Talk:Vowel length

From Wikipedia, the free encyclopedia

This article is part of WikiProject Phonetics, an attempt to build a comprehensive and detailed guide to phonetics and descriptive phonology on Wikipedia. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and see a list of open tasks.

Contents

[edit] Australian English

In Australian English it is vowel length (not in the traditional but the phonetic sense) which distinguishes such minimal pairs as the following. Note: phonetic symbols are in X-SAMPA

[6] VERSES [6:]

"pus" vs. "pass",
"come" vs. "calm",
"lust" vs. "last",
"cup" vs. "carp" and
"hut" vs. "heart"

[I] VERSES [I:]

"bid" vs. "beard",
"lid" vs. "leered" and
"Sirius" vs. "serious"

[e] VERSES [e:]

"head" vs. "haired",
"dead" vs. "dared" and
"ferry" vs. "fairy"

For more information on the vowels of Australian English refer to this Australian Monophthong Vowel Chartor the following Description of the Acoustic Characteristics of /hVd/ Vowels in Australian English. Note: the New Zealand accent is similar: New Zealand Monophthong Vowel Chart.

I'm moving this from the article page. The phonemic nature of vowel length in AusE and NZE is worth a mention but the fine details and lists of examples would probably be better on the Australian English page. Also the links don't work any more: see Talk: Australian English.

[edit] Vowel length and other features

An anonymous commenter has just added two paragraphs to the first section (renamed accordingly) of the article:

Vowel length is perceptively the most direct way of describing certain contrasting features of vowels. However, the concept of the perceived phonemic "length" of a vowel cannot be precisely defined. For example, if the tempo of speech is varied, the physical length must vary. The time duration of a vowel, as measured in seconds or milliseconds, cannot simply be the fundamental cue for distinguishing short vowels from long vowels. The distinction of phonemic long vowels vs. short vowels therefore must be derived secondarily from physical length and other phonological features.
Linguists have come to use alternative measures to describe the vowel-length distinction more precisely. Some of these measures include: Stress, Tenseness, Tongue-Root Position and Tone Contour, or the ratios of lengths with respect to other sounds [1]. Hence, we have the distinction of Stressed/Unstressed, Tense/Lax, Advanced-Tongue-Root (ATR)/Retracted-Tongue-Root (RTR), Leveled-Tone/Contour-Tone, etc. The features that distinguish long and short vowels in various languages (or dialects) may be quite different. Japanese long and short vowels, for instance, have little to do with English's long and short vowels.

I don't know that that's relevant. The two paragraphs tell us that (1) vowel length isn't absolute and (2) different languages distinguish pairs of similar vowels in different ways. But in the context of this article (established at least by implication in the introduction and the section below on English), only the actual distinction of time is relevant, and I think it's fairly obvious that it's a proportionate distinction that's relevant, but if we want to emphasise that, the word "proportionate" in the lead definition is probably more useful.

I think maybe the way to describe it is as a strawman; "the time duration of a vowel as measured in seconds" isn't what vowel length is, vowel length is "the time duration of a vowel as measured in proportion to other vowels of the speaker".

As for the rest, they do not apply to vowel length, but perhaps to binary distinctions of similar vowel-sounds. (Is there a category which contains distinctions like length/tenseness/ATR etc.?) I therefore think that the two paras should be removed. Can anyone explain why they shouldn't?

—Felix the Cassowary (ɑe hɪː jɐ) 02:20, 31 October 2005 (UTC)

