‘rough’ < *g̑hₑr’su̯‑: Gk. χέρσος, Skr.hr̥ṣitáḥ ‘bristling’, Av.zarštva- ‘stone’, Lat.horreo, hirsūtus, √g̑heres‑,§ 95 iv (3).—W.aber, O. W.aper ‘confluence’,aberth ‘sacrifice’ < *abber- < *ad-bher‑, √bher‑.

There seems no good reason to suppose thatgd,db could giveᵹẟ, ẟf in W.  W.gŵydd ‘goose’ cannot come from Stokes’s *gegda (ifg were not assimilated,eg would giveei, notw͡y, in W.), and Pedersen’sbreuddwyd < *brogd- (Gr. i 109) is not convincing. W.ẟf can only come fromzb, orzg§ 97 iii,iv, or fromdm; words likeaddfwyn, addfain come fromad‑m- (mwyn ‘gentle’,main ‘slender’), not from *ad‑b‑. ¶ Two soft spirants coming together, where no vowel has fallen out between them, can only occur when the first was already the spirant <z in Brit., or when the second was the sonantm.

iii. (1) Ar.tt becametˢt, and Ar.dd(h) becamedᶻd(h),§ 87 ii,§ 91 ii, giving W.s (ss) andth respectively. But whend + t ort + t came together in Kelt., they becamett, which, like Lat.tt, appears in W. asth; thus W.athech ‘skulking’ < *ad-teg‑s‑: W.techu ‘to skulk, lie hidden’, √(s)theg-§ 92 i.—W.saeth ‘arrow’ < Lat.sagitta.—Fortt + liquid see§ 99 v (4).

Similarlyd‑d when they came together in Kelt. > Brit.t > W.d; as inedifar ‘repentant’ < *ad-dī-bar‑: W.bâr ‘indignation’, Ir.bara: Lat.ferio.—W.credaf ‘I believe’, Ir.cretim (td‑d) < *kred d‑: Skr.s̑rad dhā- ‘confide, believe’.

Ar. *k̑red dhē- lit. ‘set (one’s) heart (on)’ was not a fast compound (cf. Skr.s̑rád asmāi dhatta ‘believe in him’); thus the W.credaf is explained by thed‑d coming permanently together in Kelt. (for Ar.d‑dh > W.th§ 91 ii), Brugmann² I 670, 691. Lat.crēdo is also irregular, as if * ‘give’ had been substituted for *dhē ‘put’, Sommer 251.

Whend‑d came together later in Brit., they seem to have been simplified tod giving W., as inaẟysg ‘education’ < Lat.addisc‑; so W.aẟef ‘home’ < *ad-dem‑, √demā-§ 91 i.

(2) The change of the firstt intt to the affricative was perhaps due to the tendency in Ar. to avoid double consonants, which in other cases seem to have been simplified. Gemination however was a special characteristic of diminutives and hypocoristic or pet names, and of child-language, which was in a sense a language apart ; and in these even tt remained unchanged. Thus Gk. Νικοττώ (for Νικοτέλεια), Δικκώ, Θεοκκώ, Φίλλιος, Κρίττις, Σθέννις, O. H. G.Sicco (forSigerīch orSigbertus), Lat.Varrō (besideVārus), Brit.Commios (besideComux, Gaul.Comus), W.Iol-lo (with doublel in Ml. W.§ 22 ii, forIorwerth),Gutto (forGruffuẟ);—Gk. ἄττα, Lat.atta ‘papa’; Skr.akkā ‘mama’, Gk. Ἀκκώ, Lat.Acca Lārentia (: W.y nawfed ach ‘the ninth degree of consanguinity’, lit. ‘the ninth *mother’, cf. “the 4th mother”§ 123 v;ach ac edryd ‘descent’, lit. ‘*mat- and pat-ernity’;achoedd, achau ‘lineage’). As the above examples show, the habit of doubling in such forms persisted in new creations, and may account for theqq in the ogammaqqi, and for thett in Brit. *genettā > W.genethr.p. 1359 ‘girl’. So in tribal names:Brittones besideBritannī;Gallī beside Γαλάται. Also in names of animals: Lat.vacca; W.bwch ‘buck’ (ch <kk), Skr.bukkas id.; Gaul.cattos, W.cath; Ml. W.buch ‘cow’ < *boukkā; W.mochyn ‘pig’, Ir.mucc, Germ. dial.mocke ‘sow’; Ir.socc, W.hwch ‘pig, sow’; O. E.dogga ‘dog’; Persson, IP. xxvi 68.

The Spirants.

§ 94. i. Ar.s was of very frequent occurrence. It remained generally in Pr. Kelt. Initially Ar.s before a vowel (Lat.s, Gk. ῾, Germ.s, Lith.s, Skr.s) appears in Ir. ass‑, in W. generally ash‑, sometimes ass‑. Examples: Ir.samail ‘likeness’, W.hafal ‘like’ < *sₑmₑl‑: Lat.similis, Gk. ὁμαλός, √sem- ‘one’.—Ir.sam, W.haf ‘summer’: O. H. G.sumar, E.summer, Skr.sámā ‘year’.—Ir.sen, W.hên ‘old’: Lat.senex, Gk. ἕνος, Skr.sána‑ḥ ‘old’, Lith.sẽnas ‘old’.—W.had: Lat.satus§ 63 vi (1).—W.hun ‘sleep’: Lat.somnus, Gk. ὕπνος,§ 63 viii (1).—W.hynt ‘way’, Ir.sēt,§ 65 iii.—W.hîr ‘long’, Ir.sīr: Lat.sērus,§ 72.— — Ir.secht n‑, W.saith ‘seven’: Lat.septem, Gk. ἑπτά, etc. < Ar. *septm̥§ 86 ii (1).—W.sīl ‘progeny, seed’, besidehīl < *sē‑l‑, √sē-§ 63 vi (1).—W.serr, Ir.serr,§ 86 i (5).—W.saer, Ir.sāer < *sapero‑?: Lat.sapio.—W.sugnaf, Ir.sūgim ‘I suck’: Lat.sūcus, sūgo, O. E.sūgan, sūcan ‘suck’.

ii. Medially between vowels Ar.s remained after the separation of the P and Q divisions; and is found in Gaulish, as inIsarno‑. In Ir. and W. it becameh, and generally disappeared, except where it became initial by metathesis, as in W.haearn, though it is in some cases still written in Ml. W.; thus W.eog, Ml. W.ehawc, Ir.eo, gen.iach < Kelt. *esāk- < *esōk‑, Lat. esox < Kelt. The reduction of vowel-flankeds gave rise to new diphthongs in Brit., which developed largely like original diphthongs; see§ 75 i,ii,vi,vii,§ 76 ii (3).

iii. The change ofs toh differs from the soft mutation; in the latter a voiceless consonant becomes voiced, thust >d; the corresponding change ofs would be toz. Buts did not become voiced; it remained voiceless, but was pronounced loosely, and ultimately becameh. It must have been loosened already in the Roman period, for Lat. intervocalics introduced at that period remains, as incaws <cāseus. Now Lat. explosives undergo the soft mutation; the loosening of Brit.s is therefore earlier, and so the interchanges/h does not enter into that system. Before such a system of interchanges was organized it was natural to choose one or the other sound for the same word; and the postvocalic reduceds was chosen for most in Brit., the postconsonantal fulls for others. It is quite possible that the two forms persisted in many words for a considerable period, so that we have e.g. W.Hafren beside Brit. (‑Lat.)Sabrĭna. There is only one certain example of Lat. initials- givingh‑; that ishestawr <sextarius; this either was a trade term borrowed early, or has followed the analogy of words likeHafren. Possibly a transition stage is represented byIxarninus, Isxarninus besideIsarninus Rhys LWPh.² 418. (The Ir. reduction ofs is independent, and is included in the Ir. system of initial mutation.)

iv. Ar.su̯- remained in Pr. Kelt., and givess in Ir.,chw̯‑, hw̯- in W.,§ 26 vi. Thus Ar. *s̯uesōr > Ir.siur, W.chw̯aer ‘sister’§ 75 vii (2).—Ar. *su̯id‑t- > *su̯itˢt- > W.chw̯ŷs ‘sweat’: Skr.svídyati ‘sweats’: Lat.sūdor < *su̯oid‑: E.sweat.—Ar. *su̯ek̑s > W.chw̯ech, Ir.: Gk. ῾ϝέξ,§ 101 ii (2).—Ar. *su̯ek̑(u)r- > W.chw̯egrwn ‘father-in-law’,chw̯egr ‘mother-in-law’: Lat.socer, socrus, Gk. ἑκυρός, ἑκυρᾱ́, Skr.s̑vás̑uraḥ, s̑vas̑rū́ḥ.—W.chwi ‘you’ < *s‑u̯es: Lat.vōs§ 159 iv. Beforeɔ fromā it was unrounded toh, as inhawdd < *su̯ād-§ 148 i (6).

Medial‑u̯s- >h > W.§ 76 ii (3).

§ 95. i. Ar.sm‑, sn‑, sl‑, sr- remained in Pr. Kelt, and appear in Ir. unchanged, in W. asm‑, n‑, ll‑, rh‑. Thus,sm‑: Ir.smēr ‘blackberry’, W.mwyar ‘blackberries’§ 75 vi (2).—Ir.smir gen.smera ‘marrow’, W.mêr id.: Gk. σμυρίζω, μυρίζω ‘I anoint’, E.smear, Lith.smarsas ‘fat’.—sn‑: Ir.snechta, W.nyf ‘snow’: Lat.ninguit, O. H. G.snēo, E.snow§ 92 iii.—Ir.snāim ‘I swim’, W.nawf ‘swimming’: Lat.nāre, Skr.snā́ti ‘bathes’.—Ar. √senē()- ‘thread’: Ir.snīim, W.nyddaf ‘I spin’, Ir.snāthat, W.nodwydd ‘needle’: Lat.nēre, E.snare, Skt.snā́yu ‘bowstring’.—sl‑: Ir.slemun, W.llyfn ‘smooth’: Lat.lūbricus < *sloibricos, E.slip.—Ir.slūag, W.llu ‘retinue’: O. Bulg.sluga ‘servant’.—sr‑: Ar. *sreu̯‑: Ir.sruth ‘stream’, W.rhwd ‘dung-water’ (rhwd tomydd I.G. 238),rhewyn ‘gutter’: Lith.srutà ‘dung-water’, Gk. ῥυτός, ῥεῦμα, etc.§ 58 vi,§ 76 iv (1).—Ir.srōn ‘nose’ < *srokn‑, W.rhoch ‘snore’ < *srokn-§ 99 vi (3): Gk. ῥέγχω, ῥέγκω ‘I snore’, ῥόγχος ‘snoring’, ῥύγχος ‘pig’s snout’,§ 97 v (3).

Ass- before a vowel sometimes remains in W., so a few examples occur ofs- before a sonant, as (y)snoden ‘band, lace’, Ir.snāthe gl. filum < *sn̥̄t‑, √senē()‑;—(y)slath besidellath ‘lath’, Ir.slat: E. ‘lath’, O. H.G.latta withouts‑. The N.W. dial.slyw̯en ‘eel’ is prob. for *sỿllỿwen: Corn.selyas, syllyes ‘eels’, Bret.silienn (stlaoñenn) ‘eel’; the Mn. lit. W.llỿsỿwen, S. W. dial.llỿsw̄́en, seems to be a metathesized form; prob. √selei‑: Lat.līmax. The second element is perhaps‑onɡhu̯‑: Ir.esc-ung ‘eel’: Gk. ἔγχελυς ‘eel’ (the root has many forms, see Walde² s.v.anguis).

ii. (1) Medial‑sm‑, ‑sn‑, ‑sl‑, ‑sr- probably remained in Pr. Kelt., but became‑mm‑, ‑nn‑, ‑ll‑, ‑rr- in both Ir. and W. (In W.‑mm- is written‑m-, andll is now the voicelessƚƚ, properly doubleƚƚ§ 54 i (2)). Examples:sm: W.twymyn ‘fever’ < *tepes-men-§ 86 i (3).—W.ym ‘we are’, Ir.ammi < Kelt. *ésmesi§ 179 ix (3).—sn: W.onn-en ‘ash’, Ir.huinn-ius < *os‑n‑: Lat.ornus < *osinus, O. H. G.as‑k, E.ash.—W.bronn ‘breast’, Ir.bruinne id. < *brus‑n‑: O. H. G.brus‑t ‘breast’.—sl: W.coll ‘hazel’, Ir.coll < *qos‑l‑: Lat.corulus < *cosulus, O. H. G.hasal, E.hazel, Lith.kasulas ‘spear’.—sr: W.fferru ‘to congeal’ < *spis‑r‑: Lat.spissus ‘thick’. After a long vowel or diphthongn orr is simplified, as inffūn ‘breath’ < *spois‑n-§ 96 iv (1);—gw̯awr ‘dawn’ < *u̯ōs‑r‑: Lat.vēr ‘spring’ < *u̯ēs‑r, √eu̯es‑. But the simplification took place too late to give *f, *l form, ll intwymyn, pwyll, etc.; and‑m, ‑ll remained double after simple vowels and shortened them, as indrŭm§ 100 v,dŭll (2) below.

(2) An explosive before one of the above groups simply disappears; thus *prə-t-snā > W.rhann§ 63 vii (2);—*tuk-slo‑s > W.twll§ 86 ii (3);—*dr̥k-smā > W.drem ‘sight’, √derk̑-§ 61 i;—W.rhwym ‘band’ < *reig-smen > √reig̑‑: Lat.corrigia;—W.pwyll, Ir.cīall ‘thought’ < *qeit‑sl‑: Skr.cit-tá‑m ‘thought’,caityaḥ ‘soul’;—W.dull ‘manner, appearance’ < *doik‑sl‑, √deik̑‑: Gk. δείκνυμι.

(3) But a sonant in the above position remains. Examples: W.garm ‘shout’, Ir.gairm < *g̑ar-smn̥, √g̑ā̆r‑: Lat.garrio;—W.telm ‘snare’, Ir.tailm, gen.telma < *tel‑sm‑: Gk. τελαμών ‘thong’;—Ml. W.anmyneẟ (nowamynedd), Ir.ainmne ‘patience’ < *n̥-smenii̯ā, √menēi̯ ‘thought’, pref.n̥- ‘in’;—W.mymryn ‘a little bit’, Ir.mīr ‘a bit of flesh’ < *mēmsro‑m (ī shortened in Brit.,m lost in Ir.): Lat.membrum < *mēmsrom, Gk. μηρός < *mēmsros or *mēsros, Skr.mās ‘flesh’;—W.cern ‘back of cheek’ < *k̑ersn‑: Lat.cernuus < *k̑ersn‑, Gk. κάρηνον < *k̑ₑrasnom) Lat.cerebrum < *k̑erasrom; W.carr yr ên ‘jawbone’ either < *k̑ₑr’s‑r- (: cf. Lat.cerebrum) or simply *k̑ₑr’s‑;—W.amnaid ‘nod’ (for *anmeid), O. W. pl.enmeituou, O. Bret.enmetiam gl. innuo < *en-smet‑: Ir.smētim ‘I nod’ < *sment‑. It is to be observed thatm in these groups =mm, and is not mutated tof.

iii. (1) Ar.‑ms‑, ‑ns- became‑ss- in Pr. Kelt., and appear so in Gaul., Ir., and W. Thus Gaul.esseda ‘war-chariot’ < *en-sed-ā§ 63 ii; and acc. pl.‑ass inartuass (like Lat.‑ās) < *‑āns. In W., where‑ss- became final by loss of the ending, it became‑s early; but medially it is still double, though now written‑s-§ 54 i (2). Examples: W.crasu, Ml. W.crassu ‘to bake’,crās ‘baked’ < *krams- < *qrm̥‑s‑, √qerem‑: Lat.cremo, Gk. κέραμος, W.cramwyth ‘pancake’ < *kram-pok-tī;—W.mīs ‘month’, Ir. gen.mīs < *mēnsis: Lat.mensis, Gk. μήν, Lith.mė́nů, mė́nesis ‘moon, month’;—W.gw̯rēs ‘heat’ < *ɡhrens-os) √ɡher‑,§ 92 iii: Skr.ghrąsáḥ ‘heat of the sun’ < *ɡhrens-ós;—Ml. W.cysseẟ ‘sitting together’ < *kon-sed‑.

(2) The same change takes place before an explosive; thusnst >st;nsq >sp; as W.cystadl, cystal ‘as good’§ 96 ii (3);cosp < *konsq-§ 96 iii (5).

(3) The nasal also disappears when an explosive came between it and thes, as in W.cysefin ‘primitive’, Ml. W.cyssefin < *kint’samīnos, besidecyntaf ‘first’§ 106 iii (3),cyntefin ‘Spring’ < *kintu-samīno‑.

iv. (1) Ar.‑ls‑, ‑rs- probably became‑ll‑, ‑rr- in Pr. Kelt. Examples of the former are uncertain in W., because‑ln‑, ´‑li̯- also give W.ll; perhaps W.pell ‘far’ < *qel‑s‑: Gk. τέλος.—W.carr, Ir.carr, Gaul.carr-(us) < *qₑr’sos§ 63 iii; W.twrr ‘crowd’ (b. b. 44, 45), ‘heap’ < *tur’‑s‑,ur <u̯ₑr§ 63 viii, √tu̯er‑: Lat.turba, turma (W.torf < Lat.).

(2) An explosive between the two sounds disappears, giving the same result; probably the majority of W.rr’s come from such groups as‑rks‑, ‑rts‑. Examples: W.gyrr ‘a drove’ (of cattle) < *gerks- < *gerg‑s‑: Gk. γέργερα· πολλά Hes., Lat.grex, W.gre;—W.torri ‘to break, cut’ < *torq‑s‑, √tereq‑: Lat.truncus< *tronqos, W.trwch ‘broken, cut’ < *tronqos;—W.carreg ‘stone’ < *k̑ₑr’q-s-ikā, √k̑ereq‑: Skr.s̑árkaraḥ ‘pebble’, Gk. κροκάλη ‘pebble’, W.crogen ‘shell’,craig ‘rock’ < *k̑roqi̯‑;—W.torr ‘belly’ (generally of an animal),torrog ‘pregnant’, Ir.torrach ‘pregnant’ < *torks‑: Lat.tergus ‘body of an animal, hide’;—W.gwarr ‘upper part of back’,gwarr hëolg. 300 ‘ridge of the roadway’ < *u̯ort‑s‑: Lat.vortex, W.gwarthaf ‘summit’ < *u̯ortₑmo‑;—W.corr ‘dwarf’ < *qort‑s‑: Lat.curtus, Ir.cert ‘little’, √(s)qer‑.—Possibly we havell from‑lks- in W.callestr ‘flint’ < *qel’qs‑: Lat.calx, Gk. χάλιξ, √q(h)eleiq- parallel to √k̑ereq- above.

(3) An explosive following the group remains, and thes disappears; thus W.torth ‘loaf’, Ir.tort < *torst- ‘baked’: Lat.tostus < *tors(i)tos:torreo < *torseiō; W.tarth ‘vapour, mist’ (tarth mwg Act. ii 19 ‘vapour of smoke’,tan twym tarthb.t. 38 ‘hot scorching fire’) < *tₑr’s‑t‑: Gk. τερσαίνω, √teres- ‘dry up’;—W.garth ‘promontory, hill’, Ir.gart < *g̑hₑr’st‑: Gk. χέρσος, √g̑heres-§ 93 ii (3) (not to be confused withgarth ‘enclosure’: Lat.hortus§ 99 vi (1),§ 76 vi (2)).

§ 96. i. Ar.s + tenuis remained in Pr. Kelt. In Brit. the group either remained or became a double spirant; thussk gave either (1)sk or (2) χχ; andst gave either (1)st or (2) a sound betweenþþ andss, which becamess. It is probable that form (1) occurred after a consonant, and form (2) after a vowel, being caused by a loose pronunciation of thes. Both forms occur initially and medially, and in the latter case form (1) can be shown in a large number of cases to have followed a consonant now vanished. In Ir.st gavess, initiallys‑, and the other groups remained unchanged.

Tenuis +s also became a double spirant in Brit. A media befores had become a tenuis in Ar., and gives the same result. An aspirated media befores changed it toz in Ar., thusdhs >dhz (dzh); the group became tenuis +s in Kelt., with the same result.

Whens is combined with two explosives in any order it is the first explosive that drops: thusllost < *lompst-ii (3);asgwrn < *ast-korn-ii (4);nos < *nots< *noqtsii (5). The same simplification took place later in words borrowed from Lat.: W.estron ‘stranger’ <extrāneus,astrus <abstrūsus, etc.,§ 103 i (5).

ii. (1) Ar.st- becames- in Ir.,st- ors- in Bret., Corn., and W. Examples: Ir.sāl, W.sawdl, Bret.seul ‘heel’ < *stā‑tl-63 vi (1); Bret.steren, Corn.steren, W.seren ‘star’: Lat.stella < *ster-lā, Gk. ἀστήρ, O. H. G.sterno, E.star: Ar. *stē̆r‑;—Bret.staon ‘palate’, W.safn ‘mouth’: Gk. στόμα;—Ir.sere, W.serch ‘love’, Bret.serc’h ‘concubine’: Gk. στέργω: Ar. *sterk/g‑;—W. (y)starn, Bret.starn, stern ‘harness’ beside W.sarn ‘causeway’63 vii (2), √sterō- ‘spread out’. It is not to be supposed thatst- becames- in W. inseren etc. after the separation of W. and Corn., since Lat.st- generally remains (not always;swmbwl § 66 ii (1)); but rather thatst- ands- existed side by side, and one form or the other prevailed; cf.§ 94 iii. The lisped formþ- is attested in Gaul. in the nameĐirona, also speltSirona (? star-goddess, < *stēr‑).

(2) Medial‑st- gave Ir.ss, Bret., Corn., W.ss. Whenss became final in W. it was simplified early; but it remained double medially, and is still double after the accent, though now writtens§ 54 i (2). Examples: Ir.ross ‘promontory, forest’, W.rhos ‘mountain meadow’ (Richards), ‘moor’ < *pro‑sth‑: Skr.prasthaḥ ‘table-land on a mountain, plain’, √sthā- ‘stand’;—Ir.cas-achtach ‘cough’, W.pas ‘whooping-cough’, Bret.pas ‘cough’ < *qəst‑: O. E.hwōsta, Germ.Husten ‘cough’: Lith.kósėti ‘to cough’, Skr.kā́sate ‘coughs’;—Ir.foss ‘servant’, W.gwas ‘servant’,gwasanaeth ‘service’ < *upo-sthā-n-ā́kt-§ 203 i (4): Skr.upa-sthā́-na‑m ‘attendance, service’;—W.gwasb.t. 4 ‘abode’, Ir.foss ‘rest, stay’ < *u̯ost‑: Gk. ἄστυ < ϝάστυ, Skr.vā́stu ‘dwelling-place, homestead’.—The alternative lisped formþþ is attested in Brit.Aθθedomaros beside gen.Assedomari CIL. iii 5291 (Rhys CB.² 277), W.Gwynn-assedb.b. 67, withaθθ‑, ass- perhaps < *ast‑: Gk. ὀστέον, Skr.ásthi ‘bone’, W.asen ‘rib’,ais ‘breast’.

(3) When‑st- is preceded by a nasal or explosive or both, the whole group gives W.st. Examples: W.cystal, oldercystadl ‘as good’ < *kom-sthə-dhlo- ‘standing together’: Lat.stabulum < *sthə-dhlo‑m;—W.trwst ‘tumult’ < *trum‑st- (ru <u̯r̥§ 63 viii (1)), √tu̯er‑: Lat.turma, turba, Gk. σύρβη, Att. τύρβη;—Ir.loss, los(i. erball) ‘tail’, Bret.lost ‘tail’,lostenn ‘petticoat’,lostek ‘tailed, trailing’, W.llost ‘tail’ inllost-lydan ‘beaver’,arllost ‘the butt end of a spear’ < *lomp‑st‑, √leb- ‘hang down’: Skr.lámbate ‘hangs down’, Lat.limbus ‘hem of a garment’ < *lembos, E.lop inlop-eared, lop-sided: W.llusgo ‘to trail, drag behind’ < *lop‑sq‑;—W.cynllwst ‘kennel’ < *kuno-loq‑st‑, √leɡh- ‘lie’;—W.gast ‘bitch’ < *ganst- for *kan‑st-§ 101 iii (2) < *()ₑn- ‘dog’§ 76 v (1);—W.clust ‘ear’, Ir.cluass < *kleut‑st‑, a Kelt. formation < Ar. *k̑léutom ‘hearing’: Av.sraotə‑m, Goth.hliuþ.—(For the group after a liquid, see§ 95 iv (3).)

After a prefix both forms occur: W.gwa-sarn ‘litter’, √sterō‑;gwa-stad ‘level’, √sthā- ‘stand’;di-serch ‘unlovely’, √sterk/g- (1) above;di-stadl ‘insignificant’, lit. ‘without standing’, cf.cystadl above.

We have perhaps to assume *u̯os- (cf. Lat.sus‑) besideu̯o- and *dēs- beside *dē‑, giving *‑sst- beside *‑st‑, resulting in‑st- beside‑s‑. It is however to be borne in mind that forms with prefixes were not originally fast compounds; and thus the form after a prefix may represent the old initial.

