Sanskrit: Pronunciation of obscure vowels ऋ, ॠ, ऌ and ॡ.

tarkshya

Senior Member
Marwari
Standard Sanskrit alphabet lists 4 vowels, whose correct pronunciation remains a mystery to me. These are :
ऋ, ॠ, ऌ and ॡ.​

Can somebody explain the correct way to pronounce these vowels, preferably with the help of audio.
Wikipedia lists all the vowels and consonants recognized by IPA, along with the audio files here.

en.wikipedia.org/wiki/IPA_vowels_chart_with_audio
en.wikipedia.org/wiki/IPA_pulmonic_consonants_chart_with_audio

Can somebody tell me if these 4 Sanskrit vowels are part of these IPA charts. I guess not, but would like to confirm.

The reason I ask is that I suspect that originally these 4 vowels represented 4 distinct vowels sounds spoken by Indo Aryans, or even before them, the speakers of PIE. However, to me it appears that these sounds were lost forever and nobody now knows for sure how there vowels were pronounced originally. All modern pronunciations of these vowels are just a best guess. Is my hypothesis correct?
 
Last edited by a moderator:
  • Qureshpor

    Senior Member
    Panjabi, Urdu پنجابی، اردو

    tarkshya

    Senior Member
    Marwari
    Thanks for the reply Qureshpor. The last thread you listed (http://forum.wordreference.com/showthread.php?t=1215393) exactly describes my problem. I am still at loss to understand why these letters will be called "vowels" when the received pronunciations of these letters are some variations of "r" and "l" sound. Could it be that the original vowel sound is lost to us?

    Does any one know of any cognates in other IE languages which uses variations of these sounds? This should give us some clue.
     

    tonyspeed

    Senior Member
    English & Creole - Jamaica
    Thanks for the reply Qureshpor. The last thread you listed (http://forum.wordreference.com/showthread.php?t=1215393) exactly describes my problem. I am still at loss to understand why these letters will be called "vowels" when the received pronunciations of these letters are some variations of "r" and "l" sound. Could it be that the original vowel sound is lost to us?

    Does any one know of any cognates in other IE languages which uses variations of these sounds? This should give us some clue.

    Or the idea of what is a vowel has changed over time. It is claimed the original ऋ sound was more similar to an English rhotic 'r'. It is easy to see how this could be perceived as a vowel.
     

    Au101

    Senior Member
    England, English (UK)
    Hello. Sorry for resurrecting this one, I've been a bit busy.

    I can't easily provide you with an audio, unfortunately. It isn't too hard to find some good recordings of the first hymn of the Ṛg-Veda on youtube. You might like to listen to it, while reading along. Listen out for how the ऋ is pronounced in the first verse:

    [FONT=Sanskrit 2003, sans-serif]अ॒ग्निमी॑ळे पु॒रोहि॑तं य॒ज्ञस्य॑ दे॒वमृ॒त्विज॑म् ।[/FONT]
    [FONT=Sanskrit 2003, sans-serif]होता॑रं रत्न॒धात॑मम् ॥ १ ॥[/FONT]


    Agním īḷe puróhitaṃ yajñásya devám tvíjam;
    Hótāraṃ ratnadhā́tamam.

    Hmmm...are these vowels part of the IPA chart? The IPA does not provide a separate symbol for every conceivable variant of every conceivable sound - that would be absurd, of course. Like numerous natural language alphabets, the IPA uses diacritics. These vowels are represented in the IPA using the 'syllabic' diacritic: http://en.wikipedia.org/wiki/Syllabic_consonant. This link contains a lot of good information, which will come in handy later on. The IPA representation of the vocalic pronunciation of these letters would be as follows:

    ऋ = /r̩/
    ॠ = /r̩ː/
    ऌ = /l̩/
    ॡ = /l̩ː/

    But no, none of them has a single, separate IPA symbol to itself, that's true.

    I feel right now would be the best time to talk about what constitutes a vowel. A vowel is not actually that easy to define from a technical point of view, but it's quite easy to separate vowels from consonants intuitively. Vowels are sounds made without obstructing the airflow to a significant degree. Unlike consonants, where you use articulators (such as the tongue, lips and various structures in your mouth) to really get in the way of air coming up from the lungs, with vowels, the airflow is merely shaped and directed by the articulators. As it is put in "An Introduction to Language," by Victoria Fromkin, Robert Rodman and Nina Hyams, a vowel is:

    A sound produced without significant constriction of the air flowing through the oral cavity.