Remove. I think the writer was confused by the use of the phrases 'long' and 'short vowel' in languages like English. In languages which actually have phonemic vowel length, it is a parameter independent from any of the others that were mentioned. As for your def, Felix, you're getting there, but I think the comparison should be with what could be expected from a vowel of different length in the same position of the same utterance - a rather hypothetical factor, I admit, but I don't think you're going to find an absolute measure. For example, vowels are shorter at the beginning of a prosodic unit than at the end, and of course vary by emotion, emphasis, whether the speaker's trying to get a word in edgewise or holding off an interuption, or trying to sound important, etc., so you can hardly go by speaker average - though averaging the vowels of adjacent words should be a good enough measuring stick, at least in a language where long vowels are 50% longer or more than short vowels. kwami 02:45, 31 October 2005 (UTC)
Not true. Average vowel duration is not useful for distinguishing short vowels from long vowels. Consider the expression BEAT IT! in English, especially when shouted. The duration of the final short vowel in IT can be several times longer than the duration of the long vowel in BEAT, yet no native speaker will ever confuse it with BIT EAT. Clearly, vowel length (time duration) is NOT the primary differentiator between BEAT IT and BIT EAT, in fact, it is a useless differentiator. We must recognize that native speakers rely on other phonetic features (tenseness and/or tongue-root position) as primary differentiators for aurally recognizing the vowels in these expressions. (preceding unsigned comment by 69.172.251.228 (talk • contribs) 15:33, 4 November 2005)
(Please sign your posts by leaving four tildes: ~~~~ if nothing else to help understand the timings of the posts easily.)
There's a few problems with your argument:
  1. Has anyone argued that "Beat it" and "bit eat" form pure length distinctions in English? I speak the dialect of native English which probably has the largest number of pure length distinctions (based on my understand of English dialects, and so I could be wrong). For me, there is a substantial qualitative difference between the long E of "beat" (which has a diphthongal quality of considerably variable duration) and a short I (which is a durationally short monophthong). It is of no use to use vowels of two different qualities to prove a distinction of length; you might as well use "good" and "guard" to prove your point for all the good "bit" and "beat" do. (And on a completely different note, I cannot imagine anyone saying "beat it", no matter how shouted, with a substantially longer second vowel than first—if I try, it sounds like "beat art".)
  2. "bit eat" doesn't make sense. Why would you shout at someone "bit eat"? If you shout "beat it" at someone, you're likely annoyed at them and they will know what you're about to say before you've even said half of it. If you're saying "bit eat", you're probably mad anyway. Redundancy would prohibit a misunderstanding. (Likewise, if you said to me (in an Australian accent): "I'm going to make my [beːd]", I wouldn't think you said that you're "going to make your bared", which doesn't make sense.)
Let's think instead of a distinction that is always considered to be primarily or exclusively length, and use sentences that are actually ambiguous. The only language I know sufficiently well that I can satisfy these two requirements is my own native dialect of Australian English, and the pair will be the STRUT vowel /a/ vs the BATH/START /a:/. (I apologise in advance for the profanity in the example but the context of shouting and so forth makes it seem like an obvious example.)
If I was angry at someone and wanted to deny them permission (or something), I might shout at them "YOU CAN'T", which I'd say as /jə kaːnt/, and the /a:/ would be substantially longer than the /n/. If however in my anger I contracted the /a:/ and/or lengthened the /n/ so that the difference in length between these two elements was much less, but I didn't change the quality of sounds at all, my audience would likely be highly offended (I would've called them a cunt—highlight that if you want to read it). Given how basic the distinction of length is to me and other Australians, I would also have a hard time trying to argue that it was an accident and they misunderstood me... Vice-versa it's similar; if I was trying to offend them, I must keep my vowel short and clipped because regardless of the quality (so long as its open, unrounded and central-to-back), if I lengthen it, I'll be misheard as saying "can't".
—Felix the Cassowary (ɑe hɪː jɐ) 13:35, 4 November 2005 (UTC)
You're talking about English, which is a language generally that DOES NOT DISTINGUISH vowel length as an independent phonemic feature. Whatever conclusions you draw will be necessarily incorrect, because they are based on a false premise ("English distinguishes vowel lenght"). As indicated in the article already, languages with two phonemic lenghts can have five different, distinctive phonemic lengths, which covers the allophonic variation. --Vuo 13:55, 4 November 2005 (UTC)
Previous comment was addressing the futility of using the concept of average length to distinguish long vowels from short vowels. What you mean by a language that distinguishes vowel lengths (like Japanese) actually distinguishes them by other phonetic features, just like English does not use vowel length as primary differentiator. In the case of Japanese, the primary differentiator is the tone/pitch contour, which goes undetected by speakers of non-tonal languages. Speakers of non-tonal languages only realize that the long vowel is longer, but do not realize that the tone/pitch contour is different. Long vowels are produced with contour tone/pitch. (This difference is detected subconsciously by human speakers or intentionally detected in speech recognition software.) I don't know Finnish, but I would venture it is the same case. Duration is never the primary differentiator for distinguishing short vowels from long vowels. Other phonetic qualities are. Please find out what is the real feature that works in the Finnish case (Google may help). My first guess would be tone/pitch contour. ~~~~
Good news. Finnish long/short vowel distinction indeed is based on pitch contour, as I have been able to personally verify from the sound files at: http://donnerwetter.kielikeskus.helsinki.fi/FinnishForForeigners/ch1-en/ch1-gr-vokaalit.htm and http://www.ddg.com/LIS/InfoDesignF97/paivir/finnish/pronunc.html, including minimum-pair tuples like taka 'back' (prefix), takaa 'from behind', takka 'fireplace' and taakka 'burden'. The pitch contours are all dramatically different. Case closed. ~~~~
It appears that you've confused accent with vowel length. In Finnish, tonal variation is used for emphasis, but not for anything else, as in the case of those sound recordings. They are deliberately pronounced in this "tonal" way. This does not reflect normal speech. Manipulating the vowel sound in a recording produces a reliable, logarithmic response curve w.r.t. to length. --Vuo 13:47, 10 November 2005 (UTC)