(4) Beforer orl, Ar.st remains in all positions in W. Thus W.ystrad < *strə‑t‑, √sterō-§ 63 vii (2);—W.ystrew, trew ‘sneeze’ < *streus-§ 76 ii (2), √pstereu‑: Lat.sternuo, Gk. πτάρνυμι;—W.ystlys ‘side’, Ir.sliss ‘side’ < *stl̥t‑s‑: Lat.latus < *stlət-os, √stel(ā)‑;—W.arwestr ‘band, (apron‑)​string’ < *are-u̯est-rā: Gk. Dor. ϝέστρᾱ (γέστρα· στολή Hes.), Lat.vestis;—W.rhwystr ‘obstacle’ < *reig-s-tro- ‘*snare’: W.rhwym§ 95 ii (2);—W.bustl ‘gall’ (u fory§ 77 vii (2)), Corn.bistel, Bret.bestl < *bis‑tl‑: Lat.bīlis < *bis-lis (different suffixes‑tl‑:‑l‑);—W.destl ‘neat, trim’,di-ddestl ‘clumsy, unskilful’ D.G. 196, 240 < *deks‑tl‑: Lat.dexter, Gk. δεξιός, W.dehau ‘right’, etc. It is seen that a consonant before the group drops.

On the other hand whenst came before an explosive thet dropped; thusstk >sk, as in W.asgwrn, Ml.ascwrn ‘bone’ < *ast-korn: Gk. ὀστέον, see(2) above (initiala/o altern.§ 63 v (2)); andllosgwrn ‘tail’ similarly formed from *lompst‑, see(3);—W.gwisg ‘dress’ < *u̯ēst‑q‑,di-osg ‘to undress’ < *dē-u̯ost‑q‑, √u̯es‑: Lat.vestis, etc.

(5) Ar.ts givesss in Ir. and W. Originalds anddhz becamets, giving the same result.—W.blys ‘strong desire’ < *mlit‑s‑, noun in‑s- besidemelys ‘sweet’ participle in‑t-§ 87 ii, base *meleit‑;—W.llys ‘court’, Ml. Bret.les, Ir.liss, less < *(p)l̥t‑’s‑. with an‑s- suffix which lost its vowel, added to *pl̥th-§ 63 viii (1);—W.aswy, Ml.W.asswy, asseu ‘left (hand)’ < *at-soui̯ó‑s < *ad-seu̯i̯ós: Skr.savyáḥ ‘left’.—An explosive before the group drops; thus W.nos ‘night’ < nom. *not‑s < *noqt‑s besidenoeth intrannoeth ‘the following day’,heno, O. W.henoid (≡henoyth) ‘to-night’ from oblique cases *nokt‑; soglas- ‘milk’ < *gləkt‑s§ 63 vii (3);tes ‘heat’ < *tekts < *tep‑t‑s: Lat.tepeo, etc. A nasal before the group drops,§ 95 iii (3); but a liquid remains, and the group becomesll orrr,§ 95 iv (2).

iii. (1) Ar.sk̑- appears assc- in Ir., assc- orh- (< χ) in W. In W.sc has becomesg, and initiallyysg‑,§ 23 ii. Thus W.ysgḯen, Ir.scīan ‘knife’, √sk̑hē()‑: Skr.chyáti ‘cuts off’;—Ml. W.isgaudb.b. 35 ‘darkness’, Ir.scāth ‘shadow’ < *skāt‑: Goth.skadus, E.shade, Gk. σκότος, Skr.chādáyati ‘covers’;—W.hegl ‘shank’ < *skek‑l‑: E.shanksk̑eq/ɡ‑: W.ysgogi ‘stir, shake’, E.shake, Lith.szókti ‘to leap, dance’.

After a prefix: W.cysgod,gwasgod ‘shade’ < *skāt‑, as above.

(2) Ar.sq- gives Ir.sc‑, W.sc- (ysg‑) orchw̯- (or before a round vowelh‑). Thus W.ysgwyd ‘shield’, Ir.scīath < *sqeit-om: Lat.scūtum < *sqoit-om, O. Bulg.štitŭ ‘shield’ < *sqeit-om;—W.ysgar ‘to separate’, Ir.scaraim, √sqer‑: Lith.skìrti ‘to separate’;—W.chwith ‘left (hand)’ < *sqī‑tn‑,chwidr ‘perverse, fickle’ < *sqī‑tr‑, Mn. Ir.ciotach ‘left-handed’ <*sqi‑tn‑, W.ysgoewan f. ‘fickle one’ < *sqai‑u̯‑, all R-grades of *sqēi̯- ‘left, oblique’: Lat.scaevus, Gk. σκαι(ϝ)ός, E.shy;—W.chwalu ‘to scatter’, Bret.skula, Ir.scāilim ‘I scatter’, √sqel-§ 101 iv (2):hollt ‘split’iv (1) (β).

With a prefix: W.gwa-sgar-af ‘I scatter’, √sqer-§ 101 iv (2);—cy-chw̯ynn-af ‘I rise, start’, Ir.scendim: Lat.scando, Skr.skándati ‘leaps, bounds’, √sqend‑;—W.osgo ‘slant’,nyt osco-esb.t. 25 ‘he swerved not’ < *op-sqaiu̯‑; Ml. W.amry-scoyw̯, Mn. W.amrosgo ‘diagonal, awkward’ < *sqaiu̯‑: Lat.scaevus, see above;—W.cy-huddo ‘to accuse’: Icel.skúta ‘a taunt’,§ 156 i (9).

skl‑, skr‑, where they remained in Brit., survived in W., nowysgl- ysgr‑, asysglyfaeth§ 101 iv (2),ysgrafell ‘rasp’: E.scrape,iv (3). But these were mostly reduced early to *sl‑,sr‑,§ 101 ii (3). Medially we may have‑chl‑,‑chr‑,§ 156 i (11),(13).

(3) Ar.sq- gives Ir.sc‑, W.chw̯‑. Thus Ir.scēl, W.chw̯edl, Corn.whethl ‘news, a tale’ < *sq-e-tlo- > √seq- ‘say’. With a prefix: Ml. W.ky-chwedɏlb.t. 38 ‘news’ = Ml. Bret.quehezl, Bret.kel;—W.dym-chwel-af ‘I overthrow’: Gk. σφάλλω, Skr.skhálati ‘stumbles’, √sqhel‑;—W.dy-chwel-af ‘I return’ < *do-sqel‑, √qel- ‘turn’,§ 101 iv (2).—sp in the old compoundcosp, see(5).

(4) Medially between vowels Ar.‑sk̑- > W.ch, but is hardly to be found except in old compounds likegochel ‘to guard (against)’,ym-ochel ‘to take shelter’ < *upo-s-k̑el‑, √k̑el-§ 63 iii.—Ar.‑sq‑,‑sq- gave χ, generally unrounded toch; in Ir. all appear asss. Thus Ar. verbal suffix *‑sqe- (: Skr.‑ccha‑, Gk. ‑σκω, Lat.‑sco), appears asch in W.chwenychaf; finally‑wch < *‑yχ < *‑i‑sq- (: Gk. ‑ι-σκω)§ 201 iii (2);—Ml. W.amkawẟw.m. 453 ‘replied’ < *am-χ‑awẟ§ 156 i (4) < *m̥bi‑sq, √seq ‘say’; suffix§ 182 iii.

(5) After an explosive or nasal, however, Ar.‑sk̑‑,‑sq‑ > W.‑sc‑ (‑sg‑), and Ar.‑sq > W.‑sp‑; in Ir.‑sc‑. Thus W.mysgu,cymysgu ‘to mix’, Ir.mescaim ‘I mix’ < *mik̑‑sq‑: Lat.misceo, Gk. μίγνυμι, Skr.mis̑rá‑ḥ ‘mixed’, √meik̑/g̑‑;—W.llusgo ‘to drag’ < *lop‑sq-ii (3) above;—W.hesg ‘sedges’, Ir.sescenn ‘swamp’ < *seq‑sq‑: E.sedge, O. E.secgseq/ɡ- ‘cut’: Lat.seco etc.;—W.llesg ‘languid, infirm, sluggish’, Ir.lesc ‘slothful’ < *leq‑sq‑, √(s)lē̆g‑: Skr.laŋga‑ḥ ‘lame’ < *lenɡ‑, Lat.langueo < *lənɡ‑, Gk. λαγαρός;—W.gw̯rysg ‘twigs’ < *u̯r̥d‑sq‑: Lat.rāmus < *u̯r̥̄d-mo‑s, √u̯erō̆d-§ 91;—W.diaspad f. ‘a cry’ < *dē-ad-sq‑ətā, √seq, suff.§ 143 iii (18); W.cosp ‘punishment’, Ir.cosc ‘correction, reprimand’ < *kon‑sq ‘talk with’.

As the group‑sku̯- or‑squ̯- contains three distinct consonants, it gives‑sp- in W. (not‑ch‑); thus W.hysp ‘dry’ (without milk),di-hysb-yddu ‘to bail’ (a boat, a well, etc.),di-hysb-ydd ‘inexhaustible’ < *sisq-u̯o- redupl. of √seiq- ‘dry’: Avest.hišku- f.hiškvī‑, Lat.siccus < *sīcos (W.sych, Ir.secc < Lat. ?).

(6) Ar.‑ks‑,‑qs‑,‑qs‑ give Ir.ss, W. Bret. Corn.‑ch- or‑h‑. Thus Ir.dess ‘right (hand)’ < *dek̑s‑, W.deheu ‘right, south’ < *deksou̯i̯os, Gaul.Dexsiva dea: Lat.dexter, Gk. δεξιός, Goth.taihswa, O. H. G.zësawa;—Ir.ess‑, W.eh‑,ech-§ 156 i (15): Lat.ex, Gk. ἐξ;—W.ych ‘ox’ (Ml. Ir.oss) < *uqsō: Skr.ukṣā, O. H. G.ohso,§ 69 v.—So finally: W.chw̯ech ‘six’, Ir.,sess- < *su̯eks: Av.xšvaš, Gk. ἕξ (῾ϝεξ), Lat.sex, Goth.saihs, E.six < Ar. *su̯ek̑s, *sek̑s§ 101 ii (2).

As beforets, an explosive or nasal before the group dropped; but in that case‑ks- probably, like‑sk‑, did not become χ, but remained and developed like Lat.‑x‑; so perhapstrais ‘oppression’ < *treks- < *trenk‑s‑: W.trenn, Ger.streng§ 148 i (13). A liquid before the group remains,§ 95 iv (2);‑ksl‑,‑ksm‑ etc.,§ 95 ii (2).

iv. Afters, Ar.p in Kelt. either (α) became *f as usual; or (β) was altered toq and developed accordingly.

(1) (α) Ar.sp(h)‑ > W.ff‑, Ir.s- (mutated tof‑). Thus W.ffun ‘breath’ < *spois‑n‑: Lat.spīro < *speis-ō;—W.ffêr ‘ankle’, Ir.seir ‘heel’ (acc. du. di pherid) < *sper‑: Gk. σφυρόν ‘ankle, heel’ < *sphu̯ₑr‑: Lat.perna, Gk. πτέρνα < *pu̯er‑n- (Jacobsohn, KZ. xlii 275), √sphu̯erē- see(2) below;—W.ffonn ‘stick’, Ir.sonn ‘stake’ < *spondh‑: E.spoon, O. E.spōn ‘chip of wood’, Icel.spānn,spōnn ‘chip’, Gk. σπάθη ‘spatula’, σφήν ‘wedge’, √sp(h)ē-,spend- ‘hew’. — — Similarly before a liquid: W.ffraeth ‘eloquent, witty’ < *sphrəkt‑, √spherē̆ɡ‑: Germ.sprechen, O. E.sprecan; E.speak, see§ 97 v (3);—W.ffrwst ‘haste’ < *sprut-st-: Goth.sprautō ‘quickly’, W.ffrwd§ 101 ii (3);—W.fflochen ‘splinter’ < *sphloq‑n‑: Skr.phálakam ‘board, plank’, √sp(h)el‑: Germ.spalten, E.split, cf. W.talch§ 86 ii (3).

(β)sp(h)- > Kelt.sq > W.chw̯- (h‑) orsp‑, Ir.sc‑. Thus W.chw̯ynn ‘weeds’ (prob. originally ‘furze’, as E.whin which comes from it) < *sqinn- < *spid‑sn-: Lat.pinna < *pid-snā; Ir.scē gen. pl.sciad, W.yspyddad ‘hawthorn’ < *sqíi̯-at‑: Lat.spīna,spīca, √spei‑;—W.chw̯ydu ‘to vomit’,chw̯ŷd ‘vomit’§ 100 ii (3), √spei̯eu̯‑: Lat.spuo, E.spew, etc.;—W.hollt ‘split’,hollti ‘to split’, beside (α) Bret.faouta ‘to split’ < *spol‑t‑, √spel-§ 101 iv (2);—W.yspar ‘spear’, Bret.sparr: Lat.sparus, O. H. G.spër, E.spear, √sphu̯erē-§ 97 v (3);—W.chw̯yrn ‘swift’ < *sphern‑,hwrẟ ‘a violent push’ < *sphuri̯-§ 100 iii (2).

(2) Medially, Ar.‑sp- gives (α) W.‑ff‑, or (β) W.‑ch‑, Ir.‑sc‑. Thus W. dual (α)uffarnau (β)ucharnau ‘ankles’ < *u̯i-sp()ₑr‑n‑: sg.ffêr, Lat.perna above;—(β) W.ucher ‘evening’, Ir.fescor: Lat.vesper, Gk. ἕσπερος§ 66 iii.

After a consonant (α)‑sp- > W.ff; unlike‑st‑,‑sk‑, which preserve the explosive,sp had become‑sf‑, and there was no explosive to preserve. Thus W.effro ‘awake’ < *eksprog- dissim. from *eks-pro‑gr‑: Lat.expergiscor for *ex-pro-grīscor (Walde, s. v.): Av.fra-γrisəmnō ‘waking’, Skr.járate ‘wakes’, Gk. ἐγείρω, √ɡer‑,ɡerēi̯‑.

(3) Ar.‑ps- also gives (α) W.‑ff‑, or (β) W.‑ch‑, but Ir.‑ss‑. Thus (α) W.craff ‘sharp, keen’ < *qrap‑s- < *qrab‑s‑: Icel.skarpr, O. E.scearp, E.sharp, E.scrape, W.crafu ‘to scratch’;—W.praff ‘burly’ <qrₑp‑s‑: Lat.corpus, etc.;—(β) W.uwch ‘higher’,uchel ‘high’, Ir.ūasal,uassal, Gaul. Uxello-dunum < *ups‑, *upsel‑: Lat.sus‑, Gk. ὕψι, ὑψηλός ‘high’, ὑψίων ‘higher’;—W.crych ‘curly’, Gaul.Crixus,Crixsus: Lat.crispus (prob. < *cripsos): Lith.kreĩpti ‘to turn’, √qer- ‘turn’, extd. *qreip‑;—W.llachar ‘bright’, Ir.lassair < *lapsₑr‑: Gk. λάμπω;—W.crach ‘scabs’ < *qrap‑s‑:craff above, see§ 101 ii (2). As in the case of‑ks‑, seeiii (6), the *‑ch- may become‑h‑, as incah-el besidecaff-el < *qap‑s-§ 188 iv.

§ 97. i. Before a media or aspirated media,s had becomez medially in Pr. Ar. Thus the V-grade of √sed- was‑zd‑. Ar.z became in Pr. Kelt. This remained in Brit., and the media following it was reduced later to the corresponding voiced spirant.

ii. Ar.‑zd- > Kelt.ẟd. In W. this becameth, throughẟẟ; in Ir. it appears ast,tt (≡d‑d), Mn. Ir.d. Thus Ar. *nizdos ‘nest’ > Ir.net,nett, Mn. Ir.nead, W.nyth: Lat.nīdus, O. H. G.nest, E.nest, Skr.nīḍá‑ḥ, √sed-§ 63 ii;—W.syth 'upright',sythu ‘set erect’, Ir.seta ‘tall’ < *sizd‑: Lat.sīdo < *sizdō, Skr.sī́dati ‘sits’ for *sīḍati < *sizd‑, Gk. ἵζω < *sizdō, √sed‑, redupl. *sizd‑;—W.gŵyth ‘anger’,ad-wyth ‘hurt, mischief, misfortune’ < *g̑heizd‑, Ml. Ir.goet ‘wound’ < *g̑hoizd‑: Skr.héḍa‑ḥ ‘anger’ < *g̑heizd-os,héḍati ‘angers, vexes, hurts’, Lith.žáizda ‘wound’,žeidžiù ‘I wound’, Av.zōižda- ‘hateful’;—W.brathu ‘to stab, bite’,brath ‘a stab, a bite’ < *bhrazd(h)‑: Russ.brozdá ‘bit, bridle’ < *bhrazd(h)‑, O. Bulg.brŭzda id. < *bhr̥zd(h)‑: with‑st‑, Skr.bhr̥ṣṭí‑ḥ ‘tooth, point’, Lat.fastīgium for *farsti- (< *frasti‑?), √bhera‑s‑? Walde² 275, extension of √bher- ‘prick’: W.bêr ‘spear, spit’;—‑d- presents: W.chw̯ythaf ‘I blow’ < *su̯iz‑d‑, Ir.sētim id. < *su̯eiz‑d‑: Skr.kṣveḍati ‘utters an inarticulate sound, hisses, hums’ < *ksu̯eiz‑d‑: with‑t‑, O. Bulg.svistati ‘sibilare’.

After a consonant the result is the same, for the consonant had dropped in Brit., and thoughst of that period remains (e.g. Lat.‑st‑), the mutationd > is later, so that Brit.‑ẟd >ẟẟ >th. Thus the prefix *eks- +d- gave *e(g)zd- > *eẟd- >eth- as inethol ‘to elect’ < *egz-dol-: E.tale, Ger.Zahl ‘number’, W.didoli ‘to segregate’, Skr.dálam ‘piece’, Lith.dalìs ‘part’, √dē̆l- ‘divide’.

iii. Ar.‑zg̑(h)‑,‑zg(h)- > Kelt.‑ẟg‑; in Ir. it appears asdg (≡ẟᵹ); in W. *ẟᵹ becamei̯ẟ by met.; afterw, *ẟᵹ >ẟf. Thus W.maidd ‘whey’ < *meᵹ̑ẟ‑, met. for *meẟᵹ‑, Ir.medg ‘whey’, Gallo-Lat.mesga (s for? cf.§ 96 ii (1)): Lat.mergo, Lith.mazgóti ‘to wash’, Skr.majjati ‘sinks’ < *mezɡ‑;—W.haidd ‘barley’ < *se‑zg‑, redupl. of *seg‑: Lat.seges;—perhaps W.twddf ‘a swelling’ for *tuẟᵹ- < *tuzg‑,s-stem of √teu̯ā- (: Goth.þūs‑) +‑g- suff.: Lat.turgeo (Walde² rejects his first suggestion that this is from *tuzg- in favour of Solmsen’s *tūrigo,IF. xxvi 112ff., with‑igo (:ago), though this is usually 1st conj., asnavigāre).

W.gwẟf ‘throat’, N. W. dial.gwẟw, pl.gyẟfe,gyẟfa, S. W. dial.gwẟwg, pl.gyẟge,gythce, Bret.gouzoug, with‑g for‑ᵹ,§ 111 vii (4), seems to require *guzg‑; ?g̑hu‑s‑, √g̑hēu‑, (: Lat.fauces) +‑g‑, as inmwn‑g ‘mane’.

iv. Ar.‑zb(h)- > Kelt.ẟb > Ir.db, W.ẟf. Thus W.oddf ‘knag, knot, nodule’, Ir.odb: Gk. ὀσφύς (< *ost-bhu‑?).

v. (1) The above groups are found only medially. Initially Ar.s- did not becomez‑, but changed a following media to a tenuis ; thussb- >sp‑, *sbh- >sph, etc., Siebs, KZ. xxxvii 277 fF. Hence the initial alternationsb‑:sp- anddh‑:sth‑, etc., as in Germ. dumm, E. dumb < *dh‑: Germ.stumm, W.di-staw <sth‑,§ 156 i (11).

(2) Ass- could be prefixed or dropped in Ar. and for a long time after the dispersion,§ 101 ii (1), Siebs l. c. holds that the above explains the initial alternation of a media and tenuis. In a large number of cases it undoubtedly does so. Where the media is general and the tenuis exceptional, it affords a satisfactory explanation, as in the case of the Kelt.t- intafod ‘tongue’ corresponding tod- elsewhere (O. Lat.dingua), which is parallel to thet intaw! ‘be silent’ (s still kept indi-staw) corresponding to the *dh- which gives thed- of E.dumb. But it hardly explains the alternation when the tenuis is general and the media exceptional, as in W.craidd, Lat.cord‑, Lith.szirdìs, E.heart, Gk. καρδία < *k̑‑: Skr.hŕ̥d‑, Av.zərədā < *g̑h‑, since <sk̑h, without a trace of thes- in the whole of Europe, is improbable. But whatever the explanation may be, the fact of the alternation can hardly be called in question.

(3) As an example of the variety of forms produced by variables‑, we may take √bhu̯erē‑, extd. *bhu̯erē̆‑ɡ‑/​‑ɡh‑/​‑q‑, orig. meaning 1. ‘hurl’, 2. ‘smite’; hence from 1. ‘sprinkle, cast (seed); roar, snore; rattle ; talk’; from 2. ‘break; crash, break out, burst; smell’.bh‑: W.bwrw ‘hurl, smite’,bwrw glaw ‘to rain’,bwrw had ‘to cast seed’ < *bhur’ɡ- (ur <u̯ₑr); Lat.frango < *bhrənɡ‑,frāgor < *bhr̥̄ɡ‑,frā- grāre, E.break,burst, W.brych,brith ‘speckled’;—sph‑: W.hwrẟ§ 100 iii (2),chw̯yrn ‘swift’§ 90 iv (1),chw̯yrnu ‘to roar, snore’; Skr.sphuráti ‘spurns, darts, bounds’,sphū́rjati ‘rumbles, roars, rattles, crashes’ < *sphu̯r̯̄ɡ‑; Lat.sperno,spargo; E.spurn,sprinkle; Gk. σφάραγος; W.ffraeth§ 96 iv (1),ffroen < *sphruɡ-nā (ru <u̯r̥);—p(h)‑: W.erch ‘speckled’, Gk. περκνός; W.arch-fa ‘stench’ < *phₑr’q‑;arogleu ‘a smell’, compoundpₑroqo-prāɡ‑?—(p…​ɡ >t…​ɡ§ 86 ii (3))trywyẟ ‘scent’,trwyn ‘nose’ < *pruɡ-no‑,trawaf ‘I strike’ < *prug- (ru <u̯r̥);—spr >sr§ 101 ii (3): W.rhuo ‘roar, talk loudly’ < *srogi̯‑, Gk. ῥέγχω, ῥέγκω, ῥόγχος, ῥύγχος, W.rhoch ‘snore’.

§ 98. i. (1) In Gk. and Kelt, a dental explosive sometimes appears after a guttural where the other languages haves; this is explained by the supposition that Ar. possessed after gutturals another spirant, similar to E.th inthink, W.th, which is writtenþ. After an aspirated media, ass becamez,§ 96 i, soþ becameð; thusghþ >ghð (gðh). Brugmann² I 790 ff.

(2) Ar.k̑þ- (Lat.s‑, Gk. κτ‑, Skr.kṣ‑) gave Kelt.t‑. Thus W.tydwet,tydwed[17]b.b. 20, 36 ‘soil, land’ < *tit‑: Lat.situs ‘site’, Gk. κτίσις ‘settlement’, κτίζω ‘I found’, Skr.kṣití‑ḥ ‘abode, earth, land’: √k̑þei̯- ‘earth’, see(3) below.

Ar.‑k̑þ- (Lat.‑x‑, Gk. ‑κτ‑, Skr.‑kṣ‑) gave Kelt.‑kt‑. Thus W.arth ‘bear’, Ir.art < *artos < *arktos: Gk. ἄρκτος, Lat.ursus < *urcsos, Skr.ŕ̥kṣaḥ: Ar. *ark̑þos, *r̥k̑þos§ 63 v (2).

(3) Ar.g̑hð- (Lat.h‑, Gk. χθ‑, Skr.h‑, Germ.g‑, Lith.ž‑) gave Kelt.d‑. Thus Ir.indhe, W.doe ‘yesterday’ < *desī = Lat.heri: Gk. χθές, Skr.hyáḥ,§ 75 vii (2); this occurs medially in W.neithi̯w͡yr ‘last night’78 i (2) for *neith-ẟi̯w͡yr < *nokti di̯eserāi (assuming the case to be loc.): O. H. G.gestaron, E.yester‑, Lat.hesternus: Ar. *g̑hði̯es‑, suff. *‑ero‑/​‑tero‑.—W.ty-ẟyn ‘a measure of land, a small farm’ lit. ‘*house-land’,tref-ẟynb.t. 14,gwely-ẟyn (gwelitinb.b. 64), Ml. pl.tyẟynneu for *‑ẟy̆́ni̯eu < *domi̯‑: Lat.humus, Gk. χθών: Ar. *g̑hðem- ‘earth’; allied to this as meaning ‘terrestrial’ are the names for ‘man’: W.dyn, Ir.duine < *doni̯o- < *g̑hðomi̯‑: Lat.homo, Lith.žmů̃,žmo-gùs pl.žmónės, Goth,guma pl.gumans: Ar. *g̑hðem‑. This may be for *g̑hði̯em- as Pedersen suggests, Gr. i 89–90; in that case the root must be *g̑hðei̯‑, which therefore must be the same as k̑þei̯- above, with Ar. alternationk̑‑/​g̑h‑; hence W.daear ‘earth’ < *g̑hðii̯-ₑrā, √g̑hðei̯‑.

