Talk:Phoneme

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] Serbian orthography

Rephrasing to avoid saying that the Serbo-Croatian orthography is perfectly phonemic; it's not, as it fails to represent stress, pitch, and length on vowels.

[edit] /buts/

I just removed this:

A sound that is a single phoneme in one language may be a phoneme cluster in another. For instance, /buts/ means leg-covering footwear in English and consists of four phonemes /b u t s/; but in Hebrew it means a kind of cloth and consists of only three phonemes /b u ts/.

In fact, /ts/ in English is two phonemes /t/ and /s/, while IIUC Hebrew /t_s/ ("t-s ligature") is one phoneme. It's not the same "thing" interpreted as a phoneme cluster in English and as a single phoneme in Hebrew. The Hebrew sound is an affricate, like English "ch".

I understand that some very common clusters may be interpreted as single phonemes in certain languages, for functional purposes. The Georgian "harmonic clusters" come to mind, but I'm really not sure, so I'd be thankful to anyone who can enlighten me on this. Pablo D. Flores 11:45, 9 Sep 2004 (UTC)

[edit] Ligatures

Someone added a comment directly in the article asking for /tʃ/ to be written as the ligature /ʧ/. Ligatures are not required for affricates, though they may clarify things. It is best not to use this symbol, I'd say, because if the user's computer is not properly configured, they will at least see the "t". With the ligature symbol, they won't see anything. Also, unless we force another font, it won't even appear to be different from the non-ligatured version even on properly-configured machines. — Chameleon My page/My talk 07:03, 5 Jul 2004 (UTC)

Agreed. Say no to ligatures.--[[User:HamYoyo|HamYoyo|TALK]] 08:14, Jul 5, 2004 (UTC)
I disagree. There is a phonetic difference between the sequence and the affricate. Some languages have a contrast between these sounds. An example from Klallam:


[k̕ʷə́nts] 'he looks at it'   vs.   [k̕ʷə́nʦ] 'look at me'


(IPA doesnt really give a satisfactory symbolization of affricates). The issue mentioned above is irrelevent in many languages — but it is not with respect to Klallam. But even if it is irrelevant to English, it seems nice to symbolize phonemes in ways that show their similarities to the phonetic manifestation. Peace - Ish ishwar 07:47, 2005 Jan 26 (UTC)
I believe the Handbook of the International Phonetic Association says that affricate ligatures should only be used if necessary. Nohat 09:03, 26 Jan 2005 (UTC)
Yes, I believe you are right (I dont have a copy of the handbook at home but this sems to be what I remember). However, other systems of phonetic transcription provide different symbols for affricates & their respective sequences. For example, in the system commonly used to transcribe Native American languages the alveolar affricate is [c] while the sequence is [ts]. There is generally a principle in phonetic/phonological transcription where only one symbol is used to represent each phoneme. Many linguists are dissatisfied with the IPA so they either (1) ignore it & create their own system or (2) join the IPA & vote for change (or of course they just deal it & leave it alone). (I just want to mention that the IPA is not without its controversies and within the entire corpus of linguistic literature the reader will find many other things besides IPA.)
So with this said, I like the digraph (and at the least the ligature) because it abheres to this one symbol = one phoneme principle (the ligatured affricate adheres marginally so). You can note that unicoded IPA doesnt have digraph characters for lateral affricates, but I point out that other transcription systems do have unitary symbols for these as well. But then there are the practical issues with improper font rendering & the trend in wikipedia to only use IPA. I dont know. I am just dissatisfied with IPA in this regard. There is a phonetic difference and sometimes a phonological difference, but IPA is a bit clunky in its symbolization. Thanks! --Ish ishwar 20:31, 2005 Jan 26 (UTC) - Ish ishwar 23:25, 2005 Jan 26 (UTC)

[edit] Czech r-hacek

The text says:

Possibly the rarest sound is the one represented by "r hacek" (found in the name Dvorak) in the Czech language; it appears to be unique to the language.

I think this sound (ř) is similar or equal to the one of French 'j'. It is also close (and related) to Polish 'rz'. -- AdSR

/ř/ is not equivalent to French /j/; French /j/ is the same as the English /s/ in measure, correct? Czech has the letter ž for that. /ř/ is similar, but it's rolled. I believe the Polish rz is closer to the Czech ž than to ř. —Bkell 02:18, 18 Jan 2005 (UTC)

A Czech taught me to pronounce it once. It seemed to me to be /r/ and /ʒ/ articulated simultaneously. — Chameleon My page/My talk 07:03, 5 Jul 2004 (UTC)

Some Spanish speakers have the r-hacek sound for <rr>, although it is nonstandard and not terribly common.

Which speakers do that? Chameleon 18:17, 7 Nov 2004 (UTC)
Spanish speakers in the north-western parts of Argentina (provinces of La Rioja and Tucumán, at least, come to mind). Former president Carlos Menem (a curse be upon him) has this pronunciation, as well as famous folkloric singer Mercedes Sosa. I have a different dialect but the r-hacek is not that difficult to make, really. --Pablo D. Flores 01:40, 18 Jan 2005 (UTC)

Czech is certainly NOT the only language to have a phoneme unique to it. Dozens if not hundreds of languages share this property. Czech r-hacek just happens to be the example that is most well-known to Western linguists.

[edit] Slant brackets

The text says:

For technical reasons (slant brackets interfere with the mechanism of Wikipedia), it is advisable to square brackets for phonemic as well as phonetic transcription in Wikipedia entries.

I think this "feature" has been removed from Wikipedia. Should this para be excised? Hari

Yes, it should be excised. Slant brackets around lowercase letters have never been a problem on Wikipedia. The mess that most of them are in was caused by the use of a buggy conversion script, which thankfully was a once-only event. --Zundark, 2002 Mar 11

25 vowels in Punjabi? I find that hard to believe. Nagari has 13 vowel letters, of which one (vocalic 'l') is used only in one Sanskrit word; the others are six vowels (aiueor) with short/long/diphthong variations. -phma

Don't confuse language and writing system. The English alphabet only contains five vowels (a, e, i, o, u) but (depending on dialect), the language itself contains more (bat, bath, bait, bet, beat, knit, night, not, note, boy, naught, soot, suit, such, aout, about) Hippietrail 07:25, 9 Mar 2004 (UTC)

[edit] Allophony

The sounds /z/ and /s/ are distinct phonemes in English, but allophones in Spanish.