Just because there's a difference in pitch contour doesn't mean it's the primary difference in perception. You'd need to get a Finn to listen to a recording with the wrong length–pitch contour relationship, and see what they thought of it. (As for the recordings, I can't work out how to get them to play.) I suggest that the case is indeed not closed. (And in any case, Finnish and Japanese aren't the only languages that have length distinctions, so you'd have to say something about all the many others that linguists have generally described as having a primary distinction of length.)

If you have any references to published literature that supports your claim that no language has a primary distinction of length, I would be interested to read them.

BTW, Just to clarify: In the example link I gave you for signing your post, you weren't meant to put the "<nowiki>" tags around it, it's just the four tildes. (I put in the nowiki tag so that it wouldn't convert it from four tildes to my signature.)

—Felix the Cassowary (ɑe hɪː jɐ) 06:57, 5 November 2005 (UTC)

Reference literature in cutting-edge development is always hard to come by. Part of the fault is the Western European heritage of being oblivious or plainly ignorant on tonal/pitch features. Hopefully this situation will change soon enough with Chinese becoming a more popularly taught language. Here is an article on Finnish that can help to elucidate what I am talking about. See http://www.ling.lu.se/disseminations/pdf/49/bidrag39.pdf, or for a fuller version http://www.mediateam.oulu.fi/publications/pdf/364.pdf. Take the short version for instance. Everything becomes clear in Fig. 1 and Fig. 2. Pay attention to the syllabic division (between points 3&4 in Fig. 1, between points 5&6 in Fig. 2.) The author (Suomi) has shown that Finnish words with CVCV structure and CVVCV structure actually follow quite similar word-wise pitch pattern. However, the first vowel, if short, has a straight rising slope in the basic frequency F0 (that is, it has a rising pitch/tone). On the other hand, if the first vowel is long, it follows a rising+falling pitch/tone contour. Aurally, human ears respond to spectral variations (frequency variations) very efficiently. The recognition of long vs. short vowels is no doubt based on this spectral difference. [[~~~~]] 18:47, 5 November 2005 (UTC)
There is evidence [1] [2] that although other prosodic cues are concomitant with vowel length, reliable distinction is retained even if the physical length is digitally manipulated. However, what is still missing from the article are the observations that not length per se, but its ratio to the tempo of speech is the primary cue [3]. Furthermore, on the significance of physical length, the Finnish timing system uses length to denote accent and stress, and as such, the timing system is very complex and doesn't follow syllable-timed, stress-timed nor mora-timed patterns. [4] --Vuo 12:45, 18 November 2005 (UTC)

