This glossary provides a list ofsound laws that have been formulated by linguists for the various Indo-European languages. Any sound law which affects any of the major branches of the Indo-European family or more than one descendant language are included.
The word-medial sequence*-mn- is simplified after long vowels and diphthongs or after a short vowel if the sequence wastautosyllabic and preceded by a consonant. The*n was deleted if the vocalic sequence following thecluster was accented, as inAncient Greekθερμός (thermós, 'warm') fromProto-Indo-European*gʷʰermnós ('warm'); otherwise, the*m was deleted, as inSanskritअश्नः (áśnaḥ) from Proto-Indo-European*h₂éḱmnes ('anvil [gen. sg.]'). The sequence remains if the*-mn- sequence is heterosyllabic, such as in Ancient Greekπρύμνος (prýmnos, 'prominent'). The law was first discovered byJohannes Schmidt in 1895 and is named for theAvestan reflex𐬀𐬯𐬥𐬋 (asnō).[5][6]
aspirate throwback
(Ancient Greek, Sanskrit)Also,aspiration throwback. When a root-final aspirated stop loses its aspiration for whatever reason, typically due to another process, the aspiration is retracted to the initial consonant whenever that initial consonant is capable of taking an aspirated quality.[7][8] One example includes the Ancient Greek rootτρίχ- (tríkh-, 'hair'), which becomesθρίξ (thríx) in thenominative form.[7] The process was mentioned earlier by the Sanskrit scholarPāṇini, but brought to modern scholarship in the first clause of a two-part law proposed byHermann Grassmann in 1863,[9] though the name "aspirate throwback" appears later.[7][a] The second clause is now referred to alone asGrassmann's law.[7]
Also,Buddha rule. If a cluster of two or moreobstruents contains at least onevoicedaspirated consonant, the whole cluster becomes voiced and aspirated. The process may have been inherited from Proto-Indo-European, though this is not universally accepted.[10] The law is named after the German linguistChristian Bartholomae who discussed outcomes of the process in the variousIndo-Iranian languages in 1882.[11] The alternative name stems from the fact that the etymon for the Sanskrit wordबुद्ध (buddhá; 'awake, enlightened') is affected by this process, derived from*bʰewdʰ, meaning 'to be awake', and*-tó-, the passivepast participle suffix.[12][13]
Labiovelars lose their labialization and become plain velars when preceded or followed by*w or*u.[14][15] Thisdissimilatory process explains the reflex in words like Ancient Greekβουκόλος (boukólos, 'cowherd'), derived from Proto-Indo-European*gʷoukʷólos. The expected form*βουπόλος (*boupólos) does not appear because the initial*kʷ in*kʷólos is preceded by the*u in*gʷou-, whereas inαιπόλος (aipólos, 'goatherd'), the expected form-πόlos (-pólos) is attested, derived from Proto-Indo-European*ai(ǵ)kʷólos.[14] This process remained productive intoProto-Germanic, where it also came to apply to labiovelars preceded by*-un- through anassimilatory process which caused*n to have a labializedallophone.[15] Examples of this include Proto-Germanic*tungōn- from Proto-Indo-European*dn̥ǵʰwéh₂- both meaning 'tongue', whence bothLatinlingua (fromOld Latindengua) and Englishtongue (fromOld Englishtunge).[16]
double-dental rule
When twodental consonants form a consonant cluster, a sibilant isepenthesized between the dental consonants.[17] Examples include Proto-Indo-European*witsto- ('seen, known'), which isunderlyingly*wid-tó-.[18] Whether this should be interpreted as*s-insertion oraffrication is debated.[19][20]
Laryngeal consonants are lost between a vowel and any other consonant in pretonic syllables.[21][22] Examples of this includeProto-Celtic*wiro- (whenceOld Irishfer, meaning 'man'), Latinvir ('man'), and Old Englishwer ('man'), all of which are derived from Proto-Indo-European*wiHró-. If the vowel is long before the process occurs, it is shortened.[21][23] The process likely did not take place in Proto-Indo-European as it only affected the western Indo-European languages; examples from eastern families show laryngeal retention, such asLithuanianvýras ('man, husband') and Sanskritवीरः (vīráḥ; 'man, hero').[24] The law is named forVladimir Dybo, who published work on the topic in 1961.[25]
(Ancient Greek, Sanskrit)Also,ha-ha rule;breath dissimilation. When an aspirated consonant is followed by another aspirated consonant in same or following syllable, the first consonant loses its aspiration. Examples of pairs affected by this process include Ancient Greekθρίξ (thríx, 'hair') in the nominative case, butτριχός (trikhós) in thegenitive case.[26][27] Hermann Grassmann first proposed this process in 1863 as the second clause of a two-part law. The first clause is now known as theaspirate throwback.[26] The law remained productive after the Greek devoicing of aspirates and /h/,[b] from earlier*s, which behaved as an aspirated stop.[29]
Kortlandt effect
