Talk:Bengali phonology

Aspirated and murmured
According to a 1921 article by Chatterji, there are no voiceless or voiced aspirates. This article seems well-written if unsourced. If there is no sourcing soon, I will remove all references to aspiration here and at Romanization of Bengali. — Æµ§œš¹  [aɪm ˈfɻɛ̃ⁿdˡi] 17:10, 27 July 2009 (UTC)

বিচ্ছিরি [bittʃʰiri] and বিশ্রী [bisri]
"For some speakers, [s] can be analyzed as an allophone of either /ʃ/ or /tʃʰ/ ([ʃalam] for সালাম [salam] 'greetings' or বিচ্ছিরি [bittʃʰiri] for বিশ্রী [bisri] 'ugly')."

I don't think this is right. First, [bittʃʰiri] is probably not the correct IPA transcription. It's more likely to be [bitʃʰ:iri]. Secondly, even so, the [tʃʰ] in [bitʃʰ:iri] isn't really an allophone of the [s] in [bisri]. বিচ্ছিরি is a todbhobo (i.e., derived Bengali) word for the totshomo (i.e., Sanskrit) বিশ্রী. When a Bengali speaker is pronouncing বিচ্ছিরি, he knows he is pronouncing a different word than বিশ্রী, a word with a different lexeme in the Bengali lexicon. --Zaheen (talk) 19:19, 11 August 2009 (UTC)


 * Could [bittʃʰiri] be an emphatic form? Gemination is common for that. — kwami (talk) 02:30, 11 March 2012 (UTC)

Postalveolars
What's the source for Bengali's postalveolars being alveolo-palatal? And the retroflex /ʂ/? Everything I've read on the language so far (which isn't much) has said they are simply palatoalveolars like in English and Hindi. AlexanderKaras (talk) 11:50, 6 January 2011 (UTC)


 * In the notes, we say "For some speakers, [s] can be analyzed as an allophone of either /ɕ/ or /tɕʰ/". That makes no sense, given the current chart. — kwami (talk) 20:34, 10 March 2012 (UTC)

Going through Masica, it would seem there are some serious problems to correct. For example, the spirantization of g, k, kh in Dhaka. I'll try to get to this this afternoon. — kwami (talk) 20:55, 10 March 2012 (UTC)

For diphthongs, we only ref Masica and s.o. from 1926. Masica d n h /ee̯/ /eo̯/ /æo̯/ /oe̯/, so I'm moving them here for now: — kwami (talk) 02:28, 11 March 2012 (UTC)


 * I'll put some notes (clipped from Masica) on my sandbox talk page. If s.o. here wants to use them to rewrite the article, feel free. This is just sitting on my to-do list. — kwami (talk) 19:58, 10 April 2012 (UTC)

Culture of Bengal template
Recently a template was placed on this page. In most phonology related articles such templates are avoided. Again, as this template is huge in size, it reduces the pleasure of reading the article. I've simply removed it for the time being. I think, there should be an consensus before putting that template on top of this article. »  nafSadh did say 18:05, 10 March 2012 (UTC)
 * I don't think the navigational template is that much different from the one at Spanish phonology, Hungarian phonology, or Catalan phonology.
 * One option we could consider is modifying the template so it's more presentable on this and other articles. — Æµ§œš¹  [ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ]  20:24, 10 March 2012 (UTC)
 * I don't really have a firm opinion on this point. If someone has removed the template, it is his pleasure. Since there is an objection, I will also remove phonology from the template. Cheers. - Chandan Guha (talk) 03:46, 11 March 2012 (UTC)

Masica notes
These were notes I took from Masica, but never got around to integrating into the article. Now I probably won't, so use them if you can.

They're direct quotes. No page numbers, though. — kwami (talk) 22:17, 18 July 2012 (UTC)

There is a tendency to pronounce the /c/ as an alveolar (or "dental") affricate [ts] in Eastern and Northern dialects of Bengali (Dacca, Maimansing, Rajshahi)

The [ts] pronunciation of /c/ in some NIA dialects has progressed to [s], in which case we can no longer speak of a stop articulation and, barring other developments, the inventory is reduced by one. This has taken place in the Chittagong dialect of Bengali (as described by Uchida), and in Assamese.

