Pronunciation model: Indian English
View the key for Indian English here.
There is no single ‘Indian English’ (IndE) variety and there are few extensive descriptions of IndE phonology. As a second (or third or fourth) language in India, English is spoken in a range of different ways, influenced by regional variation and interference from one or more of the 200+ native Indo-Aryan, Dravidian, Austro-Asiatic and Tibeto-Burman languages, as well as sociological differences (Gargesh 2008). English is spoken by the most highly educated but also by those with little formal education, and at all points along the continuum between. While some individuals speak English with the same proficiency as with their first language, more are in professions requiring a suitable general proficiency and potentially even more have English skills specific to their domains of employment. The description below is rooted in the description of Indian English by Gargesh (2008) but with reference to Wells (1982) and others.
The past 15 years have witnessed an increasing number of descriptions of IndE, often with subtle differences of influence from different speaker populations. OED’s model is predominantly based on the description of Gargesh (2008), from the same series of descriptions as provided the basis of many other OED World English pronunciation models. While consonants are largely consistent across descriptions, for many of the vowels Gargesh notes substantial variation and this is reflected in differences across descriptions. The selected vowels are steered by Gargesh’s most common vowel as a starting point, with influence from Sailaja (2009) and Wiltshire (2020), and Pandey (2015) as a further source. There are notable dissimilarities between the sources and compromises have been sought which are consistent with OED’s general approach of reflective fairly mesolectal pronunciations. This being a more complex task than reflecting the more homogenous phonology of acrolectal speakers, as there are many subtle first-language effects on the nature of an individual’s IndE phonology (Sirsa & Redford 2013), with Indian English consultant input being invaluable in resolving these discrepancies.
Wiltshire explains how a systematic approach to Indian English (termed ‘General Indian English’, GIE) was introduced in 1972 with view to educational use in India. The fundamentals of GIE in terms of consonants remain largely current. However, the OED Indian English model is unique amongst OED’s World English pronunciation models in that the inventory and patterns described below apply to relatively few OED entries. The base model describes the features of pronunciation commonly found amongst Indian speakers of English assuming nothing of the root language(s) of any given word. However, where a word has arrived in English mostly via speakers of another language spoken in India (which applies to many of the ‘Indian English’-tagged entries in OED), Indian speakers familiar with that language are likely to use qualities outside of the ‘Indian English’ inventory. Those supplementary qualities are explained within an ‘Extensions for Languages Spoken in India’ grid (ELSI) at the end of the model. ELSI is based on languages (other than English) which according to the 2011 Indian census accounted for more than 4% or 50 million first-language speakers: Hindi, Bengali, Marathi, Telugu, Tamil, Gujarati, and Urdu. A supplement to ELSI assimilates scholarly impressions of the likely original pronunciation of the Sanskrit, recognizing that it is likely to be influenced by other languages en route (Zieba et al, 2002).
Indian English Vowels
|TRAP||æ||PALM||ɑː||CURE||oː(r) + ɔː(r)||RABBIT||ɪ|
|LOT||ɔ||NURSE||ɜː + ər||PRIDE||aɪ||BEAUTIFUL||ʊ|
*Except before another vowel-initial syllable, where this becomes /aːʋ/ as in power.
There is variability in whether length contrast is indicated in accounts of Indian English, with variation in use of length for contrast reported both on a group and individual basis, but Wiltshire (2020) concludes there is ‘widespread uniformity’ to the use of length for contrast and this feature of the description by Gargesh (2008) is represented in the OED model. It should be noted though that the OED model for Indian English vowels is a modified simplification of Gargesh’s description, with a wide range of variation reported for several of the vowel distinctions and readers are directed to Wiltshire (2020) for further detail. The expanded description below does not attempt to account for all such variation, simply the most salient qualities within OED’s key sources.