Proto-Indo-European*d undergoesdebuccalization, becoming the laryngeal*h₁, whenever it is followed by adental consonant, a consonant followed by a dental, or whenever the following syllable begins with a dental.[30] Examples include Ancient Greekἑκατόν (hekatón, 'one hundred') from*dḱm̥tom, where the simple loss of the initial*d is attested in forms like Latincentum and Sanskritशतम् (śatám), but the debuccalization of*d instead of deletion adequately explains the initial vowel in the Greek term.[31] The process also explains the relationship between the Ancient Greek termsδρέπω (drépō, 'I pluck'), from the Proto-Indo-European root*drep-, andHomeric Greek*ἐρέπτομαι (*eréptomai; 'I feed on, I munch'),[c] from the same root affected by the debuccalization (*h₁rep-).[34] The law is named afterFrederik Kortlandt, who first proposed the sound change in 1983.[35][36] The debuccalization appears to have taken place before theAnatolian languages split off from Proto-Indo-European.[37]
Laryngeals are lost in utterance-final position.[38] The process does not causecompensatory lengthening.[39] The Latin nominative singular, which ends in-a rather than the expected*-ā, is thought to be derived from an earlier usage of thevocative case, which would have invoked the law; this process did not affect the relatedSabellic languages, however.[40] The process similarly affected the vocative in Ancient Greek as well.[24] The law is named for the Dutch linguistF. B. J. Kuiper, who published work on this process in 1955.[41][42]
In a word of three syllables with a vowel pattern*é-o-V, where V is any vowel, the accent is moved forward to the middle syllable, becoming*e-ó-V. This explains the penultimate accent in terms likeVedic Sanskritचत्वारः (catvā́raḥ), the nominative plural form of 'four', from Proto-Indo-European*kʷetwóres. The law is named after this example.[43]
In a dental–dental–resonant sequence, one of the dental consonants is deleted and there is no compensatory lengthening.[18] Evidence from Latin and Gaulish seems to suggest that the second dental consonant was simply deleted, but evidence from Greek and Sanskrit indicates that the second dental consonant underwent voicing assimilation and then the resulting geminate was shortened; Proto-Germanic evidence provides examples of both.[44] The name is derived from the Greek termμέτρον (métron, 'measure'), derived from Proto-Indo-European*méd-tro-.[45]
neognós rule
Laryngeals are lost in zero-grade contexts where full-grade root contains a consonant–vowel–resonant–laryngeal string, in that order, in certain reduplicated forms and in some other compounds. Examples include the Ancient Greek termνεογνός (neognós, 'newborn'); the Greek term is derived from Proto-Indo-European*newoǵn̥h₁o- through the medial form*newoǵno-.[24][46]
When a long vowel is followed by either a liquid or nasal consonant which is itself followed by a stop consonant or*s, the vowel is shortened.[47] In addition to liquids or nasals, glides may also trigger the process in that position, though this is controversial.[47] The law is named afterHermann Osthoff, who first postulated the process in 1879, followed by two important reanalyses in 1881 and 1884, though this particular appellation was not universal until as late as 1939.[48][47]
Also,Pinault's rule. Laryngeals are dropped in word-medial position between a consonant and*y, such as in Latinsocius ('friend') from Proto-Indo-European*sokʷh₂-yo-.[22][49] Only*h₂ and*h₃ appear to have been affected,[49] though the law has been invoked to explain instances of*h₁'s disappearance in the same context, such as Proto-Celtic*gan-yo- ('to be born') from Proto-Indo-European*ǵnh₁yetor.[50] The law is named for the French linguistGeorges-Jean Pinault.[49]
(Balto-Slavic, Albanian, Armenian, Indo-Iranian)Also,RUKI;ruki;ruki rule;ruki change;iurk rule;Pedersen's law. In thesatem languages, Proto-Indo-European*s is retracted when preceded by*r,*u,*k, or*i.[51] In Indo-Iranian, the phoneme retracted toš, while in Slavic it further retracted intox in most circumstances.[52][53] This change is not generally believed to be commonly derived from an earlier common innovation, but rather were independently conditioned.[52] In Indo-Iranian, the sound change caused the retroflexion oft,d, andn in the same contexts, not justs.[52] Similarly, it does not appear that the Baltic languages underwent the full process.[52] The consonants*g and*gʰ are also sometimes included.[53] The name is derived from the constituent letters which govern the change. Its order is a Russianmnemonic device;ruki (руки) means 'hands'.[54] Although the concept was articulated as early as 1818 byRasmus Rask, the alternative name is derived fromHolger Pedersen, who wrote about the process in detail.[55]
(Not fully accepted) If ans-mobile is added to a root that begins with a voiced consonant, that consonant is devoiced. If it is aspirated, it retains its aspiration, giving a tripartite alternation between the s-mobilized, plain, and voiced unaspirated forms.[56][57] The law is named after the German linguistTheodor Siebs who first proposed the concept in 1901, though it was not published until 1904.[56]
When two consecutive nonsyllablic consonants or a long vowel and a nonsyllabic consonant precede a nonsyllabic sonorant, the sonorant becomes syllabic.[58] Examples of this process include the adjective-forming suffix*-yó-, as in Greekπεζός (pezdos, 'on foot') from*pedyós, as compared with its non-adjectival formπόδες (pódes, 'feet') from the Proto-Indo-European stem*ped-.[58] The law is named for the German philologistEduard Sievers, who first promulgated the concept in 1878.[59]
Word-finally, when a laryngeal, *y or *w is preceded by a vowel and followed by a nasal consonant, it is dropped and the preceding vowel is lengthened.