--Vuo 12:45, 18 November 2005 (UTC)

A paper on Japanese's case can be found at: http://www.asel.udel.edu/icslp/cdrom/vol4/780/a780.pdf, where it is found that syllabic duration alone is useless for native Japanese speakers to detect a long vowel in a multi-syllabic word. Those that identified long vowels from duration alone were all non-native Japanese speakers. For native Japanese speakers, the moraic structure is detected from the pitch change, not from the duration. [[~~~~]] 00:54, 6 November 2005 (UTC)

Surely in Thai, vowel length is quite independent of vowel quality. There are abundant minimal pairs where vowel length is the only difference. Unless you can show explicit research into Thai that shows this to be untrue, I think we should consider the general position stated at the beginning of this section as unfounded. Of course there are languages where length of vowel is only phonemic in combination with changes in other qualities. That may have confused the writer. −Woodstone 08:54, 10 November 2005 (UTC)

[edit] Syllabic lengthening & NPOVness

I think that the "short vowels in syllabic lengthening" section is NPOV and/or inaccurate. It makes a two claims:

  • that in quantity-based languages, a short vowel lengthened to the length of a long vowel can be distinguished from a long vowel. This is patently false, or at best very POV.
  • that Germanic languages are "tenseness assisted" languages. This presumes a particular definition of the tense vs lax opposition that is certainly not common. See the comparison between tense and lax vowels discussion on the Tenseness talk page for some objections.
  • (further, its discussion of Vietnamese is probably off-topic and would be better done on the tenseness page, I think. In any case, sources are desirable.)

I think the section should probably be removed.

—Felix the Cassowary (ɑe hɪː jɐ) 09:53, 16 November 2005 (UTC)

Agreed. The fact that Finnish was added initially is a sign of making up facts to conform to a theory. In Finnish songs, the distinction is simply neutralized. For example, the line maa kallis isien is sung maa kallis iisiien in the Finnish national anthem (which, curiously enough, was not originally Finnish). It needs to be derived from context that [iisiien] doesn't derive from /iisi/ ← English easy, but from isien "of the fathers". --Vuo 12:45, 18 November 2005 (UTC)
I just deleted it. —Felix the Cassowary (ɑe hɪː jɐ) 09:53, 19 November 2005 (UTC)

[edit] Australian English Phonemic vowel length

I have disputed that 'bid' and 'beard' are distinguishable only by vowel lengths. For me 'beard' definitely has a schwa before the final 'd' (and contrasts with 'bed' vs 'bared', where I agree the vowel length alone makes the phonemic distinction).

I don't know if I'd say either the schwa-full form or the schwa-less form are more common in speech, and my reading of the literature suggests that the schwa-less form definitely becoming more common, and may well be the norm amongst at least younger Sydneysiders by now. I've rephrased the intro so it's clear not everyone will always have all of these distinctions. [Also, you may well find that in an uncontrived utterence in casual conversation when you're not paying attention even you will frequently use the monophthongal form, tho this too is age-graded.] (—Felix the Cassowary)
Hmm, the schwa is definitely there if I speak quite casually and quickly.
Perhaps your speech is more typical of an older generation or a foreign dialect (consider your lust/last issue). If you wish to pare the list down to say shed/shared and hut/heart (the two most widely accepted) and direct the interested reader to the Australian English phonology page, I will not object (actually, sounds like a good idea to me). (—Felix the Cassowary)

Also, 'lust' and 'last' have quite different vowels, as in most other English dialects. Whether this is typical of General Australian English, I'm not entirely sure, as my own dialect potentially has some UK English influence from my parents.

