Talk:Voiced velar fricative

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.

I have two books phonetics that say that the sound occurs as an allophone of /g/ in Spanish. Svenskans fonologi by Claes Garlén (pg. 87, 210) and Fonetikens grunder by Olle Engstrand (pg. 148).

Peter ^Isotalo 04:03, Jun 7, 2005 (UTC)

Would you specify the dialect? This isn't the case for Mexican Spanish, nor for any other variety of Spanish I've heard (not that I've heard that many), nor for the standard dialects used by the Spanish-language phonologies I've read, nor for the sources used by the Spanish phonology article in Wikipedia, nor it is mentioned as a variant in the Spanish dialects and varieties article, etc.

However, it is extremely common for basic phonetics textbooks to get this wrong, just as Hebrew and Arabic are commonly but erroneously claimed to have voiced pharyngeal fricatives, perhaps because for a long time the velar fricative symbol pulled double duty for the approximant, as the pharyngeal symbol still does, and when people see the gamma, they just assume it means "fricative". (An example is Daniels & Bright, The World's Writing Systems, which shows a [ɣ]. However, their phonetics generally isn't very accurate.) Or else a book will simply use a gamma, without bothering to say that they're using it to represent an approximant. Anyway, the blanket statement that "In Spanish, [ɣ] occurs as an intervocalic allophone of /g/" is wrong. kwami 07:25, 2005 Jun 7 (UTC)

There's a sound sample of the allophone in Fonetikens grunder (written in 2004, btw) on the publishers webpage. It's a European Spanish rendition of el carro de Rodriguez, which is transcribed as [el 'kar:o ðe roðɾiɣeθ]: wav-file. It is used to illustrate the differences in /r/-realizations foremost, but also to show how the voiced plosives /b, d, g/ are realized as fricatives that are leaning towards approximants. The exact same feature is present in Portuguese and Catalan, which are both included in the IPA handbook and are described as fricatives. Galician shows similar allophones except that they seem to be decidedly approximantic. This is also described in the handbook. Whatever the situation of the whole fricative/approximant situation is, it is either not accepted by all phoneticians or you might be extrapolating the theories too far.

Since you're questioning the use confirmed by some really heavy duty sources here, could you please tell me what your competence for these kinds of assertions are? Are you a trained phonetician? Have you studied phonetics on an academic level? I know I haven't, and that's why I'm do not try to assert my own analyses over that of serious phonetic literature and the authority of the IPA.

Peter ^Isotalo 20:00, Jun 7, 2005 (UTC)

And about the Arabic and Hebrew, are you aware that both languages have phonologies in the IPA handbook and both are assigned pharyngeal fricatives? There is indeed a tendency for voiced fricatives (and plosives) to be realized as approximants in colloquial speech, but the way you describe it makes it sound as if the approximants are simply the "true sounds" with the rest being some kind of historically backward compromises in dire need of being cleansed. Again, I think you're exceeding your competence in these matters.

Peter ^Isotalo 21:19, Jun 7, 2005 (UTC)

No, in the Handbook, neither Arabic nor Hebrew is described as having a voiced pharyngeal fricative. Take a look at the charts: each language has only a single (voiceless) phoneme [ħ] in that cell. The Arabic ayn is described as being a pharyngealized glottal stop, [ʔˁ], whereas the Hebrew ayn is described as being a pharyngeal approximant, and is placed in the approximant row of the chart. This is despite the fact that both Arabic and Hebrew are transcribed using the symbol ʕ. That is, in the IPA's own handbook, the symbol ʕ is not reliably a fricative!

True, European Portuguese is described as having a voiced velar fricative allophone for /g/. However, in Galician, which is basically a northern Portuguese dialect that's been influenced by Spanish, it says that "they have approximant variants [β̞ ð̞ ɰ]". Also, in Catalan, although the fricative symbol ɣ is used, it's then specified as being an approximant, not a fricative! The exact words are, "The three voiced plosives have approximant variants [β], [ð], [ɣ] in onset positions ..." That is, it seems that Catalan is just like Galician. Actually, thanks for pointing that out: it shows that none of the voiced fricative symbols can be relied upon to be fricatives, even in official IPA usage.

I've checked out your sample, but the segment is so short it's hard to tell what it is. It's not obviously a fricative (though not obviously an approximant either). I've looked at it in Praat, and there is no trace of frication in the spectrogram up to 7700 Hz, after which the signal has been filtered out. The waveform is just slightly irregular, but not what I'd expect from a fricative. What it looks like to me is a voiced plosive with very little occlusion -- which isn't a bad definition for an approximant. However, there's enough background noise that I wouldn't want to swear just from this soundfile that it's not a fricative.

I have studied phonetics at a graduate level, though I'm certainly not a phonetician. However, I've seen enough inaccuracies and inconsistancies in reputable sources that I'm leery of claims that aren't corroborated by laboratory measurements. Good linguists miss the lateral flap all the time, for instance, simply because they aren't trained to hear it. (That's the r sound in Japanese and the l in much of West Africa.) Also, when giving a phonemic breakdown of a language, it's very common to take a "good enough" attitude. Often a linguist's aim will be to show the phonemic contrasts, and isn't particularly concerned about the exact phonetic realizations of those phonemes. There's even a common attitude in Chomskyan circles that language description isn't a worthwhile endeavor, that it's the theory that matters. As for there being historical residues "in dire need of being cleansed", I do believe that. French /t/ is almost always described as being dental, for example, even though it's alveolar, because that's the convention. (It's just laminal, unlike English, where /t/ is apical.)