(4)ɡhð- (Gk. φθ‑) gave Kelt.d‑. Thus W.dar-fod ‘to waste away, perish’,dar-fodedigaeth ‘phthisis’ < *dar- < *ɡhðₑr‑: Gk. φθείρω < *ɡðer‑; W.dyddfu ‘to pine, waste away’ < *di‑d‑m- redupl.,‑m- suff.: Gk. φθῑ́ω, ἀπο-φθίθω; in Skr. with *qþ‑, askṣárati ‘flows, passes away, perishes’,kṣī́yate ‘decreases, wanes’.

ii. In Gk. we sometimes find ζ- where the other languages havei̯‑. This equation is held to imply an Ar. palatal spirantj (the sound which is written ᵹ̑, i. e. palatal ᵹ, in other connexions in this book; it differs from in being pronounced with more friction of the breath). Examples are W.i̯au ‘yoke’, Lat.jugum, Skt.yugá‑m, Gk. ζυγόν, all < Ar. *jugóm;—W.i̯ās ‘a seething’, Skr.yásyati ‘seethes, bubbles’, Gk. ζέω: Ar. √jes‑;—W.uwd ‘porridge’, Ml. W.i̯wt§ 37 ii, Bret.iot, Lat.jūs, Skr.yūṣa‑m ‘broth’, Gk. ζῡ́μη: Ar. √jēu‑; W.i̯wrch, O. Corn.yorch: Gk. ζόρξ§ 65 iii (2);—W.i̯oli: Gk. ζῆλος§ 201 iii (2).

The Sonants.

§ 99. i. Initially before vowels, and medially between vowels, Ar.l,r,m,n (so in most of the languages, butr- > ἐρ- in Gk.) remained unchanged in Pr. Kelt. In W. initiall- andr- becamell- andrh‑,§&nsp;103 i (4). Many examples occur in the above sections; as W.llost < *lompst-§ 96 ii (3); W.halen ‘salt’§ 58 ii; W.rhwym, √reig̑-§ 95 ii (2); W.adferaf, √bher-§ 58 iii; W.mis ‘month’§ 95 iii (1); W.haf, Ir.sam ‘summer’§ 94 i; W.naw ‘nine’§ 76 iii (1); W.ychen ‘oxen’§ 69 v. The treatment of these sonants in combination withs has been discussed in§ 95, and in combination withs and an explosive in§ 96. There remains the combination of sonants with one another and with explosives.

ii. (1) Ar.ml‑,mr- remained in Pr. Kelt., but in Brit. they becamebl‑,br- and appear so in W.; in Ir. bothm- andb- appear. Thus W.blys < *mlit‑s-§ 96 ii (5);—W.bro ‘region’, Ir.mruig ‘boundary’ < *mrog‑: Lat.margo, O. H. G.marka, O. E.mearc, E.march§ 65 ii (1);—W.brag ‘malt’, Ir.mraich < *mrəq‑, W.braenu ‘to rot’ < *mrəq‑n‑, √merāˣq- ‘decay’: Lat.fracēs ‘oil-dregs’, Gk. ἀμόργη (< *ἀμόρκᾱ, whence Lat.amurca Walde² 464).—Similarly Ar.m- before or, short or long: W.blith ‘milk, milch’, Ir.mlicht,blicht < *ml̥k̑t‑,§ 61 i; W.blawd ‘flour’ < *ml̥̄t-§ 61 ii.—The same change probably took place medially also; in that position bothm andb would now appear asf, but in O. W. fromm is writtenm, whilev fromb appears asb; and such a form asamcibretox. < *m̥bi-kom-(p)ro-ret-§ 156 i (9) impliesv <b; so Brit. Sabrina probably contains *sam‑. In the Coligny calendartio-cobrextio very probably contains *kom-rekt- = W.cyfraith, Rhys CG. 16. But. W.cyṽ- < *kom- persisted by analogy:cymreith (mv)l.l. 120; cf.§ 16 iv (3). (Lat.m…​l becamemb…​l incumulus,stimulus§ 66 ii (1).)

(2) Ar. medial‑lm‑,‑rm- remained in Pr. Kelt., and‑lmp‑,‑rmp- became‑lm‑,‑rm‑; they appear so in Ir.; in W. them appears asf or. Thus W.celfydd ‘skilful’,celfyddyd ‘craft’, O. Bret.celmed gl. efficax, Ir.calma ‘doughty’ < *qₑl’mp‑: Lat.scalpo, Lith.sklempiù ‘I polish’, Skr.kalpanā ‘fashioning, invention’,kl̥ptáḥ ‘arranged, trimmed, cut’: E.skill, Goth.skilja ‘butcher’; √(s)qel‑, extd. *(s)qelep‑;—W.cwrf,cwrw̯, Ml. W.kwrɏf, coll.cwrw forcwrwf orcwrw ‘beer’, Ir.cuirm, Gaul. κοῦρμι, < *korm‑: Lat.cremor ‘thick juice obtained from vegetables’; lit. ‘*decoction’, √qerem-§ 95 iii (1);—W.serfyll ‘prostrate’ < *stₑrm‑: Lat.strāmen, Gk. στρῶμα, Skr.stárīman- ‘strewing’, √sterō-§ 63 vii (2).—So in old compounds: W.gorfynt ‘envy’, Bret.gourvent, Ir.format < *u̯er-ment‑: Lat. gen.mentis, E.mind: Gk. ὑπερ-μεν-ής with same pref. and root: √men‑; but later compounds may haverm, asgor-moẟ ‘too much’.

Probably them was already somewhat loose in Brit., as Gaul.ceruesia ‘beer’ beside κοῦρμι shows it to have been in Gaul. Hence new formations with a newm might be treated differently. Thus, in Lat. loanwords, while we have usuallylf,rf, as inpalf <palma,terfyn <terminus, we may havelm,rm, as inGarmon <Germānus,salm <psalmus, prob. borrowed later.

iii. (1) Ar.‑nl‑,‑nr- became‑ll‑,‑rr- respectively in Pr. Kelt. Thus W.gwall ‘want, defect’,gwallusỺ.A. 154 ‘negligent’, now ‘faulty’, Bret.gwall ‘defect’ < *u̯an-lo‑, √u̯ā̆n‑: Lat.vānus, E.want;—W.garr ‘knee’, Bret.garr ‘jambe’ < *gan‑r-§ 63 vii (4). But in compounds in which the sounds came together after the Brit. period, then remains, and the group becomes‑nll‑,‑nrh- in W., as inan-llad,an-rheg§ 111 i (1).

(2) Ar.‑ln- also became‑ll- in Pr. Kelt. Thus W.dall ‘blind’, Ir.dall ‘blind’,cluas-dall ‘deaf’ < *dh()al’-no‑: Goth.dwals ‘foolish’, O. E.ge-dwelan ‘to err’, √dhu̯elāˣ‑. But‑rn- remained, as in W.chw̯yrn ‘swift’ < *sphern-§ 96 iv (1); W.carn ‘hoof’, Bret.karn, Galat. κάρνον· τὴν σάλπιγγα, Hes. < *k̑ₑr’n‑, √k̑erāˣu̯‑; W.darn,sarn, etc.§ 63 iii; Kelt. suffix *‑arn- < *‑ₑr’n‑, as in W.haearn,cadarn.

iv. (1) Ar.‑mn‑,‑nm- remained in Pr. Kelt., and appear so in Ir. (or with an epenthetic vowel); in W. the mutated formf (or§ 102 iii (1)) takes the place ofm. Thus W.safn ‘mouth’, Bret.staoñ ‘palate’ < *stom‑n‑: Gk. στόμα§ 76 vii (4);—W.cyfnesaf ‘kinsman’ < *kom-nessam‑,§ 148 i (1);—Ir.ainm ‘name’, O. W.anu < *an’mn̥§ 63 v (2); W.menw̯-yd ‘mind, pleasure’, Ir.menme ‘mind’ < *men‑m‑: Skr.mánman- ‘mind, thought’;—W.an-fad ‘atrocious’ (:mad ‘good’), Gaul. (Sequ.)anmat… ‘unlucky’ < *n̥-mat‑: Lat.mātūrus orig. ‘in good time’ Walde² 470.

An explosive probably dropped before the group: W.pythefnos,pythewnos ‘fortnight’ lit. ‘15 nights’ for *pymtheṽnoeth (dissim. of nasals) < *pempede(k)m-noktes < Kelt. *qeŋqedekm̥ noktes.

(2) Ar.‑rl- and‑lr- can hardly be traced; we should expect them to give‑ll- and‑rr-. Late‑rl- gave‑rll-§ 111 i (1).

v. (1) A group consisting ofl,r,m orn and a single explosive remained in Pr. Kelt. (except thatp dropped,§ 86, and a nasal assumed the position of a following explosive). The further development of such groups in W. is dealt with in§§ 104–6.

(2) When a liquid came before two explosives the first explosive dropped; thus W.perth ‘bush’ < *pertā < *qerq‑t‑: Lat.quercus < *perqus§ 86 ii (2): O. H. G.forha, O. E.furh, E.fir, Skr.parkaṭī ‘ficus religiosa’;—W.cellt ‘flint’ < *qelq‑t‑: Lat.calx§ 95 iv (2);—W.arth, Ir.art< *arktos§ 98 i (2).

(3) But when a nasal came before two explosives, the nasal dropped; thus W.trwyth ‘wash, lye, urine’ < *tronkt‑: W.trwnc ‘urine’ < *tronq‑: Lith.trenkù ‘I wash’ (W.trochi ‘to bathe’ < *tronq‑, seevi (3)) : Lat.stercus, Bret.stroñk ‘excrement’. It is seen that the loss is later than the changeonk >unk§ 65 iii (1) ; it also takes place in Lat. loanwords, as W.pwyth ‘stitch’ <punctum; but in the later of these the first explosive drops, as insant <sanctus.

(4) When two explosives came before a liquid or nasal, the group remained in Pr. Kelt.; thus W.eithr ‘except’, Ir.echtar < *ektro‑s: Lat.exterus,extrā, Osc.ehtrad (‑x- for *‑c- is a Lat. innovation, Walde² 263);—W.aethn-en ‘aspen’ < *aktn- < *aptn‑: Lith.apuszė ‘aspen’, O. H. G.apsa, O. E.æps, E.asp: Lat.pōpulus < *ptō̆ptol‑, Gk. πτελέα ‘elm’.

But a double explosive before a sonant was not distinguished in Ar. from a single; thusettre was not distinct frometre, Meillet, Intr.² 102. In Homer and the Veda the first syllable is metrically long; in Plautus and Aristophanes, short; ordinarily in Gk. and Lat., doubtful. In old Kelt. formations we have onet for two, as in Gaul.Atrebates, W.adref ‘homewards’ < *atreb- < *attr- < *ad‑tr‑. In later formations the double consonant remained, as in W.athrist ‘sad’ < *attrīstis < *ad- + Lat.trīstis.kr,tr may develop askkr,ttr in W. as inochr,rhuthr§ 104 iii (2). A double media in Brit. is treated regularly as a single tenuis in W., as inedrych ‘to look’ < *etr- < *ed‑dr- < *ad‑dr- or *eg‑dr‑; once as a double tenuis; see l. c.

vi. (1) A group of the formnt ornd, followed immediately or mediately by a liquid or nasal, has tended from an early period in Kelt. to become a double explosivett ordd with nasalization of the preceding vowel. In Ir. the double consonant was simplified before the sonant; seecēol,abra,cobrith (bƀ) below. The change, being a case of dissimilation of the continuants, does not take place regularly,§ 102 i; it often exists side by side with the regular development of the group. Thus O.W.ithr ‘between’, Bret, etre, Van. itre, Ir. eter (not *ēt- the regular Ir. for *ent‑) beside Bret.eñtre, Corn.yntre: Lat.inter, Skr.antár;—W.athrugar ‘pitiless’ < *ąttr- beside Ir.ētrōcar < *entr‑, both < *n̥-trougākaros;—W.cathl ‘song’ < *kąttlo‑, Ir.cēol id. < *kęt(t)lo‑, O. W.centhliat,centhiliat (enę) gl. canorum, beside Ir.cētal < *kentlo‑, Bret.keñtel ‘lesson’;—W.allwedd f. ‘key’ for *alchwedd, Bret.alc’houez metath. for *achlweẟ < *n̥-ql(ə)u̯-íiā (‘unlocker’, cf.agoriad ‘opener’ used instead in N. W.), alsoallwydd m. <‑íios: Lat.claudo,clāvis, Gk. κληίς, etc.;—W.achles ‘shelter’ < *n̥-kl̥-stā (n̥- ‘in’), √k̑el- ‘hide’: O. H. G.hulst ‘cover’, W.clyd§ 63 iii;—W.achenog ‘needy’,achen ‘need’, beside W.anghenog,angen, Ir.ēcen ‘need’ < *n̥k-en‑: Gk. ἀνάγκη.

Mediae: W.adyn ‘wretch’ < *ąddoni̯os < *n̥-doni̯os ‘not-man’, beside the laterannyn ‘wretch’,annynol ‘inhuman’, Mn. Ir.anduine;—W.agor ‘to open’ < *ąggor- < *n̥-ghor- (n̥- negative), besideegor id. < *eggor- (pref. *ek‑), √g̑her- ‘enclose’: Lat.hortus, Gk. χόρτος, W.garth;—W.w͡ybren ‘cloud, sky’, O. Corn.huibren, Ml. Corn.ebron, Bret. Van.ebr, beside Ir.imrim ‘storm’: Lat.imber,§ 100 v;—W.hebrwng ‘to accompany, convey’, O. Corn.hebrenchiat, Mn. Corn.hembronk, Ml. Bret.hambrouk < *sem-broŋk‑: Skr.sam- ‘with’, Goth.briggan, E.bring;—Bret.abrant ‘eyebrow’, Corn.abrans < *abbr‑, Ir.abra < *abr‑, beside W.amrant < *am-brant- (n̥- ‘in’): Lat. gen.front-is;—Ir.cobrith ‘help’, beside W.cymryd ‘to take’ < *kom-bhr̥‑t‑. The nasalized vowel sometimes develops a new nasal, resulting in a newnd, etc., which does not becomenn; thus W.enderig ‘steer’, O. W.enderic gl. vitulus, beside W.anner ‘heifer’ which contains oldnd; Gwyn. dial,ắŋ-gar ‘hot breath, steam’ for lit. W.ager ‘steam’ < *ągger‑, besideangerdd (ngŋŋ) < *‑aŋger‑, all < *n̥-ɡher-§ 92 v.

Similarlyltr > *ttr >thr inathro§ 76 v (5).

(2) It has been conjectured that an explosive +n sometimes became a double explosive in Kelt.; Pedersen, Gr. i 158, suggests that this took place immediately before the accent. Thus Ir.brecc, W.brych ‘speckled’ < *brikkos < *bhr̥knós: Gk. περκνός§ 101 iii (2); as‑cc occurs in Ir., the doubling here is not Brit.r̥kk <r̥k§ 61 i (1);—W.crwth a kind of fiddle,croth ‘womb’, Ir.cruit ‘harp, hump’ < *qrutn‑: Lith.krūtìs ‘woman’s breast’,krūtìnė ‘breast’.—But many doublings attributed to this cause are due to other causes; see Thurneysen Gr. 88.

(3) It seems as ifn + explosive coming after a sonant might become a double explosive, as in W.rhoch ‘snore’: Gk. ῥόγχος, ῥέγκω§ 97 v (3). We havenk >kk >c’h after a nasal in the Bret. mutation afterma ‘my’,nao ‘nine’, asva c’haloun ‘my heart’,nao c’hant ‘900’; but the development is regular in W.

§ 100. i. (1) Ar.i̯- (Lat.j‑, Gk. ῾, Germ.j, Lith.j, Skr.y‑) remained in Pr. Kelt.; it disappears in Ir., but remains in W. Thus W.i̯euanc, Bret.iaouank, Corn.iouenc, Ir.ōac,ōc: Lat. juvencus, O. H. G.jung, E.young, Skr.yuvas̑áḥ ‘youthful’ < Ar.i̯uu̯n̥k̑os;—W.i̯aith ‘language’, Bret.iez < *i̯ek‑t‑: O. H. G.jehan ‘to say’; O. W.I̯ud- ‘*warrior’, W.udd ‘lord’ < *i̯eudh‑;i̯ôn,i̯ôr ‘lord’ < Kelt. *i̯ud-nós,i̯ud-rós§ 66 v: Gk. ὑσμῑ́νη ‘battle’, Skr.yodháḥ ‘warrior’,yúdh id.,yúdhyati ‘fights’; √i̯eudh‑.

(2) Ar.u̯- (Lat.v‑, Gk. ϝ- (lost), Germ.w‑, Lith.v‑, Skr.v‑) remained in Pr. Kelt.; it appears in Ir. asf‑, in W. asgw̯‑. Thus W.gw̯aith f. ‘fois’ (tair gwaith ‘3 times’), Ir.fecht id. < *u̯ekt‑, W.ar-w̯ain ‘to lead’ < *ari-u̯eg‑n-§ 203 iv: Lat.veho, Gk. ἔχος Hes., ὄχος, Skr.váhati ‘conveys, draws, leads’, O. H. G.wagan, E.wain, way; √u̯eg̑h‑;—W.gw̯īr ‘true’, Ir.fīr: Lat.vērus, O. H. G.wār; Ar. *u̯ēros;—W.gw̯edd,gw̯ŷs§ 63 iv;gw̯all§ 99 iii (1).—So beforel orr: W.gw̯lyb§ 58 iv,gw̯lad§ 63 vii (2),gw̯raidd§ 91.

Thoughgw̯r- generally remains, it becamegw̯n- ingw̯nā́ ‘make, do’: Bret.gra, Corn.gwra < *u̯rag‑: cf. Corn.gwreans ‘work’,gw̯rear ‘worker’ < *u̯reɡ‑. In the Oldest W.r remains:guragun tagc (≡gw̯raᵹwn taŋc)b.s.ch. 2 ‘let us make peace‘,wreithb.a. 22 ‘was made’ < *u̯rekt‑; latergwnechl.l. 120,bt. 64 ‘may do’ < *u̯rek‑s‑; Ml. W.goreu ‘did’ < *u̯erāg- < perf. *u̯e-u̯rōɡ-e; √u̯ereg‑: E.work, Gk. ἔργον (ϝέργον). Also ingw̯nī́o ‘to sew’: Bret.gria id., Corn.gwry ‘seam’ < *u̯rēɡ‑, same root; cf. Ir.fracc ‘needle’,fraig ‘osier’: Gk. ῥῆγος, etc. (orig. meaning ‘bend’, hence ‘weave’, hence ‘work’; see Walde s. v.vergo).

Whengw̯r- orgw̯l- is followed by a rounded vowel or-diphthong, it may becomegr- orgl- by dissimilation: W.grug forgw̯rug§ 75 ii;glyw forgw̯lyw§ 102 iii (2).

(3) Ar.‑i̯- and‑u̯- between vowels remained in Pr. Kelt.; they disappear in Ir., but generally remain in W., though sometimes altered; see§§ 75,76, andiii (1) below.

ii. (1) After an initial consonant or was liable to drop from the earliest period§ 101 ii (2); thus W.doe, Lat.heri, Gk. χθές: Skr.hyáḥ§ 98 i (3);—W.dall: Goth.dwals§ 99 iii (2). But remained in Brit. after guttural mediae,§ 92 iv, and afters-§ 94 iv; and remained in some forms. In W. in this position generally becamei; thus W.dī́eu ‘days’ fordi̯eu as in Mn. W.trĭ́di̯au ‘3 days’ (the accentuation implies O. W.di̯‑) < Brit. *di̯ou̯es, < *di̯éu̯es (i̯ou̯ > W.i̯eu§ 76 iii (3)). The hesitation between andi must go back to O. W. when the accent was on the ult. and thei would be unaccented. Lat.i became early, and we have diawl monosyll.§ 34 ii <diab(o)lus, but pl.di|ef|yl[W 5] 3 syll.m.a. i 192a for *di̯efyl <diabolī.

After medial consonants and remained, as in W.pedw̯ar ‘four’ 63 vii (4);—W.celwydd ‘lie’ < *kalu̯íi̯o‑: Lat.calumnia < *calu̯omniā;—W.dedwydd ‘happy’ < *do-tu̯íi̯os: Lat.tuēri,tūtus, O. Icel.þȳða ‘friendship’, Goth.þiuþ ‘good’ noun, √teu̯ē()- (not √teu̯ā- ‘swell, increase’ according to Walde s. v.tueor);—W. pl. ending‑i̯on§ 121 i; verbal suffix‑i-§ 201 iii (6); see alsoiii (2) below.

(2) Between two consonants and had dropped in Brit.; thus W.garr ‘knee’ < *ganr- < *g̑ₑn()r-§ 63 vii (4);—chwann-en < *sqond- < *s‑qon()d- ib.;—golchi < *u̯olk- < *u̯ol()q-§ 89 ii (2).—On‑w̯- which came later between consonants in W., see§ 42.

(3) Betweeni or and a consonant, dropped; as inchw̯ŷd ‘vomit’ < *spi()t‑, √spei̯eu̯-§ 96 iv (1);—W.hoed ‘grief’, Ir.saeth < *sai̯()t‑: Lat.saevus (orig. ‘sore, sad’, see Walde s.v.);—W.oed ‘age’ < *ai̯()t‑: Lat.aetas, olderaevitas. Hence while W. has final‑yw,‑oyw it has no‑ywd,‑oywd,‑ywg, etc.

iii. (1) In Brit., in the diphthongii̯ (ei̯,ai̯), when accented or following the accent, became a spirant probably like Frenchj, which became, and appears so in W. Thus‑íi̯os >‑yẟ,‑íi̯ā >‑eẟ;´‑ii̯- >‑oeẟ§ 75 iv. But the change did not take place inoi̯ orīi̯.

(2) The same change took place afterl orr following the accent; thus´li̯ > *lẟ > W.ll; and´ri̯ > *rẟ ≡ W.rẟ. Examples:li̯: W.gallaf ‘I can’: Lith.galiù ‘I can’;—W.all- inall-fro ‘foreigner’, Gaul.Allo-broges < *ali̯o‑: Lat.alius, Gk. ἄλλος < *álios;—W.gwell ‘better’: Skr.várya‑ḥ ‘eligible’,várīyān ‘better’: O. E.wel, E.well, orig. ‘choice’, √u̯el- ‘wish’.—ri̯: W.arddaf ‘I plough’: Lith.ariù ‘I plough’, Goth.arjan ‘to plough’;—Pr. Kelt.Iu̯ér-i̯on‑,‑iann- > W.Iwerddon ‘Ireland’, Ir. gen.Ērenn;—W.morddwyd ‘thigh’: O. H. G.muriot ‘thigh’;—W.hwrdd ‘a violent push’ < *spuri̯- (ur <u̯ₑr§ 63 viii (1)) √sphu̯erē- ‘hurl, smite’§ 96 iv (1): Lith.spiriù ‘I kick’ (ir <ₑr§ 63 iii); also possibly W.g‑ordd fem. ‘mallet’ (g- excrescent§ 112 ii (2)), O. W.ordox. 2, Bret.orz < *púri̯-ā ‘smiter’: Gk. σφῦρα ‘mallet’ < *σφυρι̯ᾱ; in that case Ir.ordd is from British (a not improbable borrowing, cf. Pedersen Gr. i 22–4).

(3) The change of to * in the above cases took place before the Roman period, for there is no example of it in any word borrowed from Lat. The alteration was therefore earlier than the period of vowel affection, and the * could not affect; hencearẟaf, not *eirẟaf, etc.