Out of curiosity, what sound environment makes a Spanish speaker say [z]? Off the top of my head, it seems like every use of the letter "z" should be pronounced as [s]. Not that spelling and phones/phonemes need to be very intimately related, of course; the main point is: I can't think of any Spanish word with a [z] sound. Maybe it's a feature of some particular pockets of Spanish? -- Ryguasu

What relation does <z> have with /z/ in Spanish? None. <z> is always the phoneme /θ/, with the voiced allophone [ð] before certain voiced consonants. Similarly, <s> is always the phoneme /s/, with the voiced allophone [z] before certain voiced consonants, e.g. <desde> /'desde/ ['dezde]. — Chameleon My page/My talk 07:03, 5 Jul 2004 (UTC)

You'll occasionally hear Spanish speakers using the /z/ sound for an S that appears before a voiced consonant in a phrase like ¿Quién es David?, but certainly never in a word by itself, so it probably isn't a great example of an allophone. --LDC
Depends on dialect. There are many allophones of /s/. Most Spanish dialects have two: One at the beginning of syllables, one at the end of syllables. There may be different pronunciations at the ends of words than at the ends of syllables. The syllable-final allophones may be /h/ or /z/ before voiced (as in /desde/, /mismo/, or a lengthening of the previous vowel. Some dialects don't have syllable-final /s/ at all. The majority of the Spanish speakers don't have a distinct /T/ phoneme, but only /s/. If such a dialect voices syllable-final /s/, then <z> may be [z], e.g. in <feroz> /fe'ros/ [fe'roz]. J. 'mach' wust 13:16, 12 Oct 2004 (UTC)
I think German is a better example of [s] and [z] being allophones of /s/. Hippietrail 07:25, 9 Mar 2004 (UTC)
Wrong. German has both phonemes.

as far as I know, you are correct. By the way, in some cases z and s are allophones in English (for the plural -- or maybe it is /ez/) Slrubenstein

Right. For example, chairs ends in [z], while cats ends in [s]. -- Ryguasu

Allophones or allomorphs? --Brion

Yes, those are allomorphs. A better example of allophones in English would be the aspirated T at the beginning of "tip" and the non-aspirated T at the end of "pit", which English treats as the same sound in all contexts. --LDC

Also light versus dark "l" and the several realizations of /r/ depending on dialect and phonetic context. Hippietrail 07:25, 9 Mar 2004 (UTC)

Isn't it more than a bit confusing to say phonemes aren't sounds, and then in a section like this refer to them as sounds?

[edit] How many points of articulation used?

Surely Dyirbal doesn't have the most places of articulation? I can think of languages that use more than six. Ubykh, for instance, contrasts voiceless fricatives in labiodental, alveolar, postalveolar, alveolopalatal, retroflex, velar, uvular and glottal classes: fa to eat, sa sword, ʂa head, ʃa arrow, ɕa three, xa testis, χa to knit, haj no, and there is an additional bilabial class not represented in the fricatives. thefamouseccles 00:41, 10 May 2004 (UTC)

If I understand the points of articulation correctly, my dialect of English uses eight: bilabial (/p/), labiodental (/f/), dental (/T/), alveolar (/s/), palatal (/j/), retroflex (the "er" in "butter"; not sure what its IPA or SAMPA sign is), velar (/k/), and glottal (/h/). --Jim Henry 15:51, 5 Aug 2004 (UTC)

Good observation! It seems to me that for this kind of figure of the articulation points, there need to be samples of the same manner of articulation, normally stops (?) (many languages have more places of articulation for fricatives). J. 'mach' wust 13:05, 12 Oct 2004 (UTC)

This discussion is a good indicator of just how confused and confusing the slippery slope of the phoneme concept is. If the phoneme is what we say at the wikipedia pages it is, then why do we need to refer to points or manners of articulation whatsover?

[edit] English has 40 phonemes

Where does this figure come from? Doesn't it depend on dialect, due mainly to differences in vowel inventories but also to whether, e.g., "whales" and "Wales" are pronounced differently? Josh Cherry 22:15, 5 Jul 2004 (UTC)

  • It's an approximate figure, but is pretty standard in linguistic texts. It's meant to include all sounds that are phonemes in any English dialect. It doesn't depend on specific word-pairs, but on whether a given sound (e.g. "wh") is the sole difference in at least one word-pair in at least one dialect. - Nunh-huh 22:24, 5 Jul 2004 (UTC)
    • OK, but perhaps this can be clarified. For one thing, it would tend to inflate the figure, and would be misleading in a comparison to a language with less diversity. Also, the Oxford Companion to the English Language says that Received Pronunciation alone has 24 consonants and 20 vowels. Josh Cherry 22:35, 5 Jul 2004 (UTC)
      • English has 24 consonants including affricates and excluding foreign sounds such as /x/. It has 12 simple vowels and 8 diphthongs. Of course, not everyone speaks like this. Where I'm from, /h/ is deleted, /θ/ becomes [f] and /ð/ becomes [v]... but "English" without adjectives means "standard English". — Chameleon My page/My talk 22:51, 5 Jul 2004 (UTC)
        • I don't think it would hurt to say "standard". More important, at least on this side of the pond "standard English" does not necessarily mean RP. I doubt that the numbers are the same for standard American English, for example. I wouldn't make a statement like "English is non-rhotic" just because RP is. Josh Cherry 23:34, 5 Jul 2004 (UTC)
          • I don't take "English" without adjectives to mean "standard English". What is "standard" after all? The number of phonemes depends on your dialect. How does one say that one dialects is "standard" whilst another is not? Wouldn't it be better to avoid this whole debate and specify a well know dialect or two? Or perhaps it would be best to give a range. Whatever is to be done the following sentence from Phonemic extremes needs a rewrite. The English language itself uses a rather large set of 13 vowels, though its 27 consonants are pretty close to average. Some dialects might have as few as 13 vowels but ... 27 consonants? 40 - 13 = 27. I smell a rat. - Jimp 25Apr05 ... Okay I've changed it so as to give a range. The range I gave was 13 to 22 vowels plus 22 to 26 consonants. I'm no linguist and my range might be wrong but as far as I'm aware this is about right. At least it's more correct than what was there before. - Jimp 23May05

[edit] Phonetic languages

Jight,
Kollipy,
Proto-combobulationism,
Knick-pawn,
Spoying the waze,
Psycho-stellization.

I've just invented all these words randomly. I bet that, if you said them, your pronunciation would be the one I intended. The spelling system of English follows rules and is largely predictable. Not as much as Spanish, but much, much less than Chinese. English as an "unphonetic language" is therefore an exaggeration and a simplification.