    This means that vowels are the most prominent - or sonorant - element of a syllable. From a phonologist's point of view, a vowel is nothing more or less than the sound which occurs at the heart of a syllable. If it occurs here, it is a vowel. If it doesn't, it isn't. If you have more than one separate vowel (not diphthongs, triphthongs, etc.), you have more than one syllable. Hence:

    <cat> (/kæt/) - one vowel, one syllable. The vowel is at the heart of the syllable. We say it's in the nucleus.
    <at> (/æt/) - one vowel, one syllable. The vowel is at the heart of the syllable. Just /t/ on its own is not permissible in English, whilst just /æ/ on its own is.
    <strings> (/stɹɪŋz/) - one vowel, one syllable. The vowel is at the heart of the syllable.
    <onomatopoeia> (ˌɒnəˌmætəˈpi.ə) - six vowels, six syllables. The two vowel sounds at the end - /i/ and /ə/ are pronoucned separately, so we have two syllables. When you count out syllables intuitively, you can hear that they are all clustered around vowels.

    Other people, however, may not agree with this interpretation, but it is the most relevant to us. A sound like /r̩/ is a vocalic r. What it means is that the /r/ sound is at the heart of the syllable - it is the nucleus. This is the meaning of 'vocalic consonant'. It is a sound which is usually consonantal (/r/ involves considerable obstruction of the vocal tract, unlike vowels), but which takes on a vocalic character. Vocalic consonants are consonants which occur in the nucleus, they occur in the same places as other vowels in the language, they behave like other vowels in the language (in terms of phonology and morphology), they can carry tone (in Sanskrit, the vocalic consonants could carry tone). Vocallic consonants tend to be liquids and nasals, since these can be pronounced continuously, like a vowel. Note that you cannot pronounce a stop like /b/ continuously. If you try, you just have a long period of silence and then the /b/. (You may hear a small gulping noise as the air is damned up behind your lips, but the /b/ sound is instantaneous).

    So, to get an idea of how this can be done, try pronouncing 'gld' without putting in any vowels. Let the 'l' be the vowel, just make a long 'lllll' sound. If you've done it correctly, the sound you would've made is a vocalic l - the ऌ. If you'd like to hear how this should sound, I can tell you that Czech and Slovak contain vocalic 'r's and vocalic 'l's. There is a wonderful link on this page here: http://en.wikipedia.org/wiki/Strč_prst_skrz_krk where you can hear how a vocalic 'r' sounds. You may find it profitable to have a look into some of the pages about Czech and Slovak. In English, some speakers pronounce words like <button> with a vocalic 'n': /ˈbʌtn̩/, but some people pronounce it with /ən/: /ˈbʌt.ən/. English also has the vocalic l in words like <awful> /ˈɔːfɫ̩/ when pronounced by some people. Others may pronounce this word: /ˈɔːfəɫ/, or without a /l/ at all.

    Now: the history lesson. This might get a bit involved so I'm going to try to keep things cursory and simplistic, but this may be at the cost of rigour and absolute accuracy. Unfortunately, I myself am not that well-versed in this matter anyway.

    In old PIE there existed both *r and *l. Both sounds. In Iranian we have only r. In general, this is the case in Vedic Sanskrit too, but there are some l sounds. In Classical Sanskrit you see a lot more l and many old Vedic words with r come to have l in Classical Sanskrit. It seems - according to Burrow in 'The Sanskrit Language' - that the dialect of Sanskrit from which the language of the Vedas developed was a dialect which, like Iranian, made all/most of the old PIE *l sounds into r sounds. However, there was another dialect which kept the old *l sounds as 'l's. It seems that these two dialects would have influenced each other, mixed and merged, to produce Classical Sanskrit.

    Now, are you familiar with saṃdhi. If you aren't just know that it is a form of ablaut, which is where the quality of a vowel is used to encode some grammatical feature, as in English: sing, sang, sung. There are three 'grades' in Sanskrit and there were in PIE too. In some grammatical forms you get the zero grade, in some you get guṇa and in some you get vṛddhi. The correct grade is vital to the correct conjugation and declension of Sanskrit words and provides valuable grammatical information.