1. (Not fully accepted) In pre-Proto-Indo-European, the word-final fricatives*s and*h₂ are deleted following a vowel–resonant sequence, followed by compensatory lengthening.[5] Some linguists include in the law a process where if the resulting sequence is*-ōn, the*n is also dropped, but others describe that deletion as a separate process.[60] The law is named after the Hungarian-British linguistOswald Szemerényi who first described the process in 1956.[61]
2. Also,broad Szemerényi's law;final Szemerényi's law. In pre-Proto-Indo-European, syllable-final fricatives are deleted following a vowel–consonant sequence, followed by compensatory lengthening. The process is blocked if the consonant preceding the fricative is in a cluster or if the following syllable begins with a consonant.[62][63]
weather rule
"Weather rule" redirects here. For the utility regulation, seeCold weather rule.
Laryngeals are lost in word-medial position preceding a stop followed by a resonant and a vowel. The law is named for its reflex in English,weather, which is derived from Proto-Indo-European*weh₁dʰrom ('weather'); the laryngeal*h₁ is deleted before the sequence*-dʰro- which comprises a stop, a resonant, and a vowel, respectively.[24] The law does not cause compensatory lengthening.[39]
The palatovelar consonants*ḱ *ǵ *ǵʰ are depalatalized when preceding*r unless that*r is followed by an*i.[64] The law is named after the German linguistOskar Weise who observed the results of the change in a 1881 essay on the topic.[65]
Within a morpheme, only one nasal–stop cluster is allowed; if more than one cluster exists in theunderlying representation, any of these clusters created by morphophonological processes, such asepenthesis, are deleted.[66] Examples are found in the termskuvend ('assembly') andmbrëmë ('last night'), where the respective expected epenthetic forms*nguvend and*mbrëmbë reduced following this process.[67] The process has been likened toLyman's law in Japanese andGrassmann's law in Ancient Greek and Sanskrit.[68] The law is named for the American linguist Samuel Rosenthall who first proposed the process in 2022.[66]
At some point after Proto-Indo-European, but before the first attestation ofHittite, voiceless stops developed intogeminates in word-medial position and voiced stops devoiced.[69] Examples of the process include𒄿𒌑𒃷 (yukan, 'yoke') from Proto-Indo-European*yugóm, contrasted with𒊭𒀝𒋼𒀀 (šakkar; 'dung, excrement') from Proto-Indo-European*sóḱr̥.[70] The law is named for the American linguistEdgar Sturtevant, who first formulated the law in 1932, though he credited "the inspiration for [the] observation" to his studentC. L. Mudge.[71]
In disyllabic verbs, if the syllable-final vowel is short or has a circumflexed tone, the stress is moved from that final syllable to the first syllable.[75] The process appears to have been resistant to homophony, however.[76] The law is named for the Dutch linguistCarl Ebeling who first postulated the process in 1963, with a revised form in 1967; it was later revised further byFrederik Kortlandt in 1977.[76][77] Chronologically, the process occurred extremely early in the history of the language, shortly after the application ofHirt's law and before the application ofDybo's law.[76]
Lidén's law
Word-initial*w in Proto-Indo-European is lost before non-syllabic*r and*l as the language developed intoProto-Balto-Slavic.[78] The law is named for the Swedish linguistEvald Lidén [sv;no;hy], who wrote about the process in 1899.[79] The process described by the law likely occurred after the development ofHirt's law, but before the syllabification of resonants. While it is possible that the law occurred after the syllabification of resonants and only affected non-syllabic resonants,Ranko Matasović finds this "improbable on phonetic grounds".[80]
Also,Hirt–Illich-Svitych's law. If the syllable preceding the expected stressed syllable has a vowel immediately followed by a laryngeal, the stress is retracted to that syllable. Examples include comparisons of Lithuanianvýras ('man, husband') andmótė ('mother') with Sanskritवीरः (vīráḥ; 'man, hero') andमाता (mātā́, 'mother'), respectively. The law was first proposed by the German philologistHermann Hirt in 1895, but the original formulation was corrected in 1963 by the Soviet linguistVladislav Illich-Svitych.[81] The process applied prior the elimination of zero-grade stems and before laryngeal consonants underwent their merger; Ebeling demonstrated in 1967 that it must have occurred after the language underwentoxytonesis.[81] It occurred shortly before the application ofEbeling's law.[76]
In words with a mobile accent paradigm of three syllables or longer, the accent was retracted from a medial onto the initial syllable.[81][82] Examples include Lithuanianpiemuõ ('shepherd') in the nominative, butpíemenį in the accusative.[81][82] A falling tone is always found on the affected syllable when the law is not invoked.[82] The law is named afterHolger Pedersen, who first proposed it in 1933.[83][82] The process occurred very early in the Balto-Slavic period;Frederik Kortlandt considers Pedersen's law to be the oldest law of retraction and possibly the earliest law of accent in the language.[82] However, the process later reappeared in both the Slavic and Baltic languages.[84] In Lithuanian, for example, it appears to have reappeared sometime after the lengthening of stressede anda, but before the application ofde Saussure's law.[85] In Slavic, it occurred after the application ofEbeling's law.[83]
