It depends on their status in the syllable: the sound at the beginning of "yes" is almost the same as the sound in the middle of "keep", but the one in "yes" is shorter and less prominent [j] as opposed to [i:].

But there are very short vowels, too, f.e. the "i" in horror film. And the nucleus in the syllable can be a consonant, f.e. within the second sylllable of "syllable".

Maybe the answer is in the distinction between phonetics and phonology: *phonetically* (e.g. according to an audio analysis of that particular fragment of sound out of context) the "i" in horror film could be practically identical to an approximant [j], but *phonologically* someone who knows English will hear one (in context) as a nucleus and the other not (and someone more familiar with another language might hear the same thing in a different way).

Btw I'm not sure that an approximant in English would really be less closed than an /i/ or /u/ vowel... Although phonetics/phonology is not my specialty.

I think the terms "vocoid" and "contoid" can help in illustrating this distinction. So from a phonetic point of view [i] can be seen as a vocoid (syllabic, nucleus) in the English word "seat". However, when you try to analyse the second element of the diphthong /ei/ in the word "say" , we may say that /i/ here is a contoid, i.e. [j]. So vocoids/contoids are phonetic while vowels/consonats are phonology.

But if we make a distinction between vocoid and contoid as phonetical terms there must be a phonetical (material) difference between vocoid and contoid.

I think both /i/ and /j/ are realized by a vocoid (i.e. non-obstruent) sound. "A vocoid is a sound made with an open oral cavity such that there is little audible friction in the mouth." (http://www.sil.org/linguistics/GlossaryOfLinguisticTerms/WhatIsAVocoid.htm). As Abdullah says, vocoid is a phonetic term, and the phonetic quality of pronunciations of /i/ and /j/ is pretty much the same. However, the glossary page I just cited goes on to say that a vocoid "has the potential to be analyzed phonemically as a vowel". That is, the vocoid sound *may be* interpreted by the hearer as a vowel /i/. But if it is slightly shorter, slightly quieter etc *in relation to other neighbouring sounds* it may instead be interpreted as a consonant (e.g. in "yes").

a vocoid "has the potential to be analyzed phonemically as a vowel".

But the glossary page you cited also says that [l] is a vocoid. Could [l] be analyzed as a vowel? I don't believe that. So the glossary seems to be not very reliable.

The glossary is talking about *syllabic* [l'], which I suppose could be "analyzed" (by a non-linguist, i.e. heard) as /əl/ (schwa + l).

The introduction of this Wikipedia entry teases out the intersections/conflict between the phonetic and phonological terms bit:

http://en.wikipedia.org/wiki/Vowel

1. I'm not sure that something can be "heard" or "analyzed" which does not occur ([ə]). (There are some fantasies in phonologist's ways to argue.)

2. What about the [ l̩] in Plzeň (Czech pronunciation: [ˈpl̩.zɛɲ])?

Just fyi, i consider ``r'' as an occasional vowel also - herd, bird, word, burn - it is clear to me that the ``vowel'' in each of these words is irrelevant, and that the ``r'' is the actual vowel, which is also true for the island of krk in the adriatic (and in other slavic words of the region)

The difference between vocoids ([i, u] and also [ɪ, e; y; ʏ, ø; ɨ; ʊ, o]) and contoids ([j, w] and also respectively [j̞; ɥ; ɥ̞; j̈; w̞]) is in the stronger dynamic feature of the second ones which imply that the first ones, namely the vocoids, are very much more likely to receive the syllable prominence. The prominence is a combination of volume, pitch and (relative) duration giving the intensity which begets the syllable nucleus. As Klaus Schuricht pointed out [l] isn't a vowel. Nevertheless, non-obstruent contoids are capable to receive prominence and constitute a syllable nucleus. Take care that [j, j̞; ɥ, ɥ̞; j̈; w, w̞] and other possible approximants or semi-approximants (ie approximants with very slack articulation) are not shorter than other contoids, nor they're shorter than «(relatively) normal-length» vocoids! For example, a transcription like [ja], when reliable, represents something having about the same duration as either [ça] or [ca]; and [j] has about the same total duration as the total duration of [ç] and [c], which is also about the same duration of the [a].

Actually, very rarely, stops also may receive certain prominence, as in the possible English [ˈk̩˺kju̘] «thank you».

