Tuesday 30 November 2010

vowel inventories

It is good from time to time to stand back and consider what the phonetics of familiar languages look like from the perspective of language universals, and how they might appear to those who don’t know them.

Vowel systems of different languages vary widely in size.

The ever-fascinating World Atlas of Language Structures Online devotes the second of its 142 “feature” articles to Vowel Quality Inventories. There is (i) a narrative discussion and (ii) a customizable map of the world plotting vowel systems with unusually small (2–4) or unusually large (7–14) inventories. Various theoretical or procedural issues, such as the status of diphthongs, have to be settled before you start counting. That done, though, the average number of vowels in a language turns out to be just fractionally below 6. The smallest vowel quality inventory recorded is 2 and the largest 14.(Click here for the full-size map.)

And which language would hold that record number of fourteen vowels? Standard German! (Count them: iː ɪ eː ɛ aː a yː ʏ øː œ uː ʊ oː ɔ, to which you could of course add ɛː for some speakers. The weak vowels ə ɐ are not counted, nor the nasalized vowels used in borrowed French words.) The “variety of British English included here” is reckoned to come in equal second place, with thirteen. Its posited inventory presumably comprises iː ɪ e æ ɑː ɒ ɔː ʊ uː ʌ ɜː eɪ əʊ, the last two being regarded as unitary. Not only the remaining diphthongs but also schwa are excluded.

Considerably more languages have an inventory of five vowels than any other number — just over a third of the sample. Familiar examples would be Spanish, Greek, Japanese and Swahili, all with just i e a o u.

Polish adds one more, ɨ, giving six vowels (ignoring the nasalized ones, which we can arguably analyse away). Korean adds two, ʌ ɯ, giving seven.

Four languages in the sample have only two contrasting vowel qualities.

You can see why many foreign learners of English (but not speakers of other Germanic languages) would find the English vowel contrasts so difficult to master. Not to mention foreign learners of German.

_ _ _

No posting tomorrow. Next: 2 December.

51 comments:

  1. The set of languages is of course far from complete, which leads to some easy misses. Standard Swedish has 17 vowel phonemes (if you count vowel length as phonemic, which most people do, and which was also done for German), three more than German, while Danish has considerably more. (See for instance http://en.wikipedia.org/wiki/Swedish_phonology#Vowels).

    ReplyDelete
  2. Though Wikipedia indeed gives those 14 vowels for German phonology, either Wikipedia is wrong, or WALS is (or WALS counts the schwa). In the accompanying text, WALS says "long and short variants of the same vowel are always counted once", and therefore /a/ and /a:/ should be counted as a single vowel, leaving 13 vowels for German. It is also unclear why /eɪ/ would be considered a separate vowel in English, as botu /e/ and /ɪ/ are part of its vowel inventory, and as WALS explains, "The diphthongs (...) can be resolved as combinations of two of these basic vowels occurring within a single syllable." /əʊ/ then would be a vowel, as /ə/ does not occur in the English vowel inventory. Dutch, though not a language in the WALS sample, arguably has more vowels: 12 or 13 simple vowels (/a ɑ e ɛ o ɔ i ɪ y ʏ ø u (ə)/ and 2 or 3 diphthongs that classify as separate vowels according to WALS's rules as they have at least one element not found in the simple vowel sequence /œy ʌu (ɛi)/. However, I think counting vowels like this doesn't really serve anything, as phonemes are lumped together: just because standard Dutch has diphthongized /œy/ from original /u/ in most, but not all, environments, it gains a vowel, while dialects that have not diphthongized do not, though they have the same phonemic distinctions.

    @David: WALS does not count vowel lengths, see the accompanying text.

    ReplyDelete
  3. Why are weak vowels not included? Imagine trying to speak English without schwa.

    ReplyDelete
  4. Kilian makes some good points - it's a bit arbitrary. And why isn't /ɛː/ (long ä) included in Standard German?

    But still, John's point is that some Western European languages in general have larger-than-average vowel inventories. Standard French has 15 by my reckoning: a ɑ ə e ɛ ɔ o y u œ ø ɑ̃ ɛ̃ ɔ̃ œ̃ (hope tose tildes have come out OK). Or are nasal vowels not counted separately?