Short vowels with non-acute accents are lengthened before unaspirated voiced stops (*b,*d,*g, but not*ǵ). The newly lengthened vowel receives the acute accent.[86] The law is named after the German linguistWerner Winter who wrote his proposal in 1976, though it was not published until 1978.[87] Frederik Kortlandt has dated the law to the final years of the Balto-Slavic period.[88]
(Lithuanian)Also,Saussure's law;Fortunatov–de Saussure's law. If a word's stress is on a syllable without a falling tone, the syllable moves to the closest following syllable with an acute vowel.[89][90] The law is named after the Swiss linguistFerdinand de Saussure who built on previous works on the subject byFilipp Fortunatov andAdalbert Bezzenberger in 1894 and 1896.[91] The process was preceded byPedersen's law.[89][90]
Hjelmslev's law
(Lithuanian) When a vowel receives an accent, it takes on the intonation of the following syllable.[92] The law is named for the Danish linguistLouis Hjelmslev, who proposed the process in his doctoral thesis in 1932.[93] Kortlandt states that, if the ictus retracts to a laryngealized vowel, the laryngeal is deleted and the result is a rising tone, butN. E. Collinge argues this is beyond the scope of process and that both Hjelmslev and de Saussure allow for a falling tone in their analyses.[94][89]
When theProto-Baltic sequence*-ás is found word-finally, it is reduced to*-əs and loses its ability to carry stress. As a result, final stress is retracted to the penultimate syllable. This change explains the accentuation paradigm ofo-stem nouns. Examples includebáltas ('white') andvar̃nas ('raven') from earlier*langás and*warnás, respectively. This is contrasted with mobile paradigms, where the accent is final, such as ingalvà 'head' andžvėrìs 'beast'.[98] The law is named after the Finnish linguistEino Nieminen [fi;lt;eo;pl] who proposed the law in 1922.[98][99] The process certainly precedesLeskien's law in Lithuanian, but has been suggested to go at least as far back asProto-Baltic based on evidence fromOld Prussian.[100]
Also,Vasilyev–Dolobko's law. In a word with mobile stress and a final encliticized morpheme, the stress moves from the initial syllable to the final syllable.[101][102] If the encliticized morpheme has no syllable, the stress moves to the preceding syllable.[101] This process occurred sometime between the end of the Balto-Slavic period and the loss of intervocalic*j, which places it sometime beforeDybo's law.[101] The law is named afterMily Dolobko [ru] who wrote about the law in 1927. Because its first formulation was byLeonid Vasilyev [ru] in 1905, it is sometimes known by the joint name.[102]
Also,Dybo–Illich-Svitych's law;Illich-Svitych's law. If a neoacute syllable was accented, the accent shifted to the following syllable. The process wasblocked if the accent wasunderlyingly mobile.[103] Examples include*blъxy ('fleas'), with stress on the second syllable, as compared with Lithuanianblùsos and Greekψῠ́λλαι (psýllai), where stress falls on the first.[103] Though it always shifts right, the resulting accent differed based on the conditions of the following syllable. For example, the underlying form**dòbrota became*dobròta ('goodness'), while**žènica became*ženı̋ca (diminutive of 'wife').[103] The process has been dated to around the turn of the 9th century and followed shortly byStang's law of Slavic, but is preceded byVan Wijk's law.[104][105] The law is named for the Russian linguistsVladimir Dybo andVladislav Illich-Svitych, who first discovered the process. Dybo published an article on the topic in 1962, followed by another in 1963 which wasco-authored with Illich-Svitych and another the same year by Illich-Svitych alone.[103][106]
Following the results ofHavlík's law, weak yers were deleted. Examples are found inOld East Slavic, where earlier forms for the word 'prince' demonstrate a retained yer (кънѧзь [kъnjazь]), but later forms show it without (кнѧзь [knjazь]); the retained forms are dated to around 1075, while loss became widespread between the 1120s and 1210s.[107]
In lateProto-Slavic, the vowels represented by*ь and*ъ, calledyers, become "weak" – that is, subject to moraic shortening – in final position or if they are followed by a vowel other than another yer. A yer becomes "strong" if the following syllable contained a weak yer. In words with three successive syllables all containing yers, the final yer was weak, causing the penultimate syllable to be strong and the antepenultimate weak.[107] This process represents the first phase of the largeryer shift; the second phase is known as thefall of the weak yers, where weak yers were deleted.[107]
Also,Ivšić's retraction;Stang's law;Stang–Ivšić's law. Accented weak yers, as according toHavlík's law, lost their accent to the preceding syllable, which received a "neoacute" accent.[citation needed] The law is named forStjepan Ivšić, who first articulated the concept in 1937.[108]
Also,opening of syllables;tendency to rising sonority. Word- and syllable-final obstruents and obstruent clusters are deleted. Finals nasals are lost after short vowels and nasalized after long vowels.[109]
In words with a mobile accent paradigm, if the first syllable is accented with a rising (acute) accent in Proto-Balto-Slavic, it is converted into a falling (circumflex) accent in Proto-Slavic. Examples of this includeSerbo-Croatiangláva in the nominative case, butglȃvu in the accusative. The process appears to be somewhat resistant to analogical leveling.[110] The law is named after the French linguistAntoine Meillet who first described the law in 1902.[110]