The fact that the change does not take place initially corroborates the view that it did not happen before an accented vowel. All forms that occur can be explained under this supposition; thusall- < *áli̯o‑, butail ‘second’ < *ali̯ós, etc.; see§ 165 vi.

iv. Ar.‑mi̯- became‑ni̯- in Pr. Kelt.; as W.dyn ‘man’, Ir.duine < *g̑hðomi̯o‑,§ 98 i (3),§ 121 i;—W.myned, ‘to go’, Ml. Bret.monet, Corn.mones < *momi̯- for *mami̯-§ 65 v (2), by assim. for *bam‑i̯- < *ɡₑm‑i̯‑, √ɡem‑: Lat.venio, Gk. βαίνω both < *ɡₑmi̯ō, Goth.qiman, E.come. The‑i- disappeared before the‑e- of the suffix; the suffix may have been‑at‑,§ 203 ii, which following the accent would become‑et- after, see§ 65 vi (1). The was lost in the compoundsan-fon,dan-fon ‘to accompany, send’, prefix§ 156 ii (1).

v. In some cases metathesis of took place in Brit. Thus Ir.suide ‘soot’ comes from *sodi̯o‑, but W.hudd- inhuddygl ‘soot’ implies *soi̯d‑; O. E.sōt, Lith.sů́džiai ‘soot’ have L°-grade; so W.suddaf ‘I sink’ < *soi̯d- < *sodi̯- beside W.soddaf ‘I sink’,sawdd ‘subsidence’ < *sōd‑, √sed-§ 63 ii.—W.drum ‘ridge’ < *droimm- < *drommi̯- < *dros‑mi‑: Ir.druimm < *drommi- (i-stem): Lat.dorsum < *dr̥s-so‑m, Gk. δειράς < *ders-ad‑, Skr.dr̥ṣ-ád ‘rock, millstone’, √deres‑;—W.turi̯o ‘to delve’ < *toirg- < *torgi̯‑: Lat.porca§ 101 iii (1);—W.ar-o-fun ‘intend’,dam-(f)un-aw,dym-un-o ‘desire’, with‑fun- < *moin- < *moni̯‑: Lat.moneo, √menēi̯‑, extension of √men- ‘mind’;—W.ulw ‘ashes, powder’ < *oi̯lu- < *polu̯i̯‑: Lat.pulvis < *polu̯is;—W.Urien, O. W.Urb-gen§ 25 i < *oirbo-gen- < *orbi̯o‑: Gaul.Orbius ‘heir’, Lat.orbus, Gk. ὀρφανός;—W.wyneb ‘face’, in comp.wynab-r.m. 30 < *einep‑, *einap‑ < *eni̯-əq- (§ 65 vi (1)): Skr.ánīkam ‘face’ < *eni-əq-, √ōq; the un-metathesized form is seen in O. W.einepp, whereein- is from *en()-§ 70 v, since oldei had then becomeui ≡ Mn.wy; O. W.enep, Corn.eneb Bret.enep, Ir.enech show lost, which occurs beforee in Brit., seevi below, and cf.§ 35 ii (2), and is usual in Ir., cf.i above;—W.wybr,wybren ‘cloud’Ỻ.A. 104, 91, ‘sky’, O. Corn.huibren gl. nubes < *eibbr- < *embhri-§ 99 vi (1): Lat.imber gen.imbris (i-stem) < *embhri- (: Gk. ἀφρός ‘foam’, Ir.imrim ‘storm’); without metathesis and with lost, Bret. Van.ebr, Corn.ebron,ebbarn; again, with metath., W.nwyf-re ‘sky’ < *neib- < *nebhi̯o‑; the root is *enebh‑, of which *embh- is FV, and *nebh- is VF; with‑l- suffix,§ 90. (W.nef ‘heaven’ is however from √nem- ‘curve’ hence ‘vault’, as shown by Bret.neñv, Ir.nem; also seen in W.nant ‘vale’ < *nm̥‑t‑.)

vi. drops beforei ore, seeiv, v, above; cf.§ 75 ii (2).


Interchange of consonants

Consonant Alternation.

§ 101. i. Comparison of the derived languages points to certain alternations of consonants in Pr. Aryan; they are mostly the result of dialectal variation, and of the accidents of consonant combination. The same causes produced the same results after the dispersion; and while some of the alternations mentioned below may be primitive, others are certainly later, and some comparatively recent. Three kinds of alternations may be distinguished: (1) the consonant alternates with zero; (2) the manner of articulation varies; (3) the place of articulation varies.

ii. The cases where the consonant alternates with zero are the following:

(1) Initials- before a consonant is variable; thus Gk. στέγος, Lith.stógas ‘roof’, Skr.sthágati ‘conceals’: Gk. τέγος, Lat.tego, W.to ‘roof’; √(s)theɡ‑,—Ir.scaraim, W.ysgaraf ‘I separate’ Lith.skiriù id.: Lat.caro ‘flesh’, orig. ‘piece (of flesh)’, Gk. κείρω, Skr.kr̥ntáti ‘cuts’: √(s)qer‑;—W.chwech ‘six’ < *su̯ek̑s: Armen.vec̣ < *u̯ek̑s;—Lat.spargo, E.sprinkle: Gk. περκνός, W.erch ‘speckled, grey’ < *perq‑,§ 97 v (3). This treatment ofs- persisted long after the dispersion; and many of the examples found are undoubtedly cases of the dropping or the adding ofs- in the derived languages. In Kelt.s- seems to have been added and dropped with a freedom hardly equalled elsewhere.—As‑s was an extremely common ending in Ar., it is natural to suppose that‑s st- would be confused with‑s t‑, so that it would not always be easy to decide whether the initial hads- or not. But some scholars regard thes- as a “preformative” or more or less meaningless prefix; see Schrijnen KZ. xlii 97 ff.

(2) A consonantal sonant after an initial consonant was sometimes dropped. Thus W.chwech, Gk. ῾ϝεξ < *su̯ek̑s: Lat.sex, Goth.saihs < *sek̑s;—Gk. πλατύς, W.llydan, √plethē- ‘spread out, stretch’: without‑l‑, Lat.patēre, Gk. πετάννυμι, W.edau ‘thread’;—W.brau ‘brittle’ < *bhrāɡ‑, Lat.frango, E.break: Skr.bhanákti ‘breaks’, Ir.com-boing ‘confringit’, Armen.bek ‘broken’;—W.cryg ‘hoarse’ < *qri‑q‑,ysgrech ‘scream’ < *s‑qriq-nā, Gk. κρίζω, κριγή, E.shriek, Lat.crīmen, √qrei‑: without‑r‑, W.cwyn ‘complaint’ < *qei-no‑, Ir.cōinim ‘I mourn’, Germ.heiser ‘hoarse’, O. E.hās > E.hoarse (intrusiver);—W.craff ‘sharp’,crafu ‘to scratch’,crach ‘scabs’, E.scrape: without‑r‑, W.cafn ‘trough’ (scooped out), E.scab,shave,shape, Gk. σκάπτω, σκάφος, Lat.scabo, Lith.skabùs ‘sharp’: *sqra‑b‑/​‑bh‑/​‑p‑;—Lat.brevis < *breg̑hu̯is, Gk. βραχύς < *br̥g̑hus: without‑r‑, Ir.berr, W.byrr, Corn.ber, Bret.berr ‘short’ < *bek’-s-ro‑s (with‑ro- suff. like W.hīr ‘long’ < *sē-ro-s); Ir.bec(c) ‘small’ < *beggos with dimin. gemination; W.bach ‘small’ < *bₑg̑h(u)so‑;bychan ‘small’, O. W.bichan, Bret., Corn.,bichan < *biksogno- < *briks- < *br̥g̑h(u)so‑;bechan < *beg̑h(u)so‑, assumed to be f. in W.—Later examples of lost‑r- are E.speak: O. E.sprecan, Germ.sprechen;—W.gw̯aith ‘work’: (g)w̯reith§ 100 i (2);—Guto (ttt) hypocoristic form ofGruffudd.

(3) Between initials- and a sonant, a labial or guttural was liable to drop; thusspr:sr, andsql:sl, etc., Siebs,KZ. xxxvii 285 ff.—W.cleddyf ‘sword’,ar-choll ‘wound’§ 156 i (6),clais ‘bruise’ < *qləd-ti‑,claddu ‘to bury’, √qolād- ‘strike, cut, dig’: W.lladd ‘kill, cut off, mow’, Ir.slaidim ‘I strike, cut’ < *slad- < *sqləd‑;—W.ffrwd ‘stream’,ffrydio ‘to gush’ < *spru‑t‑, Germ.Sprudel ‘fount, gush, flow of water’: W.rhwd,rhewyn, etc.,§ 95 i, < *sru‑;—W.ffroen f. ‘nostril’, Ir.srōn f. ‘nose’ < *sprugnā; withouts- (p…​g >t…​g 86 ii (3)), W.trwyn m. ‘nose’ < *prugno‑s,trywyẟ ‘scent’ < *prugíi̯o‑: Gk. ῥύγχος ‘pig’s snout’ < *srunɡhos§ 97 v (3).—So prob. Lat.scaevus, W.chwith§ 96 iii (2) < *sq‑, by(2) above for *sql‑: Lat.laevus, Gk. λαιός < *sl‑; by(2) *sl- > *s‑, whence W.asswy < *ad-sou̯i̯‑, Skr.savyáḥ; assk̑- alternates withsq‑, seeiv (1), the simple root is perhaps *k̑lei‑: Lat.clīno,clīvus, W.cledd ‘left (hand)’,go-gledd ‘north’. So perhaps Lat.lact- for *slact- for *sqlact‑: Gk. γάλα, W.glas-dwr§ 63 vii (3);—W.ffreub.b. 37 ‘fruit’ < *sprāg‑: Lat.frāgum < *srāg‑.

(4) A semivowel after a long vowel was often dropped: Skr.aṣṭā́u ‘eight’, Goth.ahtau: Skr.aṣṭā́, Gk. ὀκτώ, Lat.octō. The reduced grade may come from either form ; see √uerē()-§ 63 vii (5).

Other sonants might disappear finally after long vowels, as Gk. κύων: Skr.s̑vā́ ‘dog’, Lith.szů̃, Ir., W.ci;—Gk. μήτηρ: Skr.mātā́.

iii. While the place of articulation remained the same, the mode of articulation might vary.

(1) At the end of a root a tenuis frequently alternated with a media. Thus O. E.dȳfan, E.dive < *dheup‑: W.dwfn ‘deep’, Gaul.dubno‑, Lith.dubùs ‘deep’ < *dhub‑, √dheup/b‑;—Lat. gen.pācis: Lat.pangopā̆k̑/g̑‑;—Lat.sparg-o: Gk. περκ-νός, W.erch,ii (1) above;—Lat.plancus, W.talch: E.flake, √pelāq/ɡ-§ 86 ii (3);—Lat. lūceo, Gk. λευκός, W.llug ‘light’: W.go-leu ‘light’, Gaul.Lugu‑, √leuq/ɡ‑—So Lat.porca, W.rhych ‘furrow’ < *pr̯k̑: W.turio ‘to delve’ < *torg̑i̯- (t- forp-§ 86 ii (3)); W.tyrchio ‘to delve’ is a late form fromtwrch = Lat.porcus, prob. allied to the above words despite Armen.herk ‘newly ploughed land’ which implies‑q‑; (Lith.par̃szas ‘pig’ implies‑k̑‑); see§ iv (1).

In the same position an aspirated media alternated with a media:—W.oen ‘lamb’, O. E.ēanian ‘yean’ < *hn‑: Gk. ἀμνός < *n‑: Lat.agnus ambiguous;—Skr.budhná‑ḥ ‘bottom’, Gk. πυθμήν < *bhudh‑: O. E.botm < *bhud‑: W.bôn ‘bottom’ < Kelt. *budn‑ó- ambiguous.

An aspirated tenuis alternated with an aspirated media:—Skr.nakhá‑ḥ ‘nail’: Ir.ingen, W.ewin, Lat.unguis, Lith.nãgas ‘nail’.

(2) Initially a tenuis alternated with an aspirated media, more rarely with a media. Thus W.craidd, Lat.cord‑, Gk. καρδία, Lith.szirdìs, E.heart, Sk.s̑rad‑, all fromk̑‑: Skr.hŕ̥d‑, Av.zərədā, from *g̑h‑;—Ir.cingim ‘I go, stride’, W.rhy-gyngu ‘to amble’, Ir.cēimm ‘stride’, W.cam id. < *k̑n̥ɡh-smen‑: Germ.Gang, E.gang-way, Gk. κοχώνη for *καχώνη < *g̑hn̥ɡh‑, Lith.žengiù ‘I step, stride’ < *g̑henɡh‑; cf. √sk̑eɡ-§ 96 iii (1);—Lat.porcus, Ir.orc,torc, W.twrch, O. H. G.far(a)h < *p‑: O. H. G.barah, O. E.bearh < *bh‑;—O.Lat.dingua, O. H. G.zunga, E.tongue < *d‑: Ir.tenge, W.tafod, Corn.tavot, Bret.teod < *t‑, see§ 92 v,§ 97 v (2);—W.erch ‘grey, speckled’, Gk. περκνός: W.brych,brith ‘speckled’,bwrw ‘cast, sprinkle’, see§ 97 v (3). As in the last equation, several examples occur in W. and Ir. ofb- forp- pointing to the alternation ofp‑:b(h)- before the disappearance ofp- in Kelt. Thus Lat.pūs,puter, Gk. πῦον, πῡ́θομαι, Goth.fūls, E.foul, Skr.pū́yati ‘putrefies, stinks’, √peu(āˣ)‑,pēu‑: W.baw ‘dirt’ < *b(h)eu‑,budr ‘dirty’ < *b(h)eu‑tr‑; also with for,iv (1), Lat.paedor < *pai̯‑d‑, √pēi̯‑: W.baeddu ‘to dirty’ < *b(h)ai‑d- (‑d- present);—Lith.plùskos ‘hair’, O. E.flēos, E.fleece, Ger.Fliess, √pleus‑: W.blew ‘hair’ (mostly of animals, not of man’s head in W., as in Corn, and Bret.) < *b(h)leus‑;—Lat.pasco, Gk. πατέομαι, Goth.fōdjan, E.food, W.yd ‘corn’, Ir.ith id., Skr.pitú‑ḥ ‘food’, √()‑: O. W.bit ‘food’ < *b(h)it‑, Ir.bïad id. < *b(h)ii̯‑, W.bwyd do. < *b(h)ei‑t‑;—Lat.piget, Lith.peĩkti ‘to blame’, O. E.ficol, E.fickle, √peiq/ɡ‑: W.bai ‘blame, fault’ < acc. *b(h)igi̯m̥;—Gk. πέπρωται, ἔπορον, Lat.pars, W.rhan, √perō-§ 63 vii (2): W.barn ‘judgement’ < *b(h)ₑr’n‑,brawd id., Ir.brāth id. < *b(h)r̥̄t- (for meaning cf. Germ.Teil ‘part’:Urteil ‘judgement’). The above alternation may be accompanied by a similar alternation medially; thus Lat.caper, Gk. κάπρος, W.caer-i̯wrch ‘roebuck’, all < *qap(e)r‑: W.gafr ‘goat’, Ir.gabor,gabur, Gaul.Gabro- < *g(h)ab(h)r‑;—Lat.capio, Goth.hafjan, W.caffel ‘to get’ < *qap‑: Lat.habeo, W.gaf-el ‘to take hold (of)’ < *ɡhabh‑.

There seems to have been a later tendency to substitute a media for a tenuis initially before a sonant in Brit. and Goidelic; as in Brit.Britan- for *Pritan-§ 3 iii;—so W.brig ‘top (of a tree), crest (of a wave), hair of the head, border (of a country)’,briger ‘hair of the head’ < *brīk- for *prīk‑, metath. for *krīp- > W.crib ‘comb, crest, ridge (of a roof)’: Ir.crīch ‘boundary of a country’ < *qrī-q-u̯o- broken redupl., √qerēi- ‘separate, divide, cut off’: Lat.crēna ‘notch’,crista ‘crest’,crīnis ‘hair of the head’;—Ir.droch ‘wheel’: W.tro ‘turn’;—Ir.gēc: W.cainc ‘branch’ < *k̑n̥q‑: Skr.s̑ā́khā ‘branch’;—W.gast ‘bitch’:ci ‘dog’§ 96 ii (3).—Cf. W.Grawys,Garawys ‘Lent’§ 138; < Lat.quadragēsima. Still later is the softening of the initial of an adverb, and of a proclitic, asdy ‘thy’; these are regarded as mutated forms, and are not mutated further (except occasionally by false analogy).

(3) Alternations like the above occur also in suffixes; as *‑tro‑: *‑dhro- and *‑tlo‑: *‑dhlo‑.

(4) Thoughl andr are not mixed indiscriminately, several doublets occur in which they alternate, as √g̑hu̯er- /g̑hu̯el-§ 92 iv. These alternations may have originated, as suggested by Meillet, Intr.² 143, in reduplicated forms in which, by dissimilation,r may becomel, or evenn. Thus √ɡerē- ‘devour’ gives *ɡer-ɡel‑, *ɡₑn‑ɡr‑, etc., also withɡ forɡ by dissim.; thus Gk. βιβρώσκω, Lat.vorāre, W.barus ‘greedy’ < *ɡₑr‑: (broken redupl.) Gk. ἔβροξε, Ml. H. G.krage, Ir.brāge, W.breuant ‘windpipe’ < *ɡr̥̄ɡ-n̯t‑: (full redupl.) Lat.gurgulio, O. H. G.querechela, Gk. γάγγραινα: Lat.gula.

iv. The place of articulation might vary.

(1) The different gutturals sometimes alternate. Thus,q/​: √leuq/k̑‑: Skr.rócate ‘lights, shines’,roká‑ḥ ‘bright’, Lith.láukti ‘to expect’, with *‑q‑: Skr.rús̑ant- ‘bright, white’, Lith.lúszis ‘lynx’ with *‑k̑‑;—the suffix *‑qo‑: *‑k̑o‑, as Skr.maryaká‑ḥ (márya‑ḥ ‘young man’) with *‑q‑: Skr.yuvas̑á‑ḥ (yúvan- ‘young’) with *‑k̑‑: Lat.juvencus, W.ieuanc ambiguous;—√ak̑‑/​oq-§ 63 v (2);—√k̑ei‑: √qōi‑: √qei̯ē‑, see Walde s.v.civis. For a large number of examples see Brugmann² I 545 ff. Afters‑,‑q- predominates,§ 84 Note 2; and/​q alternate, as Skr.chinátti ‘cuts, severs’ < *sk̑‑: Lith.skë́dziu ‘I separate’ < *sq‑, √sk̑(h)eid‑/​sq(h)eid‑.

ɡh/​g̑h:—Lat.fī-lum ‘thread’ < *ɡhī‑: W.gī-au ‘nerves, sinews’ < *g̑hī‑;—W.gw̯res, Gk. θερμός, etc. < *ɡh‑,§ 92 iii: Lith.žarýjos ‘glowing coals’, Alb.zjar̄ ‘fire’ < *g̑h‑;—W.gw̯elw ‘pale’, Lith.geltas ‘tawny’ < *ɡh‑: Lith.želiù green, W.glas ‘green’ < *g̑h,§ 92 iii.

Exactly the same change of position as the last is involved in the alternation of and, which occurs in some roots, as √g̑hēu‑: √g̑hēi‑ ‘yawn’.

(2) The Ar. consonant seriesp,t,k,q,q is not a line withp andq as loose ends, but as it were a circle, in whichp andq approach one another.q combines the back with the lip position, and the shifting of the stop to the latter position makes itp. It is not surprising therefore thatq becamep in some languages as W., Osc.-Umb., Gk., or that under certain conditionsp >q,§ 96 iv. Already in Ar. there seem to be some cases ofp alternating withq, and even withq; this takes place beforel, and beforer when it is a variant ofl. Thus we have the parallel roots *pel‑, *quⁿel‑, *qel- ‘to turn’, also withr, *qer. Examples:—*pel‑: Lat.poples ‘bend of knee’, Ir.imb-el, W.ym-yl ‘rim, edge’ < *m̥bi-pel‑, W.cyf-yl ‘border, vicinity’ < *kom-pel‑,ol-wyn ‘wheel’, Gk. πέλομαι < *pel- (sinceqe > τε§ 89 i);—*qel‑: Lat.colo,incola, Gk. τελέθω, πολεύω, W.dy-chwel-af ‘I return’ < *do-sqel‑; redupl. Gk. κύκλος, O. E.hweohl, E.wheel;—*qel‑: Gk. κελλόν· στρεβλόν Hes., Lat.coluber;—qer‑: Lat.curvus, Gk. κορώνη, Ir.cor ‘circle’, W.côr ‘circle, close’,cored ‘round weir’, Ml. W.at-coraf ‘I return’, Ir.cruind, W.crwnn ‘round’.——So the roots *spel‑, *sqel‑, *sqel‑, *sqer- ‘to split, separate, scatter’; thus *spel‑: O. H. G.spaltan, E.split, Skr.sphāṭáyati ‘splits’, Bret.faouta ‘to split’, W.ffloehen ‘splinter’,hollti ‘to split’§ 96 iv (1);—*sqel‑, *sqel‑: Lith.skeliù ‘I split’, Bret.skula, W.chwalu ‘to scatter’, Ir.scāilim ‘I scatter’;—*sqer‑: Lith.skiriù, W.ysgar, etc.ii. (1); also in the sense of ‘snatching’; withp, Lat.spolium: withq, W.ysglyfio ‘to snatch’,ysglyfaeth ‘prey’ < *sql̥‑m‑.——So Gk. πλεύμων, πνεύμων ‘lung’, Lat.pulmo (for *plumō), O. Bulg.plušta, O. Pruss.plauti ‘lung’, the ‘light’ member (cf. E.lights ‘lungs’), W.lluman ‘banner’ < *pleus-mₑn‑: Skr.klóman- ‘right lung’ < *qleumon‑, W.ysgyfaint dual ‘lungs’ < *s‑qumₑn- (l lostii (2), see also§ 121 iv), Bret.skevent, Ml. Ir.scaman (? < Brit.), Ml. W.yscunb. b. 4 ≡ysgwn ‘light, soaring’, O. W.scamn‑, W.ysgawn,ysgafn, Bret.skañv ‘light’ < *s‑qumn-§ 76 vii (4); W.cwhwfan for *cỿ-chw̯ỿfan ‘to wave in the breeze, flutter’ < *ko-squmon‑,chw̯ŷf ‘waving’ < *squmō: √pleu‑/​(pneu‑) ‘float, waft’.

(3) The change ofp tot, which sometimes occurs is doubtless always secondary, as in Skr.ṣṭhī́vati ‘spews’ (: Lat.spuo, E.spew) where the is due to the following palatal, cf. Gk. πτύω < *pi̯ūi̯ō. In Kelt,p becameq beforeq, but sometimest before a palatal or velar§ 86 ii (3), perhaps a compromise between the labial and guttural positions.

Assimilation, Dissimilation and Metathesis.

§ 102. i. Assimilation, dissimilation and metathesis of consonants have taken place at all periods; most of the examples occurring have arisen since the Ar. dispersion. In many cases the change has become a phonetic law; but most of the changes, especially of dissimilation and metathesis, occur only accidentally.

ii. (1) Assimilation of joined consonants: (a) Ar.pd >bd etc.§ 93 i;sd >zd§ 97;ghþ >ghð§ 98.—(b) In most of the derived languagesmt >nt, etc.§ 84, Note 3.—(c) In Kelt.tk >kk, etc.§ 93,ii (2), (3);nl >ll,nr >rr,ln >l§ 99 iii;lẟ >ll§ 100 iii (2).—(d) In W.nt >nnh etc.§ 106,llt >ll§ 105;dẟ >d‑d >t§ 111 vii (2);lᵹ >l l§ 110 ii (2). In Late Mn. W.nff >nth inbenthyg < Ml. W.benffic < Lat.beneficium.

(2) Assimilation of separated consonants: Italo-Kelt.p…​q >q…​q§ 86 ii (2).—Kelt.b…​m >m…​m in *momi̯at- > W.myned§ 100 iv.

iii. (1) Dissimilation of joined consonants: (a) Ar.tt >tˢt§ 87 ii.—(b) When two continuants come together there is often a tendency to alter one of them either to an explosive or to a semi-vowel: thus in Brit.ml- >bl‑,mr- >br-§ 99 ii (1); in W.nẟ >nd as inbendith ‘blessing’,sẟ >sd,lẟ >ld >lld,llẟ >lld§ 111 vii (2);ẟl >dl as inbodlon,ẟr >dr as incadr§ 111 vii (1);mχ >mc as inamcan§ 156 i (4);nṽ >nw̯ as in O. W.anu§ 99 iv (1),rv >rw̯ as insyberw̯§ 105 ii,fl >wl§ 104 v. In many cases the spirant disappeared:fn >n§ 110 iii (4),ẟn >n§ 104 iv (1).—(c) In W.mni̯ >ml inteimlo ‘to feel’ < *teimni̯o < *tamn- < *tang-smen‑: Lat.tango.

(2) Dissimilation of separated consonants: (a) Already in Ar.r…​r >r…​l etc.§ 101 iii (4); andtr…​r >t…​r in *tisores ‘three’ fem. > W.tair, Skr.tisráḥ§ 69 iv.—(b) In Kelt.gn…​n >gl…​n in *glūn- > W.glin ‘knee’§ 63 vii (4);l…​l >r…​l in *arali̯os > W.arall ‘other’, Ir.araile.—(c) In W.gw̯…​ >g…​ inglyw ‘lord’ < *gw̯lyw̯ < *u̯li-u̯o‑s, VR of √u̯elē(i)-§ 63 vii (2);gw̯…​v >g…​v ingreẟf ‘instinct’ (greẟfu ‘to be inbred’) < *u̯r̥d-mā: Ir.frēm§ 91;r…​r >r…​l inChwefrol§ 138 i (2);l…​l >l…​r inllefrith ‘new milk’ for *lle-flith < *lo-vlith ‘*calf-milk’;th…​th >t…​th ingwrtaith ‘manure’ < *u̯er-tek‑t, √theg-§ 92 i;l…​ >l…​d in late Mn W.machlud for Ml. W.ymachluẟ etc.§ 111 vii (3);…​l >d…​l inpedol ‘horseshoe’ for *peẟawl < Lat.pedālis.

iv. (1) Metathesis of joined consonants: (a) Nasalized stems may be the result of the metathesis in Ar. of the suffix‑n- with the last consonant of the root; thus *juɡ‑n- > *junɡ- > Lat.jungo, √jeuɡ‑; if so, forms like Skr.yunákti ‘joins’ are analogical formations which arose in imitation of forms withn as part of the root; but the effect is the same as that which would be produced by an Ar. infix‑ne‑.—(b) In Brit.di̯ >i̯d, etc.§ 100 v.—(c) In W.lg >gl inannwyl ‘dear’ < *induglens < Lat.indulgens;chl >lch inallweẟ ‘key’ for *alchweẟ, Bret.alchouez, for *achl-§ 99 vi (1);nm >mn inamnaid ‘nod’ < O. W.enmeit§ 95 ii (3);dn >nd inandaw ‘listen’ for *adnaw§ 76 iii (1),andwyo§ 76 iv (4).