    ReplyDelete
  5. I wonder how easy the English vowels really are for German-speakers. The two languages may have similar vowel counts, but the German system appears more straightforward, in that it consists of seven long-vs-short, similar-quality pairs (and it surely helps that each pair corresponds to a letter of the alphabet). The English monophthongs don't seem so easily paired off with one another, for example where would ʌ fit in? Of course that's just my intuition as a native English speaker, trying to second-guess the EFL experience for Germans.

    I've always found it odd that only one pair in German, a vs. is actually identical in quality, as Kilian Hekhuis noted. One might have expected a vs. ɑː, as in Welsh. I confess that when I attempt to speak German, my vs. ʏ and øː vs. œ can only be reliably distinguished by length - but that's entirely my fault.

    Speaking of Welsh, Wikipedia gives 13 simple vowels.

    Pete - ɛː is not included for German presumably because we're only counting vowel qualities. As for nasal vowels, the link explains: "...nasalized vowels do not add to the inventory as long as a non-nasalized counterpart occurs..."

    ReplyDelete
  6. Thanks Leo - But how come /a/ and /aː/ are both counted but /ɛ/ and /ɛː/ are not?

    ReplyDelete
  7. Pete: Many speakers of Hochdeutsch do not distinguish from ɛː, but make Beet and spät rhyme. Duden Aussprachewörterbuch, 6th ed., p. 36: "[ɛː] kann durch [eː] ersetzt werden...".

    ReplyDelete
  8. You get 13 qualities for English if you just count the symbols used in Gimson's transcription (including schwa), and as WALS cites Gimson and says that only different vowel qualities count, I suspect that may be how they got the number. (That's i ɪ e æ a ɑ ɒ ɔ ʊ u ʌ ɜ ə.) If you use the same method with Upton's transcription, it goes down to 12 (gain ɛ, lose æ and ɜ).

    Similarly you still get 14 for German if you include ə but not .

    ReplyDelete
  9. The contrast between ɛː and eː in Standard German is due to spelling pronunciation, and it is far from established, e.g. even among those who do not rhyme Beet and spät, many will rhyme Meter and später.

    English vowels are not that easy for German NS, but at least they do not have to struggle with the notoriously difficult lax/tense vowel contrast.

    ReplyDelete
  10. My favourite minimal pair is "Reeder" - "Räder". (It also contrasts with "Redder", a kind of lane, though I'm only familiar with that one in placenames.)

    ReplyDelete
  11. Keeping 'British English' FACE and GOAT seems to reflect a North American analysis. Disgarding those two would leave us with eleven. We could get back up to thirteen, though, by including the usually monophthongal SQUARE vowel and the increasingly commonly monophthongal NEAR vowel. Now that's what I call a true English length contrast - KIT vs NEAR, 'bid' vs 'beard'.

    ReplyDelete
  12. ɛː screws up the symmetry of the system, as all other stressed vowels come in pairs of one long vowel and one short vowel (e.g. ɔ and oː but no ɔː).

    ə is in complementary distribution with ɛ, the former only occurring in unstressed syllables and the latter only in stressed syllables, and ɐ can be considered an allophone of /r/ (when non-syllabic) or /ɛr/ (when syllabic) -- according to Wikipedia.

    ReplyDelete
  13. Paul Carley - true about "bid" and "beard" (at least for me), but if we transcibe them bɪd and bɪːd then we can only count the ɪ vowel once, so it doesn't increase the total.

    ReplyDelete
  14. Phoneme counts are always controversial since the results depend on the principles of analysis used. Swedish was mentioned. A different solution is say 9 vowel phonemes, and quantity as a separate distinction, yielding long and short allophones of the 9 vowels. I need more time on that site, I didn't find any rankings or counts for individual languages. There was a time when you could lose your job if you suggested fewer vowel phonemes than three, the orthodox definition of "system" was that orthodox.

    ReplyDelete
    Replies
    1. The 9 'long' and 8 'short' vowel phonemes in Central Swedish are distinguished almost entirely by quality. 7 of the long ones have diphthongal proprties (e:,i:,o:,u:,y:,ö:,å:), and 2 of the long monophthongal (a,ä). But the monophthongal ones a: and ä: have significant quality differences from their short counterparts. a: is much more back than a, and ä: is more low than ä, particularly for speakers under age 30. The seminal source for Central Swedish vowels is Eklund and Traunmüller (1997). You can view the waveforms there. Few people have mentioned Danish on this thread. It actually has more 'pure' vowels than Swedish.

      Delete
  15. Leo

    Going by the German example, I thought we were including length distinctions.