    Now, in PIE, the zero grade of *l was *ḷ (vocalic l) and the zero grade of *r was *ṛ (vocalic r). It appears that the old PIE *ṛ became Vedic Sanskrit ṛ. In the early pronunciation of Vedic Sanskrit this was a proper, bona fide vowel, pronounced as described above, like the vocalic 'r's in Czech I linked to. Arthur Anthony MacDonnell in his Vedic Grammar tells us:

    "The vowel ṛ, now usually sounded as ri (an early pronunciation as shown by the confusion of ṛ and ri in ancient inscriptions and MSS.), was in the Saṃhitās pronounced as a vocalic r, somewhat like the sound in the final syllable of the French chambre. It is described in the RV. Prātiśākhya as containing an r in the middle. This agrees with ərə, the equivalent of ṛ in Old Iranian."

    As you can see, it was not long before this early pronunciation came to be replaced the modern pronunciation /ri/ we hear today. (N.B. /ru/ is also used in some parts of India and is the common pronunciation of ṛ in Gujarati, I believe). The linguistic (looking at cognates with other languages, the fact it carries tone, the fact that it undergoes ablaut, the fact that nouns can have stems in -ṛ, the distribution of the sound (the fact that it occurs where vowels occur and not where consonants occur)), orthographical (the fact it is written as a vowel) and documentary evidence of the speakers all tell us that this sound began its life in Sanskrit as a vowel sound.

    As an aside, if you like, here are some cognates as given by Thomas Burrow in the Sanskrit Language:

    Sanskrit: pccháti "asks"
    Latin: poscit < *porscit
    Old High German: forscōn

    Sanskrit: pitṛ́ṣu, locative, plural of pitár- "father"
    Greek: πατράσι (patrasi)

    Sanskrit: vttá- "turned"
    Latin: versus, vorsus

    Sanskrit: mtá- "dead", cf.:
    Latin: mortuus, mors
    Lithuanian: miti "to die"
    Old Slavonic: sŭmtĭ "death"

    It also seems that old PIE *ḷ also became ṛ in Vedic (remember what I said about l becoming r).
     

    Au101

    Senior Member
    England, English (UK)
    Now, as for ॠ, PIE had a long vocalic r, but it does not seem that PIE long vocalic r is the source of Sanksrit ॠ. PIE long vocalic r seems to have become sanskrit īr/ūr. No, Sanskrit ॠ seems to have been a grammatical peculiarity of Sanskrit. If you look at the paradigms for -अ and particularly -इ and -उ stems, we find:

    Accusative plurals: munīn, tarūn, etc.
    Genitive plurals: munīnām, tarūṇām, matīnam, dhenūnām.

    And so we get:

    netṝn, pitṝn, mātṝḥ
    netṝṇām, pitṝṇām, mātṝṇām

    We do not really find ṝ anywhere else. It appears it merely exists by grammatical analogy and is unique, in this regard, to Sanskrit.

    Its pronunciation was once what you'd expect - a long /r̩ː/. Like ऋ only longer. Nowadays, /riː/ and /ruː/ tend to be used.

    As I've noted already, old PIE *ḷ became Vedic ṛ. There are only a few forms of one single Sanskrit word: kḷp which contain ऌ and in Vedic this word appears as kṛp. Kḷp seems to have come about later in the development of Sanskrit, though it isn't completely clear why, nor can we be certain that this is the case. Regarding pronunciation it is like English <bottle> in some pronunciations (see above), or like the 'l' in gld. From MacDonell's Vedic Grammar:

    "The very rare vowel ḷ, now usually pronounced as lri, was in the Saṃhitās a vocalic l, described in the RV. [Ṛg-Veda] Prātiśākhya as corresponding to l representing an original r."

    As he says it is now generally pronounced lri (which I find very hard to say) - or, in fact, lru. It only occurs a handful of times ever.

    ॡ does not exist. It does not exist in any word anywhere in Sanskrit. At all. It is there purely for orthographical completeness, to complete the pattern. PIE did have a long vocalic l, but Sanskrit does not. If it had existed, it would presumably have been originally pronounced as a long form of ऌ, like ॠ is the long form of ऋ. That is to say: /l̩ː/.