Shakhmatov's law
(No longer widely accepted)Also,Šaxmatov's law. Stressed short falling (circumflex) accents shift to the preceding syllable.[111][112] The law is named afterAleksey Shakhmatov who described the process in 1915.[112][113]
Also,first palatalization of velars;first regressive palatalization. When the velar sounds*k,*g, and*x occur before front vowels and*y, they undergopalatalization and become*č,*ž, and*š, respectively. If the velar follows a sibilant (i.e.,*sk or*zg), the sibilant was backed to*šč and*ždž before the cluster was reduced in some dialects to*št and*žd.[114] The process was probably mediated by coronalization before a finalassibilation. The time at which the process occurred is unknown but it must have begun prior to theMigration Period and was completed sometime between the 6th and 8th centuries.[115]
Also,second palatalization of velars;second regressive palatalization. Following the monophthongization of Proto-Slavic*ai to*ě in open syllables and*i in closed syllables, velars became palatalized again. The process had different outcomes based on dialect:*k typically became*c and*g became either*dz or simply*z, but*x became*š inWest Slavic,*xʲ inNovgorodian, and*s elsewhere. The process is estimated to have happened around the same time theWest Slavs were migrating north ofCarpathia and theNovgorodians were migrating to what is now northwestern Russia.[116] Like thefirst palatalization, the process occurred in several stages: palatal coarticulation, palatalization, and assibilation. It appears that Novgorodian only went through the first stage.[117] Clusters that had originally contained*ku̯ underwent palatalization inSouth Slavic and someEast Slavic unless it was preceded by*s, but West Slavic and Novgorodian did not experience this palatalization at all.[118]
Also,progressive velar palatalization;palatalization of Baudouin de Courtenay. When Proto-Slavic*i,*ī, or*in precede a velar, the velar is palatalized and then assibilated;*k and*g become*c and*dz, respectively, in all languages, with*dz undergoing furtherlenition to*z outside of Eastern South Slavic, Slovak, andLechitic. In West Slavic,*x became*š, but*s or*sʲ inSouth andEast Slavic.[119] Examples of this process include earlier*liːkad, meaning 'face', which became*liːca (whenceOld Czechlíce), and*mεːsinkaːd, thegenitive singular form of 'moon', which became*mεːsĩːcaː (whenceOld Polishmiesięca).[119] The first stage of this process was in effect by the 6th and 7th centuries.[120]
Stang's law
Not to be confused with the Proto-Indo-European lawof the same name orIvšić's law, sometimes also called "Stang's law".
If a word-final syllable was long falling (circumflex) accented, the accent was retracted onto the preceding syllable. The originally accented syllable is shortened, and the newly-accented syllable receives a "neoacute" accent. This change applied afterDybo's law, and often "undid" it by shifting the accent back again.
van Wijk's law
When*j follows a consonant, it becomes*ь. In post-tonic syllables, the*ь assimilated to the following vowel, lengthening it.[121][105] Whether the lengthening process is a part of the larger assimilatory process is the subject of some debate, though the lengthening as a separate result is sometimes termed "van Wijk's length".[122] AlthoughAleksey Shakhmatov was the first to suggest the process in 1898, the law is named for the Dutch linguistNicolaas van Wijk, who published work on the process in 1916.[123][124]
Also,jer shift;third Slavic vowel shift;fall of the yers. The yer vowels,*ь and*ъ, underwent two-part process followed by a third step that fractured the realization of the vowel qualities. First, yer vowels underwent an alternating pattern of weakening every other yer in a word beginning with weakening the final one; this first part of the shift is referred to asHavlík's law.[125] The process is a patterned form ofcompensatory lengthening.[126] Next, the weak yers were deleted, referred to as thefall of the weak yers.[107] Following this deletion, the remaining strong yers were thereby shifted to different vowel qualities in the variousSlavic languages, which collectively are known as the vocalization of the yers.[127][128]
During the period between Proto-Indo-European andProto-Celtic, when*e is followed by a resonant then by*a, the*e assimilates to*a. In other words, in the sequence*eRa, where*R signifies any resonant,*e becomes*a, thereby becoming*aRa.[129][130] Examples of this change include Proto-Celtic*taratro- ('drill'), whenceIrishtarathar ('auger') from earlier*teratro-, derived from Proto-Indo-European*terh₁tro-, whence Ancient Greekτέρετρον (téretron; 'borer, gimlet'). The law does not affect*ā and probably did not affect environments where the*a was word-final.[129] The law also appears to have affected words where the*a was formerly a laryngeal consonant in Proto-Indo-European, such as in Proto-Celtic*banatlo- ('broom plant') which may be derived from Proto-Indo-European*bʰenH-tlo from a root meaning 'to hit, to strike'.[131] The process was expanded inWelsh include environments where the resonant is followed by a nasal, explaining the vowel quality in Welsh words likesarnu 'to trample' but notOld Irishsernaid ('to arrange, to order'), both from a Proto-Celtic*sternū/*starnati paradigm (from an older subjunctive Proto-Indo-European form*ster-nh₂-e/o, cognate with Latinsternō).[132] Similarly, another expansion of the process appears in bothBrittonic andGaulish, causing the assimilation of*o to*a in the resonant–vowel environment*oRa, thereby rendering*aRa. CompareMiddle Welshtaran ('thunder') and GaulishTaranis ('the Celtic god of thunder'), which were affected by the expanded law, with Old Irishtorann ('thunder'), which was not.[133] The law is named after Lionel Joseph who covered the topic in 1982.[129][131]