Almost all of the ‘short’ vowels have common variants in different parts of India, usually a slight raising or lowering of the tongue position from the sounds represented above (and this is apparent in e.g. how Sailaja 2009 uses /ɛ/ for dress and draws a /ɛ/-/e/ contrast, while other sources have a higher /e/ quality for this vowel). There is an /aː/ variant of lot, which also occurs as a variant quality of cloth, thought and north, but is not represented here. A potential lack of lot-thought contrast is widely reported in IndE literature despite length contrasts indicated elsewhere. The bath and start vowels seem to be long, unrounded, and open but often central if not front-central (more [aː]) – in the IndE transcriptions /ɑː/ should be interpreted as encapsulating a range of long unrounded open qualities. The nurse vowel varies widely, the main quality being /ʌ/, /ə/, or even /aː/ (Sailaya 2009 claims that the /ɜː/ is only in non-rhotic Standard Indian English, but /ar/ or /ər/ in rhotic accents; for this reason nurse is split rather than having bracketed rhoticity). force can be realized as /oːr/ or /ɔːr/ so may be identical to the quality of north, but the distinction is generally maintained (see Domange 2015). voice can be /ɔɪ/, /oe/, or /oɪ/, while the second elements of pride and mouth are not as weak as in British English. square can vary across qualities including /eː/ and /æ/. Sailaja suggests a diphthongal /eə/ for square and /ʊə/ for cure, but notes a tendency for ‘non-standard’ Indian English is to shift vowels other than mouth and pride to long vowels, with beer as /biːr/, fair becoming /feːr/, and poor /puːr/. The OED model reflects an apparent grey area between these fully-diphthongal and fully-monophthongal positions as described by Gargesh, though deviates to Sailaja’s approach for traditionally near words where a /r/ necessarily follows (e.g. hero as /ˈhiːroː/). happy can be higher and longer.
Note that IndE will often have strong vowels in syllables which would be weak in British English, ‘with “full” in the sense of both quality and duration” (Wiltshire 2020, p.49). Cricket has /-ɛt/, consider /kɔn-/, different /ˈdɪfrɛnt/ (Wells 1982, symbols modified for consistency), though duration of close front vowels is only represented as long where clearly so (hence enabled with /i/). However, schwa (/ə/) does have a significant role in Gargesh’s description and transcriptions, as well as in the recordings by OED’s IndE speakers. The representation of the added vowel as /
ɪ/ is also a vast simplification, as many such instances are far more likely to be /ɪ/ than /ə/, while in others (such as use at morpheme boundaries, e.g. with plurals) either the reverse may be true or the frequency of occurrence more even (though kit is used where there is clear evidence that a central quality would not be found). A number of vowel contrasts are ‘not clear-cut’ in varieties of Indian English, Gargesh citing /ʌ-ə/, /ɒ-ɔ/, and /ɛ-æ/ (the OED model nonetheless recognizes the third of these, while the second is condensed into an /ɔ/ quality but one sometimes distinguished by length, and the first simply to /ə/).
It is common for vowels follow orthographic cues, with pleasure having /eː/, curvature being /kə(r)ˈʋeːtʃə(r)/ and necessary being /nɛˈsɛsəri/.
Some descriptions of IndE suggest that it is generally rhotic, with a postvocalic /r/ which is often trilled (even in consonant clusters) rather than an approximant, though descriptions vary. Tapped /r/ sounds are feasible alternatives. (Note that Gargesh does not show rhoticity as an inherent part of the vowel qualities, but for consistency with the other OED models this feature is added here.) Sailaja (2009) describes Standard Indian English pronunciation as non-rhotic, with rhoticity a marker of ‘most non-standard varieties’. Wiltshire (2020) suggests that early models of Indian English accounted for both rhotic and non-rhotic varieties, and while some descriptions turn to linguistic factors as determining rhoticity, others refer to variation in regional and social terms. For simplicity and practical usability of OED’s model, rhoticity is bracketed as with other World English models where rhotic and non-rhotic forms are commonplace (note the exception of the nurse vowel, where a minor quality distinction is also made).