    Nowadays, I suppose if anyone does say it - for example when learning Devanagari, they would pronounced it lrī or lrū. However, it does not exist.

    Now: regarding whether or not we can tell how these sounds were pronounced. In theory, we can never be truly certain of how languages were spoken in the past, since archeologists have yet to find any handy recordings! So until someone invents a time machine, we have to deduce it from the available evidence. Some of our guesses, of course, are better than others. Happily, PIE reconstruction is now a relatively advanced field and there is a lot of information.

    In this specific case, though, as well as the wealth of evidence, we can be quite certain about how the sounds were pronounced, because of the central role accurate, oral recitation of the Vedas played in Vedic religion. Because of this, we have written down for us a meticulous account of the pronunciation of these sounds: the prātiśākhya and other śikṣā texts. Unfortunately, these aren't that easy to understand. However, the Vedic speakers themselves tell us quite a lot about the pronunciation of their language. We also need only note that these sounds carry tone and are arranged alongside the other vowels to be confident in their vocalic nature. Also, the fact that Vedic pronunciation has been relatively well-preserved is useful, although the practise of oral recitation is dying out and modern renditions also incorporate numerous changes.
     

    Au101

    Senior Member
    England, English (UK)
    I suppose, to put it more succinctly (I've just thought of this), a vocalic consonant - such as ṛ - is pronounced exactly as you would normally pronounce it, you just have no vowels either side. You just have it - perhaps with consonants either side.
     

    marrish

    Senior Member
    اُردو Urdu
    Thank you so much for this enlightening explanation and the work of love.
     

    tarkshya

    Senior Member
    Marwari
    Hi Au101, tons of thanks to you for such a detailed reply. I have never seen such a thorough discussion on these rare sounds of Sanskrit. You reply must be taken as the final, authoritative word on the subject.

    The only reason I am replying late is because I was reading your reply with full attention as it deserves, and trying to soak in the knowledge. And I must say, the Check/Slovak tongue twister helped a lot :)

    It was also quite interesting to know that vocalic l disappeared from Sanskrit for a while, and then reentered. I like to think that its reentry was made possible by some strict grammarian who did not like his students playing fast and loose with the language :)

    Anyway, thanks again.
     

    tarkshya

    Senior Member
    Marwari
    Let me take this discussion on /r/ and /l/ little bit further. The fact that /l/ disappeared from both Persian and Vedic Sanskrit came quite as a surprise to me. I in fact checked the texts of Rgveda, the oldest known vedic sanskrit text (It is available at sacred-texts.com) and confirmed this fact. /l/ sound is so rare in Rgveda that one can count its occurrence in the entire corpus on the fingers of one's hand, while /r/ of course is rampant. And I thought only Japanese/Koreans have problem distinguishing between /r/ and /l/ :)

    In the classical Sanskrit texts like Mahabharata and Bhagwad Geeta, I found both /l/ and /r/ in abundance.

    My question is, can you give some example words in Sanskrit, where /r/ in Vedic changed to /l/ in classical. I know you gave an example of kṛp/kḷp, but it is really an example of /ṛ/ and /ḷ/, i.e. vocalic r and l. Do you know any words with consonantal r and l where this switching of sounds happened?

    Also, you said /l/ disappeared from Persian too. Did it ever reappeared there just like Sanskrit, or did Persian lost its /l/ forever? Are there any bonafide Persian words (i.e. not Arabic or otherwise foreign) in modern Persian?
     

    Qureshpor

    Senior Member
    Panjabi, Urdu پنجابی، اردو
    Persian has plenty of words with "l". For example

    paliid, laaf, laNgiidan, polaad/folaad, laalah, malaNg and so forth.
     

    mundiya

    Senior Member
    Hindi, English, Punjabi
    Let me take this discussion on /r/ and /l/ little bit further. The fact that /l/ disappeared from both Persian and Vedic Sanskrit came quite as a surprise to me. I in fact checked the texts of Rgveda, the oldest known vedic sanskrit text (It is available at sacred-texts.com) and confirmed this fact. /l/ sound is so rare in Rgveda that one can count its occurrence in the entire corpus on the fingers of one's hand, while /r/ of course is rampant. And I thought only Japanese/Koreans have problem distinguishing between /r/ and /l/ :)

    I wouldn't consider /l/ to be that rare in the Rigveda. From a search in Monier Williams Sanskrit dictionary, there are 243 words in the Rigveda with the /l/ sound.