(Old Irish)Also,MacNeill–O'Brien's law.Lenition of Old Irishn,r, andl is lost in final unstressed syllables even though they are etymologically expected to be lenited in that position.[134][135]
1. Proto-Celtic*b and*u̯ become*β before*n word-internally. The latter change is rare, but occurs in words like Old Irishamnair ('maternal uncle') from Proto-Celtic*au̯n andomun ('fear') from*ɸoβnos. The law appears to have only occurred in contexts where there is no front vowel preceding the cluster, which accounts for apparent counterexamples like Welshclun ('hip, haunch') from Proto-Indo-European*ḱlownis and Old Irishbúan ('permanent') from*bʰewHnos, though some other etymologies account for different sources; the Welsh term may be a later borrowing from Latinclūnis, for example. The law shares a relationship to another Proto-Celtic sound change where*ɸ became*u̯ between either*a or*o and*n.[136]
2. (Old Irish) Unless the syllable is stressed, voiceless obstruents are voiced word-initially and word-finally.[137]
(Not fully accepted) Following the application ofGrimm's law, when the laryngeal consonant*h₃ is preceded by a sonorant and followed by*w, it becomes*k.[138] Purported examples of this process include*gʷih₃wós.[139] It may also have affected*h₂, as in*taikuraz from Proto-Indo-European*dayh₂wḗr ('brother-in-law'), as compared with Sanskritदेवा (devā́) and Greekδᾱήρ (dāḗr).[139] The law has been the subject of heated debate;Don Ringe, however, describes objections to the law as somewhat incoherent.[139] Dating the law has proven somewhat difficult; Ringe suggests that if the law affects*h₂, it probably precededGrimm's law and theepenthesis of*ə between nonsyllabic consonants and laryngeals.[140] If true, it likely became*g before beingdevoiced to*k by Grimm's law.[140] Irrespective of whether the law affected*h₂, it certainly preceded the merger of labiovelars and clusters of a velar consonant followed by*w.[140] The law is named for the American linguistWarren Cowgill who formulated what is considered its strongest defense over several decades, though it was first proposed byWilliam Austin, another American linguist, in 1946.[141]
Also,Primärberührung.[d] When a plosive is followed by any voiceless sound, normally *s or *t, it becomes a voiceless fricative and loses its labialization if present. A dental consonant followed by *s becomes *ss, and followed by *t becomes either *ss (in older inherited forms) or *st (in newer and productive forms).[citation needed]
Also,first Germanic sound shift;first Germanic consonant shift;Rask's rule. The three series of Proto-Indo-European plosives undergo a chain shift. The first shift causes voiceless stops –*p,*t,*k, and*kʷ – to become the voiceless fricatives*f,*θ,*x, and*xʷ, respectively.[e] Next, the plain voiced stops –*b,*d,*g, and*gʷ – devoice and become*p,*t,*k, and*kʷ, respectively. Lastly, the aspirated voiced stops –*bʰ,*dʰ,*gʰ, and*gʷʰ – become voiced stops*b,*d,*g, and*gʷ, respectively.[146] This sound change is sometimes obfuscated inOld High German as a result of theHigh German consonant shift.[147] The process did not affect the second consonant in a cluster of two adjacent obstruents. Compare two versions of theOld Frisian word for 'throat':strot- andthrot-. Both are derived from the Proto-Indo-Europeans-mobile root*(s)trewd-, the former including the s-mobile while the latter does not. In this example, the form with the s-mobile blocks theassibilation of the*t.[144] Several other exceptions are covered byVerner's law.[148] The Danish linguistRasmus Rask is credited with first articulating the law in 1818, but its more common name is from German folkloristJacob Grimm, who – after reading Rask's work – expanded on it and published it in the preface of his German grammar book in 1819 with a rewrite in 1822.[149]
(North andEast Germanic)Also,Verschärfung;sharpening;intensification. Thegeminated glides*jj and*ww arehardened into geminate plosives. In North Germanic,*jj becomes*ggj, while it became*ddj in East Germanic,*ww becomes*ggw in both.[150]
(Ingvaeonic languages) When*n is followed by a fricative, the*n is lost and the preceding vowel is lengthened to compensate.[151][152] The vowel was probably nasalized first and then experienced a loss in nasal quality before lengthening.[153]
(Controversial) FollowingGrimm's law, if a stop consonant abuts*n, the*n devoices and assimilates, creating a geminate stop.[154]Donald Ringe writes that "the problem with Kluge's suggestion is simply that the etymologies are unconvincing".[154] Frederik Kortlandt, however, supports the law, estimating that it can be dated to between Grimm's andVerner's law.[155] The law is named for the German philologistFriedrich Kluge.[154]
