Naturalization and Indian English

Read about our Indian English in the Oxford English Dictionary FAQs

“I had some lovely andouillette /ˌɒndwɪˈɛt/,” said Charlie.
“You mean /ɑ̃dujɛt/,” replied Sam.

Do you side with Charlie or Sam? English dictionaries have wrestled for over 100 years with how to represent the pronunciations of words loaned from other languages, evaluating the extent to which each word is naturalized. An English speaker who is familiar with French, and knows that andouillette is a French word, might use a French pronunciation while otherwise speaking English. This reflects a process called code-switching. On the other hand, some borrowed words, for example Italian pasta dishes, are commonly naturalized; a typical English speaker will either be unaware of the Italian pronunciations or, even if aware, will pronounce the words with differences that make them sound more ‘English’ (more ‘naturalized’) in terms of the vowels, consonants, stress, or syllabification.

OED maintains a policy of only offering primary pronunciations that are naturalized, at least in part. As an English dictionary, it would be beyond our scope to represent the pronunciations of every root language, just in case a speaker familiar with Norwegian gives a less-anglicized pronunciation of fjord. In general, we limit to the vowels and consonants of our British or U.S. English models, and where a word is only minimally naturalized by many speakers, we represent this by having as few changes as possible, choosing the English qualities that are closest to those of the root language.

Since 2015, however, this decision-making process has taken on another layer of complexity. The inclusion of more regional pronunciations where words are associated with other varieties of English – such as Scottish, East African, Philippine, Caribbean – has opened the door to some less naturalized pronunciations, but again each variety has its own ‘English’ system. A word borrowed from Kiswahili may be less naturalized in East African English than in British English, but the ‘East African English’ pronunciation is certainly not a Kiswahili pronunciation. Until now, this approach has been most tested with New Zealand English, where a great many speakers will adopt some qualities of Māori on encountering a Māori word. The OED pronunciation model accounted for this blurring of English and Māori by recognizing that some of the New Zealand English vowels may be deliberately lengthened or shortened, while sometimes also giving pronunciations which are much more anglicized.

But what of situations where a variety’s words have many different root languages, and where the variability of speakers’ familiarity with those root languages gives no straightforward geographically or demographically delineated variants? To that scenario add a continuum of English proficiency, from those with abilities specific to their employment through to those who are skilled as equally in English as in any other tongue, and the situation starts to resemble the complex situation OED’s pronunciation team encountered when we first explored adding Indian English pronunciations. Exploring the recent historical, educational, and geographic context with consultants Rohini Bakshi and Dr Meghana Kamble highlighted to us that the approach taken with OED’s other pronunciation models would not be appropriate. The key challenge was how best to represent a variety where any familiarity with a word’s root language will result in a fluent code-switch, using qualities outside their usual ‘English’ sound inventory, and where (unlike with e.g. New Zealand Māori pronunciations) there is simply no recognized ‘Indian English’ form?

After several years and publishing pronunciations in several more varieties of English, we returned to Indian English with a helpful new friend in tow – ELSI. While there are some commonalities to how English is pronounced by speakers across India when the words are not of a root language spoken in India, the same cannot be said for words which are. The given pronunciations must allow for a strongly Sanskrit-based yaksha, a Hindi-like diya, an Urdu-guided kalgi, Bengali-influenced almirah, Marathi-flavoured vada, Telugu (or Kannada) motivated pallavi, Gujarati-shaped bindaas, or Panjabi-coloured amrit. ELSI – the Extensions for Languages Spoken in India – is a set of symbols and principles which expands the model to account for differences in length of vowels or consonants, or occurrence of contrastive nasalization, rhoticity, or aspiration. An additional range of retroflex consonants (in which the tip of the tongue is curled back to articulate with the front of the hard palate) are possible, beyond the /ʈ, ɖ/ found for many speakers in tie and die. Some additional sounds are found towards the back of the mouth, such as the uvular plosive /q/ in some Urdu words. There are some symbols which have a broadened use, such as /aɪ/ and /aʊ/ reflecting not only the vowels of pride and mouth but also some of the similar but different Marathi vowels. /ʊ/ and /ə/ both capture a wider range of qualities than with their use in British or U.S. Englishes, as do the consonants /ʃ, tʃ, dʒ/ and /r/, and this will be evident on some of the audio. The transcription /nj/ represents a palatal nasal sound in words such as jnana as well as the sequence at the beginning of news.

But ELSI does not permit a free-for-all when it comes to pronunciation. The pronunciations given are constrained by our lexicographical principles in needing to be meaningful, faithful, and usable, but also representative of these words as spoken by individuals speaking English. They remain broad transcriptions, with narrow detail unspecified. They will still have stress placement indicated in manner expected of English, even if the root language is more syllable-timed or mora-timed. While ELSI allows for more blurred distinction between English and the root languages, the result is not ‘Hindi pronunciations’ or ‘Urdu pronunciations’ – there remains overarching English context and constraint in the representations, and a (perhaps minimal) extent of anglicization will usually be evident.

Consequently, the resulting pronunciation model is one that is rooted in both academic descriptions of Indian English (predominantly that of Gargesh, 2008, with notable influences of Sailaja, 2009, and Wiltshire, 2020), and also in descriptions of the Indian root languages, the input of consultant Divyanshi Shaktawat being vital in resolving many of the discrepancies. OED now uses the model, including ELSI, with an evolving transcription rubric to predict possible pronunciations of any OED entries indicated as rooted in or heavily associated with Indian English. Wherever possible, these are then checked against sources of natural language use, and in many cases then reviewed by our consultant, before being prepared for recording by genuine Indian English speakers.

ELSI’s introduction does not herald a change to OED’s approach with other varieties. The Singapore and Malaysian English pronunciations will not acquire a new arrangement for words of Malay or Mandarin Chinese, just as Canadian English will not have new variants of a francophone nature. Our research-informed impression is that there are different contexts when it comes to anglicization and naturalization of loanwords in other varieties of English, and our other varieties generally warrant an approach which is one step further from code-switching (with selected idiosyncrasies such as the specific allowances for New Zealand Māori terms or those borrowed from Afrikaans in South African English).

These Indian English pronunciations have been a long time coming, but whether you are curious as to why your favourite Indian dish is not pronounced the same by everyone, how names for yoga terms have evolved to sound different in different varieties, or simply where we commonly find the stresses in names of Hindu festivals, these new additions should address your queries.

The opinions and other information contained in the OED blog posts and comments do not necessarily reflect the opinions or positions of Oxford University Press.