    ReplyDelete
  16. army1987

    I'm not so sure about an analysis into long and short pairs. When you get down to the phonetic nitty gritty, such a neat description starts to break down. That's why I was so keen on bid/beard as a true pair based on a length distiction alone (for those who have such a NEAR). It would be the only such pair.

    ReplyDelete
  17. Keeping 'British English' FACE and GOAT seems to reflect a North American analysis.

    Why? The starting point of FACE is distinct from DRESS for the vast majority of BrE speakers (not to mention those northerners for whom it is monophthongal). I guess you could argue that the starting point of GOAT is the same as NURSE for some speakers, but I doubt it's true of that many.

    ReplyDelete
  18. It seems somewhat arbitrary to include height, backness and lip rounding and length in the basic phoneme count but not tone or nasalization.

    The IPA Handbook article on Cantonese lists 11 monophthongs, 11 diphthongs and 6 contrastive tones. I don't know whether every monophthong/diphthong can contrast all six tones, but if they do then that gives potentially (11 + 11) * 6 = 132 contrasting vowel sounds.

    ReplyDelete
  19. Paul Carley - if we go on WALS's stated terms, we are not counting length pairs: "Long and short variants of the same vowel are always counted once."

    But you are right - they are apparently counting both long and short a for German, so I'm as confused as anyone. I can't find a specific vowel count for German on their site, only an overview which states "Vowel Quality Inventories: Large (7-14)".

    ReplyDelete
  20. vp

    Many American writers treat FACE and GOAT as essentially monophthongal and give them the phonemic symbols /e/ and /o/ accordingly.

    In this sense including these two vowels in a list that doesn't include diphthongs seems to reflect an American style analysis.

    How that analysis reflects phonetic facts is not something I consider myself qualified to judge.

    Maybe John could give us his thoughts on the matter in a future blog. To what extent do US transcription habits reflect real differences from RP, and to what extent is it just a matter of differing transcription conventions?

    ReplyDelete
  21. vp

    Tone isn't usually counted as a vowel modification. More a property of a syllable.

    ReplyDelete
  22. A lot of phonetic transcriptions of German distinguish between /e/ and /ɛ/ in short form, used for e and ä respectively. For example, my Collins pocket dictionary from 2003 gives /e/ for Metall and /ɛ/ for hässlich.

    I am more surprised that /e/ was omitted than I am that /ɛ:/ was.

    ReplyDelete
  23. The size of vowel inventories can vary considerably within languages.

    For Korean, this ranges from 10 to 6. Standard pronunciation given in textbooks favours the 10-vowel system. It also allows the two front rounded vowels to be pronounced as glide-vowel sequences, giving an 8-vowel system.

    The 10-vowel system is still found in the speech of older Koreans of the Central and Southwestern regions (Standard Modern Korean was codified based on the speech of the Central region around Seoul in the early 20th century). The 8-vowel system is found in older North Koreans.

    Younger Koreans of the Central and Southwestern regions as well as Jeju Islanders have a 7-vowel system, merging the two mid front unrounded vowels. This is the type most commonly heard in the South Korean media, so it makes sense that WALS uses this. Older Jeju Islanders may have a 9-vowel system, where the ɒ of Middle Korean is preserved.

    Southeastern Koreans further merge the back unrounded vowels, leading to a 6-vowel system, and younger North Koreans lose the roundedness distinction in the back vowels completely, leading to a different 6-vowel system.

    To sum up:
    10-vowel system: i e ɛ a o u ʉ ɵ ʌ ɯ
    9-vowel system: i e ɛ a ɒ o u ʌ ɯ
    8-vowel system: i e ɛ a o u ʌ ɯ
    7-vowel system: i e a o u ʌ ɯ
    6-vowel system A: i e a o u ʌ
    6-vowel system B: i e ɛ a o u

    (ʉ ɵ ʌ ɯ are also written as y ø ə ɨ respectively depending on descriptions)

    ReplyDelete
  24. What about /o/ in German as in the end of Sakko or Saldo? Is this discarded as a weak vowel?

    ReplyDelete
  25. Have you read the following three articles:

    - “What Forced Syntax to Emerge”?
    - “Why Is There Morphology”
    - “Syntax as the Last Step in Evolution of Language”

    Although non of them actually deals with phonology, what they do show is the effect of post-glacial changes upon communities led to development of syntax and erosion of morphology.

