Tamil Transcription

(The transliteration of a word in ISO 15919 is enclosed by « » and the IPA transcription by / /)

Tamil grammar treatises usually tend to reflect historical phonology rather than the contemporary phonology of the language. The earliest parts of a widely quoted grammer treatise Tolkāppiyam is usually dated approximately 2000 years ago. School textbooks (and other books) on Tamil usually follow the 13th century grammar treatise "Naṉṉūl". However, Tamil grammar treatises have not differed much in their treatment of phonemes with respect to Tolkāppiyam.

Because of this stagnation for several centuries, most textbooks actually define the phonology of the medieval language rather than the modern language. This textbook version of Tamil phonology does not accurately represent the modern phonology of the language. We'll attempt to describe how contemporary phonology differs from the formal prescriptive grammar treatises.

Modern Tamil Phonology

Let's begin with vowels. Vowels can be either be short (1 mora) or long (2 morae). The long and short distinctions in Tamil include change in quality of the vowel as well the quantity (except «o»). For example, «kaṇ<» is /kəɳ/ but «kāṇ» is /kɑːɳ/ not /kəːɳ/. Similarly, «cey» is /sɛj/ but «cēy» is /seːj/ not /sɛːj/. Traditional treatises generally do not note the change in quality.

Now, Tamil grammars also additionally recognize several special class of vowels. These are extra-long vowels (more than 2 morae) and extra-short vowels (less than 1 mora). The extra-long vowels are denoted by an additional vowel grapheme added to the sequence (in order to account for the added mora). But due to this orthographic convention, the extra-long vowels have now been reduced to spelling pronunciation where the sequence is pronounced with a vowel hiatus. While «perātavar» should be pronounced as /pɛɾɑːˑd̪əʋər/, it is often pronounced as /pɛɾɑː.əd̪əʋər/.

As for the extra short vowels, grammar treatises recognize extra-short «i» & extra-short «u». Extra-short «i» does not exist any more. And extra-short «u» is supposed to appear only with plosives and in specific positions of a word. But at present nearly every non-final «u» (and even intervocalic u in some positions) is reduced. Again, do note that though it is grammatically considered as a "reduction", phonetically only the quality of the vowel changes. The length is retained as such. For example, «pāṭṭu» is /pɑːʈʈʉ/, «kallu» is /kəllʉ/ and «taḻuviya» is /t̪əɻʉʋɪjə/.

Now, let us move on to the consonantal part.

Historically (at least during the time of Tolkāppiyam), «ṟ» was an alveolar stop. But it is (still!) traditionally classified as a stop along with «k», «c», «ṭ», «t» and «p». Since a few centuries earlier it has merged with «ɾ» in intervocalic positions. The historical alveolar stop pronunciation is partially retained only in clusters «ṟṟ» and «ṉṟ» as /t̺t̺ʳ/ and /nd̺ʳ/ respectively. Similarly, «n» & «ṉ» have merged and are no longer distinguished. However, these historically phonemes have been retained in the orthography. Thus, word pairs that involve differentiation of «ṟ»/«r» and «n»/«ṉ» are purely orthographical and homophones. Both «maṟam» & «maram» are pronounced as /maṟam/.

Also «c» (traditionally listed as a plosive) is not longer an affricate /tʃ/ in intervocalic positions where it is realized as a dental fricative /s/. The affricate is sometimes retained in the word-initial position (in free variation with /s/), during gemination and in vowel-less situations.

Originally, «ḵ» was probably a /x/. But the sound I was taught at school fell something between /g/ and /ɣ/. It appears that in several regions, the /x/ sound is still prevalent. So, I have retained /x/ as the pronunciation for «ḵ».

Lately, the difference between «ṉ» /n/ - «ṇ» /ɳ/ & «l» /l/ - «ḷ» /ɭ/ is also being lost, with the latter phonemes getting merged with the former. This is quite evident by the informal names (two-looped n vs three-looped n and small l vs big l) by which they are referred to be distinguished when "spelling out" words. In many regions, «ḻ» /ɻ/ is getting merged with «ḷ» /ɭ/. This has resulted in hypercorrecting «ḷ» with «ḻ» and vice-versa in written Tamil.

But as such, the transcription given in the converter is fairly conservative. It does differentate between all the sounds described in the previous paragraph.

Tamil originally lacked the fricatives /s/ (at word-final position), /ʂ/ & /ɦ/. But these have now been adapted to the language and are now a part of the language's modern phonology. Similarly, /dʒ/ also gained its own grapheme at all non-post-nasal positions (because «l» is already multi-valent by representing /s/ (inter-vocalic), /dʒ/ (post-nasal) and /tʃ/ (gemination/post-consonantal) ). Similarly, «śrī» has also been borrowed and adapted to the language. These are a few linguistic puritanists & fanatists who detest and shun these "foreign" characters, but they are non-mainstream fringe elements who don't blatantly disregard the modern phonology of the language.

Apart from the phones of the literary Tamil, spoken Tamil also has nasalized vowels. In written colloquial language, the nasalization is orthographically represented using a nasal consonant. This is because words with word-final nasal consonants usually end up nasalizing the preceding vowel. Consider the following sentence:

avaṉ vantāṉ
əʋən ʋənd̪ɑːn (Literary Tamil)
əʋə̃ ʋənd̪ɑ̃ː (Spoken Tamil)

Narrow Transcription

It is to be noted that the narrow transcription of the tool is based on the recitation of classical Tamil verses by Punal K Murugaiyan in [1]. It is not based on prose. It identifies as much as 100 different phones in the language. Hence, the transcription derived from his work is quite detailed and meticulous.

Broad Transcription

I have tried to cut down the unnecessary details and retain only the basic features that are absolutely necessary for the language.

The narrow transcription uses several fricatives such as /ð/, /β/, /x/ and /ɣ/. I am not quite sure I (and few others I have personally discussed) agree with the identification of all these phones. I myself find it hard producing /ð/, /ɣ/ and other fricatives when learning other languages. But even Clayton (1934) recognizes some fricatives in the language. I suppose its a dialectal thing. To simplify things, I have replaced them with the equivalent voiced plosives. Also, the narrow transcription analyzes /ə/ being allophonic with a back vowel /ʌ/. I have fused them to /ə/ for the broad transcription.

To be fair, the narrow transcription is based on a prosodical recitation which may have affected the quality of vowels and consonants. I have only retained the obvious features of modern Tamil such palatalization of initial-vowel «e», labialization of initial-vowel «u», vocalization of inter-vocalic plosives and several other features discussed earlier and hence reducing the amount of phones represented.

References

1. Murugaiyan, K (2010). Paṉṉiru Tirumuṟai Olipeyarppu. Chennai, India: Kāntaḷakam Publishers

2. Keane, Elinor (2004). "Tamil". Journal of the International Phonetic Association 34 (1): 111–116.

3. Clayton, A.C (1934). A progressive grammar of common Tamil. India: Christian literary society