Indian English Consonants
|b||buy||ʃ||shoe||ʋ||west, vest||ŋ||sing (+/g/)|
|ɖ||die||θ (= [t̪])||thigh||j||year|
|ɡ||guy||ð (= [d̪])||though||l||low|
The consonant symbols chosen are largely consistent with those of GIE as summarised by Wiltshire (2020), though the OED system arises more directly from the description of Gargesh. For example, words with /t/ or /d/ in British or US Englishes are commonly realized with retroflex sounds /ʈ/ and /ɖ/ in IndE, with the tongue tip curled up and touching the top of the mouth further back. The sounds /ð/ and /θ/ are rarely heard as their BrE forms, sounding like a dentalized /d/ and dentalized and aspirated /t/ instead. However, for most speakers, a contrast is retained between tin and thin, den and then and for this reason the symbols /θ/ and /ð/ are used in OED’s IndE model in the same way that they are used in Irish English, highlighting the salient dental place of articulation even though plosive rather than fricative manner. The use of /θ/ and /ð/ also facilitates the contrasts required for some of the root languages between these dentalized alveolar plosives and retroflex plosives. In words with a non-Indian root, the likely presence or absence of aspiration on /p ʈ k b ɖ ɡ/ is not represented.
British English /ʒ/ is commonly realized in IndE as /dʒ/, /z/, or /j/, or even /ʃ/ (although Sailaja states that it is ‘definitely’ present as /ʒ/ in Standard Indian English Pronunciation). For consistency, OED transcribes these as /dʒ/ (with /ʒ/ only where it exists in a word of Urdu origin). Amongst speakers who pronounce /ʃ/ and/or /ʒ/, they are articulated differently from the forms in British or American English, the tongue tip being curled down (which will be heard on the recordings).
British English /f/ is often aspirated /p/ in IndE (likely dependent on the speaker’s first language), leading to confusion between fit and pit to the British ear (see the note below on the usual realization of IndE /p/). Other speakers may use /f/ or a bilabial fricative. In any event, a distinction is still retained between /f/ and /p/, hence the /f/ symbol remains the symbol of choice for OED. A major feature of IndE is the interchangeable nature of /w/ and /v/. Wells (1982) discusses various realizations of this merger, often a labiodental approximant, while Sailaya (2009) asserts that the distinction is maintained in Standard Indian English pronunciation but not non-standardly and research cited by Wiltshire (2020) supports the view that a complete lack of /v/-/w/ contrast is inaccurate (the contrast varies). Nevertheless, this is a salient feature of much Indian English pronunciation to speakers of other varieties and OED reflects this in a simplified way, using the IPA labiodental approximant symbol /ʋ/ and having no /v/ or /w/ symbols. However, it is also possible that some words with /v/ in acrolectal or British English accents are pronounced as IndE /f/ where the spelling indicates it, such as in of.
In terms of morphophonology, both Gargesh and Sailaja note the -ed ending is commonly /d/ irrespective of the preceding consonant, so missed /mɪsɖ/ and mist /mɪsʈ/ are not homophones as in British or American English. Following Sailaja, plurals follow the opposite pattern of being /s/ even in words such as dogs (/dɔɡs/) and kisses (/ˈkɪs
ɪs/). Unlike Sailaja, the OED model does not represent full vowels in separate-syllable -ed or -es morphemes (instead following Gargesh’s description of more central qualities), but other suffixes follow Sailaja in being represented with full vowels (such as -ness, -less, -est having the dress vowel, as does the first syllable of -metry).
The final consonant in British English sing is always pronounced with a final /ɡ/ in IndE, and is similarly so for all such words. There appears to be a tendency towards /n/ in some situations where other varieties would assimilate, as concrete is transcribed by Gargesh with /n/ rather than /ŋ/.