(Gothic) Spirants in unaccented syllables changed their voiced–unvoiced quality based on the quality of the preceding consonant, whereby voiced spirants appear after unvoiced consonants and voiceless spirants appear after voiced consonants. Examples of both can be found in thedative singular forms,𐌰𐌲𐌹𐍃𐌰 (agisa, 'fear') and𐍂𐌹𐌵𐌹𐌶𐌰 (riqiza, 'darkness').[156][f] In short, spirants are voiced when they are immediately preceded by a vowel without primary stress and the preceding consonant before the vowel is unvoiced. If the preceding consonant is a cluster where the second consonant is a liquid, the spirant remains unvoiced, but if the second consonant is a glide, it is voiced.[158] The process does not affect word-final spirants and the second element in a compound word where the simplex is stress-bearing. While some exceptions occur due tomorphological leveling, there are at least seven words for which Thurneysen could not supply an explanation.[159] The law is named after the Swiss linguistRudolf Thurneysen who posited the law in 1896 and published it in 1898.[160]
After the application ofGrimm's law, when a voiceless fricative is preceded by an unaccented syllable and not word-initial or abutting another voiceless consonant, it is voiced.[161] Examples of this process include Proto-Germanic*ubilaz ('evil, bad') from Proto-Indo-European*h₂upélos with the intermediate form*ufélos.[161] Following this, the mobileProto-Indo-European accent was lost and stress fell on the first syllable by default.[162][163] The process occurred before the loss of*gʷ from Proto-Germanic's phonemic inventory, though its chronological relation to vowel contraction in hiatus is unclear.[164] The law is named after the Danish linguistKarl Verner who described the process in 1876.[165]
Also,Bàrtoli's law. In short–longmetrical feet whereoxytone stress is expected, the syllable becomesparoxytone before a word boundary. The law only occurs in anapestic (short–short–long) and cretic (long–short–long) contexts. An example of this process can be seen in Ancient Greekθυγάτηρ (thygátēr, 'daughter'), in its nominative singular form, when contrasted with its accusative singular formθυγατέρα (thygatéra) and with the Sanskrit nominative singularदुहिता (duhitā́, 'daughter'). While some exceptions can be attributed to analogical change, there are still some unexplained exceptions.[166] The law is named after the Italian linguistMatteo Bartoli, whose 1930 publication contrasted Sanskrit oxytone words with their Ancient Greek paroxytone cognates.[167]
Whenever Proto-Indo-European*o occurs between a resonant and a labial, it becomes Greekυ (y).[168][169] The resonant and the labial can be on either side of the*o and produce the same output. Examples of a resonant followed by*o followed by a labial includeνύξ (nýx, 'night') from*nokʷt-, whence also Latinnox ('night'). This law is preceded by laryngeal coloring, meaning that Proto-Indo-European sequences of*h₁o,*h₃o, and*h₃e are also accounted for in the law, as are cases in which the zero-grade form was vocalized, such as inστόρνυμεν (stórnymen 'we smooth out') from Proto-Indo-European*str̥-n-h̥₃-, showing thenasal infix. Some later sound changes obfuscate the law, but there is evidence to show that the sound change still occurred. For example, Proto-Indo-European*nomn̥ ('name') gave way toProto-Greek*onuma. Although inAttic Greek the form becameὄνομα (ónoma), the expected formὄνυμα (ónyma) is found in bothDoric andAeolic Greek; the expected form is also found in derivatives such asἀνώνυμος (anṓnymos; 'nameless, inglorious').[168]
Stress cannot appear any earlier in a word than the antepenultimate syllable. If the final syllable is heavy, it cannot appear any earlier than the penultimate, but syllable-final consonants do not contribute to syllable weight.[170] The law is attested in every known Greek dialect, with the possible exception ofThessalian subdialect ofAeolic Greek.[171] Some exceptions exist in Attic and Ionic Greek as a result ofquantitative metathesis, which disrupted the original formulation.[171] The process probably predatesHomeric Greek, though it is certainly constrained to Greek and not a larger Indo-European process.[171]
Also,Vendryès's law. Anyperispomenon with a short vowel in the antepenultimate becomes proparoxytone inAttic. The law is named after the French linguistJoseph Vendryes.[172]
Also,law of dactylic retraction. Oxytone words in Proto-Indo-European become paroxytone in Ancient Greek if the word has a dactylic ending; in other words, the stress in dactylic words moved from the final syllable to the penultimate syllable.[172][173] The law counts endings such asον-on,ος-os, andοι-oi as short. Examples of this tonic retraction include cognate pairs likeποικίλος (poikílos; 'variegated, complex'), with a paroxytone, and Vedic Sanskritपेशलः (peśaláḥ). The law is named after the American philologistBenjamin Ide Wheeler who published work on the topic in 1885.[172][173]
In open, non-final syllables, the vowel*o is lengthened and becomes*ā. In all other contexts,*o becomes short*a.[174][175]Masato Kobayashi has argued that theunderlying sound change is*o to*ā, but the sound change does not occur in closed syllables to circumvent syllable-weight violations.[176] The law is named forKarl Brugmann who first articulated the process in 1879; although Brugmann applied the term "law" to his work, the use of his name was later applied by his regular collaboratorHermann Osthoff.[177][175]
(Sanskrit) When Proto-Indo-European*l precedes a dental consonant, the latter becomes aretroflex consonant and the*l is deleted. Examples includeजठर (jaṭhára, 'belly'), which is derived from Proto-Indo-European*ǵelt-, andकुठार (kuṭāra, 'ax'), derived from*kult-. The law is named after the Russian linguistFilipp Fortunatov.[178]