Furthermore, Italian is a bad example. It is far less "phonetic" than Spanish, say. Mi piace la pesca means "I like the peach" or "I like fishing" depending on whether you say /'peska/ or /'pεska/. Pizza is pronounced /pittsa/, but piza would also represent this pronunciation. Similarly, /piddza/ would also be spelt pizza or piza. It is an irregular system compared to Spanish. Chamaeleon 21:52, 17 Jan 2005 (UTC)

Nobody says English is "unphonetic", just that it is more so than most other languages sharing its alphabet. English is often unpredictable both ways: text to speach and speach to text; French only one way: mostly it is dependably pronounceable, but spelling from sound needs etymological and semantic knowledge.
"easy" examples as above are easy to construct, but:
  • how to pronounce "read" without knowing the tense intended?
  • how to write identically sounding words of "red" (colour) and "read" (past tense)
  • examples like this can be found in masses
I still maintain there is no set of rules that comes close to consistent: a majority of over 50% correct guesses is not enough to be convincing; should be at least 95%. --Woodstone 22:26, 2005 Jan 17 (UTC)
I think it is strange to use the term "phonetic language". It seems to be resulting from a confusion between a language's sounds & a language's orthography. One language cannot be more phonetic than other language because phonetic refers to the sounds occurring in languages & all langs have sounds (the definition must be extended to include signed langs). No writing system is truely phonetic: the sounds represented by [t] differ acoustically & articulatorily in every different environment. Cheers! - Ish ishwar 07:26, 2005 Jan 26 (UTC)
Of course, as the article indicates "phonetic language" is a misnomer, but is the term many people use to discuss a language having a "phonetic spelling". This should be regarded within the language itself. For a language with an ideally phonetic spelling, there would be a set of rules that would allow to write down any spoken word correctly and read out any written word correctly without needing semantic or etymologic knowledge. Because different langages usually have different phonologies they need different rule sets if they share the same writing system (unless the script carries no phonological information at all) --Woodstone 15:36, 2005 Jan 26 (UTC)
Yes. So my covert question was: Why are discussing orthography in a entry on phoneme which has to be with sound systems not writing systems? --Ish ishwar 21:20, 2005 Jan 26 (UTC)
Good point, but in order to discuss phonemes in a text media, a way to write them down is needed. In that context it is worthwhile to point out the fundamental difference (and at the same time the similarity) with orthography. That does not merit an article on its own, so it doesn't strike me as too much out of place here. --Woodstone 22:45, 2005 Jan 26 (UTC)

quote of part of discussion above>>I've just invented all these words randomly. I bet that, if you said them, your pronunciation would be the one I intended. The spelling system of English follows rules and is largely predictable. Not as much as Spanish, but much, much less than Chinese. English as an "unphonetic language" is therefore an exaggeration and a simplification.>>

Yes, but are the rules phonetically or phonemically speaking, predictable. Or does having a large English vocabulary internalized (including spelling) help you to come up with, by analogy, ways to pronounce such words? What seems 'rule based' for a fluent speaker who can read a language might be total chaos for someone simply trying to decode or recode the written language. anonymous

quote from above>>:: Yes. So my covert question was: Why are discussing orthography in a entry on phoneme which has to be with sound systems not writing systems? --Ish ishwar 21:20, 2005 Jan 26 (UTC)>>

Well one reason would be whether you are using a standard writing system or IPA symbols (either narrowly or widely), you are still trying to capture dynamic, complex spoken language in written form (hence the need for a unit like a phoneme in the first place). All efforts to do so at a phonological level are doomed to be incomplete.

Since the 'writing systems' section came up here in discussion, it is here I will point out I re-ordered the section and made a few corrections as well. Also, saying writing systems are conventional is a way saying nothing much. Almost all the writing systems under the sun today are arbitrary and conventional. Korean hangul, as an interesting contrast, can be said to be 'phonetically motivated', but as the system is used to read and write Korean, is still largely conventional and arbitrary. Non-linguists won't understand these terms anyway.


Counter-examples: pough (ow? aw? awf? etc.), chind (k? tsh? sh?), satia (tee-uh? shee-uh? shuh?), farel (accent on first or second syllable? what are the vowels?). All are plausible English words. Try finding similarly plausible native spellings with similarly ambiguous pronunciations in Swedish, say. —Simetrical (talk • contribs) 21:05, 12 March 2006 (UTC)

[edit] Phonological extremes

It seems that the comparison between Spanish "sin" and English "seen" is incorrect. English "seen" would be spelled in IPA like [si:n]. The reason of this mistake is perhaps that it is hard to find words with a short [i] in English. The point being made is correct, but the example is wrong. --Woodstone 15:43, 2005 Jan 26 (UTC)

I dont have a problem with this example. The example just is just using a broader transcription than what you write above (in fact the English vowel would probably be nasalized & many Eng. dialects have a following off-glide, so you could be even narrower). The example, I believe, is concerned with only the vowel quality not the duration. It stills illustrates the point: phonetically similar phonemes symbolized with the same character are not equivalent.
On a slightly different note, from a functional phonological perspective (from André Martinet) similar phonemes in separate languages can never be equated with each other because every language has a different phonological system & therefore the phonemes to be compared have a different set of oppositions to the other phonemes within their respective languages.
An issue that I would bring up is why do we need to mention that languages have unique sounds? Isnt this a matter of phonetic inventories & not a matter of phonology? Why does this belong? I think that we could say that every phoneme is unique to a particular language & if measured in precise acoustic detail every phonetic manifestation of a phoneme is unique to a particular language. We can only speak of similarity.
Further comment is quite welcome. Cheers! --Ish ishwar 21:15, 2005 Jan 26 (UTC)

Agree that any phonetic transcription can be done in broad or narrow ways. However, when stating an IPA transcription, the one commonly used (or the closest one from the standard IPA set) should be chosen. This still leaves a range of broadness (nasality, off-glide) that can be ignored. My OED describes the vowel explicitly as long, so representing English "seen" by /sin/ is unusual. It must be possible to find a better example. --Woodstone 22:53, 2005 Jan 26 (UTC)

It is not so clear from the article, but I think that the example is trying to illustrate that a phoneme /i/ is not equivalent between languages. Although Spanish & English both have /i/, in English /i/ is phonetically [iː] or [ij] or [ɪj] while in Spanish /i/ is phonetically [i] (?) (I dont know Spanish phonetics). So, the phonological representation is the same, but the phonetic representation is different. Most of what I have seen on English writes the English phoneme as /i/ and only when discussing phonetics is a phonetic representation given.
But if one wants an example that has phonemes that are even more phonetically similar, then I offer a comparison between Japanese, French, & Italian /i/. Japanese /i/ is less close than French /i/ and it is less front than Italian /i/ (Akamatsu 1997:30). So taking Daniel Jones' cardinal vowel [i] as a reference & using IPA diacritics, we have:
  • French /i/ = [ i̱ ]
  • Italian /i/ = [ i̞ ]
  • Japanese /i/ = [ i̱̞ ]

What does everyone think? (my browser doesnt really display the diacritics correctly in wikipedia) - Ish ishwar 23:54, 2005 Jan 26 (UTC)

References:

  • Akamatsu, Tsutomu. (1997). Japanese phonetics: Theory and practice. LINCOM studies in Asian linguistics 03. München: LINCOM EUROPA.
The long/short distinction for English vowels is indeed traditionally transcribed for British English, but not for American English, which accounts for well more than half of native English speakers. The difficulty is that most American dictionaries don't use IPA. However, if you read, for example, the front matter of Jones' English Pronouncing Dictionary, it talks about how the length difference is not considered salient at all for American English. The phonetics section at the beginning of the Merriam-Webster Collegiate dictionary, too, describes all the symbols used in the dictionary in terms of IPA symbols, and length marking is not used for [i]. Also, the American Heritage Dictionary doesn't use length marking for any of the vowel symbols in the pronunciation key which maps AHD symbols to IPA symbols. The reality is that the length difference between [i] and say [E] is very small in American English, particularly when compared to the lengthening effect of following voiced consonants, which is much greater. That is, the difference in vowel length is much greater between seed and seat (both [i]) than it is between seed and said.
But even disregarding all of this, the point I intended was about vowel quality. Spanish [i] is higher (closer) and produced with more spread lips than English [i]. Spanish [i] is also totally monophthongal, while like pretty much all English vowels, English [i] can be slightly diphthongal. In fact, when non-English speakers make fun of English by speaking in a pseudo-English babble, one of the defining characteristics of this babble is that it contains no stable vowels. I guess ultimately the point is that the symbol [i] corresponds to an area within the vowel space and not a particular point, and each language that has a sound that is transcribed as [i] doesn't have the same [i] as other languages' [i]. Of course, the exact quality of a vowel varies from speaker to speaker and even from phonetic environment to phonetic environment, but doing statistical analyses of samples of different speakers havs enabled phoneticians to pin down approximately where overall English [i] is and approximately where Spanish [i] is and they're not the same place. I'm kind of babbling on here. Perhaps some of this should go on the page. Nohat 02:44, 27 Jan 2005 (UTC)


I have gone ahead and taken the part we're discussing out of the 'Phonological Extremes' section; I don't really understand what it's doing there. Now it's up with other basic explanations in "Background and Related Ideas."

I've also:
-Clarified that we're talking about phonemics; I found the paragraph confusing because it was conflating the discussion of broad vs. narrow phonetic transcriptions with how pronunciation may vary from language to language even if we use the same phonemic transcription.
-Edited the link--before it was pointing to an irrelevant point on the page, so I got rid of the # and it now refers to the whole page (what is being cited is two different things on the page). algormortis

[edit] Chronemes

The bit about tonemes seems relevant, but do chronemes really belong here? After some googling, it appears that the term appears most often in discussions of articulatory physiology, rather than phonology. Also, I've always found references to length as a feature of phonemes, just like point of articulation, nasalization, etc. I'm not a linguist, so maybe there's a theory that considers length not as a feature but as something different... -- Pablo D. Flores 21:22, 6 Apr 2005 (UTC)

There are languages with many minimal pairs where length of either a vowel or consonant is the only differentiating factor between words. The "handbook of the international phonetic association" mentions them (as chronemes) parallel to differentiation by tone or stress. In English there are differences in length, but these are never distinctive for meaning. That may be the reason they seem outlandish. −Woodstone 21:33, 2005 Apr 6 (UTC)
I'm fully aware that length is a contrast; I was mentioning the fact that I hadn't heard of chronemes as such. Instead, I've always seen phonemes described as a set of features, where length is just one of them. For example, /i:/ = [+high][+front][-rounded][+long]. Unlike tone, I have not seen length considered apart from the description of a phoneme. But if the IPA handbook mentions chronemes, I'm fine with it. --Pablo D. Flores 14:32, 7 Apr 2005 (UTC)
See sanamuunnos for a Wiki example. It's spoonerism, where a phoneme moves, but the chroneme does not. Addition: the partitive marker in Helsinki Finnish is often a chroneme. Ei suomee, vaan stadii. --Vuo 00:49, 15 July 2005 (UTC)
I agree that length could be seen as just one more attribute of a phoneme, but then so is tone. If there is a toneme, there should be a chroneme (I have even seen a few references to stroneme for stress distinctions). Perhaps the choice of IPA to represent them both as separate symbols (both can be coded in-line) makes them lean towards a separate class. −Woodstone 14:57, 2005 Apr 7 (UTC)

The contradiction, in part, is that many embrace a concept of phoneme (underlaid by a contrastive view of meaning) as something segmental, but that some elements clearly (or not so clearly) aren't segmental. One might also wonder if extra beats are really consistent isochronous units. Anyway, should the discussion of the phoneme--with all its difficulties, inconsistencies, etc.--include or not include non-segmentals?

I would say by one criteria for determining 'phoneme' status, chroneme might better qualify than toneme because the toneme is clearly not limited to one segment or syllable, while a chroneme seems more locked into one or two beats coinciding with a segment. I think?

[edit] Historical Note

I object to the following: "The phoneme is a structuralist abstraction that was introduced by the Polish-Russian linguist Jan Niecislaw Baudouin de Courtenay (1845-1929) and elaborated in the works of Nikolai Trubetzkoi (1890-1938). It was later adapted to and formally psychologized in generative linguistics (after Chomsky and Halle). Rather than a basic mental unit of language, however, it may well be a perceptual artifact of alphabetic literacy (see the terms Phonemic awareness and Phonological awareness)."

Baudouin de Courtenay's phoneme was "psychological", not "structuralist". It was actually Saussure who introduced the structuralist concept, which was based on the work of Baudouin and Kruszewski. Baudouin had a concept that tolerated absolute phonetic neutralization between phonemes, which was not characteristic of structuralism. Sapir's phoneme was "psychological" in Baudouin's sense.

Regarding the contribution of Chomsky and Halle, it is fairly orthodox thinking that they "psychologized" the phoneme, but they themselves claimed to be following in the footsteps of Sapir. In point of fact, they did not follow Sapir's or Baudouin's concept of the "psychological phoneme". Unlike Baudouin and Sapir, Chomsky and Halle made no clear distinction between phonology (Trubetzkoy's term for Baudouin's "physiophonetics") and morphophonology (Trubetzkoy's term for Baudouin's "psychophonetics"). The physiophonetic/psychophonetic dichotomy was fundamental to Baudouin's theory of phonemes, and it was maintained (albeit less explicitly) in Sapir's writings. Generative linguists tend to recognize the dichotomy in various ways, but they do not recognize it as representing two separate components of grammatical description. While generative theory is a psychological approach to language, its level of "systematic phonemics" is far more abstract than that of earlier psychological theories. To claim that phonemic awareness is an "artifact" of literacy may make sense from a generativist point of view, since generative theory has rejected the phonology/morphonology split as a fundamental dichotomy, but that claim reverses and obfuscates the earlier viewpoint that phonemes motivated alphabetic writing, rather than vice versa.

Baudouin de Courtenay originally linked phonemes to two phenomena: alphabetic writing and rhyme. Nothing has yet been added to point out the fact that rhyme was also important to the concept of the phoneme. Perhaps that is because orthodox generative phonology is even worse at explaining the phenomenon of rhyme than it is at explaining the phenomenon of alphabets.