(2) Metathesis of separated consonants: (a) Ar. *bhudh/d- ‘bottom’ and *dhub- ‘deep’, if not originally the same, are confused in the derived languages: W.annwfn ‘hell’ < *n̥-dub‑n- for *n̥-bud‑n- ‘bottomless’: Gk. ἄ-βυσσος; cf. O. Bulg.dŭno ‘bottom’ and Armen.andundk῾ “ἄβυσσος” withd…​d forb…​d by assimil.—(b) In Kelt.n…​r >r…​n in Gaul.Taranis ‘Juppiter tonans’,Taranu-, W.taran ‘thunder’, Ir.toran ‘din’, < *taran‑, *toran- for *tₑnər- *tonər‑: Brit. (-Lat.)Tanar-o Chester insc. (re-metath.?), O. E.þunor, E.thunder, Lat.tono, Gk. στένω √(s)tenā‑;b…​g >g…‑b in Ir.goba, W.gof ‘smith’ < Kelt. *gobann- for *bog- < *bhog‑: Gk. φώγω, E.bake < *bhog-, Germ.backen < *bhog‑n-, Lat.focus bhō̆k/g‑; in early Kelt. before the loss ofp,kp >pk in W.archen ‘shoe’, Bret.archen < *park- for *karp- < *qₑr’p-qerāˣp- ‘shoe’§ 86 i (5).—(c) In Brit.nl >ln in W.telyn f. ‘harp’, Bret.telen, Corn.telein < *telenī for *ten-el-ī, √ten- ‘stretch’: W.tant ‘harpstring’, Lat.tendo, Gk. τείνω, etc.—(d) In W.l >l inmeẟal ‘soft’ for *melaẟ < *meləd‑: Lat.mollis < *moldu̯is, Skr.mr̥dú‑ḥ ‘soft’, etc. √melāˣ‑; and ineiẟil ‘feeble’ for *eiliẟ,§ 156 i (2):ymlā́ẟ§ 204 i, √lēd- ‘weary, weak’.


British and Latin consonants in Welsh

The Soft Mutation.

§ 103. i. (1) Brit. and Lat.p,t,k,b,d,g,m between vowels becameb,d,g,f,,,f respectively in W. Thus W.Cyndaf < Brit.Cunotam(os);—W.saeth ‘arrow’ < *saᵹeth < Lat.sagitta;—W.deg ‘ten’ < Brit. *dekan < Ar. *dek̑m̥;—W.cybydd ‘miser’ < Lat.cupidus;—W.llafur ‘labour’ < Lat.labōrem. Numerous examples occur in the above sections. The change is called the “soft mutation”.

(2) As the same changes took place generally between a vowel and a sonant (see the details§ 104), and as every initial consonant must be followed by a vowel or a sonant, it follows that where the preceding word ended in a vowel the initial is changed as above; thus while Brit. *oinos markos gaveun march ‘one horse’, Brit. *oinā mammā gaveun fam ‘one mother’, not *un mam.

(3) The conditions are, however, not quite the same initially as medially. Medially‑sk- became ‑χχ- by the reaction of the two sounds on one another before the period of the present changes. But in the case of final‑s and initialk- no reaction took place in the earlier period, and the sounds came down to later Brit. unchanged. It was then too late forsk to give χχ as shown by the retention of Lat.sc, see (5), and of Brit. medialsk fromksk etc.§ 96 iii (5); thus thek- remained, and the final syllable with its‑s ultimately disappeared. For similar reasons final‑s preserved an initial media orm- intact. Hence we have the radical consonant after words or classes of words which ended originally in‑s, such as mas. sg. nouns or adjectives; thus *díi̯ēus dagos >dydd da ‘good day’.

But when the final syllable of the first word was accented, its‑s combined with an initial tenuis, which thus became a spirant. For this reason we have the spirant mutation of a tenuis after Ml. W.ɥ ‘her’ (now writtenei) < *esi̯ā́s = Skr.asyā́ḥ ‘her’;tri ‘three’ < Brit. *trei̯és (for *tréi̯es would have given *trydd);a ‘with’ anda ‘and’ < Brit. *aggós§ 213 iii (1),§ 222 i (3). On the mutation afterni, see§ 217 iv (1); afterchwe§ 108 iii.

tair andpedair had the same accentuation, and in Bret.ter,peder, and alsopevar (=pedwar), cause the spirant mutation. The radical has been substituted in W., as in the majority of cases where the spirant occurred from the above cause.

(4) After final‑s initiall andr were unvoiced; cf.sl- >ll‑;sr- >rh‑,§ 95 i; but between vowelsl andr underwent no change. Thus we havell andrh now in those positions where the radical occurs of the consonants mentioned in(1) above, andl andr in those positions where the said consonants are softened. Welsh grammarians therefore speak ofll,rh as “radical”, andl,r as “mutated” consonants. Though the reverse is historically the case, it is convenient to retain the old terminology in dealing with the interchange of the sounds in the present language.

Note. The term “soft mutation”, first applied to the change where it occurred initially, is due to Dr. Davies, who called it “forma mollis” D. 26. It has also been called “vocal” and “middle”. The latter name, used by Rowland, owes its origin to the term “forma media” used by Davies as a name for the change of the tenues to the mediae; as applied to the six others it is meaningless. Continental scholars use “Lenition” as a term embracing the Welsh “soft mutation” and the corresponding Irish “aspiration”.

(5) Lat.sp,st,sc remained, as Ml. W.yspeil <spolium§ 69 iv (1),ystyr <historia ib.,escyn <ascend‑. An explosive before the group dropped in W., as inestron <extrāneus; so after the loss of an intervening vowel, as W.esgob <episcopus, W.esgud ‘active’ <exsecūtus. See further§ 111 vi (2). Except wherec dropped as above Lat.x >i̯s,108 v.

ii. (1) Medially between vowels, the soft mutation ofg, disappeared completely after the O. W. period; as insaethi (1);—maes < *maᵹes§ 29 ii (2): Gaul.‑magus;—teyrn ‘ruler’ < *tyyrn <tigirn‑; also finally, asda ‘good’ < *dag-§ 63 v (2);—ty ‘house’ <tigos§ 65 ii (3);—bro < *mrog-§ 99 ii (1);—bre (prob. f.) ‘hill’, Corn.bre f. < *brigā, Gaul.‑briga < *bhr̥ɡh‑: Germ.Berg;—bore ‘morning’, O. W.more inb.a. 17 l. 20, Bret.beure < acc. *mārig-an (< *‑m̥): Ir.imbārach, Mn. Ir.mārach < *mārig‑: Kelt. *mārig- < *mōriɡh- L°R₂ of √merē(i)q/ɡh‑: Skr.márīciḥ ‘ray of light’, Goth.maurgins, E.morn.—Already in O. W. we findnertheint (<‑eᵹint), besidescamnhegint (g).

ig givesy, affected toe, as above; it is often assimilated to the following vowel, as indylḗd < Ml. W.dylyet < *dliget-§ 82 ii (3); Ml. W.breenhin ‘king’ < *brigant-īn‑: Skr. acc.br̥hánt-am, gen.br̥hat-áḥ ‘high, great’ < *bhr̥ɡh-ént‑,‑n̥t‑. Beforeei it was lost, as inbraint ‘privilege’, Ml. W.breint < O. W.bryeintl.l. 120 < *briganti̯‑; Ml. W.Seint < *Sigonti̯on ‘Segontium’.—w͡y comes not fromig, but fromeig, as inmod-rwy ‘ring’ < F-grade *reig‑, as inrhwym§ 95 ii (2);mor-dwy ‘sea-voyage’ < *teig‑, Ir.tīagu ‘I go’: Gk. στείχω; socanhorthwy ‘assistance’ < *kanta-u̯er-teig-, lit. ‘*go over with’.—āg gaveeu,au,§ 71 iii.

Initially disappeared completely; but as the initial of the second element of a compound it often becameᵹ̑ > after a dental (d,,n,l,r), asLlwyd-i̯arth < *leito-garto-§ 95 iv (3);Pen-i̯arth < *penno-garto‑,mil-i̯ast D.G. 278 besidemil-ast ‘greyhound bitch’;arw-floedd-i̯ast§ 157 ii (1);Mor-i̯en, O.W.Mor-gen *‘sea-born’;Ur-i̯en, O. W.Urb-gen§ 100 v.

For before and after sonants see§ 104 ii,§ 105 ii,§ 110 ii.

(2) The soft mutation ofm was originally the nasalized spirant. The nasalization generally remains medially in Bret., but disappeared in W. towards the end of the O. W. period. Asf was thereafter the soft mutation of bothb andm, there has always been the possibility of its being referred to the wrong radical. This probably accounts for the substitution in some cases of one for the other, as inbawd ‘thumb’, O. W.maut f. (y fawd ‘the thumb’), still withm- inmod-rwy orig. ‘thumb-ring’. In a few casesm- andb- interchange, asbath andmath (y fath ‘the kind of’),baeddu andmaeddu ‘to dirty’.

Nid adwaen, iawn yw dwedyd,
Weithian ibathyn y byd.—G.I.H.

‘I know not, it is right to say it, her like now in the world.’

Och imi! pe marw chwemwy,
O bydd imathmewn bedd mwy.—D. N.,f.n. 90,c.c. 267.

‘Woe is me! though six times more died, [I doubt] if her like will ever more be in a grave.’

Inbore formore we may have dissim., as inmr- >br‑.

iii. In O. W. softened consonants were represented by the corresponding radicals; see§ 18 i,§ 19 i. It would be wrong to conclude from this that the softening had not then taken place, for its occurrence initially is due in almost every case to a vocalic ending which was then already lost. The difference between the radicalm inun march and the softf inun fam cannot be accounted for if assumed to have taken place since the O. W. period when ‘one’ wasun; it must be referred to the Brit. m. *oinos, f. *oinā. The O. W. spelling was doubtless a survival from the time when the mutated consonant could still be regarded as a debased pronunciation of the radical. On the Ml. final tenues see§ 111 v.

§ 104. i. The mutable consonants,p,t,k,b,d,g,m normally underwent the soft mutation between a vowel and a sonant; thuspr >br in W.Ebrill < Lat.Aprīlis; W.go-bryn-af ‘I merit’ < Brit. *u̯o-prinami, √qrei̯ā-§ 201 i (4);—pl >bl in W.pobl < Lat.pop’lus;—tn >dn in W.edn ‘bird’ < *pet-no-§ 86 i;—tu̯ >dw in W.pedwar < Brit. *petu̯ares§ 63 vii (4);—kr >gr in W.gogr,gwagr ‘sieve’ < *u̯o‑kr‑, √qerēi̯‑: Lat.crībrum;—br >fr in W.dwfr ‘water’§ 90;—bn >fn in W.dwfn ‘deep’ ib.; W.cefn ‘back’ < *kebn‑: Gaul.Cebenna ‘les Cévennes’ (*qeb- allied to *qamb/p-§ 106 ii (1));—dm >ẟf, seeiv (2).

ii. (1)g beforel,r,n gaveᵹ̑, which became forming a diphthong with the preceding vowel. The Mn. developments are as follows:ag >ae;eg >ei orai;ig >i;og >oe;ug >w͡y;āg >eu orau;īg >i. Thus W.aer ‘battle’, Ir.ār ‘slaughter’ < *agr‑: Gk. ἄγρᾱ;—W.draen ‘thorn’ < *dragn- < *dhrₑghn‑: Gk. τρέχνος;—W.tail ‘manure’ < *tegl-§ 35 ii (3), √(s)theɡ- ‘cover’§ 92 i, cf.gwrtaith ‘manure’ < *u̯er-tekt‑;—W.oen ‘lamb’, Ir.ūan < *ognos§ 65 ii (2); W.oer ‘cold’, Ir.ūar < *ogr‑: Gaul. (Seq.)Ogron… name of a month;—W.annwyl§ 102 iv (1);—W. ceulo < *cāgl-§ 71 iii. Examples ofgm are uncertain. Onswyn < Lat.signum see§ 72 ii.

Following the accent,g aftera became and disappeared; as in the suffix‑agno‑, < *´‑o-gno- (*‑o- is the stem vowel, which becomesa in Ir., and when unacc. beforeg in Brit.), as seen in Brit.-Lat.Maglagni,Corbagni,Broccagni giving W.Maelan,Carfan,Brychan; Ir.-ān asBroccān; so O. W.bichan, W.bychan, Ir.becān.

For the affected forms of the above groups see§§ 69,70.

(2)gi̯ >ᵹ̑ >; thus W.cae ‘enclosure, field’ < *kagi̯o‑, Gaul. 5th cent.caium, whence Fr.quai, √kagh‑/​kogh‑: Lat.cohus, E.hedge, Germ.Hecke;—Ml. W.daeoni ‘goodness’ < *dag-i̯ono-gnīm- (re-formed asda-ioni in Mn. W.). It is seen that the vowel is not affected by the, but it may be by a followingī <ō; thus W.llai, Ml. W.llei ‘less’, Ir.laigin, both < *lagi̯ōs < *lₑɡh()i̯ōs: Lat.levis, Gk. ἐλαχύς;—W. ‑(h)ai, Ml. ‑(h)ei < *‑sagi̯ō§ 121 i,§ 201 iii (4).—Soigi̯ affected bya gaveegi̯ becoming‑ei,‑ai, as W.tai, Ml.tei ‘houses’ < *tigi̯a < *tigesa, pl. of *tigos ‘house’;—W.carrai ‘lace’ < Lat.corrigia. When unaffected,igi̯ gaveii̯ >ī; as inbrī ‘honour’ < *brigi̯o‑:brenin,braint§ 103 ii (1); andllī́on in Ml. W.Kaer-llion < *ligi̯ŏnos, Brit. gen. for Lat.legiōnis.

Similarlyogi̯ >oe >‑o,§ 78 i (1), into ‘root’ < *togi̯o‑: Ir.tuige gl. stramen, andamdo ‘shroud’ < *m̥bi-togi̯o‑: Ir.im-thuige ‘clothing’: Lat.toga, √(s)theɡ‑.—ugi̯ >w͡y >-w,§ 78 i (2), inllw ‘oath’ < *lugi̯on: Ir.luige,lugae < *lugii̯on. (Ml. W. pl.llyeu,llyein, Mn.llwon, dial.llyfon are all analogical formations.)

(3)gu̯ > >w: W.tew ‘thick’ < *tegu̯-§ 76 viii (2).—og, >ou̯i̯ >eu ineuod ‘worms in sheep’ < *ogi̯- < Ar. *hi̯‑: Gk. ὄφις, Skr.áhiḥ ‘snake’.

iii. (1) Beforen Brit.k >χ̑ >, so thatkn gives the same result asgn; thus W.dwyn ‘to bring’ < *duk‑n-§ 203 iv (3);—braenu ‘to rot’ < *brakn- < *mrəq‑n-§ 99 ii (1);—croen ‘hide, rind’ < *krokn‑, Bret.croc’hen, Ir.crocenn < *krokn- (kn >kk) < *qroq‑, VF° of *qereq- broken redupl. of √qer- ‘divide, rip’: Lat.corium,cortex, O. Bulg. (s)kora ‘rind’,korĭcĭ a kind of vessel, W.cwrwgl ‘coracle’;—W.gwaun < *u̯ākn- < *u̯o-akn‑: W.ochr see below;—W.tīn ‘buttock’ < *tīknā < *tūqnā, Ir.tōn < *tūknā: E.thigh O. H. G.dioh. This may be due to gemination ofk, see(2) below; in many caseskn >gn regularly; thus W.sugno ‘to suck’ < *seuk‑n‑, √seuq/ɡ‑: Lat.sūcus,sūgo, E.suck, etc.;—W.dygn ‘grievous’ < *dikn- < *deŋgn‑: Ir.dingim ‘I press down’, O. E.tengan ‘to press’;—W.rhygnu ‘to rub’ < *rukn‑: Gk. ῥυκάνη;—W.dogn ‘portion, dose’ < *dok‑n‑, √dek̑‑: Gk. δέκομαι, δοκάνη· θήκη.

(2) Beforer,k,t giveg,d regularly, as ingogri above;—chwegr < *su̯ekr-§ 94 iv;—W.deigr ‘tear’ < *dakrū§ 120 iii (1);—W.aradr < Ar. *arətrom§ 87 i;—W.modryb§ 69 ii (4); etc.

But W.ochr ‘edge, side’ beside Ir.ochar < *okr‑, √ak̑‑/oq‑, W.rhuthr ‘rush’ beside Ir.rūathar < *reu-tro‑, √reu̯‑: Lat.ruo, implykkr,ttr forkr,tr§ 99 v (4). Compounds likego-chrwm:crwm ‘bent’ may owe theirch to this, or tos beforek.

An example ofk <gg giving the same result is Ml. W.achreawdɏrb.t. 9 ‘gathering’ < Lat.aggregātio, with excrescent‑r; cf.cyngreawdr <congregātio inCyngreawdɏr Fynydd (‘Mount of Assembly’) ‘The Great Orme’.[18] Similarlyg beforer may be treated asgg and giveg, as inllygru ‘to injure, violate, corrupt’: Gk. λυγρός, Lat.lugeo, Skr.rujáti ‘breaks’, Lith.lúžti ‘to break’, √leuɡ/‑.

iv. (1) Brit.dn > W.n (not *nn); as in W.bôn ‘stem’ < *bud-nó‑,bonedd ‘nobility’ < *budníi̯ā: Ar. *bhudh- ‘bottom’§ 102 iv (2);—W.blynedd < *blidníi̯ās§ 125 v (i).

(2) Brit.dm > W.ẟf; as W.greddf ‘instinct’§ 102 iii (2);—W.deddf ‘law’ < *dedmā < *dhedh-mā, √dhē‑: Gk. τεθμός, θεθμός < *dhedh-mos;—W.add-fwyn etc.§ 93 ii (3), q. v.

(3) Brit.dl,dr after a back vowel becameẟl,ẟr; the remained after the accent, and was provected tod, ashadl,cadr§ 111 vii (1), and disappeared before the accent, as iniôr < *i̯ud-rós§ 66 v. After a front voweldl,dr >gl,gr, and developed accordingly,ii (1); thus W.cadair, Ml.kadeir < Lat.cat(h)edra;—W.eirif ‘number’ < *ed-rīm- < *ad-rim‑[W 6]: Ir.āram;—W.i waered ‘downwards’ < *di woiret < *do upo-ped-ret‑;gwael ‘base’ < *upo-ped-los, √ped- ‘foot’;—W.aelwyd ‘hearth’, Bret.oaled, O. Corn.oilet < *aidh-l-eti̯‑: Gk. αἴθαλος ‘soot’, Lat.aedes, √aidh- ‘burn’, cf.§ 78 ii (3).

v.bl >fl orw̯l, as ingafl ‘fork’: Ir.gabul, Lat.gabalus < Kelt.; Ml. W.nywl§ 90,di̯awl§ 100 ii (1).—ml,mr§ 99 ii.—mn§ 76 vii,§ 99 iv.

Other groups of explosive + sonant are regular.

§ 105. i. Afterr Brit. and Lat.p,t,k become respectivelyff,th,ch; thus W.corff < Lat.corpus;—W.gorffwys§ 89 ii (2);—W.porth < Lat.portus;—W.archaf§ 63 iii, etc.

lk >lch, as W.golchi§ 89 ii (2);—W.calch < Lat.calc-em.—lp >lff, as W.Elffin < Gallo-Lat.Alpīnus.—lt >llt, as in Ml. W.kyfeillt ‘friend’ = Ir.comalte ‘foster-brother’ < *kom-alt(i)i̯os; W.allt ‘declivity; grove’ < *alt‑, √al- ‘grow, nourish’: Lat.alo,altus; medially it becomesll as in W.cyllell ‘knife’ < Lat.cultellus; W.di-wylli̯o ‘to cultivate’:gwyllt ‘wild’§ 92 iv; except in re-formations, as inhollti ‘to split’ fromhollt§ 96 iv (1); thet is sometimes lost finally in an unstressed syllable, as in Mn. W.cyfaill, Ml. and Mn.deall§ 75 vi (4).

ii.rb >rf, as in W.barf ‘beard’ < Lat.barba; alsorw̯, as insýberw̯ ‘proud’ < Lat.superbus.—rd >rẟ, as inbardd < Brit. *bardos (βαρδοί· ἀοιδοὶ παρὰ Γαλάταις, Hesych.).—Mediallyrg >ri̯ as inari̯an ‘silver’ = Ir. airget < Kelt. *argn̥t-om: Lat.argentum, Skr.rajatá‑m: Gk. ἄργυρος, √areg̑‑. Finallyrg >‑r,‑rỿ,‑ra,‑rw̯§ 110 ii.

lb >lf, as ingylfin ‘beak’, O. W.gilbin: Ir.gulban id. < Kelt. *gulb‑.—Mediallylg >li̯, as indali̯af§ 110 ii (2); for finallg see ib.—Mediallyld >ll as incallawr ‘caldron’ < Lat. ´caldārium;—finallyllt as inswllt ‘money, shilling’ < Lat.sol’dus.

iii.rm >rf orrw̯§ 99 ii (2);—lm >lf, ib.;—nm >nf ornw̯§ 99 iv (1).

The Nasal Mutation.

§ 106. i. (1) A nasal before an explosive was assimilated to it in position where it differed; thus Ar.k̑m̥tóm ‘100’ > Brit. *kanton; Ar. *penqe ‘5’ > Kelt. *qeŋqe > Brit. *pempe. This may be assumed to have taken place in Late Brit. when the nasal ended one word and the explosive began the next if the syntactical connexion was a close one. Subsequently a media, or (later) a tenuis, was assimilated to the nasal, becoming itself a nasal. This is called the “nasal mutation” of the explosive.

The order of the changes was the following:yn ‘in’ +Bangor first becameym Bangor, and thenym Mangor. The recent spellingyn Mangor is therefore not only a misrepresentation of the present sound, but a falsification of its history.

(2) There is a sporadic assimilation ofn toi in the groupsĭn orein, then becomingŋ̑; thuspringhafr.p. 1278, spv. ofprĭn ‘scarce’ ib. 1280 (< *qrīt-sno‑s:prid ‘precious’, √qreiā- ‘buy’);meithring (‑iŋ̑) D.G. 69 formeithrin ‘to nourish’;Einion is often writtenEingion orEingnioneiŋ̑on, which has becomeeŋŋan in Gwynedd, e.g.Llan-engan near Pwllheli.

ii. (1) Brit.mb,nd,ŋg became respectivelymm,nn,ŋŋ: they remain so in W.,mm being generally writtenm;nn finally written-n (but-nn in monosyllables in Ml. W.);ŋŋ writtenng (and Ml. W.gg org); see§ 51 iv,§ 54 i (2). Thus W.cwm ‘valley’ < Brit. *kumbo-, √qeub/p- ‘curve’: Lat.cūpa,‑cumbo, Gk. κύμβος, etc.;—W.cam ‘bent, crooked’ < Brit. *kambo-: Gaul.Cambo-dūnum, Gallo-Lat.cambiāre: Lat.campus (orig. ‘vale’), Gk. καμπή, κάμπτω, √qamb/p- ‘curve’;—W.twnn ‘bruised, broken’ f.tonn < Brit. *tund-os,: Lat.tundo, Skr.tundate ‘strikes’, √(s)teu‑d‑;—W.tonn ‘wave’ < Brit. *tundā < *tum-dā: Lat.tumeo, W.tyfu, √teu̯āˣ- ‘swell’;—W.cann ‘white’,cannu ‘to whiten’,llóer-gan ‘moon-lit’ < *qand‑: Lat.candeo, Gk. κάνδαρος < *qand‑, beside W.cynneu ‘to kindle’,cynne ‘a burning’,cynnud ‘firewood’, Ir.condud < *qond‑: Skr.cand‑,s̑cand‑ ‘shine’ < *(s)qend‑: √sqand‑/​sqend‑;—W.llong ‘ship’ < Lat.longa;—W.angel < Lat.angelus.—So before a sonant, asCymro pl.Cymry < Brit. *kom-brog-os,‑ī;—W.amrwd ‘raw’:brwd§ 63 vii (4);—Cyngreawdr§ 104 iii (2);—except where the nasal has become a media§ 99 vi (1).—The double nasal was simplified after an unaccented syllable§ 27 ii, and before a sonant§ 54 i (3).