Complementing this predominantly western distribution of /ɳ / is a northeastern focus of occurrence of a phonemic velar nasal /ŋ/, which characterizes Bengali, Assamese, Nepali, Maithili, and Bhojpuri. There seems to be no disagreement over its phonemicity in these languages; Chaudhuri (1940-44) found that the North Bengali Rajshahi dialect contained an /ɲ/ phoneme (and not the usual Bengali /ŋ/).

The retroflex flap [ɽ] is often taken as an allophone of /ɖ /, with which it often stands in complementary distribution: initial, geminate, and postnasal for [ɖ ]; intervocalic, final, and before or after other consonants for [ɽ]. It has, however, come to contrast with [ɖ ] in at least some environments in ... Bengali (primarily through English loans, cf. Ferguson and Chowdhury 1960). ... The sound is absent altogether from Assamese, East and North Bengali dialects, and Bishnupriya - in all of which it has merged with /r/ ... It should also be noted that /ɽ/ is present phonemic ally in some languages (Western Hindi, Sindhi, Bengali,

In NIA the most widespread pattern consists of one voiceless sibilant, generally [s], plus /h/. In Standard Bengali, the dominant sibilant allophone is [ʃ] (becoming [s] before dental consonants).

the development of it secondary s from *ch has also introduced a contrast in the East Bengali dialects of Dacca and Chittagong, which originally had [ʃ] in all prevocalic environments. (In some East and North Bengali dialects, this [ʃ] has become [h] .)

/h/ occurs in almost all N I languages (one exception being the Chittagong dialect of Bengali)

The semivowels /y/ and /w/ are a somewhat shaky part of the N I A inventory. In a number of languages their occurrence is practically restricted to semi-predictable intervocalic glides. Their position is weakest in the east (where in Bengali the two are confused in writing), There is a phonetic as well as a historical difference between the Eastern glides, late in origin and sometimes optional, However, both /y/ and /w/ do exist in Hindi outside of diphthongs (i.e. initially), albeit only in deictics (again, leaving aside loans). In Bengali neither occurs.

An s/z opposition has also developed in Chittagong Bengali.

A more significant variation is the absence of the voiced aspirate series in some languages, reducing the system to a three-way contrast /p , ph , b/. It is also reported from the East Bengali of Dacca (Pal 1966a) and the dialect of Chittagong (UCida 1970). Absence (in the sense of elimination) of the voiced aspirates is often correlated with the presence of T O N E (e.g., both in Punjabi and in the Bengali dialects mentioned above).

There is a natural tendency (cf. Greek, Iranian) for aspirates to evolve into fricatives, especially voiceless aspirates. On the whole this tendency has been resisted in N I A, Bengali and its dialects constitute an exception.

Even in Standard Bengali there is a tendency to pronounce /ph/ as a bilabial fricative [ɸ], and also /bh/ as its voiced equivalent [β]. In the East Bengali dialects and especially in the dialect of Chittagong that is carried much further, to the velar stops and to both aspirated and unaspirated stops (*kh, k > [χ] *ph, p > [ɸ], *g > [ɣ]).

In contrast to all these languages, where aspirated nasals are very much part of colloquial speech, in Bengali they belong to an artificial acrolectal pronunciation and are ordinarily converted into plain geminated nasals ( *brɔmhɔ = [brɔmmho] > /brɔmmo/). ... These sounds (lh, rh, r.h) occur in Bengali under the same restricted circumstances as described earlier for mh, nh, ŋh.

a set of glottalized pressure stops (ejectives) found in some East Bengali dialects (and Bishnupriya) in place of the voiced aspirate series : /bˀ, dˀ, gˀ/ in the Maimansingh dialect, /bˀ, dˀ, ɖˀ, dzˀ, gˀ/ in the Chittagong dialect (according to Goswami 1940-4). Glottalization is often connected with T O N E and in the East Bengali cases seems to be related to the evolution of tone from the voiced aspirates. Indeed, Uchida's analysis perceives tone rather than glottalized consonants in the Chittagong dialect. (Chatterji claimed that the Dacca dialect also had glottalized voiced stops, a point which is disputed by Pal [1966a] .)

