Published online by Cambridge University Press: 22 April 2026
The classical Prakrit vowel system has been chosen as a starting point to study the history of the vowel system of Gujarati. The evidence for the intervening stages is supplied by documents, and it is proposed to show how many phonemic systems have to be postulated for these intervening stages. The changes in the phonemic systems have, in turn, been followed by significant changes in the grammatical systems, thus providing us with more reliable data on the stages in the history of the Gujarati language. Statements about changes in the consonants have been excluded mainly because they are relatively few, and do not play a significant role in the changes in the grammatical system. Where the changes in consonant and vowel systems are interdependent, statements about consonant changes have been made.
1 Materials for the general history of Indo-Aryan are collected from R. Pischel, Grammatik der Prakrit-Sprachen, 1900; J. Bloch, ~U indo-aryen du veda aux temps modernes, 1934; S. K. Chatterji, Origin and development of the Bengali language, 1926; R. L. Turner, Gujarati Phonology, Journal of the Royal Asiatic Society 1921.329–65, 505–44; R. L. Turner, A comparative and etymological dictionary of the Nepali language, 1931. Data on Old and Middle Gujarati are collected from L. P. Tessitori, Notes on the grammar of old western Rajasthani with special reference to Apabhramša and Gujarati and Marwari, Indian antiquary 1914–16; T. N. Dave, A study of the Gujarati language in the 16th century (V.S.) with special reference to the ms. Bālāvabodha to Upadešamālā, 1935; E. Bender, The Nalarāya Davadantī Carita, a work in old Gujarati, 1951; K. H. Dhruva, Kādambarī of Bhālaṇa2, 1935; P. B. Pandit, A study of the Gujarati language in the 14th century with selections from Ṣāvašyaka vṛtti of Taruṇaprabha critically edited (unpublished dissertation, London University), 1949.
The chronology in this paper should be interpreted as relative, i.e. Old Gujarati (OG) refers to data of the 12th-15th centuries, Middle Gujarati (MG) refers to data of the 15th-18th centuries, and the New Gujarati (NG) to the modern colloquial Gujarati; but it is quite probable that the data represent an earlier and archaic form of the language, thus pushing back the OG period by three or four centuries and shifting the span of each period on either side (Proto-Gujarati on the one side and MG on the other). The Gujarati forms are in phonemic transcription.
Abbreviations: Sk. Sanskrit, Pa. Pali, Pkt. Prakrit, OIA Old Indo-Aryan, representing Vedic and Classical Sanskrit, MIA Middle Indo-Aryan, representing Pali and Prakrits, NIA New Indo-Aryan, representing modern Indo-Aryan languages.
I am deeply indebted to Sir Ralph Turner and Gordon Fairbanks for making valuable suggestions.
2 It is possible to interpret length of the stop consonants as subphonemic in MIA, i.e. initially they occur as single, intervocalically as geminated; for the suggestion see N. S. Troubetzkoy, Principes de phonologie 305 (1949). If this suggestion is followed, consonant length would be phonemic only in those nasals and spirants (m, ṇ, s, l, v) which can occur single or geminated intervocalically. r and y do not occur geminated; n and ṇ do not contrast, since the former occurs initially and the latter intervocalically; but nn and nṇ occur intervocalically and are therefore contrastive. Length will be significant in the case of the retroflex voiced stop ḍ because initially it occurs as single but intervocalically single and geminated.
If the MIA consonant system is reinterpreted through the reconstruction of the NIA languages, it would be possible to refine the above interpretation. Then, possibly, intervocalic -l-, -ḍ- may be assigned to different phonemes ḷ and ṛ; intervocalic geminates -ll-, -ḍḍ- may be considered allophones of l and ḍ Initial n- and intervocalic -nn- and -ṇṇ- may be assigned to the same phoneme n. Similar reinterpretation may be possible for m and s also, so that intervocalic -m-, -ṇ-, -ḍ-, -l-, and -s- would be assigned respectively to the phonemes ṽ, ṇ, ḍ, ḷ, h; initial m-, n-, ḍ-, l-, s- or intervocalic geminates -mm-, -nn- or -ṇṇ-, -ḍḍ-, -ll-, ss- would be assigned respectively to the phonemes m, n, ḍ, l, s. This interpretation is very tentative and could only be profitably discussed if various stages of the split of the Indo-Aryan dialects are specified.
3 Bloch, L'indo-aryen 41.
4 For assigning MIA e, o to the set of long vowels see A. M. Ghatage, Phonetics and phonemics in linguistics, Indian linguistics, Turner jubilee volume 182–87 (1958).
5 Final nasalisation develops as a special case in some terminational elements. Thus, -ṇ-> -m- > ~ : MIA inst. sg. -ēa > Late MIA -ẽ > OG -i; MIA nom. acc. n. pl. -āại > Late MIA -ai > OG -a; MIA present indicative 1st pers. sg. and pl. -āmi, -āmo > Late MIA -aũ > OG -aũ (dialectally -āmo > -a). Loss of final nasalisation of -ẽ > -i is early, because inst. sg. -i is regular in OG. Loss of nasalization of -aĩ (inst. sg. of extended nouns, e.g. ghoṭakena > ghoṣaeṇa > ghooḍaẽ > ghoạaĩ, and 3rd pers. pl. of present indicative e.g. karaĩ) is later; it takes place in the MG period. OG -au > MG -o but OG -aũ > MG -ũ; this indicates that the nasalization of -aũ was distinctive in MG. Final nasalization of -ũ is lost in some dialects of NG after the MG -aũ > -ũ has worked out.
6 Pischel, Grammatik 535; Chatterji, ODBL 653.
7 Dhruva, Kādambarī 203; Dave, A study 46–7.
8 After vowel length ceases to be distinctive, there is considerable confusion in MG mss. in marking short and long i and u; this confusion itself is symptomatic of the change. For a more detailed discussion of this change see P. B. Pandit, e and o in Gujarati, Indian linguistics 15.15–44 (1955).
9 Retention of OG -i- in open syllable cannot be explained. For some special changes in the terminations of the future, see R. L. Turner, Future stem in Ašokan, Bulletin of the School of Oriental Studies 6.533 (1931).
10 On the model of nouns in -īu m. and -īũ n. in OG (§5), new derivative suffixes -īũ m. and īũ n. are formed, e.g. maraṇa ‘death’, maraṇīu m. ‘ready to die’; patãga ‘sun’, patãgīũ n. ‘butterfly’ (see Dave, A study 6) ; these suffixes are inherited in MG with the replacement of -u by -o, and the MG derivative suffixes develop as -io m., -iũ n. NG inherits these as productive derivative suffixes, e.g. khānṇ ‘mine’, khāį m. ‘miner’; dānṇo ‘day’, dāį m. ‘working on daily wages’; baḷad ‘bull’, baḷadio m. “bull'; maphat ‘free’, maphatio m. ‘that which is free of charge’; thaḍ ‘trunk of a tree’, thaḍiũ n. ‘source, trunk of a tree’; pāṭ ‘bench’, pāṭiũ ‘wooden plank’; məhũ ‘face’, məhiũ n. ‘lid of an oil lamp’.
11 The source of this -o is not clear; like the n. sg. -ũ which was generalized as a plural suffix in Saurashtra, m. sg. -o became generalized as a plural suffix in NG dialects of the central and northern Gujarat. (In OG, there are some examples of m. sg. -u used for voc. pl., e.g. heraka ‘spy’, herakau ‘O spies’.)