Other variety-specific phenomena include /ɡh-/ in words such as ghost, /ʋh-/ in words such as where, and a stopping of /tʃ/ and /dʒ/ when followed by another (e.g. watch-chain as /ʋɔttʃeːn/). Other geminates (doubled consonants) are if anything more common than in British, occurring in words such as cannot and fully, within and across morpheme boundaries. As in Welsh English, OED’s Indian English model acknowledges the presence of lengthened consonants, as in sattu /ˈsəθːuː/, but limited to where this is a feature of the root language of a word and that root is commonly spoken as a first language in India (e.g. Hindi, Urdu).
Syllabic consonants are also not a feature of IndE, a non-optional schwa being present in all cases. Linking /r/ is found, but never intrusive /r/.
Several highly distinctive but more detailed features of Indian English will be heard on the recordings. For example, /p/, /t/, and /k/ are not aspirated in syllable-initial position as in British English, so to a British ear could sound more similar to /b/, /d/, and /ɡ/ respectively. Where /n/ precedes /ʈ/ or /ɖ/ it may be made retroflex (e.g. sand), and by some speakers from Southern India made retroflex in all contexts, but this is not represented in the transcriptions. /l/ sounds are always ‘clear’ as in British light and never ‘dark’ as in British ball (though may be retroflex for some speakers).
Much of the regional variation cannot be captured in this model, since the OED IndE pronunciations are intended to be indicative rather than comprehensive. Features such as Bengal’s use of /ʃ/ for /s/ or /dʒ/ for /z/ are not reflected here, nor are the low-tone or glide substitutions for /h/ found in various parts of India, nor South India’s ‘euphonic’ /j/ and /w/ at the start of words such as every and only. Similarly, region-specific processes such as vowel insertion to avoid initial consonant clusters, or consonant deletion to avoid final clusters, or varying distributions of glides, are not shown.
IndE is more syllable-timed than stress-timed, with each syllable having similar prominence (but unlike e.g. Philippine English, IndE is not represented with extended use of secondary stress). Function words such as to and her are not likely to be reduced to a ‘weak form’ as in British English, nor are contracted forms common (e.g. The cat’s running >> The cat is running). Within individual words, accentuation is influenced by native languages (for around 40% of the population, this will be Hindi). Gargesh explains the structure as it applies in Hindi, which broadly transfers to IndE. It distinguishes between light ((C)V), heavy ((C)V: / VC) and extra-heavy ((C)V:C / (C)VCC) syllable weights. OED first assigns primary stress according to natural-use evidence, then (in absence of such evidence) according to the key features of Gargesh’s Hindi description: bisyllabic words have primary accent on the first syllable unless the second syllable is extra-heavy, and trisyllabic words have primary accent on the middle syllable if heavy but first syllable if not. From Gargesh’s examples, it appears that the weight distinctions can be more complex than this would suggest, however, and Wells (1982) suggests that many stress placements may be idiosyncratic. With longer entries, compounds usually retain the left-hand stress, while there is often no word-class accentuation distinction (e.g. permit both /ˈpɜːmɪt, ˈpərmɪt/, protest both /proːˈtest/). Further detail on the complexities of IndE stress can be found in Sailaja (2009), Pandey (2015), and Wiltshire (2020).
Extensions for Languages Spoken in India (ELSI)
Vowels are largely the monophthong vowel qualities of the broader Indian English model, with the following addition and adaptations:
- Length: uniquely for OED pronunciation models, variations of length by adding or removing /ː/ marks are freely permitted based on the salience of such qualities in the root language(s)
- Nasalization (/~/): whereas nasalized vowels are limited in other varieties, any vowel of any length may be nasalized if contrastively nasalized in root language(s) (i.e. not allophonic)
- Rhoticity: vowels other than those explicitly noted in the main Indian English model can be realized with post-vocalic /r/.