[w] [j] in chart cuz phonetic only

Otherwise the quintessential seven-vowel system is that of Bengali, with differentiation of lower-mid vowels both front and back: /i, e, æ; ɑ; ɔ, o, u/. Some dialects (Rajshahi, Maimansing, perhaps Chittagong, but not Dacca) appear to have reduced systems. (Ucida finds additional long vowels /ɑː, ɔː, oː/ in the Chittagong dialect.)

The presence of "geminated vowels" (ii, ee, oo, uu, etc.) allegedly belonging to only one syllable (cf. Goswami 1966: 97) in such languages as Assamese and Bengali is a reminder that a degree of conventionality governs even these analyses. Such vowels are not usually (an exception is UCida's analysis of the Chittagong dialect) counted as "long" vowels perhaps because they are fairly recent evolutes, and are often still spelled as two vowels separated by an (unpronounced) glide.

It is possible to speak of another kind of secondary subsystem, however, consisting not of borrowed sounds but rather of native sounds of marginal status. In Bengali, for example, there are rare but undeniable contrasts between, e.g., E/e in [chEle] 'boy' and [chele] 'if (it) covers' , and between I/i in [kIntu] 'but' and [kintum] 'I used to bring/would have brought' (Clinton Seely: personal communication). Cases like these can sometimes be taken care of by positing junctures: /kin- tum/ vs. /kintu/

Another question involves so-called rising diphthongs ([u̯a, i̯a, i̯e] , etc .) vs. sequences of the "semi consonants" /y , w/ + vowel. As this is partly a matter of particular phonological doctrine (and the rigor of its application) rather than of any phonetic differences, both interpretations are found in accounts of some N I A languages. In other cases, there is more justification for the diphthongal interpretation, namely when the glides occur only postconsonantally in such combinations (e.g., C + u̯a, i̯a, etc.) and never initially. Such is the case in Bengali, for example, where moreover the glides are often treated as "non-syllabic" e̯, o̯, rather than *i̯, u.

it is clear enough that the Eastern languages have the greatest number of true diphthongs (as well as disyllabic vowel sequences): in Bengali, for instance, Ferguson and Chowdhury (1960) recognize /āi̯, āe̯, āo̯, ei̯, oi̯, ui̯, āu̯, eu̯, ou̯, iu̯, ɔe̯, æe̯, ɔo̯ (also /ii̯, oo̯/), contrasting most of them with identical disyllabic sequences.

In Bengali, where nasalization [of vowels] has only a lexical role, it is absent altogether from some East Bengali dialects (Dacca, Maimensing), and in the Calcutta Standard affects only six of the seven vowels (not /ɔ/ - or only rarely) and (marginally) only one of the many diphthongs (ɔe or oi).

Contrastive tone is reported from several N I A languages and dialects, among them Chittagong and Dacca Bengali,

To say stress is predictable does not mean it always falls on a particular syllable, as in Polish or Finnish. Chatterji (1926) does describe such a stress pattern (initial syllable of the phrase or breath group) for Standard Bengali, but emphasizes that this generalization does not seem to apply even to other Bengali dialects

An interesting form of emphatic "stress" occurs in Bengali, Marathi, and perhaps other languages (although it is not characteristic of Hindi). It involves geminating a medial consonant. Sometimes this effects a quasi-phonological change: Bengali bɔɽo 'large' > bɔɖɖo 'enormous' (Clinton Seely, personal communication).

The palatal consonants are subject to severe positional restrictions in a few languages. In Dacca Bengali -cc-, -cch-, -jj-, occur mainly only medially and geminated. On the other hand the retroflex stops occur in all positions including initial position.

