Consonants

There is not a great deal of categorical allophonic variation in Dari, though a number of consonant deletions and vowel coalescences occur at morpheme boundaries, as illustrated in the transcribed text below. The stops of Dari are unaspirated or lightly aspirated in all positions, except in careful speech. Voiced stops and affricates are optionally devoiced; in the recordings of isolated words, compare the devoiced allophones in [bʌ] ‘wind (n)’, [xu] ‘good’, [tʌ] ‘crown’, [neˈʒʌ] ‘ethnicity’ with the voiced voiced allophones in [bʊɾd] ‘he carried’, [sud] ‘interest’, [ob] ‘wood’, and [sɐg] ‘dog’. The phoneme /ɾ/ may be trilled to [r] in emphatic speech, as occurs in the words [ˈɐɡɐr] ‘if’, [ˈrɐfɐ] ‘satisfaction’, [ɣʌr] ‘cave, hole’, [rʌ] ‘road’ in the word list above; compare the flap [ɾ] which occurs in the transcribed text. The phoneme /h/ is frequently lost in everyday speech, though less often in formal contexts. Variation in the (formally produced) word list is illustrated above, with the word-final [h] preserved in [gʌh] ‘time’ and [koh] ‘mountain’, but dropped in [rʌ] ‘road’.

In relation to Iranian Persian, it can be noted that Dari has distinct /q/ and /ɣ/, which are merged in Iranian Persian but similarly distinct in Tajik Persian (KiefferReference Kieffer1985; Windfuhr & PerryReference Windfuhr, Perry and Windfuhr2009: 427). Dari always has the labiovelar approximant [w] instead of the Iranian or Tajik Persian [v]; [v] is never heard, even in formal speech.

Vowels

Dari has a larger inventory of vowels than Iranian Persian (see the six-vowel system indicated by Majidi & TernesReference Majidi and Ternes1999, and also more ‘traditional’ grammars like ThackstonReference Thackston1993 and MaceReference Mace2003). The additional contrasts – reported previously by Henderson (Reference Henderson1972) and Glassman (Reference Glassman2000) – are between the front round vowels [e] and [ɛ], and the back round vowels [o] and [ʊ]. These contrasts are similar to vowel contrasts present in Middle and Early Modern Persian (WindfuhrReference Windfuhr, Kaye and Daniels1997) though a more precise statement of the relation awaits further etymological study.

The distinction between the back rounded vowels is illustrated by the words [xu] ‘good’, [xʊd] ‘self’, and [ob] ‘wood’. The phonemes /e/ and /ɛ/ are distinguished by the similar pair [bel] ‘shovel’ and [gɛl] ‘mud’.Footnote¹ Several examples of each of these vowels are given above. As discussed below, the difference between [e] and [ɛ], and [o] and [ʊ], is reinforced with a length contrast.

The vowels [], [], [], and [] are analyzed here as diphthongs, but could alternately be analyzed as a monophthong–glide sequence. There is evidence in favor of both analyses. Although the maximal syllable template of Dari is CVCC, diphthongs do not occur in syllables with two coda consonants. This could be taken as an indication that one consonant position is being taken up by a glide. On the other hand, syllables with two consonants are not so frequent as to rule out the possibility of an accidental gap in the lexicon. Conversely, glides do not occur freely in coda position: a monophthongal analysis would then have to stipulate a list of permissible vowel–consonant sequences (i.e. [ʌj], [ɐj], [ɐw], [uj]), which in a sense merely recapitulates the diphthongal analysis. Therefore, though the phonological facts do not force one analysis over another, the diphthongal analysis is adopted here.

In the diagram above, the monophthongs are shown in their approximate locations in F2-F1 space. Two measurements are given for the diphthongs, corresponding to steady states towards the beginning and ends of the diphthongs.

Traditional analyses of Persian have distinguished between long vowels (as transcribed here, [i ʌu]) and short vowels ([ɛ ɐo]). Dari does not make a length distinction independent of vowel quality, though there are differences in the lengths of vowels. For purposes of this Illustration, duration measurements were made for the monophthongal vowels of the words produced in isolation. The results are shown inTable 1.

Table 1 Vowel duration measurements of monophthongal vowels produced in isolation in the present paper.

An ANOVA shows a general effect of vowel identity on vowel duration:F(7,135) = 3.78,p < .001. The difference in length between traditional long vowels of [i ʌu] and the short vowels [ɛ ɐo] is also significant:F(1,141) = 19.20 (p < .001). Finally, the difference is again significant if the Dari long vowel [e] and short vowel [ʊ] are included in the test:F(1,131) = 13.35,p < .001. In summary, then, there are length distinctions in Dari, which are redundant with vowel quality. These correspond to the traditional analysis of Persian short and long vowels.

Syllable structure follows a maximal syllable template of CVCC. The onset is optional; the obligatory epenthetic glottal stop reported by Windfuhr & Perry (Reference Windfuhr, Perry and Windfuhr2009) is not found in Dari (compare the words [ɐˈrɐ] ‘saw (n)’ and [ɐˈdʒɐl] ‘fate’). Windfuhr & Perry (Reference Windfuhr, Perry and Windfuhr2009) note that Arabic loanwords are a source of many complex codas. On nouns, stress is typically syllable-final, including any suffixes and clitics. On verbs, stress falls on the first prefix, or on the last syllable of the root if there is no prefix.

Transcription of recorded passage

The story ‘The North Wind and the Sun’ was adapted by the language consultant from Majidi & Ternes's (Reference Majidi and Ternes1999) Farsi version, to reflect Afghan word choice and usage.