I have never once heard the pronunciation /lʊst/ for "lust", except maybe amongst people who have "put" and "putt" as homophones. Nevermind, the correct thing to do there is use another word, because you don't pronounce "hut" (vs heart) as /hʊt/ (unless you do, in which case your speech is very much not General Australian English or your IPA is wrong). No need to dispute a technicality if the fact that's trying to be illustrated is true.
Also I should add that I've checked the pronunciation of an American dialect, and it gives "lust" as the American equivalent of /last/ (i.e. /lʌst/), and never even mentions an alternative. If the British and American pronunciations are different, this particular dictionary will usually mention them, so I think your claim of "most" is somewhat dubious.
—Felix the Cassowary 22:35, 2 May 2006 (UTC)
You're right, /ʊ/ is wrong, for me I believe it is /ʌ/. There's no question 'lust' and 'hut' have a quite different vowel to 'last' and 'heart', for me, and for several work colleaugues that I quizzed: the mouth shape is even quite visibly different. It's true the vowel in 'last' is temporally longer, but if I stretch out 'lust' to the length of the vowel in 'last', it still doesn't sound like 'last'. But the article now just says "the following are minimal pairs for many speakers", which isn't really saying much: how much is 'many', and how typical is it of most Australian speakers?
Evidently you must not be speaking General Australian English then. Research on the pronunciation of Australian English has shown since at least the 1960s that the Australian English vowel in "hut" is low and central and short; and the Australian English vowel in "heart" is low and central and long. The fact that this pair is distinguished solely by length is in no way remarkable in Australian English, because it was the first pair (the only pair in 1960s research before then [eə] was monophthongised) ... well, possibly except for the æ(ː) opposition which is frequently overlooked perhaps partly because it was the actual first pair where length was unpredictable from quality.
Except like I said it's not just me: the work colleauges I quizzed had the same distinction. I'm by no means an "older" generation speaker, and the UK english influence from my father is minimal. I would consider the vowels I use in 'hut' and 'lust' to be reasonably central, but I widen my mouth considerably more when pronouncing 'heart' and 'last'. I also found theses links: http://classweb.gmu.edu/accent/english3.html & http://classweb.gmu.edu/accent/english77.html, which transcribe the 'a' in ask differently to the 'o' in brother, which, as far as I can tell, exactly correspond to the two vowel sounds in question.
English 77 provides no IPA transcription (at least none I can see). The transcription of English 3 reflects no native accent if it's trying to use the IPA in a narrow (phonetic) way; the choice of vowel glyph looks symbolic to me, with length marks added/subtracted and the tacks included perhaps relative to her normal pronunciation. (e.g. [bluː] rather than [blʉ:], [snəʊ] rather than [snəʉ]/[snɔʉ], the same glyph (o) is used for the vowels /ɔ/, /oː/ and /əʉ/ in the words "frog" and "also", "station" is transcribed as [steʃɪn], which most Aussies would interpret as "stesh in" or "steshing".) If the vowel transcriptions are accurate, this speaker would have a strong accent. Of course, as they don't provide a recording I can't say which with any accuracy.
Consult the sources referenced in or linked to from Australian English phonology. The diagram in [5] is particularly useful, and (in the placement of the "hard" and "hud" vowels) is reflective of a number of different studies since the late 1960s (look at the second graph even!), as well as when I've looked at my own vowels. (Note also that the only purely length distinction in New Zealand English is "hard" and "hud", which doesn't mean anything but from it I'd hypothesise that the two vowels have been pairs distinguished only by length for a long time.)
I cannot hope to know why you and your colleagues diverge from what seems to be the norm for Australian English. I've never once seen anything that indicates anything different, and I've never heard anything different. Though if the distinction is between say a long low central vowel and a short near-low or low-mid central vowel, I couldn't hope to hear the difference conversationally; I have low-mid, near-low and low in free variation for all four of /æ(:)/ and /a(:)/. (This might be a part of it, actually...)
—Felix the Cassowary 13:17, 4 May 2006 (UTC)
Do you have anywhere an audio sample of any native Australian speaker using the same vowel (length aside) for 'lust' and 'last'? The audio samples I provided to me do, as far as I can tell, definitely use a different vowel for 'brother' and 'ask': there are more samples at that site, and only one (the Sydney one) appears to use at all similar sounds, but of course I can't necessarily extrapolate that 'brother' does correspond to 'lust', and 'ask' to 'last' for most AusE speakers. I'm happy to provide a recording of my own voice, from which you can judge whether my accent is indeed General Austalian.
Where's the samples on that site? There's no navigation (I'm guessing they're meant to be in a frameset?) so I don't know where I'm meant to look.
You're right in that "brother" and "lust" ought to have the same vowels, as should "ask" and "last". Very occasionally an Australian however will use a vowel more like [��] or [æː] in "ask" and "last" so they sound different from how "arsk" and "larst" would sound (like castle vs carstle). I've heard this from maybe two speakers in my life. I've heard some people claim it's common in Brisbane but I've never been there, and no Queenslanders I've heard've done it. I've never seen it comment on in studies of Australian accents.
The only recordings I've got are my own which is not very independent but I could still give them to you if you wanted... I'm happy to trust you that you say the words the way you say you say them. If you say them differently, it means that in this regard you don't speak General Australian (as it's been documented) and it could be you're using a different sociolect (e.g. Cultivated Australian) or the beginnings of a new regional accent... (Where are you from? I'm Melburnian. BTW, could you also actually sign your posts by typing four tildes? Puts down your username and the time, makes it convenient for tracking conversations.)
—Felix the Cassowary 02:05, 5 May 2006 (UTC)
"Many" ranges from "without exception known to me" for hut/heart to "excluding only older speakers" for shed/shared to "contextually and age dependent" for bid/beard. The span/span distinction is poorly studied, and there's also full/fool which is regional (and age dependent). All of these are typical within my generation if personal experience is a guide, and no study to my knowledge based on the phonetics done recently has lacked the first two. If you prefer, if we reduce it to hut and shed, we can say "most".
—Felix the Cassowary 00:46, 3 May 2006 (UTC)