    What I firmly suspect is that this process was accompanied by an increase in the number of vowel phonemes in languages. It is clear - more so than shown here in WALS that languages with very high morphological complexity, such as those of northern North America, tend to have very small vowel inventories. Languages with large vowel inventories tend to be analytic or nearly so. This is probably because complex morphology tends to involve phonological processes that are likely to make it impossible to specifically contrast vowel phonemes. In strictly isolating languages like Vietnamese, and even in languages like Swedish, extremely marked vowel sounds involving roundedness contrasts alone can survive in a way impossible in highly polysynthetic languages of hunter/gatherer economies.

    ReplyDelete
  26. @Paul Carley:

    Why isn't tone (where phonologically distinctive) treated as a property of the vowel segment?

    ReplyDelete
  27. @Paul Carley: I was speaking about German not English.

    ReplyDelete
  28. OT: For a next post, would you mind spending a few words on the famous quote by that great man who just passed away? Surely it has some phonetic interest.

    ReplyDelete
  29. @Pete: Nasal vowels are indeed not counted separately (as they can be analyzed as simple vowel [+nasal]), except when the simpel vowel is not found as a non-nasal one.

    @Leo: "for example where would ʌ fit in" - It would fit nicely with German ɐ (albeit it's a bit higher), as ʌ is much more centralized than its symbol would lead one to expect.

    @jpbenney: Though interesting, it has very little to do with the topic at hand.

    @vp: Tone is not treated by WALS as distinctive, for the same reason as length and nasality are not. See above.

    ReplyDelete
  30. @Kilian: In lots of languages nasal vowels can be analysed as oral vowel + nasal consonant. Japanese, for example. But in French this analysis doesn't hold water as demonstrated by the minimal pairs
    'bon' - 'bonne': /bɔ̃ / - /bɔn/
    'on' - 'homme': /ɔ̃ / - /ɔm/
    'saint' - 'saigne': /sɛ̃ / /sɛɲ/

    Still, the WALS notes make it clear that nasal vowels are only counted separately if their oral counterparts haven't already been counted, so fair enough.

    ReplyDelete
  31. But in French /ɑ̃ ɛ̃ ɔ̃ œ̃/ don't have the same quality as /ɑ ɛ ɔ œ/: for example, ɑ̃ is rounded and ɑ isn't,

    ReplyDelete
  32. @Kilian Hekhuis: Leo was talking about pairing English monophthongs with each other.

    (I guess one could consider KIT/NEAR, DRESS/SQUARE, STRUT/NURSE, FOOT/CURE, LOT/NORTH and TRAP/START to be pairs; but for the last few the vowel qualities are definitely not the same; and many of those speakers who rhyme mirror/nearer, merry/Mary, hurry/furry and courier/curia don't merge sorry/Laura, and few (if any) of them merge carry/starry; so such an analysis would definitely be too artificial especially for the last few pairs.)

    ReplyDelete
  33. army1987 - exactly, I don't think ʌ makes a good pair with any other vowel. People sometimes call it "short u", but that name could just as easily be given to ʊ - particularly by native German-speakers. As you say, the German system has a symmetry which the English just doesn't. And if we did have to make pairs for English, KIT would surely go with FLEECE - again, that would be particularly natural from the viewpoint of German, with its ɪ and pair.

    Going back to German a and - when I studied the language it seemed very natural to map onto my English PALM. Now, young British speakers like me are often said to have a TRAP vowel that tends towards a. If that's true for me, I should really be mapping German onto the long version of my English TRAP - æː - which I have natively in some words due to the "bad-lad split". But I really feel that German is closer to my PALM than my TRAP. Perhaps it is ɑː at least for some speakers?

    ReplyDelete
  34. @Leo: Perhaps it is ɑː at least for some speakers?

    Well, this just shows how traditions in transcription can muddle things up. If you look up formant values for the two German vowels, they seem to be in the general area of 700 Hz for F1 and 1200 Hz for F2. This is indeed more like SBE PALM/START than TRAP in e.g. Deterding's data. Don't let yourself be led astray by symbol choices...

    ReplyDelete
  35. @army1987

    So that's why I was so confused by your comment!

    ReplyDelete
  36. I would say a similar thing about the common German transcription of /aʊ/. It is more like /ɑʊ/ for most Germans.