Kelt.ng (< Ar.h) was unrounded and gaveŋŋ, as inllyngyr,angerdd§ 92 v. Whenŋŋ came before a sonant, including, it was first simplified toŋ and then lost, as inewin,tafod, see ib. So we havenawraẟr.p. 1331,g.r. [372] ‘nine degrees’ < Brit. *nou̯aŋ-grad- (navg̃raẟb.b. 42 may have oldŋ, but is prob. analogical);—W.cyni (onen) ‘trouble’ < *koŋnīm- < *kon-gnī-mu-§ 203 vii (4);—W.aren ‘kidney’, Ir.āru < *aŋr- < Kelt. *aŋgr- < *anɡhr-, √aneɡh‑: Gk. νεφρός, Lat. Praenest.nefrōnēs, Lanuv.nebrundinēs (: Lat.inguen withg, Walde s.v.). But aftere ori and beforer orl, theŋ becameŋ̑ and gave, as ineirin Deut. xxiii 1 for *eiryn,§ 77 iii, < *eŋryn pl. ofaren above;—W.cilydd ‘mate’ < *ciŋ̑lyẟ (ii̯ > W.i noty, cf.§ 104 ii (2)) < *keŋglii̯os (§ 65 iii (1)) = Ir.cēle < *keŋglii̯os: Ir.cingim ‘I go’, W.cam ‘stride’, see§ 101 iii (2); for meaning, cf. Ml. W.keimat ‘mate’ <cam. The rule only applies to old formations where theŋŋ already existed in Brit.; in newer formations, and Lat. derivativesŋ remains, asCyngreawdr above.

(2) The above changes took place before the loss of Brit. syllables, fornd coming together after the loss of a syllable remains, as intrindod < Lat.trīnitāt-em. Initial mediae were assimilated to final nasals before the latter were lost; e.g.naw mlynedd ‘nine years’ < *nou̯am mlidníi̯as < *neu̯n̥ bl‑.

Every Brit.nd becamenn, so we have no words ending innd except where a vowel has been lost in the Mn. period between the sounds, as inond etc.§ 44 vi; seeiii (4).

iii. (1) Brit.mp,nt,ŋk remained finally as in W.pump,pymp < Brit. *pempe; W.cant < Brit. *kanton; W.ieuanc < Brit. *i̯ouaŋkos§ 100 i (1). For exceptions see(2). Medially they becamemmh,nnh,ŋŋh respectively, as in Ml. W.ymherawdɏr < Lat.imperātor; W.cynhesu ‘to warm’ < Brit. *kon-tess‑, √tep‑,§ 96 ii (5); W.angheuol ‘deadly’ < Brit. *aŋkou̯‑, √anek̑‑: Lat.neco, Gk. νέκυς, νεκρός, etc. After an unaccented vowel the nasal is simplified as in the above examples,§ 27 ii; after an accented vowel the aspiration was lost, as incynnes ­‘warm’,angeu (≡aŋŋeu) ‘death’§ 48 ii,iv.

(2) Finalnt,mp are mutated ingan ‘with’ ≡gann§ 211 iv (1); incant ‘100’,pump ‘5’ which appear ascann,pum before nouns; inugeint ‘20’ which appears asugeẏn as early asa.l. ms. a. see i 4, 8, 12, etc., and isugain in Mn. W.; inarẏanta.l. i 6, nowari̯an ‘silver’; indiffrint (i ≡ y)b.b. 91 ‘vale’ ( < *dyfr-hynt ‘water-way’),alreadydiffrin inb.b. 74, Mn. W.dỿffrɥn; incymaint sometimes, especially in the phrasecymain un Eph. v 33; and often in poetry, as always in the spoken language, in the 3rd pl. of verbs and prepositions§ 173 x,§ 208 iii (2). It is seen in these examples that theh of the nasal mutations oft andp is lost finally; this is because it follows the accent of the word, see(1) above. But the aspirate was often retained before a word beginning with an accented vowel, askymein hunỺ.A. 116 ‘every one’;can hwrw.m. 136 ‘100 men’;Pum heryr ‘5 eagles’ G.Gl.m 1/606.

Gwledd echdoe a doe’n i dy,
Gwleddcann hannedd cyn hynny.—G.Gl.m 146/278.

‘A feast yesterday and the day before in his house, the feast of a hundred dwellings before that.’

Llyfr Ofydd a fydd i ferch,
Ag yn hwnugain hannerch.—B.Br.,Ỻ.H. ii 99.

‘The maid shall have a book of Ovid, and in it a hundred greetings.’

Final‑ŋc was often mutated in Ml. W. where the tenuis was generally retained, and survives in Mn. W.; e.g.ceingw.m. 108, Mn. W.cainc ‘branch’.ŋc is often writtenngc (cf.§ 18 iii), butnc is adequate and unambiguous, asnk in Eng.bank.

(3) Medialnt, etc. remain when originally followed byh as incyntedd ‘porch’ for *cynt-heẟ < *kintu-sed-§ 63 ii;cyntaf ‘first’ < *cynt-haf < *kint-isamos; and in newer formations, asplentyn ‘child’ fromplant,llanciau ‘lads’, sg.llanc. Some vocables, with mutation in Ml. W., are re-formed without mutation in Mn. W., asamranneuw.m. 41,amrantau Job xvi 16;seinnẏeu§ 128 ii, Mn. W.seintiau ‘saints’;gwynnoeẟỺ.A. 5,gwyntoedd Matt. vii 25;heinẏeuỺ.A. 123,heintiau Luc xxi 11;ceig̃heu,ceingeuỺ.A. 144,ceingciau Can. vii 8.

(4) The nasal mutation of the tenues does not date from the Brit. period, for the nasal endings of *nouan ‘nine’, *dekan ‘ten’, etc., while they mutated initial mediae, did not mutate initialp,t,k; thusnaw cant ‘900’,deg pwys ‘10 lbs.’ The mutation of the tenues was caused by nasals which survived the loss of the Brit. endings; it takes place after the prefixesan‑,cyn‑, and in other cases wheremp,nt,ŋk occurred medially.

There is no trace in O. W. of an wnmutated media; we find e.g.am- for Mn. W.am- < *m̥bi‑,scribennm.c. < Lat.scrībend‑,crunn-m.c. ‘round’ (: Ir.cruind), etc., but nomb,nd. But the tenues are found unmutated, as intantou, Mn. W.tannau, sometimes mutated as inbronannoum.c., pl. ofbreuant ‘windpipe’. Inpimphetox. ‘fifth’,hantherox. ‘half’ is perhaps reflected the transition stage in which, as thep andt were disappearing, theh was becoming more noticeable; see§ 107 v (1). In any case it is safe to conclude that this mutation came about in the O. W. period.

In Ml. W. the tenuis is mutated, as inbreenhinb.b. 75,§ 103 ii (1),ag̃heu,ag̃henb.b. 23,emen etc.§ 24 i. Though often written unmutated after a prefix and afteryn, there is evidence that it was in fact mutated,§ 107 iii,v.

iv. The nasal mutation of an explosive does not mean its disappearance, but its conversion into a nasal by the loosening of its stop. Inannoeth ‘unwise’ < Brit.-Lat. *an-doct- thed became a continuation of then, so thatnn represents ann which is continued during the time it took to pronounce the originalnd. As the W. tenues are really aspirated, that istt‑h, see§ 84 Note 1, when the stop was loosened the aspirate remained; thusnt, properlynt‑h, becamennh. That Early Ml. W.nh as insynhuir§ 48 iv is short fornnh, is proved (1) by such spellings asmorcannhuc,brennhinl.l. 120, and (2) by the fact that when it lost its aspirate after the accent it appeared as nn, assynnwyrr.m. 13,w.m. 20, whilebreenhin in whichnn had becomen after the long vowel, isbrenin (not *brennin), and an original singlen +h always givesn, as inglánaf forglánhaf, superlative ofglân ‘clean’. It is clear therefore that the mutation ofnt is strictlyn‑nh, notn‑h.

§ 107. i. While initial mediae are nasalized after several numerals, initial tenues are nasalized only afteryn ‘in’ andfy ‘my’, and this mutation is not original afterfy.

ii. Taken in conjunction with the following noun,yn ‘in’ ( < Brit. *en) has a secondary accent, butfy ‘my’ (< Brit. *men < Ar. *mene gen. sg. of the 1st pers. pron.) is wholly unaccented—the emphasis when required is thrown on an auxiliary pronoun : ‘my head’ is not *fymhen, butfy mheni. This difference betweenyn andfy is old, for Brit. *en has kept its‑n, but *men (already a proclitic in Brit.§ 113 ii) had lost its‑n before the O.W. period. This is clearly seen is phrases where the following word began with a vowel or an immutable initial; thusyn:ynn lannl.l. 120,in alldb.b. 64,in llan do. 63, 64,yn amgant do. 66,in llurv do. 65, etc.; butfy:mi-hunm.c.,vy argluitb.b. 51,wi-llav-e (≡fy llaw i) do. 50,vy llen do. 59, 62, etc. Thusyn before a consonant is necessarily a closed syllable, closed by its‑n, whilefy is an open syllable, ending with its vowel. The O.W.nyl.l. 120 ‘in its’ is probablyn̥ y, with syllabic orn̥n, a pronunciation still often heard.

iii. Afteryn in Early Ml.mss.,b andd are generally mutated, and probablyg is to be readŋ. Thus inb.b. we findinnechreu 29,innvfin (≡yn nwfn) 87,inyffrin 65,inyganhvy 47,yg godir,ygodir 63; ina.l. ms. a.eniokel (≡yn niogel) i 46, 50,emon e kolouen (≡ỿm môn ỿ golofn) i 10. Non-mutation is rarer:ym brinb.b. 33,in diffrin 47, 48. On the other handp,t,c are rarely mutated, the usual forms beingin tyno,im popb.b. 33,ym pob 87,im pen 42, 57,impell 82,yg coed, 49;en ty e-clochyda.l. i 52,en-tal e-ueig̃ 72. But examples of mutation also occur,mh,nh,ngh appearing at first asm,n,g24 i, asymlithb.b. 20,in hal art do. 49,eghyd (≡ỿnghɥ̄́d)a.l. i 40,emop lle do. 60. These examples show that the mutation had already taken place, and that the written radical was a survival of O.W. spelling. It is to be noted that then ofyn is in every case assimilated in position to the explosive, even where that is unmutated. So beforem, asim monb.b. 61,im minit eidin do. 95.

iv. Sinceyn kept its nasal, it is natural that it should mutate tenues as well as mediae; but asfy lost its nasal ending early, we should expect it to mutate the mediae but not the tenues, likenaw, which givesnawmlynedd ‘9 years’, butnawpwys ‘9 Ibs.’ In O.W. and Early Ml. W. this is, in fact, the case. Thus in O. W. we havemi-telu ‘my household’,mi coueidid ‘my company’,juv. sk. (9th cent.); and inb.b. we findvy tud 13,vy perchen,vy parch 42,wy clun 49,vy pen,vy crawn 62,vy penhid 81,vy ki 99 ; the formwympechaud 83 is a rare exception, and in no case is the tenuis nasalized. Butb andd are generally nasalized inb.b.,g being also probably forŋ; thusvy nruc 24,wy-uragon 51,vi-mrid (≡fy mryd) 82,wi-nvywron (≡fy nw͡yfron) 100,wy-nihenit 50,vy martrin 67. The occurrence of a number of examples likevy martrin 67,wy duu 82,vy dewis,vy Devs 42, is probably due to the influence of the regular non-mutation ofp,t. We do not seem to meet with such forms asvyn drwc,vym bryd which appear in latermss.;vy is written as an open syllable, andp,t,k are not mutated after it. The later mutation of these is analogical; the mutation caused byfy in the mediae was extended to the tenues in imitation of the complete and consistent system of mutation afteryn.

But in spite of the levelling of the mutation after the two words, the difference between the words themselves—the closedyn and the openfy—remained, and persists in the ordinary spelling of to-day, as inyn nhy fy nhad ‘in my father’s house’.

v. (1) The representation of the nasal initial mutation afteryn andfy has presented considerable difficulty to writers of the language. In Late Ml. W.mss.p,t,k appear unmutated, andfy is treated asfyn; thusyn ty vynntat iỺ.A. 35. That this is a conventional spelling is shown by the fact that scribes so rendered forms already mutated in their copies. Thus wherea.l. ms. a. hasemen i 84, the laterms. b. hasem pen. Similarly ther.b. scribe writes down the radical of a consonant mutated in the same passage in thew.b., asvyg̃hofw.m. 104 =vyg̃ cofr.m. 76,vymhechawtw.m. 399 =vym pechawtr.m. 255, etc. Further, the cynghanedd always implies the mutated form; as

yn-trugareẟ yn r͑i gwirion,r.p. 1216;
o syrth ym-perigɏl swrth amharawt,—do. 1250;

wherentr is to be readnhr to correspond tonr͑, andmp must bemh to answermh. Inw.m. andw. we sometimes find a survival of the curious transitional form met with in O. W.§ 106 iii (4); thusymphenw.m. 256,vyg̃chret do. 390;vyg̃ khofw. 7b. The last example shows that what is meant is not the voiceless spirant, for χ is never writtenkh.

(2) The mediaeb,d also are frequently written unmutated, especially afteryn; thusyn dibenw.m. 129 madeyn niben inr.m. 202; converselyymlaenw.m. 54 madeym blaen inr.m. 38; both haveymon colofɏnw.m. 181,r.m. 84. Here again the cynghanedd belies the non-mutation, as in

yg-karcharyn-daear ynyt,—r.p. 1168,

where we must readyn naear (to give/ as required by the cynghanedd sain). Withyn,g is generally doubled, as inyg̃g̃ovot,yg̃g̃wyẟw.m. 123, but is sometimes single, esp. before, as inyg̃ualesw.m. 57; in all cases it is doubtless to be readŋ. Afterfy the single nasal is used; thus inw.m. we havevy mot 32,vy marɏf 59,vy mrawt 62,vy-g̃wreic 62,vy ni waradwyẟaw 43; more rarely the nasal and mute, asvym-brawt 51,vyn da 459. It is seen that in spite of inconsistencies, the difference between closedyn and openfy is unconsciously reflected in these spellings.

(3) Inmss. of the 15th and 16th cent. the consonant is regularly mutated, and the two words are generally joined; thus in the Report on the Peniarthmss., we findynghaer llion 50/90,ymyellt,ynghaer 53/126,ymorgannwg 54/37,vymod 54/21,vyngwallt 54/280,ymhob 54/209,vymhennadur 57/27. Sometimes the words are separated; thusyn nef 75/172;ym hob 54/250, 61/18, 67/330;y mendith (’y forfy) 54/78 ;vy nolur 56/72.

(4) Salesbury wrotevi-dew,vi-popul forfy Nuw,fy mhobl, “to saue the word the les maimed,” as he explains (1586 Pb. Preface). G.R. mutated the consonants and joined the words,fynhy 41,ynnhy 79; he states thatm is double—“ymhob a leissiir ymmhob” 80 (see54 i (2)). His reason for joiningfy appears to be thatng cannot be initial, “canys rhy anoẟ yw sillafu fy ngwaith, fy nghaws” 42. Dr. Morgan separated the words in the case ofn andm; asfy nhŷ Job xix 15,yn nhŷ do. i 13,fy mhen xxix 3,ym mha beth vi 24; but he appears to think like G.R. thatng cannot be initial, and writesfyng-halon xxxvii i,yng-hilfach xxxviii 16, thus missing the distinction which he elsewhere observes betweenyn andfy, and wrongly representingfy as a closed syllable. The prejudice against initialng was overcome in the 1620 Bible, andfy nghalon was written as freely asfy nhy. That settled the matter as far asfy was concerned.

But the representation ofyn in the same combination still presented a difficulty. Theng (≡ŋ) was part of the preposition; at the same timengh orng was the initial of the noun, and Dr. M.’s hyphen in the middle of the trigraphngh was absurd ; the 1620 Bible therefore usedynghilfachau, returning to thems. forms. Hereng does double duty, the inconvenience of which appears when the noun requires a capital initial. Dr. M. wroteyng-Hrist; M.K. hasyngHymry p. [iv]; the 1620 BibleynGhrist 1 Cor. xv 18, 19, 22; so in the Bibles of 1677 and 1690. Later, we findyng Haerlŷdd T.J. title (1688);yn Ghymrurh.b.s. dedic. (1701);Yngroeg S.R. 16 (1728). In all these the capital is misplaced by being either put in the middle of the trigraph or transferred to the preposition. The formyn Ngh- which appears about this time, seeb.cw. Ixxv, grew out ofyn Gh- because it was felt that the initial wasNgh‑; it is objectionable because n is not accepted as a symbol forŋ except beforek org. The later formy’ Ngwyneddd.g. 41 (1789) misrepresents the preposition as an open syllable. Pughe adoptedyn Ng‑,yn M‑, because, in the teeth of all the facts, he denied that then ofyn was mutable. This unphonetic spelling, which stultifies the history of the nasal mutation,§ 106 i, has predominated since his day.

J.J. wroteyng ŋolaup 312/iv/1r., and Dr. Davies pointed out in 1621 thatynghanol was short foryng-nghanol D. 202; but it was not until about a hundred years later that the formyng Ng(h)- came into regular use. We findyng Nghrist in the 1717 Bible, and subsequently in those of 1727, 1746, 1752, and nearly all later editions. This form has been used and advocated by most of the Welsh scholars of the 19th cent., including Iolo Morgannwg (who denounces “dull ffiaidd Mr Owen Pughec.b.y.p. 237), R. I. Prys, T. Stephens, T. Rowland, and Silvan Evans.

(5)Fy being unaccented, the following nasal, though of double origin, is simplified, and belongs to the second syllable§ 27 ii, i; thus the syllabic division isfy|núw. As words are separated in modern orthography, the usual spellingfy Nuw is in every way correct. Similarlyfy merch,fy ngardd. Butyn is accented, and the double consonant remains, extending to both syllables§ 27 i; henceýn|núw, ordinarily and correctly writtenyn Nuw. In the same way we have ym Mangor,yŋ Ŋwynedd. With our present alphabet we have to write the lastyng Ngwynedd; soyng Nghadelling. It is objected to this that it is clumsy; but that is the fault of the alphabet. It is the only way of expressing the sound fully and correctly, and is the exact equivalent in modern characters of the Ml. W.yggwynedw.m. 108,yg gadelligw. 90,§ 24 i.

(6) There are, however, a number of adverbial and prepositional expressions, in whichyn, followed by the nasal mutation, is wholly unaccented. In this case the nasal is single, as afterfy; and the preposition is naturally joined to its noun, exactly like thein in the Eng.indeed. These expressions areynghyd,ynghylch,ynglŷn,yngholl,ynghudd,ymhell,ymhlith,ymysg,ymron,ymlaen,ymhen,yngham,ymhellach,ynghynt, etc. No principle of accentuation is violated in this spelling, as asserted by Silvan Evans, Llythyraeth 50, who recommendsyng nghyd etc. See above§ 47 ii.

The Spirant Mutation.

§ 108. i. Brit. or Lat.pp,tt,kk gave W.ff,th,ch respectively. Thus W.cyff ‘stem’ < Lat.cippus;Brython < Brit.Brittones;pechod < Lat.peccātum;hwch: Ir.socc, etc.,§ 93 iii (2). It occurs when an initial tenuis follows an explosive in word-composition, as inachas§ 93 ii (2),athech§ 93 iii (1),athrist§ 99 v (4). This is called the “spirant mutation” of the tenuis.

ii. In Brit.s + tenuis had already become a double spirant§ 96 i; and original oxytones ending in‑s caused the spirant mutation of a following initial tenuis§ 103 i (3), astrichant ‘300’. In this caseth- andph- were chosen as the mutations oft- andp‑, as their relation to the radicals is clearer than that of the alternative formss, χ.

iii. The spirant mutation afterchwe ‘six’ is irregular. From Kelt. *su̯eks kantom we should expect *chwe cant, sinceksk givessk, and final‑s would drop. But the independent form of *su̯eks was already *χeχ in Brit.; and we may assume that this was generalized, so that thech- inchwe chant comes from ‑χk‑.

iv. (1) Brit. or Lat.kt > *χ̑t > *χ̑þ >i̯þ; the formsi-diphthongs§ 29 i, cf.§ 104 ii (1); thusakt >aeth;okt >oeth;ukt >w͡yth;ekt >eith, Mn.aith;ikt >īth. Thus W.caeth < Brit. *kaktos§ 86 ii (1);doeth < Lat.doctus;ffrwyth < Lat.fructus;saith Brit. *sehtan < Ar. *septm̥;perffaith < Lat.perfectus;brith < Brit. *briktos < *bhr̥ktos§ 101 iii (2);eithin ‘furze’ < *ektīn- < *ak-tīn‑, √ak̑‑/oq‑;seithug ‘fruitless’; < *sek-tonk- < *seq- ‘without’ + *teu‑q‑, √teu̯āˣ- ‘increase’ ;eithaf ‘extreme’ < *ek-tₑm-os: Lat.extimus.

(2) In Ml. W. there was a tendency to voice thisth to, as inperffeiẟẏawỺ. A. 19 fromperffeith, now re-formed asperffeithio ‘to perfect’;arhwaeddont do. 32 ‘they may taste’ (:chweith ‘taste’). The survives incynysgaeẟu fromcynysgaeth ‘endowment’. Inaeth + vb. ‘to be’ forming old perfects and pluperfects, the diphthong was simplified, givingath‑, affected toeth‑, asethywỺ.A. 82, more commonlyeẟyw ‘went’; soaẟoeẟ ‘had gone’, etc.,§ 193 vi (3),(5).—Final so produced disappeared inheno,yna, etc.§ 78 i (1).

v. Lat.x >*χ̑s >i̯s; thusax >aes, etc. ; as W.llaes ‘trailing’ <laxus;pais, Ml. W.peis <pexa (tunica);coes ‘leg’ <coxa. SoSaeson <Saxones,Sais <Saxō§ 69 ii (2). Similarly Brit.‑ks- from‑nks- etc.,§ 96 iii (6).

Initial Mutation.

§ 109. We have seen that Welsh has nine mutable consonants. Initially the radical and mutated forms exist side by side in the living language. The use of the various mutations is determined by syntactical rules which have sprung from generalizations of prevalent forms. Thus an adjective after a fem. sg. noun has its soft initial because most fem. sg. nouns ended in a vowel.

The following table shows all the mutations of the nine mutable consonants:

Radicalptcbdgmllrh
Softbdgfddflr
NasalmhnhnghmnngNo change
SpirantphthchNo changeNo change

The words “No change” in the table mean that the consonants under which they are placed retain their radical forms in those positions where the others undergo the respective mutations. Thus afteryn, which nasalizes the explosives,m,ll, and rh remain unchanged; and words which cause the tenues to become spirants do not alter the other six. This is always understood when the nasal or spirant mutation is named, and there is no need to particularize except in case of irregularity.

Strictly speaking, of course, words which caused the nasal and spirant mutations changedI,r toII andrh. But for practical purposes it is simpler to treat the changes as above; see§ 103 i (4).


Later Consonant Changes.
Loss of Voiced Spirants and Sonants.

§ 110. i. The soft mutations ofb,d,g,m have all tended to be softened to the vanishing point. Being very soft “buzzes” andf were liable to be confused; and so we find one substituted for another as incuddygl (kuẟygɏlw.m. 140,r.m. 211) ‘cell’ for *cufygl < Lat.cubic’lum (prob. influenced bycudd ‘hidden’);Eiẟẏonyẟr.p. 1287 forEifionydd (eiwonitb.b. 69) ; Late Mn. W.Caer Dydd forCaer Dyf ‘Cardiff’; or two metathesized, as inclefyẟeur.m. 182 forcleẟyveu do. 126, and inclefytaud (t)b.b. 48 forcleẟyfawd: W.cleddyf,§ 76 viii (2) (Ir.claideb ‘sword’ < W.).—S.V. (P.Ỻ. xci) says of the lineKawn vedd rhad kyneddvau Rhys (by H.K., seec.c. 344) that it pleases the ear though it violates the rule. The ear does not notice the inversionv ẟ / ẟ v.

ii. (1) The soft mutation ofg has uniformly disappeared as an initial sound. Thus *dy ᵹardd has becomedy ardd ‘thy garden’. Medially it disappears or becomes before a vowel, or beforeI,r orn§ 103 ii (1),§ 104 ii. Medialnᵹn >n, as inynad§ 62 ii; cf.§ 106 ii (1).

(2) Medially afterI orr it appears as,§ 105 ii, which is lost beforey, as incŏ́lyn < O.W.colginn§ 54 ii. This palatalization of toᵹ̑ > after a liquid is comparatively late, for it does not take place finally; in that position remained dark, and became non-syllabicỿ, as in Ml. W.daly (1 syll.) ‘to hold’; this was either assimilated to thel as in N. W.dăl (<*dal-l, doubleI, not), or was lowered toa and became syllabic, as in S. W.dala; from Brit. *dalg- < *dₑl’gh‑, √delāˣgh‑: Skr.dīrgháḥ ‘long’, Lat.indulgeo,longus. Medially it is from the same stem, as indali̯af ‘I hold, maintain, continue’. So we have Ml. W.hely ‘to hunt’, N. W.hĕl ‘collect’, S. W.hela; Ml. W.boly bag, belly ', N. W.bŏl, S. W.bola; Ml. W.gwaly, Mn. W.gwala ‘sufficiency’; Ml. W.eiry 'snow', Mn. W. (N. and S.)eira, andeir ineir-law ‘sleet’,ces-air ‘hail’; Ml. W.llary ‘generous’ < Lat.largus, Mn. W.llari̯aidd. The form‑a appears in writing as early as theb.b., e.g.llara 7, where, however, the word counts as only one syllable in the metre.