I've also added 'can' (able to) vs 'can' (tin can): of course in many cases the verb 'can' is unstressed and gets a schwa, but in the sentence "He's trying to open the can: do you think he can?" the former is definitely a longer vowel length than the latter. This is discussed at http://en.wikipedia.org/wiki/Phonological_history_of_English_short_A#Bad-lad_split

The article currently describes Australian English as having [a] in 'hat', which is clearly incorrect. Are there any examples of pure length contrast involving /a:/ in AusEng, or should that line be deleted? Makerowner 02:48, 13 August 2007 (UTC)

[edit] Examples

I think this article would be much easier to understand to non-linguists with some sound files. As a speaker of a language without vowel length distinctions (as American English is, right?) it's hard to imagine vowel length. Could someone with a good understanding of the concept add some sound files? 24.196.98.61 07:06, 21 July 2006 (UTC)

It is honestly not a very exciting thing to hear, but I've uploaded some examples sounds to the commons, with some words in isolation and a sentence containing these words. The recordings are of low quality, so I don't think I'll put them into the article, but it's good enough for curiosity. —Felix the Cassowary 10:30, 21 July 2006 (UTC)
As a native speaker of a language that always differentiates vowel length, I have to criticize that "merry" /meəri/ and "Mary" /mæri/ are not minimal pairs by vowel length, but the rest are. --Vuo 17:52, 21 July 2006 (UTC)
As it happens, I pronounced "merry" and "Mary" backwards in the isolated recording (when I thought of the pair and wrote them down, I treated it as a noun phrase, rather than two words in isolation), so that whereas with the others the long vowel was first, in "merry Mary" the short vowel was first. So I will assume that means your transcriptions are actually "merry" /mæri/ and "Mary" /meəri/. But I can assure you that the vowel in "merry" is not /æ/, which I contrast in for instance "merry" vs "marry" (vs frequently [æː] in "Māori"). Perhaps the fact that the stress is a little different between "merry Mary" (as I said, I've treated it as a noun phrase) caused a bit of undershoot in the unstressed "merry", or something ... in any case, the stressed vowels of "merry, bed, shed" on the one hand and those of "Mary, bared, shared" sound the same to me, but a non-native's ears will always help in distinguishing between contrastive and non-contrastive variation! —Felix the Cassowary 11:09, π Approximation Day 2006 (UTC).
Thanks. That honestly made it a lot easier to understand. 24.196.98.61 18:57, 21 July 2006 (UTC)
No worries. —Felix the Cassowary 11:10, 22 July 2006 (UTC)