    ReplyDelete
  37. @nrdl: I agree that Leslie Nielsen was a great man. I don't think I've ever known anyone who has "Shirley" and "surely" as regular homophones, so it is strange how the joke managed to get so big.

    (Here in Britain) A lot of people at my work said, "I never knew he was Canadian." We're not very good at distinguishing Canadian and American accents on this side of the Atlantic.

    ReplyDelete
  38. @wjarek: Perhaps they should just introduce a symbol for the vowel halfway between cardinals 4 and 5 (what about a small-capital A?). Now someone will answer that no language has a three-way /a A ɑ/ contrast, but then what language has a four-way /ɘ ə ɜ ɐ/ contrast?

    ReplyDelete
  39. David Marjanović2 December 2010 at 02:15

    I see I picked the right moment to return to this blog...

    Part 1 of maybe 4:

    German /a aː/ contrast: Maybe it was counted because these two phonemes have different qualities in some accents that count as standard. (Keep in mind that the pronunciation of Standard German isn't any more standardized than "that" of Standard English.) In my Standard accent*, the quality difference is tiny and inconsistent, so that [a] and [aː] are very good descriptions, but various Wikipedia articles and discussion pages make clear that they are [ɑ aː] and [a ɑː] for others. (I suspect the central instead of the back vowel is meant, but I haven't got much experience with listening attentively to the accents of Germans.) Maybe that's why they're both counted.

    * Never mind my dialect, which lacks phonemic vowel length and handles the low vowels in an entirely different way.

    German alleged /ɛː/: The number of people who have such a phoneme is small and reportedly shrinking, and AFAIK the feature is restricted to part of western Germany. That's certainly why the WALS doesn't count it, even though it's still not considered nonstandard.

    Anyway, it may or may not mean anything, but I have never encountered anybody who pronounces it [ɛː]. What I have heard is a wide-open [æː]. On one occasion, this completely threw me off – it took me a second or three to convince myself that the people in question were in fact speaking German; only then did I understand what they said!

    I don't see a reason to declare it a spelling-pronunciation. After all, there is no /ɛ æ/ distinction in any Standard accent. I think it's just another development from the Middle High German system of "e-like vowels", which was (or so I've read) /æ æː ɛ ɛː e/ (and lacked /eː/). Just for comparison, the result in my dialect is /a ɛ e/ plus the Central Bavarian E Confusion (a strictly defined set of words that have [ɛ], [e], or anything in between depending on very subtle issues of stress, surrounding words with such vowels, and maybe the phases of the moon or something).

    Oh, one thing: everyone uses [æː] as the name of the letter ä. But that, for most people, is it.

    German /aʊ/: yes, it starts with the central vowel, not the front one, in most or all Standard accents. The front vowel is used in Styrian* and Swabian dialects.

    * Schwarzenegger!

    German alleged /ə ɐ/: I hereby flat-out deny the existence of /ə/. It is unstressed /ɛ/, and in Austrian Standard German it is, in fact, unstressed [ɛ] (Hände [ˈhɛnd̥ɛ] has twice the same vowel). In other Standard accents, it is [ɵ] or [ɘ]; I don't think [ə] even occurs... but, anyway, there are no minimal pairs to /ɛ/. (If you disagree, show me one, and I'll take it apart.) The WALS is right not to count it.

    As mentioned above, [ɐ] can most likely be regarded as an allophone of /r/. That must be why the WALS doesn't count it.

    German alleged short /e i o u ø y/: In order to be long, a German vowel must bear at least secondary stress. When a morphophonemically long vowel finds itself in an unstressed position – this happens in Latin and Greek loans and with the prefix ge-* –, it first loses its length and then becomes open, so that the intermediate stage ([e i o u ø y]) is actually pretty common. However, I don't think there are minimal pairs. Anyone?

    Over here, Sakko is stressed on the last syllable, so it ends in /oː/. I've heard Germans putting the stress on the first syllable; I'm pretty sure they ended the word in [ɔ]. Saldo is not in my active vocabulary at all, but I'd use [ɔ], too...

    * Old High German gi-. Perhaps that explains why this prefix, which is never stressed, behaves as if it were underlyingly /geː/.