I agree that this was poorly worded. Why start the a definition of a term by refuting it? I made a new section and moved the "anti-phoneme" arguments down there. I also cleaned up and expanded this section. Squidley 19:57, 18 November 2005 (UTC)

quote from above>>Baudouin de Courtenay originally linked phonemes to two phenomena: alphabetic writing and rhyme. Nothing has yet been added to point out the fact that rhyme was also important to the concept of the phoneme. Perhaps that is because orthodox generative phonology is even worse at explaining the phenomenon of rhyme than it is at explaining the phenomenon of alphabets.>>

Well how psychological is that? And if the original concept was psychological, was it psycholinguistic or not? Standard accounts of the development of the concept usually (in my experience) don't go back that far (only as far back as the structuralists with all their bifurcations and stuff about a contrastive theory of meaning) or are not clear as to who was and who was not structuralist. Also, it's one thing to say this or that concept is psychological, but it's another to show that it is. Earlier versions of this article that made it sound like a scientific fact were simply silly. The generative turn of development was anti-structuralist and anti-behaviourist, so one has to wonder why Chomsky and Halle, if they were so original or critical, even adopted the concept at all.

Baudouin De Courtenay, by the way, might well be the originator of the idea that if you are going to set down spoken language into writing you need an abstract phonological unit that simplifies and categorizes the phonetic complexity of a spoken language (how it got psychologized or at all supported in psychological research is totally unclear to me). What he meant by rhyme is here ambiguous: poetic rhyme or phonological rime?

quote from above>>Baudouin de Courtenay's phoneme was "psychological", not "structuralist". It was actually Saussure who introduced the structuralist concept, which was based on the work of Baudouin and Kruszewski. Baudouin had a concept that tolerated absolute phonetic neutralization between phonemes, which was not characteristic of structuralism. Sapir's phoneme was "psychological" in Baudouin's sense.>>

I'm really confused because at the wikipedia entry for the man himself, it says that he treated phonology as physiological and that psychology comes into play with higher levels of language.

Perhaps something else we need to consider before wasting anymore time exposing our belief systems about phonology and the phoneme would be to re-consider the word 'structuralist'. The Baudouin concept could be structuralist in the sense of showing a social or psychological structure in control of language--so the term would be used not in the strict sense of the school(s) of thought called 'structuralist'.

[edit] Allophones are not "real". They're abstracted, too.

Allophones are not "real". They are abstractions. A single allophone can be pronounced in slightly different ways. It is less abstracted than a phoneme, though. Every section in the article that says that allophones are "sounds", or "real", or something similar, should be changed, because they just confuse the reader. An allophone is abstracted from patterns of sound waves observed on measuring instruments. If you tell me that an allophone is "real", you're just waffling. The same goes for a phoneme. A phoneme is not a "sound", but a group of allophones that are somehow related. 2004-12-29T22:45Z 02:39, 15 July 2005 (UTC)

Actually going from oscillographic data to sounds is a real reach itself. People who study phonology hope that there is some sort of reality to phonology in at least two ways: articulatory phonology and perceptual phonology. I don't see how phonology as part of mental control of language could be imposed on sound, only on the production or perception of it. Allophones as dealt with in standard college textbook treatment of phonology (such as introduction to linguistics) are indeed as much abstractions as phonemes. For a start, many variations of a categorical sound don't get elucidated. And where does a dialect start and an idiolect begin? And just how does one, for example, capture all that is truly invariable across two instantiations of the same (theoretical) phoneme in the same word (same word said twice)? One would do better, phonologically speaking, by saying the word a third time than by trying to transcribe anything.

[edit] Inclusion of sign language

In my view the inclusion of sign language in the definition of a phoneme is an unnecessary complication. The word itself refers to the Greek φωνη voice. Analysis is based on sound: recordings and production manner in the mouth. Equally clear is the concept chereme from Greek (χειρ hand) an element of hand signs. Merging one concept on the other is just creating a muddle. −Woodstone 10:25, July 24, 2005 (UTC)

i disagree very much. The phoneme (or any underlying representation) is a part of language. To exclude these languages would be very unfortunate. They also have phonologies, syntaxes, and all of the other pieces of grammars. Linguistics has already been biased against them in the past. They are not as well-understood as spoken languages, but that is all the more reason to include them in our scientific investigations. What does the etymology of phoneme have to do with the content of the article? Phoneme does not refer to voice or anything in the Greek language. It refers to a linguistic concept. (the Japanese word for 'steering wheel' is derived from English handle but this Japanese word does not refer to a handle but a steering wheel. I dont understand the relevance of this...). peace – ishwar  (speak) 15:21, 2005 July 24 (UTC)

Of course etymologie does not matter normally, but the words in question did not arise naturally, but were chosen by liguists to fit what they had in mind. The techniques to study sounds (and their production), necessarily differ substantially from the ones to study gestures. Keeping the two concepts separate, as alternatives in language expression, allows clear subjects, methods and conclusions. If you merge chereme into phoneme, a need will arise to create a word to indicate phonemes that are not cheremes. −Woodstone 19:22, July 24, 2005 (UTC)

The terms are merged because they're merged in the linguistic literature. I believe it was Stokoe who coined the term 'cheireme', but it's now considered obsolete. Since the term 'phoneme' is used in linguistics for both oral and signed languages, that should be explained here. A phoneme is a cognitive concept, not an element of physical sound. If that's muddled, then it should be clarified. (The whole article is muddled, frankly; maybe some day I'll get to it.) kwami 20:03, 2005 July 24 (UTC)

It's circular to say that a phoneme is a cognitive concept that relates or acts a set for sound variations and then use relationships among somewhat similar sounds to deduce a phoneme.

[edit] Language the most vocalic

In all texts I've found about Punjabi, the 25 "vowels" include diphthongs. How many pure vowels does Punjabi have? Estonian has 26 diphthongs. While I think that Estonian has three phonemic lengths - and it is illustrated that this is the case for /a/, for example - are there any restrictions for some vowels? A parallel is that Finnish consonant /v/ does not have distinctive length, although all other consonants do, so does this limitation exist for some vowels in Estonian? One could be 'õ', but I can't know. --Vuo 23:51, 25 July 2005 (UTC)

[edit] underspecification & archiphonemes

perhaps an introductory note about underspecification, archiphonemes, & neutralization would be in order? – ishwar  (speak) 22:04, 2005 August 10 (UTC)