    Also, you said /l/ disappeared from Persian too. Did it ever reappeared there just like Sanskrit, or did Persian lost its /l/ forever? Are there any bonafide Persian words (i.e. not Arabic or otherwise foreign) in modern Persian?

    It was completely absent in Old Iranian (Old Persian and Avestan), but /l/ appeared by the time of Middle Persian.

    rohita/lohita is a very commonly given example, I'm sure there are many more.

    The first is attested in Rigveda, the second in Atharvaveda.
     

    Dib

    Senior Member
    Bengali (India)
    My question is, can you give some example words in Sanskrit, where /r/ in Vedic changed to /l/ in classical. I know you gave an example of kṛp/kḷp, but it is really an example of /ṛ/ and /ḷ/, i.e. vocalic r and l. Do you know any words with consonantal r and l where this switching of sounds happened?

    One small disambiguation may be useful here. It may not be proper to say /r/ in Vedic changed to /l/ in Classical Sanskrit. CS often contains both words/roots - sometimes as synonyms, sometimes in slightly different meanings. The situation is usually taken to imply that Vedic and CS had different base-dialects, not necessarily a temporal evolution of some of the Vedic r's into CS l. Anyways, Au101 has already pointed out lohita-rohita. There are tons of other examples too, e.g. the root ruc- (to shine; Cf. Persian rooz, day) > ruci- (light, beauty) / locana- (brightening, eye) / lokana (seeing); gara- (swallowing > drink) ~ gala (throat); cal- ~ car- (to move); probably shru- (to hear) > shloka- (a kind of verse); shrath- ~ shlath- (to become loose); etc. come to mind.


    Also, you said /l/ disappeared from Persian too. Did it ever reappeared there just like Sanskrit, or did Persian lost its /l/ forever? Are there any bonafide Persian words (i.e. not Arabic or otherwise foreign) in modern Persian?

    Indeed, no native Old Persian word had an "l" in it, and even foreign words often had an "r" instead, e.g. bābiru- for Babylon. But as QP pointed out, modern Persian is full of l. Some of this is indeed the result of native evolution within the history of Persian, e.g. Old Persian rd > New Persian l.
    - So, Old Persian has "thar(a)d-" (year) (Cf. Skt. sharad- = autumn/year) which has become "sāl" in New Persian.
    - Another interesting word is gol (= Hindi-Urdu "gul") = rose, which most likely comes from an older "*ward-" or "*wṛd-" (with vocalic r), which as far as I know have been attested only as loanwords in the neighboring languages (E.g. Arabic "ward", rose (through Aramaic, I suppose?)).
    - Similarly, Modern "boland" (tall, high) is probably from another older root *ward- (Cf. Skt. vṛdh- = to grow).
    - NP del (= H-U dil) also has l where Sanskrit has rd "hṛd", as would be expected in Old Persian too.
    - NP māl- (to sweep) < OP *mard- (Cf. Skt mṛj- "to wipe").
    The problem is that the Old Persian corpus is rather small and lacks variety, leading to lack of attestation of many of these expected forms, as they can be derived from circumstantial evidence (especially comparing with the Sanskrit and Avestan data, and loanwords in neighbouring languages).
     

    Dib

    Senior Member
    Bengali (India)
    I wouldn't consider /l/ to be that rare in the Rigveda. From a search in Monier Williams Sanskrit dictionary, there are 243 words in the Rigveda with the /l/ sound.

    It would be an interesting exercise to find out what proportion of these words have proper IE etymology, what proportion is most likely borrowed, and what proportion can go either way. How did you manage to get this number, by the way? I will be really grateful if you told me - maybe by PM, as it is rather off topic.
     

    tarkshya

    Senior Member
    Marwari
    Thanks everybody for their input. Even though the discussion has digressed a little, I got the answer to my original question and also gained valuable knowledge of sound shifts between various dialects of Sanskrit. I still have more questions on the sounds of many other vowels/consonants of Sanskrit, but I guess I will open separate threads for them to keep the discussions neatly sorted.
     
    Top