Also,Palatalgesetz. Proto-Indo-European*e palatalizes velar stops and becomesProto-Indo-Iranian*a.[179] The process is preceded by the delabialization of the labiovelar consonants.[180][181] Although several linguists have attempted to identify the originator of the law and several have declared a supposed originator, no consensus has been reached.[182] According to N. E. Collinge, it appears that six different linguists –Hermann Collitz,Ferdinand de Saussure,Johannes Schmidt,Esaias Tegnér Jr.,Vilhelm Thomsen, andKarl Verner – discovered the law "roughly simultaneously" and "in entire independence" from one another.[183]
ThePraeneste fibula is a goldenbrooch which contains an inscription considered to be the oldest surviving example ofOld Latin, dated to the first half of the 7th century BC.[184]
In a word with four or more syllables, if the second and third syllable are light, then the vowel of the second syllable is syncopated. Some examples do not appear to work unless the entire paradigm is considered.[185] Vowels which occur as a result ofanaptyxis do not undergo the process as they precede the law chronologically.[185] The law is named for the British classicistCharles Exon who first formulated the law in 1906, though its current form was conceived by the American linguistAndrew Sihler.[186]
When a short vowel is followed by an underlyingly voiced stop followed by a voiceless stop, it is lengthened. The process explains the differences between verbal forms, such asagō ('I drive') andcadō ('I fall'), and their respective derivatives, such asāctus ('made, done') andcāsus ('a fall').[187] Although the law was popularized byPaul Kiparsky,[187] it is named for the German classicistKarl Lachmann who wrote about the process in 1850.[188] It is unclear if the law applied to the whole Italic language family, but it applied at least to Latin.[187]
Also,Thurneysen's law. TheProto-Italic diphthong*ūy, derived from a Proto-Indo-European sequence of*u followed by a laryngeal and*i, is fronted to*īy. Examples includeOscanpiíhiúí ('pious [dative singular]') and Latinpius ('pious'), both from Proto-Italic*pīyo- derived from Proto-Indo-European*puh₂yo- 'pure'.[189] The root is shared with the Latinpūrus ('pure'), derived from Proto-Indo-European*puH-ro- ('clean'), which did not undergo this change.[190] Warren Cowgill argued that this law also occurred in Celtic in attempting to unify both Italic and Celtic into a double-jointedItalo-Celtic subfamily.[191]
The Proto-Italic diphthong*ow becomes*aw before a vowel.[189][192] This process precedes the Proto-Italic rounding of the diphthong*ew to*ow. Examples include Latincaueō ('I am weary of'), derived from Proto-Indo-European*kowh₁-eyo-, whence Ancient Greekκοέω (koéō, 'I am aware').[189] There appears to be some variation based on Romansocial register and the process may also affect the inverted sequence*wo. For example, the Latin termvacuus exists alongsidevocīuus in the work of some comedians, including inCasina, a play written byPlautus. The law is named for the Swiss linguistRudolf Thurneysen and the French classicistLouis Havet, who appear to have developed the concept largely independently of each other in 1884 and 1885, respectively.[193]
^Trask (2000) creditsCollinge (1985) with naming the process, but in the work cited, Collinge states that others have "fashionably called" it by this name prior to publication.[8]
^Initial /h/ in Ancient Greek is represented by the diacritical mark /◌̔/ appearing over the following vowel.[28] For further discussion, seeRough breathing.
^The verb is attested only in its present participle formἐρέπτόμενοι (eréptómenoi).[32] The term is attested in theOdyssey where the men pluck andeat the lotus.[33]
^Different authors prefer different symbols to represent the Proto-Germanic fricative sounds produced by Grimm's law.Collinge (1985) prefers the symbol*þ – derived from the runethorn, used in the alphabets of some Germanic languages to representdental fricatives[142] – in lieu of*θ, but the sound both symbols represent are equivalent.[143]Stiles (2018) uses thorn as well, but replaces*x and*xʷ with*χ and*χʷ.[144]Clackson (2007) also prefers the*h convention.[145] This glossary follows the convention used byTrask (2000).[146]
^Thenominative singular forms are𐌰𐌲𐌹𐍃 (agis) and𐍂𐌹𐌵𐌹𐍃 (riqis), respectively.[157]
Ringe 2006, p. 20: "It is possible, but not certain, that the rule was inhereited from PIE."
Collinge 1985, p. 8: "Some take the law to be of PIE date, even so. Kuryłowicz regularly opines so [...] Bartholomae himself suggested a wider domain [...] So does Szemerényi [...]"
Lubotsky 2018, p. 1879: "Bartholomae's Law, which is most probably of IE date [...]"
Beekes 2011, p. 130: "But the evidence in favour of a PIE date is not very convincing. Worth considering, however, is the explanation of suffix variants like*-tro-/*-dʰro- [where]*-dʰ- would have arisen from*-t- after an aspirate [...]".
Fortson 2010, p. 69: "Bartholomae's law [...] is reflected most clearly in Indo-Iranian; whether it was of PIE date is controversial."
For the Proto-Germanic form, seeRinge 2006, p. 92 andSkeat 1879, p. 711. Skeat uses older terminology; here, "Teuton" signifies what is now known as Proto-Germanic.
For the Latin and Old Latin etymologies, seeSihler 1995, p. 39 andSkeat 1879, p. 711.
_en.svg)