I do not contest the validity of abstract theoretical analysis along the lines presented, but the examples are pretty bad.
  • In many (most?) English variants fame and fang are not a minimal pair (different vowel).
  • Why on earth would you theorise "amchor*"? You should find an example where a related word with m exists. That is why it's called archiphoneme: it comes from somewhere.
−Woodstone 21:12:49, 2005-09-09 (UTC)
Ok. spit, pit, bit is probably better for many. Actually, it doesnt really matter as the concept could be explained using any English variety. But, yeah, it may be more accessible if many can relate to it.
In anchor, you cant determine what phoneme it would be unless you simply stipulate it. So, if you didnt use archiphonemes in your analysis, then you could just as well choose amchor, anchor, or angchor (the phonetic realization results from an automatic assimilation). You wont find a contrast between the 3 nasals before /k/. – ishwar  (speak) 03:28, 2005 September 10 (UTC)
Hi, Ish. I thought I'd already posted this reply, but looks like it never got saved:
The word "archiphoneme" does not mean that it comes from somewhere. If it does, you can usually deduce the underlying representation, and present the rest as assimilation rather than as an archiphoneme. For example, and surfaces as /əŋ/ in lock 'n key, as /ən/ in this 'n that, and as /əm/ in bye 'n bye. However, since it is /ən/ before vowels and /h/ (hearth 'n home), it can be assumed to be underlyingly |ən|, and there is no need to speculate about it being an underspecified |əN|. Although I suppose you could make that argument, people reading this article would why you wonder would want to. It's precisely in those cases where there is no morphological clue as to what the underlying sound is that choosing a particular phoneme is arbitrary.
Anyway, we kept getting into edit conflicts. I've restored a nasal example, because I think it's a good one. I also changed the plosive example to a k, g contrast, because I like the example the sky vs. this guy. I suppose I could have used the spit vs. this bit, but what makes the other work so well is the high frequency of the phrase this guy. Go ahead and revert if you like. kwami 03:53, 2005 September 10 (UTC)
No, yours is probly better. I was going to put in a vowel example, but the 2 consonant examples are enough, I think. Thanks for making it clearer. peace – ishwar  (speak) 04:16, 2005 September 10 (UTC)
I think the current examples are very good, though I touched up a few details. I also added a section that argues against archiphonemes, etc. Please remember that even if you personally use an archiphonemic analysis, not everyone does, and the non-use of archiphonemes is very much in the mainstream. So, if you feel the refutation section needs work--and it probably does--kindly edit accordingly. Squidley 19:53, 18 November 2005 (UTC) (a phonologist with a Ph.D. in Linguistics)

[edit] Criticism

Such a method for determining phonemes has the profound weakness of circular logic: phonemes are used to delimit phonologically the semantic realm of language (lexical or higher level meaning), but semantic means (minimal pairs of words, such as 'light' vs. 'right' or 'pay' vs. 'bay') are then used to delimit the phonological realm. Moreover, if phonemes and minimal pairs were such a precise tool, why would they result in such large variations of the sound inventories of languages (such as anywhere from 40-48 phonemes for counts of English)? Also, it is the case that most words (regardless of homophones like 'right' and 'write', or minimal pairs like 'right' and 'light') differentiate meaning on much more information than a contrast between two sounds.

The above may perhaps be true, but the way it is formulated is not conmprehensible, nor encyclopedic in style. Should be thoroughly improved and cleaned up. −Woodstone 21:29, 30 September 2005 (UTC)

Will take a crack at it. It's sound thinking, and gets to the heart of why so much of phonology is going back to articulatory concerns or in the opposite direction, and totally away from the unworkable phonemic segment (which tries to do both, take in articulation and take in abstract linguistic cognition, and does neither very well).

[edit] Original research?

The new section Arguments against Archiphomemes and Underspecification appears to be original research. I personally don't buy archiphonemes except as a neat trick on the part of the linguist to add 'logic' to a language, but the argumentation is sloppy, and not the sort I would expect of a published source. (Speakers of many languages, for example Japanese, hear allophonic distinctions clearly, etc.) Is there a source to back this up? Otherwise, I suggest we delete, or at least pare it down substantially. kwami 20:38, 18 November 2005 (UTC)

Agreed. Rejection of a concept off-hand because it doesn't follow "obvious" conclusions doesn't appear very scientific. The text is also patronizing and full of questions meant to be rhetoric questions, but which are far from undisputed. It also appears that the writer doesn't know much about vowel harmony — Finnish speakers, for example, don't natively perceive the [y] in /olympia/ as an /y/, but as an /u/, and produce it as such, unless "corrected" by prescriptivists. But, /y/ and /u/ contrast very strongly in Finnish, consider kuu "moon", kyy "adder", so you can't say that Finnish speakers don't perceive these as the same phoneme. The reason is that frontness or backness is a phonemic feature on the level of an entire word, not at a level of a single phoneme. The Turkish-based orthography argument is bogus. Consider the Finnish case: should you have a different frontness or backness in writing, this would be a compound word boundary. Like, pii·maa "diatomaceous earth", vs. piimää "some sour milk". --Vuo 23:39, 18 November 2005 (UTC)

I just deleted it. I don't think it could be saved. All the arguments were spurious. Better to just start again from scratch. kwami 00:43, 19 November 2005 (UTC)

What we see so far is a clash of belief systems about inherited phonology. If you do real phonology, most of this article wouldn't be of any interest at all, except for its cringe-worthiness.

[edit] Restricted phonemes

  • /ŋ/, as in sing, can occur only at the end of a syllable or word, and can never occur at the beginning of a word.

What about certain African languages? E.g. Swahili ng'ombe = cow. If this and the following examples refer only to English, perhaps it should be stated. --Simonbarne 06:17, 8 December 2005 (UTC)

Okay, I cleaned up that section. kwami 07:44, 8 December 2005 (UTC)

[edit] reverted section

I reverted the following section (second half of the paragraph, mostly). It doesn't belong in the intro, but in the section on criticism; it is incomplete (a reader who isn't familiar with the concept already will have no idea what it's talking about); and it needs to be substantiated or at least illustrated. kwami 09:57, 12 December 2005 (UTC)

Although it is fundamental to description of language at a sub-lexical level in most approaches to phonology, some linguists reject the theoretical validity of the phoneme. Some believe that belief in phonemes is a product of literacy; others that the mind processes sub-phonemic elements of speech in meaningful ways. One problem with the phoneme is that of abstraction: once it is abstracted from real speech and the reality of phonetics in order to make generalizations, there really seems to be no need for any real-world considerations in delimiting the concept. A phoneme, then, is any phonological unit that has, theoretically speaking, a reality beneath a syllable. It seems to be all too convenient that proponents of the phoneme get to pick and choose phonetic properties when it suits the concept but ignore other phonetic aspects when it contradicts the concept.

I doubt if most readers will understand the article anyway. It seems that up to a point, the concept of the phoneme was established to exist because of minimal pairs and complementary distribution. The article really doesn't explain the latter concept very well at all. Most traditional treatments of this for explanation go from the phoneme to the feature in a fairly predictable segue. This article doesn't. Perhaps in the case of a Wikipedia article, simpler would have been better?

About changes and reversions to the introduction. I agree the introduction needs to be clearer and avoid too much jargon (or introduce it better). But, for example, how can a concept be fundamental to phonology while largely rejected by phonology? Have you read any phonology from about the late 1980s on? The phoneme seems to belong to the introductory texts only, not active phonological research. Sorry.

From the most recent version:>>In human language, a phoneme is a set of phones (speech sounds or sign elements) that are cognitively equivalent. It is the basic unit that distinguishes words and morphemes. That is, changing an element of a word from one phoneme to another produces either a different word or obvious nonsense; whereas changing an element from one phone to another, when both belong to the same phoneme, produces the same word with an odd or incomprehensible pronunciation.>>

Does anyone have a rough and ready measure to distinguish clearly between 'obvious nonsense' and 'incomprehensible pronunciation'? I would say in the case of trying to put positional allophones into positions where they don't belong, phonologically impossible as well.