|ʂ||voiceless retroflex fricative, particularly found in words of Marathi, Telugu, or Sanskrit origin|
|ɳ||voiced retroflex nasal (also used for English words where /n/ precedes /ʈ, ɖ/)|
|ɽ||voiced retroflex flap, also used for voiced retroflex approximant|
|ɭ||voiced retroflex lateral approximant|
|q||voiceless uvular plosive, only for words of Urdu origin|
|x||voiceless velar fricative|
|ɣ||voiced velar fricative, only for words of Urdu or Punjabi origin|
|ʒ||as in British English beige, only for words of Urdu origin|
Broadened symbol uses in ELSI
|h||used for [h] and its voiced equivalent, but also to indicate salient voiced or voiceless aspiration of a preceding consonant (crucial phonemic contrast in several languages of India)*|
|r||used for trill [r], tap [ɾ] or approximant [ɹ] qualities (the latter in words of Bengali origin); in words derived directly from Sanskrit, this symbol is used for the sound likely pronounced as either [ɽ] or [ɹ]|
|tʃ||used for palato-alveolar [tʃ] and alveolo-palatal [tɕ]|
|dʒ||used for palato-alveolar [dʒ] and alveolo-palatal [dʑ]|
|ʃ||includes potential [ɕ]|
|nj||this sequence takes the place of the voiced palatal nasal|
|ː||consonants can be followed by length marks where geminated|
|~||nasalization diacritic can be applied to consonants depending on root|
|ə||represents central vowels from mid down to open-mid or even verging on open|
|aɪ, aʊ||the Indian English PRIDE and MOUTH vowels are also used to represent the Marathi closing diphthongs (which some scholars suggest may be vowel sequences rather than diphthongal)|
*Note aspiration is only indicated in the ELSI words, and not in every case. In words of British/US origin adopted into India, voiceless stops may (a) have aspiration patterning more variably (‘unsystematically’); (b) generally be articulated without aspiration; or (c) follow specific patterns (such as the voiceless dental plosive being commonly aspirated but the voiced dental plosive not; however, in ELSI words from Indian languages with a voiced dental plosive, such a /ðh/ transcription is feasible in OED). The ‘prevoicing’ of voiced stops is not addressed here, though readers are directed to Wiltshire (2020) for a useful explanation.
Key sources for Indian English
Cowie, C. and Elliott, Z. 2019. Lexical set membership in contact varieties of English: the re-organisation of BATH and TRAP in Indian English. Poster presentation to UKLVC12, 4th September 2019.
Dhongde, R.V. and Wali, K. 2009. Marathi. Amsterdam: John Benjamins.
Domange, R. 2015. A language contact perspective on Indian English phonology. World Englishes, 34(4), 533–556.
Gargesh, R. 2008. Indian English. In: R. Mesthrie, ed. Varieties of English 4: Africa, South and Southeast Asia. Berlin: Mouton de Gruyter, pp.231-243.
Grolman, M.B., Biktagirova, Z.A. and Kasimov, O. H. 2021. Phonetic peculiarities of the English language in India. International Journal of Society, Culture and Language, 9(1), pp.102–110.
Maxwell, O. and Fletcher, J. 2009. Acoustic and durational properties of Indian English vowels. World Englishes, 28(1), pp.52–69.
Pandey, P. 2015. Indian English pronunciation. In: M. Reed & J.M. Levis, eds. The Handbook of English Pronunciation. Hoboken, NJ: John Wiley and Sons, pp.301–319.
Sailaja, P. 2009. Indian English. Edinburgh: Edinburgh University Press.
Sailaja, P. 2012. Indian English: Features and Sociolinguistic Aspects. Linguistics and Language Compass, 6(6), pp.359–370.
Sirsa, H. and Redford, M.A. 2013. The effects of native language on Indian English sounds and timing patterns. Journal of Phonetics, 41(6), pp.393–406.
Wiltshire, C.R. 2020. Uniformity and variability in the Indian English accent. Cambridge: Cambridge University Press.
Zieba, M., Stiehl, U., Mayrhofer, M. and others. 2002. The original pronunciation of Sanskrit [Online document]. Retrieved 21st July 2021 from http://www.sanskritweb.net/deutsch/ipa_sans.pdf