The distinction between vocoids and contoids isn't a matter of phonemic interpretation but it's in fact a matter of different way the sound is produced. The area in which the tong produces contoids overlaps with that where vocoids are produced. Semi-approximants are slacker (then lower) than the highest vocoids [i, ɨ, u], but still, their dynamic component (contrasting with the relative staticity of vocoids) avoids prominence, making them recognisably contoids. Obviously, an intense [j], ie [j̩] is actually [i]. Moreover there is no reason, from a phonetic point of view, to consider the second element of a diphthong being a contoid. A reliably transcribed [ei] is in fact a sequence of [e] + [i]. If it was [eɪ] it was a sequence of [e] + [ɪ] and [ee̝] a sequence of [e] and [e̝], the very same way [eɛ] would be a sequence of [e] + [ɛ] or [eə] a sequence of [e] and [ə]. We can even find sequences of the very same timbre: [ee] = [e] + [e] (still different from [eː]). In all these examples, [i, ɪ, e̝, e, ɛ, ə] are certainly vocoids. In some languages (like Arabic) we might find sequences of a vocoid and a contoid: [ej]. In Swedish as well we might find those sequences, even though the contoids are often semi-contrictives (ie «semi-affricates»), instead of approximants (so the second element should be transcribed either [j̝] or [ç̞]).

"The distinction between vocoids and contoids isn't a matter of phonemic interpretation". But the difference between "vowel" and "consonant" is phonemic, right?

The difference between contoids and vocoids is a phonetic (the different ways the sound are produced) but, yes David, it obviously also ends having a phonemic implication, because «vowels» contrast both with themselves and with «consonants» at the phonemic level.

I forgot:

«1. I'm not sure that something can be "heard" or "analyzed" which does not occur ([ə]). (There are some fantasies in phonologist's ways to argue.)» by Klaus Schuricht, Feb 6, 2013, modified Feb 13, 2013 by the author

Untrained ears can hear (or hear not!) whatever they want/believe, including vocoids which aren't actually pronounced.

Just fyi, Doris Mücke provided evidence that there is no accoustic difference (neither qualitative nor quantitative) between [ ja ] (the German word ja ‚yes’) and [ ai ̯] (the German word Ei ‚egg’) – except that they are mirrored -, and that between [ j ] = [ i ̯ ] and [ i ] there is only a quantitative difference. So ja and Ei – both are diphthongs - can be transcribed in 4 ways:

(1) [ ja ] and [ ai ̯ ] ; (2) [ i ̯a ] and [ aj ]; (3) [ i ̯a ] and [ ai ̯ ]; (4) [ ja ] and [ aj ].

(http://www.uni-koeln.de/phil-fak/phonetik/Institut/Mitarbeiter/dmuecke/muecke/publ-talks/Muecke_Glides_Akustik.pdf)

It is worth noting that a language / languages such as Arabic write the vocoids [i:] /[u:] and the contoids [j]/ [w] with the same graphemes, i.e. and respectively. For both [i:] and [j], where the grapheme is used, the allographs {يـ} and {ـي} are used when the sound is non-syllabic, i.e. word-intial and word-final. For syllabic , the allograph {ـيـ} is used. Similarly, [u:]/[w] are written with with the allographs {و} for initial position and {ـو} for both syllabic and final positions.

Another author: Catford wrote: “The vowel [i], you will recall, is a typical approximant, and the same is true of [u]. The semivowels are similar approximants, except that they are ultra-short.” (Catford, J.C., 1990. A Practical Introduction to Phonetics. Oxford: Clarendon Press, p. 72)

/I/ and /u/ are two approximant phonemes, which distinguish: "key" and "coo" or "seep" and "soup" and many other examples where the two function as vowel phonemes. Each can be a phonetic "offglide" in words like "my' and "cow," where the are NOT stressed and make no phonemic contribution to the phonemic status of these final diphthongs and are the "non-syllabic" portions of the /aI/ in "my" and the /aU/ of "cow." In the words "hoot" and "heat" the voiceless [u] is an allophone of /h/ and in "heat," the voiceless [I] is another allophone of /h/. So, when you as this question, you must expect several answsers as to the FUNCTION of each in these different environnments. Furthermore, John O'hala uses these two glides to knock down the strong version of the sonority principle since both have the same sonority markings on the hierarchy, but that does NOT predict, therefore, that the word "usual" is unmarked for the sequence /yu-/ and the word "weep" is unmarked for the sequence /wi-/ Occurrences of /yi- and /wu-/ are rare indeed. But, the two glides have the same markings in sonority hierarchical modeling. So, /u/ and /I/ are cool

Dear Mr. Schuricht, unfortunately I can't speak German thus «unfortunately» I can't read Doris Mücke's paper. In any case it is not my fault not to be able to distinguish between a vocoid and a contoid. Apparently she found such a regional speaker who recorded [ai] or even [aj] for neutral German [ae]. But the difference between [Vi] and [Vj] (or between [Vu] and [Vw]) still remains clear to me. It's like you pretended there is no difference, say, between [Vɫ̩] and [Vɫ]. Or, inversely, it's like you pretended there is no difference between [iV] and [jV] or [uV] and [wV]. Even if one's ear is untrained to such differences, one could easily infer theoretically the difference between a double nuclear syllable and a syllable consisting of a nucleus and a consonant. I don't find it too hard to get and I stop it here to avoid falling into explanatory tautology.