Wouldn't it be better to start over with this article? A minimalist approach might be best, with maximum linkage to articles covering all the other concepts. The core of an article like this really should be: first appearances of the concept, influential appearances of the concept, minimal pairs, complementary distribution, segue to features and articulatory gestures, explanation as to why most modern phonology today doesn't concern itself with the phoneme. Period.

The section you point out could certainly use editing. However, I don't agree on a minimalist approach, at least not for everything. The phoneme is a difficult concept for a lot of students in intro linguistics, and they may come here for help. Our article should read in such a way that someone with no linguistic background, and no particular ability with languages, could get an intuitive grasp of what a phoneme is, without having to read a dozen other articles first. A smart 10-year-old should be able to follow it. kwami 18:38, 14 December 2005 (UTC)

I'm talking about the entire article. Some editorial principles might have helped with this; instead this article has evolved into something rather unworkable. If I were writing the article from scratch, I would give a brief, as workable as possible definition of the concept. Then I would cover the all-important complementary distribution and categorical perception (and minimal pairs) and then point out the issues with such attempts to establish the concept. Then I would cover the development of feature (in very brief form, since the point would be to link to an article on feature). Then I would cover the use of the concept in the generative program and how this resulted in , for the most part, an abandonment of linear approaches to phonology, with an accompanying burgeoning of alternative approaches to phonology, such as articulatory phonology, auto-segmental phonology, prosodic phonology, metrical phonology, etc. (That is, why phonemics is no longer the major concern of phonology). I would then point out how the concept is still established in some areas of psycholinguistics (though see recent discussion where Chomsky takes Pinker to task on things like categorical perception), the pscyhology of developmental reading in alphabetic languages, and foreign and second language teaching (where structuralism still largely prevails).

Other things: I would stick with mostly English examples for the major points, and not bring so many other languages into the discussion unless absolutely necessary. I would clear other phonemic concepts out to other linked articles, for sure.

As it is now, the article isn't very good for beginners (including people with backgrounds in linguistics but who are weak at phonology, which is actually rather typical) and it would be of no interest whatsoever to a practicing phonologist (unless they wanted to lose a lot of hair reading it).

Knock yourself out! The phoneme is also considered doubtful among linguists who reject the generative/OT program (for example in casual speech which devoices final plosives, nasal vowels are contrastive at the same time they're predictable, such as at [æt] vs. ant [æ̃t] vs. and [æ̃nt]). kwami 08:25, 15 December 2005 (UTC)

I think most all types of phonology nowadays rejects Chomsky and Halle's SPE treatment of phoneme. Even Chomsky does. It's funny, though, to see how a field like reading simply treats the concept as if it were real? Is it even psychologically real when reading a complex alphabetic language like English? I highly doubt it.

[edit] Opinion to be removed

Last sentence of the third paragraph of Writing Systems:

"Hindi's writing system, however, probably ultimately descends from the same ancient Middle Eastern sources that gave the world the Roman, Cyrillic and Arabic scripts."

I feel that this does not belong here as it is clearly an opinion and a wrong one at that. The Cyrillic script was not an ancient Middle Eastern source but Saint Cyril only 1200 years ago. Maybe the author meant Greek, which could have been reduced to Phoenician, which also would have made Roman (Latin) redundant.

I feel this should be deleted, or posted here, on the talk page. Any defenders to this statement? JesseRafe 03:03, 19 December 2005 (UTC)

I think you're misinterpreting it. It's not saying the Cyrillic script was an ancient Middle Eastern source, but that Arabic, Roman and Cyrillic scripts were derived from an ancient Middle Eastern source. That is, it's saying that Cyrillic and the Hindi writing system (Devanagari) are cousins, not that one is an ancestor of the other. Yes, the Modern Greek, Cyrillic and Latin scripts were all derived via Ancient Greeks scripts so it mightn't be necessary to say both, but it can't hurt to—I can't imagine anyone thinking of Greece as Middle Eastern.
Also, I don't think it's clearly wrong. The Wikipedia article on Brahmi, an ancestor of Devanagari, says "Brāhmī is generally believed to be derived from a Semitic script such as the Imperial Aramaic alphabet". I think most peolpe would accept Aramaic as representing an ancient Middle Eastern source.
Hence, I don't think it needs to be removed. (One might claim it's unnecessary though, but you didn't, and at this stage I'm only defending against your accusations.)
(I moved this to the bottom of the page, where new comments are expected.)
—Felix the Cassowary (ɑe hɪː jɐ) 06:10, 19 December 2005 (UTC)
Thank you for moving this comment to its appropriate place, I'm new to contributing and Talk Pages. My main issue with it was that it was expressing an opinion - or at least phrased like an opinion. "Some scholars claim that..." sounds much better than "probably [anything]". Secondly I feel it is wrong, and poorly stated. It could have mentioned Ugaritic or Akkadian Cuneiform or some example of this ancient source as you did.
I too made a mistake and did not mean that Cyrillic was not an ancient middle eastern script, but that it was not from one per se.
JesseRafe 06:21, 19 December 2005 (UTC)
Handy hint: If you want to start a new discussion on a talk page, use the + next to the "edit this page" link. It starts a new section, and adds it to the bottom of the document. It was one of those things I never really saw, even though it was always on my screen, till one day I did see it.
I think the statement is intending to express that there is a consensus on the topic. If you said "some scholars claim that" you make it sound less certain, as if the scholars doing the claim are being innovative. And really, you shouldn't say "some scholars", but rather which scholars, and I imagine that list would get rather long in this case. But then, I don't actually know that much about the origin of these scripts so if you know better, then do tell :)
As to your correction: Cyrillic was not derived from an ancient Middle Eastern script directly, but it certainly does come from one. Cyrillic comes from a relatively recent form of the Greek alphabet, an ancient form of which comes from the Phœnician alphabet. Your objection seems somewhat pedantic to me, if not entirely wrong: I'd compare it to an objection to the statement "I have some of my grandmother's genes". Sure, my grandma wasn't my mother, but to my knowledge I am a genetic descendent of her, so it's reasonable to say that I have some of her genes.
—Felix the Cassowary (ɑe hɪː jɐ) 09:45, 19 December 2005 (UTC)

The reason I added this thought was to point out how culturally biased we might be if we viewed an Indic script as exotic when ultimately it comes from the same sources as European (and Middle Eastern ones). If you can word it better go ahead. I tried tow word so no one would object, but I realize now I'm in wikiworld, where anything is possible.