[edit] Misleading intro

The following paragraph currently in the introduction is misleading

English does indeed rely on vowel length. Listening to German speakers speak English makes this very apparent. They say "pot" and "pod" with the same vowel length (also "kit / kid", "lab / lap"). A native English speaker uses a longer vowel with words ending in a voiced consonant. (For example, pick/pig, cease / seize etc.)

While this may be described as a length distinction, it's clearly allophonic. When most people describe a language as having "a length distinction", they are thinking of phonemes. FilipeS 18:57, 15 November 2006 (UTC)

[edit] Japanese phonemes

(Moved from User talk:TAKASUGI Shinji#Vowel length)

You [TAKASUGI Shinji] wrote:

Japanese has never had /wu/; some consider the intervocalic /ɸ/ was voiced, which became [w] later; p>ɸ may have occurred before o and ö merged

You missed the point. It is precisely because Japanese never had */wu/ that */wu/ automatically becomes [u]. If there was a /wu/, then 今日 would have stopped at kewu and never developed into kyo:. As for 通る, /wo/ did exist, but it soon became [o] and hence toworu > tooru > to:ru. The change from keɸu > keu is phonetically implausible without the -w- stage. ɸ > w is the standard development and is needed to account for the change.

Also, stating "/h/ (originally /p/)" is redundant. A single phoneme, which needs a name, is realized as *[p], [ɸ], [w], or [h] depending on the environment. As a historical Japanese linguist, I am fine calling this /p/, but then it becomes difficult to talk about modern ハ・ヒ・フ・ヘ・ホ and then even harder to contrast it with パ・ピ・プ・ペ・ポ. 123.224.226.72 (talk) 11:32, 3 February 2008 (UTC)

A few other notes. /h/ was certainly once realized as [p]; however, there is general agreement that during the OJ stage (Nara period) it was [ɸ]. Thus, we talk about *[p] (note the *) for proto-Japanese. Also, regarding what you call "o and ö" (more specifically 上代特殊仮名遣). The original theory was an 8 vowel system, but since the 70s that has been mostly abandoned by the academic community in favor with a glide theory. However, to be fair in absence of a consensus, we are using subscript 1 and 2 here. 123.224.226.72 (talk) 11:52, 3 February 2008 (UTC)
Sorry, one more note. Regarding /wu/. Morphologically the phoneme did in fact exist. However, it was never phonetically realized as *[wu] but as [u]. Consider the verbs suw- (据う), uw- (植う) and uw- (飢う). Pay particular attention to how the conjugate. Here is the first one:
suw-: suwe, suwe, suu > *suwu, suuru > *suwuru, suure > *suwure, suwe
123.224.226.72 (talk) 12:00, 3 February 2008 (UTC)
Hi.
You missed the point. It is precisely because Japanese never had */wu/ that */wu/ automatically becomes [u].
You are confusing generative phonology and historical linguistics, perhaps? In generative phonology, you can and should write /wu/ > /u/ to explain inflection, but you cannot do so in historical linguistics, because /wu/ has never existed as a surface form in Japanese.
"/h/ (originally /p/)" is redundant
Yes, I know. It should be called /p/. The modern /p/ is a direct discendant of the ancient /p/, and you don't need to distinguish them. I just kept /h/ intact for your sake.
Also, regarding what you call "o and ö" (more specifically 上代特殊仮名遣). The original theory was an 8 vowel system, but since the 70s that has been mostly abandoned by the academic community in favor with a glide theory.
Again, I know that. But you can't use subscripts in the edit summary line. - TAKASUGI Shinji (talk) 12:20, 3 February 2008 (UTC)