(Catford was a great Linguist but as for D. Jones, some of their theories are nowadays dated.)

Oh, thanks to Mr. Buckingham to mention [ji] and [wu]: it's another practical example to prove [i, u] and [j, w] are different «things». The sequences [ji, wu] are quite rare in Europe/Western languages but they're quite frequent in Asia. In Czech [ji] is frequent, even among highly used words (diachronically derived from [ju]).

It's also true that [hV] is kind the same as [V̥V], i.e. a devoiced realisation of a vocoid having the same articulation point in the vowel space as the following vowel. Using the symbol [h] is better because it is not part of the syllable nucleus.

Yes Mr. Al-Saqqaf, Arabic languages uses the same graphemes (the consonantal = «y» and = «w») also where long vowels and diphtongs are to be found; nevertheless that has nothing to do really with their actual realisations. Neutral Arabic long V are just [Vː]; neutral Arabic diphthongs can receive various realisations: long vowels [Vː] (in fact [eː, oː]), diphtongs [VV] or sequences of a vowel and an approximant or even a contrictive contoid. Let's also consider French which sometimes practically reaches [ˈwiç].

Question about my answer to the niceness of "we" and "you" but not to the Asian frequency of the name "Woo.":

How did you like my boogie boarding picture? Hugh Buckingham

To Hugh: What is a boogie boarding picture? -

To what you wrote before:

I understand that /I/ and /u/ have two funktions: 1. vowel; 2. glide. I know this way to solve the problems of representing the phonetic phenomena [i] and [u]. As they can function as glides aren’t they identic with [j] and [w]? And therefore: Isn’t it arbitrary wether in phonetics we put them into the vowel or into the consonant system? Isn’t it a artificial complication to have double symbols: [i] and [j] resp. [u] and [w].

If the most important phonetical criterion is that vowels are more open than consonants I think we should put [i] and [u] into the consonant system (like, f.e., [l], which can function as a vowel, too), because they are very much closed, especially [i].

Furthermore: You wrote

„In the words "hoot" and "heat" the voiceless [u] is an allophone of /h/ and in "heat," the voiceless [I] is another allophone of /h/.“

Would you say the same if there were no written form of English at all?

And you wrote

„but that does NOT predict, therefore, that the word "usual" is unmarked for the sequence /yu-/ and the word "weep" is unmarked for the sequence /wi-/“

What do you mean with „unmarked“?

Excuse me for all these questions. Klaus

To Jákup:

I doubt that "[i, u] and [j, w] are different «things»". I understood from Hugh that they are the same "things" in different functions. So why shouldn't there be diphthongs like [ji, wu]? One can write them as [ i ̯ i, u ̯ u ], too.

A diphthong is a sequence of two vocoids. [ji, wu] aren't diphthongs; this is luckily true for the English orthography as well. [ˈjɪist] (but [ˈji̞i̝st] looks less ambiguous as I'm using official IPA) , exactly «CVVCC»: [ˈjɪ-] (or [ˈji̞-]) isn't a diphthong, [ˈ-ɪi-] (or [ˈ-i̞i̝-]) certainly is. And yes, you might transcribe [i̯, u̯] to show their relation with [i, u] the same way you may transcribe [V̯̊] for [h] or possibly even [b̃, d̃, ɡ̃] for [m, n, ŋ]... the non-syllabic diacritic makes [i̯, u̯] mean contoids.

One can "doubt that "[i, u] and [j, w] are different «things»"" or one can collect some recording, from various languages, and listen carefully to all the realisations can be found, and then make his own idea.

Diphthongs in standard Albanian language:

Descendant diphthongs áu, éu, íu (when the accented vowel precedes the semivowel ‘u’ or ‘i’) – as in words such as: káu ‘(the) ox’, théu ‘(he/she) broke’, aríu ‘(the) bear’, fshíu ‘(he/she) wiped’, ftói ‘(he/she) invited”, punói ‘(he/she) worked’.

Ascendant diphthongs uá, uó, iá (when the accented vowel follows the semivowel ‘u’ or ‘i’) – as in words such as: kuádër ‘frame’, kuárc ‘quartz’, kuótë ‘quota’, piáno ‘piano’.

Dear Professor Caka,

I would transcribe alb. "kau" by [kaw] a.s.o., and alb. "ftoi" by [ftɔj] a.s.o.