BTW, although the person who said it might be on my side, I have to say, in terms of ancient Greece, I think it's as much ME as European. This whole European vs. Asian distinction is mostly propaganda added by later biases. The Greeks were no more modern Europeans or 'westerners' than the were Persians 'Asians' or 'Asiatics' (both cultures being Indo-European), and the Phoenicians were all over the ME and Mediterranean. Alexander the Great didn't follow the distinction and you see where he ended up. And then the Roman Empire was as much E. Mediterranean, ME, and Asian as it was anything else.


quote from one of the objections>>Thank you for moving this comment to its appropriate place, I'm new to contributing and Talk Pages. My main issue with it was that it was expressing an opinion - or at least phrased like an opinion. "Some scholars claim that..." sounds much better than "probably [anything]". Secondly I feel it is wrong, and poorly stated. It could have mentioned Ugaritic or Akkadian Cuneiform or some example of this ancient source as you did. >>

Geez, it's phrased in a way to imply that the issue might be controversial. I know Indian nationalists who claim that the Indic writing systems (not the Arabic-based one for Urdu) are independent of ME influences or syncretism. If you can think of a better way of saying, go ahead. And don't you think the linguistic detail you are talking about would be better with an article on 'writing systems' or 'alphabets and syllabaries'?

[edit] Neutralization of /m/, /n/ and /N/ before the stops

Quote-In English there are three nasal phonemes, /m, n, ŋ/, as shown by the minimal triplet,

/sʌm/ sum
/sʌn/ sun
/sʌŋ/ sung

However, these sounds are not contrastive before plosives such as /p, t, k/. Although all three phones appear before plosives, for example in limp, lint, link, only one of these may appear before each of the plosives. That is, the /m, n, ŋ/ distinction is neutralized before each of the plosives /p, t, k/:

  • Only /m/ occurs before /p/,
  • only /n/ before /t/, and
  • only /ŋ/ before /k/.

Thus these phonemes are not contrastive in these environments, and according to some theorists, there is no evidence as to what the underlying representation might be. If we hypothesize that we are dealing with only a single underlying nasal, there is no reason to pick one of the three phonemes /m, n, ŋ/ over the other two.

(In some languages there is only one phonemic nasal anywhere, and due to obligatory assimilation, it surfaces as [m, n, ŋ] in just these environments, so this idea is not as far-fetched as it might seem at first glance.)

In certain schools of phonology, such a neutralized distinction is known as an archiphoneme (Nikolai Trubetzkoy of the Prague school is often associated with this analysis.). Archiphonemes are often notated with a capital letter. Following this convention, the neutralization of /m, n, ŋ/ before /p, t, k/ could be notated as |N|, and limp, lint, link would be represented as |lɪNp, lɪNt, lɪNk|. (The |pipes| indicate underlying representation.) Other ways this archiphoneme could be notated are |m-n-ŋ|, {m, n, ŋ}, or |n*|.

I've removed this, because it's not true. /t/ can occur after /m/. The word dreamt has /t/ after /m/ i.e. /drEmt/. 64.194.44.220 01:58, 20 December 2005 (UTC)

It's a standard example, usually with the proviso that it only holds within morphemes (dream-t is bimorphemic), but pumpkin (/ˈpɐmkɪn/ in my pronunciation) is a monomorphemic example, and we can probably come up with others. Someone may want to return it clarifying that it's a very strong tendency in English rather than absolute. kwami 02:19, 20 December 2005 (UTC)
Dreamt is not bimorphemic anymore than drank is. dreamt is an irregular past tense of dream. 64.194.44.220 18:50, 20 December 2005 (UTC)
Certainly it's bimorphemic: the [t] is the past tense. Some would argue that drank is bimorphemic as well, though that's a harder sell. Drank is problematic for the same reason that the phoneme itself is problematic: structuralism is a simplification that only works well in limited circumstances. But this whole article is about a structuralist concept. kwami 19:53, 20 December 2005 (UTC)
Irregular past tenses are not bimorphemic. Only regular past tenses that end in -ed are bimorphemic. 64.194.44.220 20:13, 20 December 2005 (UTC)
Let's be a little more open here, shall we? While I personally agree with 64.194.44.220, and even go a step further and reject the entire notion of morphemes (I employ a word-based approach), it would be folly of me to refuse to acknowledge that there are many, many people who analyze dreamt as dream + t and a vowel change (or something like that). I think there's a way to include the cut part--but the existence of words like dreamt cuts the legs right out from under archiphonemics, doesn't it? (forgot to sign in first) 22 Dec 2005
Why should it? If dreamt is anything, it applies only to English there could be a great many other languages out there where dreamt would be pronounced [dʒɹempt] or [dʒɹent] or [dʒɹemp] or what have you, and there could be a great many other examples of archiphonemes in English, such as the merger of voiced and aspirated stops after /s/. —Felix the Cassowary (ɑe hɪː jɐ) 00:55, 23 December 2005 (UTC)
Plenty of phonological theories consider morphological boundaries (thus morphophonemics). Dreamt only illustrates that the three English nasals are distinct at the ends of morphemes regardless of what follows, but it says nothing about whether they're distinct in NC clusters within morphemes. This is a common enough situation. That's why I brought up the pumpkin example. But even if you accept that pumpkin is /p@mkIn/, the fact remains that NC clusters in English morphemes are overwhelmingly homorganic. Perhaps pumpkin is a rare example of a word with a specific nasal before another stop, as in the polymorphemic cases, whereas nearly all monomorphemes have an archiphoneme |N|. I don't buy it, but it's not an unreasonable idea. kwami 01:32, 23 December 2005 (UTC)

[edit] What about a section on distribution?

Doesn't anyone else think the bulk of some material could be cut, but that a classic discussion of the phoneme needs to cover distribution as its own section? And better than the current article does.

[edit] /ŋ/ as English phoneme

Surely [ŋ] is just a phone of /n + g/ in English? I've removed the example for now (there are probably enough anyway). It was added by 82.84.164.233 (talk • contribs) a year and a half ago; if I'm wrong, by all means revert. —Simetrical (talk • contribs) 04:44, 14 March 2006 (UTC)

If [ŋ] is just a phone of /n + g/, then how do you distinguish the underlying forms of singer and finger? Nohat 05:26, 14 March 2006 (UTC)
Hmm. Interesting point. The latter could be /fɪnggər/, I suppose, but that does seem pretty weird. —Simetrical (talk • contribs) 02:11, 15 March 2006 (UTC)

[edit] Writing Systems

This section contains a number of misleading statements. For example, it claims that English's spelling is irregular whereas that of French is predictable by complex rules and context. An English word's pronunciation is predictable from its spelling roughly 80% of the time, much the same as French. Also, the spelling of an Arabic word is many times not predictable from its pronunciation, due to such confusing details as the alif maqsura, taa marbuta, and the dagger alif to name but a few. Perhaps Hebrew would be a better example here?

Can you give an example of a French word that has unpredictable spelling or an odd pronunciation? I mean other than obvious loanwords. Ƶ§œš¹ [aɪm ˈfɻɛ̃ⁿdˡi] 00:18, 2 December 2006 (UTC)
Retrieved from "http://en.wikipedia.org../../../p/h/o/Talk%7EPhoneme_3ea8.html"