As already noted, in many N I A languages initial clusters are restricted in tadbhava words to sequences of C + semivowel (Cy-, Cw-) : (In Bengali, Assamese, Oriya the semivowels are customarily interpreted as vowels, sometimes even as syllabic vowels.) Another group of languages tolerates also sequences of C + r-: i.e. tr- (even Bengali, Assamese, and Oriya have tr- in the word for 'thirty' = B. triʃ)

Final clusters, native or borrowed, are not allowed in some N I A languages, including Bengali , Sinhalese , Nepali , and Braj. [even NC]

N I A languages typically have large inventories of medial clusters, including not only the above types but many others unknown to O I A. (There are some exceptions, e.g., Sinhalese. It should be noted also that written tatsama clusters are greatly simplified in Bengali pronunciation , according to assimilatory rules recapitulating those of Prakrit.)

In the first case, one possibility is vowel harmony. N I A furnishes at least two clear-cut examples of this phenomenon, both involving not the adaptation of the vowel of a suffix to the vowel of the stem, as in Turkish, but the adaptation of the vowel of the stem to the vowel of a suffix, and consequent morphophonemic variants of stems. The simpler case involves Standard Bengali, in which many stems show a higher or lower alternate vowel i/e, e/ā, o/ɔ, u/o, depending upon whether a suffix with a high vowel (i, u) follows or otherwise: kena 'to buy'/kini 'I buy' ; nɔʈ 'actor'/noʈi 'actress' (Dimock 1957). (In the Standard Colloquial, the conditioning vowel has dropped out of certain verbal endings, producing what now must be regarded as a partly morphologically conditioned alternation.

This article is a perfect example of Crap!
Filled with original researches, no strong set of references whatsoever and with contradictory information all over this article, it is hard to comprehend for anyone -- it makes a perfect example of how crappy an article can be. I am not an expert, but I know it requires a lot of work. To any editors here in this article: If you think you can improve without putting your original research or observations in the article please do so. But please please, do not put anymore OR here. It is enough already. -- nafSadh did say 16:00, 14 May 2015 (UTC)

How is this even possible?
1) How can a language that produces a phonemic vowel chart with only 7 entries be described as containing 14-15 vowels while, at the same, describing vowel length as non-contrastive?

2) Why are we using square brackets for phonemic transcription and not slashes? (As is done in every other article on phonology that seeks to differentiate the phonemic from the phonetic...which this table isn't even doing).

3) What does how we transcribe sounds for Bengali on Wikipedia have to do with a phoneme chart of the Bengali language? It's irrelevant and more suited for an article or a sub-section entitled "How Bengali is transliterated on Wikipedia"...it's simply not part of the phonemic description of ANY language. 128.83.155.208 (talk) 16:41, 11 July 2016 (UTC)

Just thought I'd address your point: "How can a language that produces a phonemic vowel chart with only 7 entries be described as containing 14-15 vowels while, at the same, describing vowel length as non-contrastive?" While it's poorly described, it does mention that plain vowels all have distinct nasal counterparts, including them wouldn't be helpful. 87.114.85.187 (talk) 18:26, 15 July 2016 (UTC)

স্বামী
Is there a rule for pronouncing স্বামী (sbami, "husband") as ? --Anatoli (talk) 06:11, 2 April 2017 (UTC)


 * from my lame experience স্ব is probably an irregular consonant cluster that should always pronounced as . It should be similar to the Hindi ज्ञ (gy) which is really the letters jñ :-D – AWESOME meeos ！ *  (chōmtī hao /t͡ɕoːm˩˧.tiː˩˧ haw˦˥/) 06:35, 11 April 2017 (UTC)

Adding nasal vowels to the chart
Should nasal vowels be added to the chart? One of the other talk pages seems to include someone being confused and forgetting that Bengali has nasalization, so I think this would make it clearer, but nasal vowels aren’t used in certain eastern dialects (Khan, 2010) so that could be confusing in and of itself. Tyrui (talk) 00:32, 4 May 2023 (UTC)