    ReplyDelete
  40. David Marjanović2 December 2010 at 02:18

    Part 2:

    French: The transcription used by dictionaries and linguistics textbooks describes 19th-century upperclass French. There are still accents that are similar to it, but modern Parisian French is a different beast. In detail: /ɑ/ is completely gone from Paris, it has merged into /a/ so completely that the (mostly ignored) recent spelling reform tried to abolish the letter â. (The phoneme survives as [ɑ] or [ɒ] in other accents, though.) The alleged /ɔ̃/ is a cardinal [õ]; bon [bõ] and beau [bo] are a minimal pair for nasality. The alleged /ɛ̃/ is [æ̃], except in Québec where it's [ẽ]; I haven't encountered anyone for whom messe and mince are a minimal pair for nasality alone the way bon and beau are. /œ̃/ has completely merged into "/ɛ̃/" ([æ̃]), except in Québec where it's still [œ̃]. And the alleged /ɑ̃/ is pronounced that way only in Québec; elsewhere it is rounded and central for most people (let's write it [ɒ̃], but that's not quite accurate). Some even pronounce it as the very similar [ɔ̃], and I'm not sure if I haven't heard [õ] yet.

    It goes on. The traditional transcription already has [e] and [ɛ] as the same phoneme except in final syllables ([e] in open syllables, [ɛ] in closed ones); most people nowadays extend this to final syllables, so that's one fewer phonemic distinction. Similarly, the pronunciation of o is now completely predictable ([o] in open syllables, [ɔ] in closed ones, whether final or not), so claims that Nicolas has /ɔ/ are outdated at best; most people do still pronounce ô and au as [o] regardless of what follows, but some apply the rule even to these spellings, which means they have abandoned yet another phonemic distinction.

    English vowels for native speakers of German: yep, troublesome.

    First, it is very difficult to even get the idea that there might be a three-way distinction /ɒ ɔː ɵʊ/. Second, the RP /ɔː/ simply isn't [ɔ], it is very close to [o] – for a long time, I pronounced the English /ɔɪ/ as [oɪ] because the German /ɔɪ/ (eu) is quite different. Third, the spelling system of course screws it all up by actively misleading learners: /ɒ/ is spelled o, while /ɔː/ is written with digraphs that usually involve a, yet /ɒ/ is closer to [a] than /ɔː/ is.

    The English /ʌ/ is phonetically similar to identical to the German /ɐ/, but /ʌ/ is always stressed, while /c/ never is. The resulting conceptual block to equating the two is insurmountable. German-speaking learners first use [a] and then (if ever) use the right sound straight away. Even I, phonetics nerd that I am, find it remarkably difficult to pronounce a stressed or isolated [ɐ]. – Somehow, the French equate /ʌ/ with their unstressed vowel, which dictionaries write "/ə/" or even "[ə]" but which is actually [ø̞]*. The closest German equivalent is [œ], and that is used in Swiss Standard German for established English loans which wouldn't make the slightest amount of sense without the French intermediate.

    /æ/ is immediately equated with ä – but, see above, people are not used to using that sound in actual words, so they immediately substitute [ɛ]. Nowadays, most do soon learn better, but it takes a while. – Unlike the French, German speakers never get the idea of using [a].

    * I do agree it's a separate phoneme in French, however. Deux and de are not homophones, close though they are.

    ReplyDelete
  41. David Marjanović2 December 2010 at 02:20

    Part 3 of 3:

    Why are weak vowels not included? Imagine trying to speak English without schwa.

    By analogy with German, I suppose it's considered an allophone of something else. But of what? Perhaps /ʌ/?

    [...] Beet and spät rhyme

    Funny thing is, for me they don't. That's because spät undergoes the usual Austrian lenition of final /t/*, while Beet somehow doesn't. Perhaps because it isn't in my active vocabulary. ~:-|

    * This doesn't happen when it's preceded by a consonant. And yes, I'm implying there's a contrast between voiceless, unaspirated, unglottalized [d̥] and voiceless, unaspirated, unglottalized [t] that is only neutralized word-finally and then only behind vowels and in the opposite direction from the northern Auslautverhärtung. But I digress. :-)

    Standard French has 15 by my reckoning:

    That system actually has 16, because you forgot /i/. :-)

    No, 17, because you also forgot the diphthong oi [ŏɐ̯]. Yes, the traditional transcription interprets it as /wa/, but that's indefensible, IMNSHO.

    How about 18? No, I'm probably making an unholy diachronic mixture when I add the oin sound. Traditionally, and still in Québec, it's a sequence of /o/ and "/ɛ̃/", but for most people nowadays it's a unitary nasal diphthong, so that mois and moins are a minimal pair for nasality and nasality alone. Parisian French has arguably gained a phoneme while losing several others.