In the 16th cent. the sound of‑y in the above Ml. W. forms was not known. J.D.R. writes ity (≡ɥ), p. 136; but Dr. Davies compares it with Eng. final mute-e, as intake, and writes itỿ, asbolỿ,helỿ D. 19. The correctness of this transcription is -confirmed by theb.ch., where it appears ase (≡ ỿ,§ 16 iii), asdalea.l. i 20 ≡dalỿ. [ >ỿ >a forms an interesting parallel to the supposed Pre-Ar. givingə and then mostlya.]

(3) Lat.virgo > W.gw̯yry (1 syll.) D.G. 156,Ỻ.A. 84, 87, 90, etc., whencegwyrdawtr.b.b. 119, though we have alsogweryndawtỺ.A. 17, 50, 84,b.b. 40, direct fromvirginitātem. Inb.b. 70 occurs the pl.gwirion < Brit. *u̯irgones. Later we findmorwyn wyraa.l. i 518; Gwynedd dial.menɥn gw̯ɥrẟ (for *gwɥr-r cf.dal-l) ‘unsalted butter’, Dyfedmenyn gw̯ɥra, Rhys CC. 46. We also havegwyrf (1 syll.) D.G. 118,gwyrɏf vireindawl (4 syll.)r.p. 1199, andgweryẟ (2 syll.)r.p. 1200, D.G. 137, pl.gweryẟon (3 syll.)r.p. 1199,b.b. 71. The latter cannot be derived fromvirgo; no medial syllabic irrationaly is known in Early Ml. W.;gweryẟ must be Kelt. and may represent *ɡheríi̯ō, pl. *ɡheríi̯ones: Ir.gerait ‘virgin’,gerait (i. mac bec) ‘little boy’ O’Dav.: redupl., Gk. παρθένος < *ɡhr̥-ɡhén- (not: Skr.pr̥thukaḥ ‘boy, calf’, since *th > Gk. τ), Lat.virgin- < *ɡer-ɡhen‑, dissim. for *ɡher-ɡhen‑, and perhaps W.gwyrf < *ɡherɡ, which fits exactly,§ 92 iii. Dr. Davies wrongly takes Ml. W.gwyryf as a disyllablegwy|ryf, which it may have become dialectally,§ 16 v (3). The biblical pl.gwyryfon is formed from the new disyllable.

(4) Inbwrw < *burg-§ 97 v (3),llwrw < *lurg- < *lorg-§ 215 ii (7), the‑ᵹ was rounded by the precedingw, and became‑w̯. In derived forms, however, it became regularly; as Ml. W.bỿrẏaf ‘I cast down’, nowbwri̯af.

(5) Inhy ‘bold’ (< *hyᵹ < *sig- < *sego‑:§ 92 i) a finalf is now wrongly written. Thef is not pronounced, and there is no evidence of it in Ml. W. or the poets; seehyb.b.b. 265, D.G. 42, 269, 313, etc. It does not occur in old derivatives:kyn-hyets.g. 277,hy-der,hy-dab. In the dialects, however,f is inserted in new derivatives, ashyf-dra,hyfach, which, likellefydd,brofydd, dial. pl. oflle,bro, are due to false analogy. Other spurious forms likehyf occur in latemss., such asdaf,llef,brof forda,lle,bro. In none of these is thef an old substitution for; they are sham-literary forms made on the analogy oftref for the spokentre’.

iii. (1) Finalf was lost before the Ml. period afteraw, as inllaw ‘hand’ < *llawf < Kelt. *lāmā < Ar. *pl̥̄mā§ 63 vii (2);—rhaw ‘spade’ < *rhawf < *rā-mā, √arā-§ 63 ix. When a syllable is added andaw is replaced byo§ 81 i, thef reappears, as inllof-rudd ‘murderer’, lit. ‘red-handed’,llof-yn D.G. 107 ‘wisp’,lloffa ‘to glean’ < *llof-ha,rhofiau ‘spades’. SoprawỺ.A. 24,r.p. 1215 ‘proof’ forprawf a back-formation fromproviỺ.A. 38, 72 < Lat.probo. The re-introduction off inpraw is artificial, and inconsistent with the N.W. pron.prāw,§ 52 iii, Exc. (1).

Na wrthod, ferch, dy berchi;
Naphrawymadaw â mi.—D.G. 108; see 238, 240.

‘Refuse not, lady, to be honoured; do not try to leave me.’

It was lost afteriw̯ inRhiwabon ‘Ruabon’ forr͑iw vabonr.b. 1066, and afterw intw ‘growth’,dŵr ‘water’, reappearing intỿfu ‘to grow’,dỿfroedd ‘waters’, in whichw is mutated toỿ. It disappeared regularly afteru, as inplu ‘feathers’ sg.pluen < Lat.plūma;—cu ‘dear’, O.W.cum (m), Corn.cuf, Bret.kuñ,kuñv, Ir.cōim < *koi‑m‑, √k̑ei-: Skr.s̑éva‑ḥ ‘dear’ < *k̑ei-u̯‑os, Lat.cīvis;—du ‘black’, Corn.duv, Ir.dub < *dhubh‑, √dheubh‑: Gk. τυφλός;—so in derivativescu-dab,cu-ed,du-ach, etc.

f being originally bilabial,§ 19 ii (4), when it followed, w oru (≡ü), it was in effect little more than the narrowing of the lip-rounding at the end of the syllable, and so came to be disregarded. For a similar reason, whenf followedm, it was also lost or assimilated, as inmámaeth for *mám-faeth ‘foster-mother’;im ’y hun forim fy hun ‘for myself’.

Ni byddai bwn, heb ddau bâr,
Im ’y hunano’m heiniar.—I.D.tr. 138; cf. E.P. 277.

‘Without two pairs [of oxen] there would not be [even] a burden for myself of my crop.’ It remained incam-fa ‘stile’ (Gwyn. dial.camẟa, Dyfedcanfa by dissim.).

(2) Initialf often disappears infy ‘my’, especially in poetry, the following nasal mutation showing that’ỿ means ‘my’ not ‘the’; asyg̃korn (≡’y nghorn)ym neẟeirb.t. 35 ‘my horn in my hand’;’Y mamr.m. 194, l. 5 ‘my mother’ (‘the mother’ isy fam); so’Y myd wen§ 136 iii,’y mun D.G. 17 ‘my girl’,’y nghefn,’y mraint, do. 274, etc.—It is lost invab ‘son’ in patronymics, asHywel ab Einion; inychydig forfychydig, rad.bychydig.

Déuaf—myfi yw d’ ë́os—
Dïau,’ỿ nŷn,o daw nos.—D.G. 114.

‘I will come—[for] I am thy nightingale—assuredly, my lady, if night comes.’

(3) Medialf drops after an explosive, when followed by a rounded vowel or a liquid, as intestun ‘text’ for *testfun < Lat.testimōnium. Hence in compounds, where it is the initial of the second element, it is often lost, as inBod-órgan for *Bod Forgan (‘Morgan’s dwelling’),Bod-ẃrog for *Bod Fwrog, etc.;Bendigéidran§ 45 i (2) forBendigéid-Vran (Bendigeitvran, first written without thev inr.m. 26, andv inserted above the line). Between a consonant and liquid it dropped early in some cases as inyr llynedd, Gwenlliant§ 111 i (1) andHydref do.vii (1). Rarely before an explosive, as inagwyẟawr for *afgwyẟawr§ 74 i (1).

(4) Finalfn in unaccented syllables is generally reduced ton, especially after rounded vowels, as ineon foreofn ‘fearless’ § 156 i (15);únon Gr. O. 118 forún-ofn ‘one fear’;annwn forannwfn ‘hell’;dodren in the dialects, and sometimes in the bards, fordodrefn§ 82 ii (3);colon forcolofn, see example;ysgafn ‘light’ retains itsf in N.W. dial.; in S.W.ysgawn orysgon is used.

Val Samson wrthgolongynt
A fu’n rhwym yw fy nhremynt.—G.Gl.p 83/59.

‘Like Samson, who was bound to a column of old, is my condition.’

Finalfl gaveI in S.W.côl L.G.C. 280, forcofl ‘bosom, embrace.’

(5) Finalf began to disappear very early in the spoken language; we already findgwartha forgwarthaf inl.l. 196. Its earliest regular loss (apart from the cases cited in(1) above) occurred afteri, as in the v. n. termination‑i, e.g.moli ‘to praise’ for *molif, O.W.molimjuv. sk.;lli forllif ‘flood’;divrir.p. 1149 fordifrif ‘serious’;cyfri D.G. 4 forcýf-rif ‘to count’. But in the 14th cent. it had come to be freely dropped after any vowel, as the following rhymes show:ne’/bore G.Gr.d.g. 238,ydwy’/mwy D.G. 72,cry’/lesu do. 474,ha’/Efa do. 157; sowna’ D.G. 72,kynta’r.p. 1277. The word is treated in every way as a word ending in a vowel; thus it is followed by’n foryn,’r fory oryr, etc., asofnwy’r D.G. 321 forofnwyf y;ydwy’n forydwyf yn§ 125 iii ex. 1;Tre’rkastellr.p. 1210 forTref y Castell.

Finalf is not known to drop in the old wordsglaif ‘sword’,of ‘raw’,blif ‘catapult’ or in lit. W.llef ‘cry’,sef ‘that is’. It is still retained in the spoken language indof ‘tame’,rhwyf ‘oar’,bref ‘bleat’,prif ‘chief’,Taf ‘Taff’, and in borrowed words, asbraf ‘fine’: Fr.brave, E.brave.

iv. (1) Initial in O. W.di ‘to’ disappeared, giving Ml. W., Mn. W.i, ‘to’§ 65 iv (2).

(2) Medial disappears inmewn: Ir.medōn§ 215 iii (1); in the verbrhoddaf, v.n.rhoddi ‘to give’, which becamerho-af >rhôf, v.n.rhoi; seerhoist, etc.§ 33 iii (1); but the also persisted in the written language; see§ 186. Similarlyarhoaf for *arhoddaf§ 187 iii. Medial also disappears intỿddɥn >tyn in place-names of the formTɥn-ỿ-mā́es (*tỿɥn > *tɥɥn, *tɥ̄n,tɥn).

Medial is sometimes lost as the initial of the second element of a compound; thusrheg-ofydd (rec ouytm.a. i 324, 344) ‘lord of gifts’ forrheg-ẟofydd (recẟovyẟw.m. 452,r.m. 100);Duw Ofydd forDuw Ddofydd,Cred-ofydd forCred-ẟofydd, etc. It was also lost before an explosive, as inBlegywryta.l. i 338 (ms. l.) forBleẟ-gywryd (Bledeuuritl.l. 222);diwédydd (diweditb.b. 90) ‘evening’ for *diwéẟ-dyẟ;gwybed ‘flies’ forgwyẟbed (gwyẟbetr.m. 54).

(3) Final was lost in the relativeydd before a consonant,§ 162 i. It disappeared early in the 2nd sg. pres. ind. of verbs,§ 173 iii (2). It dropped inyssyẟ ‘who is’ (oftenissiyssȳ́ inb.b.), thoughsydd may still be heard as well assy. Sometimes innaw Duw!f.n. 63 fornawẟ Duw! ‘God’s protection!’ (i.e. God help us!). Ini fynydd ‘up’ the final‑ẟ was lost early, though it is sometimes found written in Ml. W., askyvodi ẏ vynyẟỺ.A. 111, and survives to this day in parts of Dyfed. With its,i fyny lost all trace of its original signification, as seen in the unconscious repetition inẏ vyny ẏ vynyẟ Oliverr.p. 1280 ‘up to Mount Olivet’. The final‑ẟ ofeisteẟ also disappeared very early; it iseiste in theb.b. andb.ch. So inw.m., e.g. 4 times in col. 449, in each case changed toeisteẟ inr.m. 293–4. The‑ẟ is deduced fromeisteẟaf, etc., and its re-insertion finally is artificial; it is not sounded ineiste in the spoken language. Final‑ẟ also disappeared inhwnnw etc.§ 78 i (1).

v. (1) The final‑r of the articleyr was lost before a consonant after the O. W. period; see§ 114 iii. So‑r after a consonant inbrawd§ 113 i (1).

(2) Final‑nn was sometimes lost in unaccented syllables; ascyfa ‘whole’, Ml. W.kyfar.p. 1285 forcyfan(n), cf.kyfannuw.m. 129;yma ‘here’ foryman(n)§ 220 ii (11); (e)felly ‘so’ < *hefel hynu ‘like this’, cf.fell hýnn§ 215 iv (2); Ml. W.ky- forkynn ‘as’ before the eqtv.§ 147 iv (4);‑fa for‑fann§ 143 iii (16). The tendency was arrested, and‑nn generally remains; it had not gone far inkynn before it was checked, and‑n(n) was restored. The loss also occurs in Corn. and Bret., so that it must be referred to an early peculiarity in the pronunciation of‑nn.

Provection.

§ 111. i. (1) Whenn orr came before a liquid after the loss of an intervening vowel, the liquid became voiceless; thusnl >nll;rl >rll;nr >nrh;rr >rrh. Examples:gwinllan ‘vineyard’ < *gwin-lann < *vīno-landā;hirllaes ‘long trailing’ for *hir-laes;penrhyn ‘promontory’ for *penn-ryn;an-rheg§ 156 ii (1);Henllan,Henllys, etc. Also in combinations in which no vowel had intervened, asgôr-!lanw̯ ‘high tide’,an-llygredig ‘incorruptible’. So initially:yn llawn foryn lawn ‘full’;yn rhad,mor llawn,mor rhad (yn andmor generally cause lenition of adjectives) ; sopur llawn ‘very full’;hēn llew Job iv 11 (1620),hēn llysp 121/35r.

This change had taken place before the loss of and as described in§ 110, and did not take place later. So where or originally stood between the sounds it did not occur. Thus we haveCýn-las < *Cyn-ᵹlas <Cuno-glasos;tôr-Ian ‘brink’ < *torr-ᵹlann ‘broken bank’;Hâr-lech <Harẟ-lechw.m. 38;cór-lan ‘fold’ < *corẟ-lann. Thusyn Iân,mor Iân fromglân ‘clean, fair’; and while we havey llăn ‘the hamlet’ from *yr lann fromllann ‘enclosure’, we havey Iăn ‘the bank’ from *yr ᵹlann fromglann ‘bank’, both nouns being fem. Butf appears in some cases to have dropped out early enough to allow of the change; as iny llynedd more fullyyr llynedd for *yr flyneẟ;Gwenllïan < *gwenn-flïant.

(2)l was palatalized and becamell in two positions: (α) after Brit. ei, Lat.ē; thuscannw͡yll < Lat.candēla;tŵyll ‘deceit’ < Lat.tēla;tywyll ‘dark’§ 38 x for *tyw̯-w͡yll§ 76 vii (2) < *temeil- < *temes-elo-s: Bret.teval,teñval for *teñvol, Corn.tiwul, Ir.temel: Lat.tenebrae < *temesrai, Skr.tamasáḥ ‘dark-coloured’; but not after Brit.ai, e. g.coel ‘omen’ < *kail- < *qai̯()l‑: O. H. G.heilisōn ‘augurari’: Ir.cēl < *keil‑.—(β) Between twoi’s, as inEbrill < Lat.Aprīlis;pebyll ‘tent’ < Lat.pāpilio.

ii. (1) Whenb‑b,d‑d,g‑g came together after the loss of a vowel they became doublep,t,c respectively, simplified before the accent, and before a sonant ; as inCatéyrn forCatté|ɥrn < *Cad-diᵹirn- < Brit.Cato-tigirn- (Rhys no. 47);meitin < *meid-din < Lat.mātūtīnum§ 70 v;wynepryd ‘countenance’ < *wyneb-bryd; and in the examplebywiócledd <bywiog gledd:

A’m bẃcled a’mbywiócledd
Yn arfau maen ar fy medd.—G.Gl.,m 146/198.

‘And my buckler and live sword as weapons of stone [carved] on my grave.’

When the explosives came together in different words they resulted in a double consonant, voiced at the implosion, but voiceless with the new impulse at the explosion. This change is not now represented in writing ; but inmss. and early printed books‑d d- etc. frequently appear as‑d t- etc. ; thusNid Toethineb heb lenp 54/356r. ‘There is no wisdom without learning’;Gwnaed tuw ag enaid howelp 63/7r. ‘Let God do with the soul of Howel’;Ygwaed ta a vac tëyrnp 52/22 ‘Good blood begets a king’;Glowed tim ond y glod tauc.c. 342 ‘To hear anything but thy praise’;i’r wlad tragwyddolb.cw. 86 ‘to the eternal land’;Y Ddraig côch ddyry cychwyng. 177 ‘The Red Dragon gives a leap’. “Two /b/ standeth in force of /p/….mab byχan most be pronounced as if ytt were wryttenmab pyχan” J. J. 144/51. In all cynghanedd prior to the 19th cent. such a combination corresponds to a tenuis. The writers of the recent period sometimes treat it as a media.

(2)ẟẟ becameth innyth,syth, etc.§ 97 ii; cf. dial.rhōth for *rhoẟẟ <rhoẟoẟ ‘gave’. Similarlyᵹᵹ becamech indichon§ 196 ii (2). But generally two voiced spirants remained, written single, as inprifarẟ forprif-farẟ ‘chief bard’.

iii. (1) When a media was followed byh the two became a double tenuis; thusateb (ttt) ‘reply’ < *ad-heb < *ati-seq, √seq- ‘say’;drycin ‘storm’§ 27 i < *dryg-hin;gw̯lypaf ‘wettest’ for *gw̯lyb-haf§ 147 ii.

When the sounds came together in different words they gave the double sounddt etc., seeii (1) above; and in all standard cynghanedd‑d h- corresponds tot,‑b h- top,‑g h- toc; asOer yw heb hwn, ŵr hypert Gr.H.G. 99.

(2) Similarly in some casesfh >ff;ẟh >th; as inlloffa ‘to glean’§ 110 iii (1),§ 201 iii (4);diwethaf ‘last’§ 149 i;rhotho§ 186 ii; bytho§ 189 ii (4). Sofr‑h >ffr indyffryn§ 106 iii (2);f‑rr >f‑rh >ffr incyffredin§ 156 i (9). But as a rule the groups remain, asdyddháu,dyfrháu; and‑f h‑,‑ẟ h- do not correspond toff,th in cynghanedd.

iv. When two similar consonants, whether explosives or spirants, one voiced and the other voiceless, came together, they became a double voiceless sound medially, simplified where double consonants are usually simplified, as before a consonant; thuspópeth (ppp) < *pób-peth ‘everything’;gwrthrych ‘object’ < *gwrth-ẟrych. In ordinary pronunciation the result is the same when the sounds occur in different words; and in Ml. W.mss.‑th ẟ- frequently appear asth only; thusathiweẟỺ.A. 157 fora’th ẟiweẟ ‘and thy end’;Athelw athwylaw ar lletr.p. 1220 ‘And Thy image with Thy hands extended’; cf. 1205 l. 34, 1321 l. 32; similarlyweinllian tec 1424 for (G)wenllïant deg;cereint ᵗḍuw 1220 (d deleted by dot,t substituted).

v. (1) When two unlike mediae came together, the group was unvoiced at the implosion, but not necessarily at the explosion. In Ml. W. both are usually written as tenues; thusdicterr.p. 1209,atkessynt 1309,hepcor 1230,dywetpwytw.m. 96,ducpwyt do. 183,attpawrb.b. 35. The second is, however, often written as a media, aso wacderr.p. 1280,atborẏon do. 1208,kytbar do. 1300,llygatgall do. 1308. In the 1620 Bible we haveatcas,datcuddiad, etc.; but the more usual spelling later wasatgas,datguddiad, etc., which perhaps represents the sound more accurately. When however the second consonant was a dental it tends more to be voiceless. In the Bible we find such forms asdigter fordicter, theg being due todig. In cynghanedd either consonant may correspond to a tenuis or a media. Pughe’s etymological spellingsadgas,udgorn,hebgor, etc., misrepresent the sound, which is as nearly as possibleatgas,utgorn,hepgor.

(2) A media was frequently, though not necessarily, unvoiced beforel,r,m,n,,f and even,. Thus in Ml. W. we findllwtlawr.p. 1222 ‘Ludlow’,atrawẟ 1251,tatmaetheur.m. 24,atnewyẟwys 93,wreicẟa 23,dynghetven 73,atwen 245,lletẏeithr.p. 1222. But whiler.m. hasgrwytraw 86, the olderw.m. has in the same passagegrwydraw 183. Inr.p. 1269, 1303 we havesygneu ‘signs’ but in 1214, 1215 it is writtensycneu. Indeed ther.b. scribe, who had no ear for cynghanedd, writes tenuis and media where they should correspond; asheidẏaw/​ehetẏatr.p. 1283,chenedloeẟ/​chynatleu 1204,dilitẏa/​dy aelodeu 1216. In the last example the sound is certainlyd, asaelodeu cannot havet. It might therefore be supposed that the sound was always a media, and that to write it a tenuis was a mere orthographical convention. But though the sound is now generally a media, there is evidence that it might be, and often was, a tenuis: (α) D.G. has such correspondences asDadliti̯a ’r/​diwyd latai p. 19,neiti̯wr/​natur 133; and (β) the tenuis has survived in a number of examples, asCoetmor (forcoed-mor <coed mawr);tycio ‘to prevail’ <twg ‘prosperity’ < *tuq‑,√teu̯āˣ‑, cf.§ 108 iv;eto foretw̯o <edwaeth § 220 ii (7);ysgatfydd ‘perhaps’;Llan Decw̯yn;caneiti̯o ‘to brighten’ (of the moon) <cannaid;cartref,pentref.

(3) The mediae were unvoiced before voiceless consonants; thusatseinb.t. 20,datseinr.m. 289,Botfforddg. 102. In Late Mn. orthography etymological spellings prevail, asadsain Ezec. vii 7,Bodffordd. The latter, the name of a place in Anglesey, is always soundedBotffordd, in spite of the spelling withd.

(4) It is seen from(1),(2) and(3) above that a media is liable to be unvoiced before any consonant in the middle of a word. But we have seen in the preceding subsections that a change which took place medially also occurred when the group belonged to different words. Hence final mediae must frequently have been sounded as tenues before an initial consonant; and this is very probably the reason why they were so commonly written as tenues, the pre-consonantal form being generalized in writing. The facts are briefly summarized in§ 18 ii.

But before an initial vowel it is certain that a final explosive, though written as a tenuis, was in fact a media in the 14th cent. In the following examples fromr.p. (which might easily be multiplied) it is seen that the finalt orc in heavy type must be pronouncedd org to correspond to a media in the other part of the line:

Digystuẟ |anrec am (decystwyll 1202,
Glotoleu |yn (glewdalu 1203,
Gwlediceurswllt |vu (gwlatagorseẟ 1208;

so before a liquid:

Temɏl ẏgrist |teu amlwcrat 1200.

Such a slip asSetlibera nos a maloỺ.A. 150 shows that the scribe was in the habit of writing finalt where the sound wasd. Cf. also§ 18 iii. That the written tenuis does not mean that the vowel was short in a monosyllable likegwac nowgw̯āg is proved by such a spelling asyn waac…y gadeir waacw.m. 449,r.m. 293. Cf.§ 55 i.

The final media before an initial consonant, however, corresponds to a tenuis in much later cynghanedd, especially when the initial is voiceless:

Hebswydd |mor (hapus a hwng. 239
Brigffydd |a bairkoffa hwn, etc.,P.Ỻ. Ixxix.

Though the explosive is now a media before an initial consonant as well, we have a trace of the tenuis inap forab (forfab§ 110 iii (2)), as inap Gwilym besideab Edmwnd.

(5) Since the explosive was a tenuis before a consonant we have‑p m- and‑t n‑; these combinations were mutated tomh andnh in the following examples, the voicelessness of the tenuis being retained after its assimilation:Amhadawcp 61/18r. forAp Madawc,Amhredyddc.c. 334 forAp Mareduẟ, am mydronb.b. 94 (mmh§ 24 i), etc.;prynhawnw.m. 70,r.m. 50,Ỻ.A. 121 forpryt nawnw.m. 162,r.m. 229. The late spellingprydnawn is an artificial reconstruction; the spoken language preserves the traditional pronunciationprynháwn.

Ag un lliw, gannwyll awyr,
Y barnwn haulbrynhawnhwyr.—I.D. 7.

‘And of the same colour I judged the late evening sun,—the candle of the sky.’ Cf.brynhawn/​bery’n hir D.G. 73,Barn hen/​brynhawn do. 428.

vi. (1) A media was unvoiced after nasal + tenuis. The following cases occur:ŋk‑d' >ŋkt orŋt, as inieuenctid ‘youth’ also writtenieuengtid;—nt‑g >ŋk, as indifancoll D.G. 387 ‘perdition’ < *difant-goll;deincryd D.G. 385,r.p. 1157 ‘gnashing of teeth’ < *deint-gryd.