(It's a pity that we missed each other during the conference. I was ill.)

In full style of pronunciation, combinations of unstressed vowel ‘i’ and ‘u’ with an another vowel are pronounced generally as group of vowels, not as diphthongs: ftoi ‘(the) quince’, kau ‘(the) ox’, dheu ‘(the) soil ’, miu ‘(the) mouse’, but combination of pronominal clitic (short form of personal pronoun) ‘i’ with an other vowel (ia dha ‘he gave it to him’, ia thashë ‘I said that to him’, iu afrua ‘he approached to him’ etc.) are pronounced as diphthongs: ja dha, ja thashë, ju afrua. (j = i̯ )

(Hopefully we’ll meet up next time.)

The Albanian language doesn’t have the phoneme /w/, therefore the English words (generally proper names) with grapheme are transcribed in Albanian with the grapheme or with the grapheme : Uashington ‘Washington’, Uells ‘Wells”, Uilson ‘Wilson’, uiski ‘whisky/whiskey’; Ajova ‘Iowa’, Havai ‘Hawaii’, Holivud ‘Hollywood’, tramvaj ‘tramway’, vat ‘watt’, vaterpolo ‘water-polo’, Vullf ‘Woolf’, etc.

I have long thought that in english, 'r' is actually a vowel in many uses: the words jerk, dirt, work, burn all have exactly the same sound, so the written vowel (e, i, o, u) is clearly irrelevant - then there is the adriatic island krk which makes it more explicit

The Serbian/Croatian/Bosnian liquid [r] can be syllabic, playing the role of a vowel, as for example in the words: brk (moustache), brz (fast), crn (black), crv (worm), grd (ugly), Grk (Greek), grm (bush), Krk (a Croatian island), krst (cross), krv (blood), mrk (dark, brown), mrš (go away), prst (finger), rt (cape), trg (square), trn (thorn), tvrd (hard, solid), Trst (city of Trieste in Italy), skrb (care), smrt (death), srž (marrow, essence), strm (steep), škrt (stingy), vrh (peak), vrt (garden), zvrk (spinning top).

In the standard Serbian/Croatian/Bosnian language, there is also a syllabic /l/, which appears only in loanwords: bicikl (bicycle, bike) debakl (debacle), Vltava (Czech river).

There are also «unexpected» instances like /pəˈlɪis/ [pəˈlɪ̘is] (or [-ˈi̞i̝-]), also [pl̩ˈɪis, -ˈi̞i̝-], and /ˈvaɔəl/ [ˈväɔ̝̽ɯ̞ɫ] also [-(ɯ̞)ɫ̩] (to say nothing about [ˈvɑ̘ˑɯ̞ɫ]).

Would you ever say that [l̩, ɫ̩] vocoids? Because of their phonemic use, you might list them among vowels but from the articulatory point of view they're «intense» (or syllabic, or «nuclear») contoids. The word krk exists also in Czech (and it means 'throat', although the name of the island means 'church' - completely different etymology), which I feel much more comfortable with, and it's /ˈkɾ̩k/ [ˈkɾ̩k]; the [ɾ̩] receives the highest syllabic intensity of this monosyllabic word which is formed by three contoids.

Prof. Luciano Canepari also provides transcriptions featuring intense stops: I haven't got any real (conscious) experience about. I'll ask him for clarifications. If his analysis is correct, it's the final proof that it is inappropriate to include vocoids among approximant contoids, i.e. that [i, u] and [j, w] can be rightly categorised as different things.

@ Jákub

„…that [i, u] and [j, w] can be rightly categorised as different things.“

I fear there is no 100% proof for that in terms of physical science. (Though there are heuristic advantages in seeing them as different things. Many of the categories of phonetics might be better understood if we look upon them as heuristic categories: segment, feature, sound, consonant, vowel, syllabel, word, sentence … There is a certain „scripticism“ affecting the „things“ or making them to become things; Linell 1982 called it the „written language bias“.)

@ Klaus

Indeed I safely wrote «rightly categorised». I wouldn't imply that «[j̍, w̩]» aren't the same thing as [i, u]. Nevertheless, from both a theoretical, and a practical point as well, it's better to avoid for [i, u] to be included among approximants and rather group them in the whole category of [V], vocoids.

The Albanian glide j (like the English consonant y) is produced like its English counterpart. “For j the tongue moves quickly to or away from a high front unrounded vowel position in respect to the neighboring vowel, depending on whether that vowel precedes or follows it, respectively. It is the consonantal (short duration) counterpart of the vowel i, and in appropriate rhythmic environments (before or after a vowel) actually has been used interchangeably with that vowel.” [Leonard Newmark, Morfologjia, Stanford University Press, 1982].