    (It also contrasts with "Redder", a kind of lane, though I'm only familiar with that one in placenames.)

    And I, a native speaker, didn't know it at all. No wonder! The dd tells you it's a Low German word. The High German consonant shift turned bb dd gg into pp tt ck.

    my Collins pocket dictionary from 2003 gives /e/ for Metall

    That's completely wrong. It has [ɛ]... if not "[ə]" (the stress is on the second syllable, as the ll shows).

    the effect of post-glacial changes upon communities led to development of syntax and erosion of morphology

    What? What do we know about glacial languages?

    languages with very high morphological complexity, such as those of northern North America, tend to have very small vowel inventories. Languages with large vowel inventories tend to be analytic or nearly so.

    Finnish, Hungarian?

    Does German count?

    How many vowel phonemes does Mandarin really have...?

    Why isn't tone (where phonologically distinctive) treated as a property of the vowel segment?

    Probably because it's not limited to the vowel – it extends to the voiced consonants of a syllable.

    ReplyDelete
  42. David Marjanović2 December 2010 at 02:24

    The English /ʌ/ is phonetically similar to identical to the German /ɐ/, but /ʌ/ is always stressed, while /c/ never is.

    Um. No palatal plosives were harmed in the making of this typo. An error must have occurred with Ctrl+C and Ctrl+V... /ɐ/ is never stressed.

    ReplyDelete
  43. @army1987 -
    vowel symbol between cardinals 4 and 5: see Barry and Trouvain in JIPA 2008

    ReplyDelete
  44. If you wanna get to 19, add maître [mɛːtʁ] mettre [mɛtʁ] to that. (That's what Wikipedia says, but for some reason my instinct tells the same, though I'm not a native speaker and I haven't spoken any French for years.)

    ReplyDelete
  45. David Marjanović: "The alleged /ɛ̃/ is [æ̃]" - yes, and so consistently that in Britain it's common knowledge even for the non-Francophone non-phonetician. Most people in this country know how to order a væ̃ blɑ̃. I never hear vɛ̃.

    Wjarek - I shall now continue to use ɑː in German with a clear conscience.

    ReplyDelete
  46. @army1987: You must not have been following this thread at all, as it has been repeated countless (ok, slight exeggeration) times that length distinction alone doesn't count.

    ReplyDelete
  47. David Marjanović4 December 2010 at 22:13

    If you wanna get to 19, add maître [mɛːtʁ] mettre [mɛtʁ] to that.

    Ah, yeah, the mythical past when the circumflex still indicated phonemically long vowels.

    I haven't encountered anybody who makes such a distinction; outside maybe of what's left of a few dialects, it's probably safe to say it's extinct.

    The use of ʁ in the same transcription is rather funny. The sound is not only relatively newfangled, it's rare; some speakers use it, but I haven't found any Parisian who does. [ʀ], [ʀ̥], [ʀ̥͡x], [x]*, [h] even**, yes, but never [ʁ]. Furthermore, nobody would keep it voiced behind [t]. I'd say [mɛtʀ̥] for both.

    * When the new tramway line 3 passes Porte d'Orléans, the taped voice says [ˌpo̞xːtø̞dɔʀleˈɒ̃].
    ** Like in Brazilian Portuguese.

    ReplyDelete
  48. David Marjanović: I wish I could figure out what the French rhotic - the one that I typically hear - really is. You can tell I'm not a phonetician. I don't usually hear a trill, but it doesn't sound like what I'd expect a voiced fricative to sound like either. And it's not the same as the German r (I mean the uvular but not trilled German r). The French r sounds awfully like a voiceless fricative to me, but I feel I must be wrong because it's never transcribed as one.

    ReplyDelete
  49. @Leo:

    Like you, I hear the French rhotic as a voiceless uvular fricative too, in all positions (even intervocalically). This would be the same development as in Brazilian Portuguese, of course.

    ReplyDelete
  50. "My favourite minimal pair is "Reeder" - "Räder"."
    The two words are homonyms in standard German. (Unless you want to stress the difference between the two words. Indeed, when I was reading your post, I was using /ɛː/ in Räder to pronounce it in my head, but after a moment of thought that this was only because my brain wanted to distinguish the two words; we don't like homonyms in German.)

    ReplyDelete

Note: only a member of this blog may post a comment.