(2) A media was generally unvoiced after a voiceless spirant; asglastwfɏrr.m. 146 forglasdwfr§ 96 ii (5);neillparth do. 148 forneillbarth;dywespwyt do. 90 ;gwnaethpwyt do. 89;gwallcob.cw. 37 forgwallgof;alltud forall-dud. On the other handp andc are voiced, sometimes even in Ml. W., afters; thus while we haveyskynr.m. 11,kyscu do. 21,ysprytỺ.A. 99, we also finddisgynnentr.m. 14,goresgyn do. 91,ysbrytỺ.A. 3,esgussawdw. 1a,pasgadur ib. Though the tenuis was commonly written up to the 18th cent., Dr. Davies’s orthography has generally prevailed since the appearance of his dictionary; in this the media is written except in the groupsst,llt,cht,fft,thp.

(3) An initial media is sometimes found written as a tenuis after a voiceless spirant:Canyscollyg̃hyw.m. 78 changed toKan nysgollynghy inr.m. 56;Bei yscuypunb.b. 81 ‘If I knew’;oskovyna.l. ii 18 ‘if he asks it’;seithpechawtỺ.A. 143 forseithbechawts.g. 36 ‘seven sins’;a’thcaledrwyddrh.b.s. 74 ‘and thy hardness.’

vii. (1)ẟl >dl, as inbodlon ‘satisfied’ < *boẟ-lawn. The recent spellingboddlon is a reconstruction due to Pughe; the natural pronunciation isbodlon (S. W.bŏ́lon); cf.Fodlon im dan fedw̯lwyn ir D.G. 172 ‘contented with me under fresh birch-trees’;Bodloni bydol annyn Gr.O. 34 ‘to satisfy a worldly wretch’;hadl ‘lying in ruins’ for *haẟl < *sₑd-lo‑, √sed- ‘settle’§ 63 ii. Similarlyẟr >dr, as incadr ‘puissant’ for *caẟr: Gaul.Belatu-cadrus ep. of Mars, O. Bret.cadr gl. decoreo, Bret.cazr,kaer ‘handsome’: Gk. κεκαδμένος, Skr.s̑ās̑ad- ‘distinguish oneself’. It took place after the loss off; thusHydref ‘October’ <hyẟfref (heẟvrefa.l. i 24,calan hyddfrefm.a. i 346b ‘Oct. 1st’),dedryd ‘verdict’ < *deẟf-fryd.

On the other handd ( < orig.t) is sometimes treated as before a sonant, and in S. W. dial, has remained or disappeared. Thuscenedl iskenetɏl inb.b. 10, 16, wheret, but in O. W. iscenetlb.s.ch. 2, wheretd (S. W. dial.cenel);hoedl (withd <t, cf. Late Brit.Vennisetli, and see§ 63 vii (5)) is treated ashoeẟl by Casnodyn,hoeẟɏl/​heẟwchr.p. 1248, cf. 1234, 1241, but G.M.D. hashyder/​hoedɏl do. 1320, cf. 1212; so I.G.,Hudol/​hoedl 310. S. W.hw̯eẟel forchwedl cannot mean that the suff. was *‑dhlo‑, for‑edl- would give ‑eil§ 104 iv (3). So S. W.gwaẟan forgwadn ‘sole’ has orig.t, sincedn givesn;gwadn < *u̯o‑t‑n- < *u̯o‑dd-no- < *upo-bd-no- ‘*under foot’, √ped‑.

The late change ofdrum,drem totrum,trem is probably due to the soft mutationẟr- becomingdr‑, and thed- being then mistaken for the soft mutation oft‑. It certainly is not a phonetic law thatdr- should becometr‑, fordrwg,drych,drain,drud, etc., all retaindr‑.

(2) >d afters,t,d, and in old formations afterl,ll,n; thustreisdwyn for *treis-ẟwyn ‘a taking by force’,dreis-dwyn/​dristytr.p. 1288;atal ‘to withhold’ < *ad-dalᵹ < *ad-ẟalᵹ < *ati-dalg‑;llygeitu forllygeid-ẟu, etc.;bendith for *ben-ẟith < Lat.benedictio;melltith ormelldith for *melẟith < Lat.maledictio.

This change also takes place initially; thusnosda ‘good night’§ 146 iii (2),nosdu Diar. vii 9 ‘black night’, for *nos ẟ- (nos being f.—the orig. mutation was rad. after *no(k)ts, but this cannot be assumed to have survived); soyr wythnosdiwethaf ‘last week’;trosDafyddg. 237,trosdaeare. xiv fortros ẟ‑;Bleẟyn tur.p. 1284 forBleẟyntdu forBleẟynt ẟu;lleiandu D.G. 20 ‘black nun’;Siwandu L.G.C. 319, 321 ‘black Joan’;holldaear do. 446.

Pan aeth Tomos ap Rhoser
AtDuw a’r saint trwy y sêr.—L.G.C. 38.

‘When Thomas ap Rhosser went up to God and the saints through the stars.’

Llyma ’r blaid lle mae’r blodau
A ’r holldawn o’u rhyw ill dau.—T.A.,c. ii 83.

‘This is the band [of children] in whom are the flowers and all the gift of their [the parents’] two natures.’

Yna nosa, myn Iesu,
Einioes dyn inegis nosdu.—G.GL,c 7/44.

‘Then, by Jesus, man’s life darkens like black night.’

(3) >d before or after the above sounds, and continuants such asm,f, even when separated from them, see§ 102 iii (2); as Late Mn. W.machlud <ym-achludd§ 44 v < Lat.occlūdo; Late Mn. W.gormod forgormoẟ the usual form in the bards;Maesyfed ‘Radnor’ forMaes Hyfeiẟ;didol < *di-ẟawl§ 156 i (11),pedol§ 102 iii (2). The change, being a form of dissimilation, is only accidental.

(4) The change of tog and off tob under similar conditions is rare:arglwyẟ 'lord' besidearlwyẟ (both inw.m. 160) < *ar-ᵹw̯lw͡yẟ < *pₑri-u̯lei- VR1e of √u̯elēi‑; cf.glyw§ 102 iii (2);cwbl for *cwfl§ 168 iii (3);parabl ‘saying’ for *parafl < Lat.parabola;cabl ‘calumny’ for *cafl < *kaml- met. for *kal’men: Lat.calumnia§ 100 ii (1); so Bret.cablus, Corn.cabal.

§ 112. i. (1) In O. W. and Early Ml.W. an initial vowel or a medial vowel in hiatus seems to have been pronounced with a distinct breathing which is often represented byh. This breathing was voiced, and so differed fromh <s, which was voiceless. Examples are, initial: O. W.ha,hocox. ‘and’;heitham do., Ml. and Mn. W.eithaf§ 108 iv (1);—Ml. W., froma.l. i,huydvet (wythfed) 58;huchof ib. ‘above me’;hun din (un dyn) 124 ‘one man’;yr hun (yr un) 256;huiui (wyf i) 114;er hyd (yr ɥd) 326 ‘the corn’;ohyd (o ɥd) 82 ‘of corn’;hercki (erchi) 152;hodyn (odyn) 78, etc. Medially it occurs not only where a soft spirant had disappeared, as indiheur.m. 181 < *di-ᵹeu, Mn. W.dïau ‘truly’;rohia.l. i 118 <roẟi; but also where no consonant ever existed, as indiheuỺ.A. 21 ‘days’;dihag̃eir.b.b. 48 ‘escaped’.

(2) Although this breathing has generally been smoothed away, it was liable to become voiceless before an accented vowel, and in that case it survived ash; thus medially indihangol ‘escaped, safe’; initially, after a vowel inpahám for *pa am ‘what for’,pahachosỺ.A. 123,pahawr do. 13; afterr inun arhugain ‘21’,yrholl§ 168 ii (3); in all positions inhogi ‘to whet’ for *ogi < *āk‑, √ak̑‑/​oq‑: W.agalen ‘whetstone’. This occurs in several cases in which an initial accented vowel was followed by two consonants, so that it was pronounced rather forcibly; thus W.hagr ‘ugly’ for *agr, Bret.akr,hakr, √ak̑‑/​oq‑;—W.hardd ‘handsome’ for *arẟ ‘high’: Ir.ard, Lat.arduus, cf.Harẟ-lech orig. quite evidently ‘high rock’;—so sometimeshenw̯ ‘name’ (henw̯ ‘noun’r.g. 1121), generally withh- in Gwyn. dial., butanwédig without it: Bret.hañv,hanu,hano, Corn.hanow; O. W.anu, Ir.ainm, seep. 81.

(3) On the other hand initialh (<s) might come to be confused with the soft breathing, and so disappear before an unacc. vowel, as ineleni ‘this year’ < *he-fleni: Bret.hevlene, with the same prefix asheẟiw ‘to-day’;yvellyw.m. 41 for *hefelly, see§ 110 v (2); O. W.anter-metetic gl. semiputata (hanner mededig).

(4) In O. W. the breathing is found (rarely) before a suffix where it was clearly marked off from the stem, as incasulheticc (casul-edig)m.c., but no trace of a breathing in such a position remains. We have, however, a medialh before an accented vowel under the following conditions:—(α) Where the vowel is followed by two consonants, ascenhédloedd Ps. ii 1;kynhel-lisr.b.b. 234,cynhali̯aeth (l‑l andli̯ < *lᵹ);cymhedrawlm.a. ii 343 (cymedrawl ib. 355);cynhyrchol Marc iv 8; but this never became a strict rule; it is carried somewhat further in the recent than in earlier periods:cynneddfau Diar. xxxi cyn. (1620),cynheddfau in late edns.—(β) Wheren stands forẟn, as inbonhéddig frombonedd < *budníi̯ā, as if the had left a soft breathing;blynyddoedd is a late formation§ 122 iv (2) and has noh.—(γ) Wherer comes aftern, as inanrheithi; this occurs even after the accent, asanrhaith§ 111 i (1).

Anh which has always been voiceless occurs before the accent (a) in the nasal mutation ofp,t,c§106 iii (1), asdanheddog for *dant-eẟawc;kymhellawẟr.b.b. 327 < Lat.compell‑;anghenus < *aŋk‑;angheuol < *aŋk‑, etc.; probablyplannhedeu§48 i followed the analogy ofplanhigi̯on < *plant‑;canhwylleur.b.b. 380 seems to be due to the treatment of Lat.nd asnt, cf. Corn.cantuil, Bret.cantol.—(b) For originals, as inanheẟeuw.m. 81,cyfanheddu do. 73,Anheẟu mewn crwynhyẟod Ỻ., fromann(h)eẟ < *n̥do-sed-§ 63 ii;glanháu,parháu etc.§ 201 iii (4); probably‑he- iniscolheicb.b. 91, pl.yscolheigonr.b.b. 235, Mn. W. sg.ysgolháig is the suffix‑ha- < *sag- see ib.‑s- between sonants disappeared, e.g.amynedd§ 95 ii (3); butkenhadeu§ 48 i may contain a reflection of it:kennad ‘message, messenger’ < *k̑ens-n‑ətā, √k̑ens- ‘speak with authority, etc.’ see Walde² 151: Lat.censeo, W.dangos§ 156 ii (1).—Theh which provects mediae always comes froms; in no case is provection caused by an “accentual h”, orh developed from a soft breathing.

ii. (1) The semivowels,,ü̯ seem to have been pronounced in Early Ml. W. with friction of the breath, which is often represented byh beforei oru, especially in theb.ch. Thusyhu (yw)a.l. i 6;Mahurth (Mawrth) 64;entehu (ynteu) 130;nehuat (neuaẟ) 78;arnehy (arnei) 100. More rarely it occurs between twou’s (uu =w), or twoi’s, as inarnauhu (arnaw) 132;doissihion (doythi̯on) 124. With such a breathing would be equivalent to back, and at an earlier period it was represented byg, which survives inenguya.l. i 100 forenw̯i ‘to name’ (which never had the mediag, as the is fromm); this also may stand between twou’s in thisms., asdim orauguenel (a w̯nel)dyn medub.ch. 120 ‘anything that a drunken man does’;auguenelhont (a w̯nelhont) do. 118. In O. W. is writtengu as inpetguarox. forpedwar ‘4’. The sound of, then, was virtually; this afterh <s becameχ, writtenchw-§ 26 vi,§ 94 iv. Initially on the analogy ofg: it becameg in the position of a radical consonant, remaining as a soft mutation; latergw̯‑:w̯‑. This had taken place before the separation of Bret. and Corn.

(2) The breathing before a vowel might also take the form, so as to give a new initialg‑; thus *orẟ§ 100 iii (2) >gorẟ f.b.t. 7 through *ᵹorẟ, this being taken for the soft mutation after the art., as inir guit (≡ỿr ᵹw͡yẟ) f.b.b. 97 ‘the goose’. Latergallt forallt f. ‘slope’;gerfydd besideherw̯ydd forerw̯yẟ§ 215 ii (5). The Mn.godidogg. 252 forodidog ‘rare’, and N.W. dial.gonest foronest probably involve a confusion of initialo- with the prefixgo-§ 156 i (16).

(3) Conversely initialg is sometimes lost, as ineuog ‘guilty’ forgeuog (geuawcỺ.A. 155 “mendaces”,gau ‘false’);elor f. ‘bier’ forgelorm.a. i 205a, met. for *gerol < Lat.gerula (elawrg. 234 is factitious).

Loss of Syllables.

§ 113. i. (1) The last syllable of every Brit. word, or Lat. word borrowed in the Brit. period, which contained more than one syllable, is lost in W. Thus W.gwynn f.gwenn ‘white’ < Brit. *u̯indos f. *u̯indā, W.ciwed < Lat.cīvitas, W.ciwdod < Lat.cīvitātem,§ 115 i. The syllable doubtless became unaccented in all cases; its vowel then became indistinct, and was ultimately lost, with the final consonant, except when the latter was a sonant. Brit. final‑l is unknown, and‑m had become‑n; the only final sonants therefore were‑r and‑n. When the syllable ended in one of these it seems to have become *‑r̥ or *‑n̥, which became non-syllabic. Final‑r remained, as in W.chwaer < Brit. *su̯esīr < *su̯esōr,§ 75 vii (2); W.ymherawdr < Lat.imperātor; but in common words it disappeared after a consonant in W., as inbrawd ‘brother’ for *brawdr (= Bret.breur) < Brit. *brāter. Final‑n nasalized a following initial media§ 106 ii (2), and was lost before other initial consonants. In the comparative it attached itself to the followingo, as inglanach no ‘cleaner than’ for *glanachn o§ 147 iv (3). It survived after a vowel innamen§ 78 ii (1),cymerwn§ 180 iii (1).

A sonant coining before the final vowel also remained, as inffenestr < Lat.fenestra,perigl < Lat.perīc’lum; later this was liable to drop where the new ultima was unaccented§ 16 v (3), and probably the vacillation between liquid and zero accounts for the development of excrescent liquids in some words:tymestl < Lat.tempestas,achreawdɏr§ 104 iii (2).

(2) The vocalic ending of the first element of a compound,§ 155 ii (1), became an obscure vowel, and disappeared; thus Brit.Maglo-cunos > W.Maelgwn; Brit. *Katu-mannos > W.Cadfan; Brit.Mori-dǖnon > W.Myrddin; Lat.bene-dictio > W.bendith. Similarly the vowel before the suffixes‑tāt‑,‑tūt‑,‑tero‑, etc., asciwdod < Lat. acc.cīvitātem,gwendid ‘weakness’ < Brit. acc. *u̯anno-tǖtan; and the‑i- in the spv. suffix *‑isamos, astecaf ‘fairest’ for *teghaf < *tek-isamos. In many words of four or more syllables the vowel of the second syllable was elided, as Ml. W.agwyẟawr < Lat.ābēcēdārium,meitin <mātūtīnum,Saesneg < *Saxonikā, etc. Stems in‑ā- had‑o- in composition; thus Kelt. *teutā ‘people’ wasTeuto- in compounds; andā̆ in the second syllable generally remains in nouns, as inCaradog < Brit.Caratā́cos,ffnrfafen < Lat.firmāmentum. But in many formations‑a- in the ante-penult was lost, as in Ml. W.karhont < *karasonti§ 183 ii (1), and the suff.‑gar < *‑ākaros§ 153 (8).

The loss of the root vowel in such forms asallweẟ < *n̥-ql’u̯‑íi̯ā§ 99 vi (1),dedw̯yẟ < *do-t’u̯‑íi̥os§ 100 ii (1) had probably already taken place in Brit. So in some cases the‑i- of the spv., as in Ml. W.nessaf < *ned’samos§ 148 i (1).

Disyllabic and compound prefixes are treated like the first element of a compound; thus Kelt. *ari- > Brit. *are- > W.ar‑; Brit. *kanta- > W.cannh-§ 156 i (6), (7); *kom-(p)ro- loses its‑o- and givescyfr- as incyfr-goll; so *u̯or-en-sed- loses its‑e- and givesgorsedd ‘high seat’, as if from *u̯ore-ssed‑.

(3) The inscribed stones (5th to 7th cent.) do not throw much light on the above changes. The ogam inscriptions are Goidelic, and those written in Roman letters are in bad Latin, while many of the names even in the latter are Goidelic in form. In some cases a name has the Lat. nom. ending‑us, asCatamanus Rhys no. 6 (LWPh.² 364),Aliortus no. 14,Veracius g, but most have the Lat. gen. ending‑i, asCunogusi hic jacit 5 ‘[the body] of C. lies here’. The names and the followingmaqqi of the ogams show that‑i is gen., and not a debased form of the Brit. nom.‑os. (The ogam‑i is the Kelt. gen. suffix *‑ī, being the Italo-Kelt. gen. of‑o- stems.) As a rule the Lat.fili agrees, but often does not, thusDervaci filius Justi ic jacit 37. Fem. nouns end in‑e, which is doubtless the ordinary late Lat.‑e for‑ae, though the noun in apposition stands in the nom., asTunccetace uxsor Daari hic jacit 77,et uxor eius Caune 20. A nom. in‑a appears inAvitoria filia Cunigni Eglwys Cymun insc. Possibly a Brit. nom. ending in‑o for‑os occurs inAliortus Elmetiaco hie jacet Rhys 14 (the only stone withjacet) andVitaliani Emereto 76. In a few cases no ending occurs:Etterni fili Victor 71, in which the legend is complete, andVictor is gen.;Velvor filia Broho 32. These and the false concords seem to indicate that the case endings were lost in the spoken language.

The stem-vowelu appears as‑u- and‑o‑, asCatu-rugi Rhys 60,Cato-tigirni 47; and‑o- appears as‑o‑,‑u‑,‑e‑,‑i‑, asCuno-gusi 5,Vendu-magli 45,Vinne-magli 21,Vende-setli 12,Venni-setli 67, pointing to‑ə- for which the Roman alphabet has no symbol. The form‑a- for‑u- or‑o‑, as inCata-manus 6, is Goidelic; cf. in bilingual stonesCuno-tami in Roman characters,Cuna-tami in ogam 75;Trene-gussi in Roman,Trena-gusu in ogam 73. In some cases the stem-vowel was preserved, and forms containing it survive beside forms in which it is lost; thusDumnagual besideDumngual both ingen. v. That the former is not merely an archaic spelling of the latter is shown by the survival of both in the Mn. language:

Mal mab iDdyfnwalMoel-mū́d
Yw Phylip braff i olud …
Mae yn llaw hilDyfnawal
Yr erwi maior a’r aur mâl.—L.G.C. 209.

‘Like a son of Dyfnwal Moelmud is Philip of vast wealth. In the hand of the descendant of Dyfnawal are the broad acres and the milled gold.’ Other similar doublets areTudwal andTudawalr.p. 1394,Dingad andDinogatb.a. 22. Theaw inDyfnawal is the regular development ofou̯ before a vowel, see§ 76 iii (1); before another consonant the‑o- remains, as seen inDinogat.

(4) The forms used in writing are always traditional, and in the above inscriptions the names have probably archaic forms preserved with the Latin in which they are embedded, since other evidence points to the loss of the terminations at this period. The re-formations consequent on the loss of the endings are largely the same in Bret. and W.; thus W.‑au, Bret.‑ou represents the pl.‑ou̯es ofu-stems,§ 120 i; these stems could not have been very numerous, and the addition of W.‑au and Bret.‑ou to nouns of all classes denoting common objects, and totad,mam and others, can hardly be an accidental coincidence, and is clearly subsequent to the breakdown of the Brit. declension. It seems therefore probable that the new language was in an advanced stage of development before the separation of the two dialects.

In the oldestms. of Bede,a.d. 737, the stem-vowels and terminations are completely lost, as inCar-legion,Ban-cor,Dinoot. The reduction was therefore an established fact in the early 8th cent.

(5) The vowel of the penult is sometimes lost after a diphthong, apparently when the accent originally fell on the ultima, as inclaer < *klii̯arós§ 75 vi (1);haul < *sāu̯eli̯ós§ 76 v (1); so probablycawr < *kou̯arós§ 76 iii (4). Withhaul ‘sun’ < *sau’li̯ós < *sāu̯eli̯ós contrast the disyllablehuan ‘sun’ < *sā́u̯anos < *sā́u̯ₑnos (withn-suff. like E.sun, cf. Walde² 721); affectedau, short because unaccented, gives W.au§ 76 v (1); and accentedā́u gives W.u§ 76 iii (5); see§ 76 v Note, p. 108.

ii. In a disyllabic proclitic a final short vowel might disappear in the Brit. period; thus Ar. *mene ‘my’ > *men, and caused the nasal mutation,§ 107 ii,iv.

iii. (1) The final consonant of a monosyllabic proclitic was lost in W.; thus Brit. *men ‘my’ gave W.fy ‘my’; but not till after it had mutated the following initial (in this case causing the nasal mutation of mediae§ 107 iv).

(2) But the consonantal ending of an accented monosyllable was in general retained; thus W.chw̯ech ‘six’ < Kelt. *su̯eks (butchwe before a noun); W.nos ‘night’ < Brit. *noss < *nots < *noqts§ 96 ii (5); W.moch ‘early’: Lat.mox; W.yn ‘in’ < Brit. *en < Ar. *en.


Notes
  1. G.R., Rowland, Silvan Evans, Tegai. Rowland'shaitch is a S.W. vulgarism.
  2. Here and in the following sections up to § 26, a letter printed in heavy type represents the written letter; a letter printed in italics represents the sound.
  3. There is one example inc.m. 5, which stands quite alone in ther.b., and so is prob. a scribal error.
  4. 4.04.1Pronounce thewyn ofwyneb so that it rhymes with thewyn ofddŵɥn,trŵɥn as the cynghanedd lusg demands.
  5. 5.05.1Nota’r gwyneb, os gwyneb, theg being ruled out by the cynghanedd.
  6. 6.06.1Notfy ngwyneb, eu gwynebau.
  7. Pughe, derivingcadwyn fromcadw̯, made itcadw̯yn, and asserted that it was masculine. He then inferred a fem.cadw̯en, which (ascadwyn is fem.) was unfortunately adopted by many 19th cent. writers. But no one has had the courage to writecadwenni for the pl.cadwyni. There is an old wordcadw̯en orcadw̯ent which means ‘a battle’, fromcad.
  8. This word has been naturalized in Welsh as in other languages, and the natural Welsh pronunciation is probably nearer the original than theṓmega now sometimes heard from the pulpit in imitation of the English fashion. The adjective is not an enclitic in ὦ μέγα. The natural accentuation, as used by the hymn-writers, is unconsciously adopted by those like A. Roberts who are not affected by a little learning.
  9. It is often supposed thatheulwen is a proper compound ofhaul andgwên, meaning the ‘smile of the sun’; but erroneously, forheulwen is the ‘sun’ itself, not ‘sunshine’.
  10. The common spellingPenmaenmawr appears to be due to popular etymology. Camden, 4th ed., 1594, p. 18, hasPen-mon maur, and the word is now pronouncedPèn-mon-máwr.
  11. The lastol ofolynol was mistaken about the middle of the last century for the adjectival termination‑ol (=‑awl), and from the supposed stemolyn an abstract nounolyni̯aeth was formed to render ‘succession’ in ‘apostolical succession’!
  12. In all the above examples the cynghanedd is either T2 or C2, which implies the accentuation indicated. See ZfCP. iv. 124, 137.
  13. The cynghanedd is S4, which implies the accentuation marked.
  14. L. G. C.’s editors printvoneddig in spite of the answeringh innyhuddo.
  15. nrh andnhr have the same sound but differ in origin:nrh =n +rh;nhr is fromn +tr. They are often confused in writing.
  16. G. Mechain (iii. 224) writing to Tegid, assents tobrenin, breninoedd “though from habit I always readbrenhinoedd with an aspirate; but the root does not warrant such reading.” His pronunciation was correct; and it just happens that the “root” does warrant it; see§ 103 ii (1).
  17. In Late W. wrongly spelttudwedd from a fancied relation totud ‘people’, whence ‘country’. The examples inb.b. both rhyme with‑ed.
  18. The identification of the name (treated as two common nouns by Silvan Evans) is the discovery of Professor J. E. Lloyd, Tr. Cym. 1899–1900, p. 158
Wikisource notes
  1. Corrected on p. xxvii to: “Kellynnawc (lll)”
  2. On p. xxvii the author says to add: “f,dd”.
  3. On p. xxvii the author says to delete “;—raccw § 210 x (3)”
  4. On p. xxvii the author adds “*” here.
  5. Corrected on p. xxvii to: “di|e|fyl
  6. Corrected on p. xxvii to “*ad-rīm‑
Retrieved from "https://en.wikisource.org/w/index.php?title=A_Welsh_Grammar,_Historical_and_Comparative/Phonology&oldid=15530394"
Hidden categories:

[8]ページ先頭

©2009-2025 Movatter.jp