I must say that, at least in Mexican and Central American Spanish, I've never noticed a velar fricative. When I pronounce the sound as an approximant, people compliment my pronunciation. This is in marked contrast to Cairene Arabic or Dutch, where the frication is unmistakable. That's why I think Arabic is a good example of a velar fricative, but Spanish is not. (There's a question as to whether the Arabic sound is velar or uvular, but I believe that's a dialectical issue.) Perhaps Castillian Spanish does have a fricative - that I wouldn't know. Or it may be accurate to say that there is slight frication in all Spanish dialects. But at least in Mexico, it's nowhere near a prototypical fricative as it is in some Dutch and Arabic dialects.

Would that be a fair compromise? That the Spanish allophone is often described as a fricative, but it's closer to being an approximant, although perhaps with some frication? kwami 07:00, 2005 Jun 8 (UTC)

I will second the observation that it tends to be phonetically naïve linguistics texts that describe Spanish intervocalic voiced stops as fricatives. They are in fact approximants, which may under certain conditions (like any approximant) be lightly fricated. Using the unmodified symbols for fricatives β and ð may make one come to the erroneous conclusion that there are a set of fricatives in spanish [β], [v], [ð], [z], [ɣ] whose production is contrasted only by place of articulation. In fact [β], [ð], [ɣ] are produced as approximants that are lightly fricated (if at all), but [z] as in de[z]de is a full fricative (I can't think of any examples with /f/ followed by a voiced consonant (there may not in fact be one, but if there were it would be pronounced as a fully fricated [v]).

Most details descriptions of Spanish phonetics I have seen include the "lowered" diacritic underneath β and ð (β̞ and ð̞). For the velar, some use ɣ̞ and others use ɰ (the velar approximant). The difficulty is that the IPA gives no symbols for bilabial and dental approximants, but it does have a symbol for a velar approximant. Some writers have taken this gap in the IPA to be significant: that the phonetic realizations of intervocalic [b, d, g] should be transcribed using an analogous set of symbols; and since there are no symbols for the bilabial and dental approximants, then one should use the symbols for the fricatives with the "lowered" diacritic, signifying less (or no) frication. Others simply view the gap as merely an oversight and use the symbol for the velar approximant that is available (ɰ) and "make do" with the symbols for fricatives with a diacritic for the other two.

Ideally, there would be IPA symbols for the bilabial approximant and the dental approximant, or at the very least there would be a diacritic that explicitly means "no turbulent airflow (frication)". That would make for an easy solution for an orthogonal set of symbols to transcribe these allophones. The sections of the chart where such symbols would go are white (not shaded) indicating that the IPA considers them to be possible articulations, just not ones which have been assigned symbols. The topic I gather is something of a sore spot in Spanish phonetics and there has probably been some pressure on the IPA to assign symbols to these sounds. However, there are certain elements of the IPA that labor under the conceit that the alphabet is actually a kind of universal phonemic alphabet, and if no language distinguishes bilabial fricatives and approximants or dental fricatives and approximants, then diacritics will suffice for transcribing these sounds.

Given that, there is another general principle for the use of the IPA that diacritics can be excluded if the level of phonetic detail doesn't necessitate them. Thus ɣ is a perfectly valid transcription for intervocalic /g/ in Spanish because one can work under the assumption that the essential difference between g and ɣ is that ɣ is a continuant and g is not. It may not be the case that use of a fricative symbol necessarily implies turbulent airflow. But then again it might—it's all very vaguely defined.

In sum, both ɣ̞ and ɰ are used to transcribe intervocalic /g/ in Spanish. There are benefits and drawbacks to both methods, as the IPA doesn't provide a "best" solution. The sound is not fricated like /s/ and /f/ (or [z] and [v] for that matter), but is very lightly fricated if at all.

As for whether this page should include Spanish among the examples, I am ambivalent. At the very least, we could mention that some phonetic descriptions of Spanish give the symbol for the "voiced velar fricative" for the sound /g/ in the environment where it is weakened, but there is little or no turbulent airflow and most detailed descriptions of Spanish phonetics use instead the symbol for the velar approximant. Nohat 08:50, 8 Jun 2005 (UTC)

P.S. I know the environment is not strictly intervocalic, but that's not important for this discussion.

From Nohat's post There are benefits and drawbacks to both methods, as the IPA doesn't provide a "best" solution.

This is exactly what I was trying to say. IPA, though pretty detailed all things considered, is a compromise. It's more precise than your standard language orthography, but can never be precise enough to cover every imaginable deviation or minor detail. This is the essence of this discussion and I think kwami is being very unreasonable about a lot his edits.

kwami, you're making a lot of edits that seem to be based on the perception that there is One True Version, which needless to say, is quite obviously not NPOV. Please don't be so eager to completely assert your views on the articles you edit. You're attempting to oversimplify very complicated issues and you're doing it in a very aggressive manner to boot. Please try to be a bit more open to criticism.

Peter ^Isotalo 13:01, Jun 8, 2005 (UTC)

I thought the idea was to describe the International Phonetic Alphabet? I hardly see a problem with trying to be accurate. Spanish does not have a velar fricative. It has a velar approximant. Taking an attitude of "fricative, approximant, what's the difference?" is doing a disservice to those of our readers who are trying to figure out what the difference is. If you wish to use gamma when describing Spanish, because you're only trying to capture the phonemic essentials and don't care about the phonetic details, fine. But Spanish as an example of a velar fricative is wrong. The same for [c], which is commonly used to represent an affricate in Indic languages. Using Hindi as an example of [c] would also be wrong, though a note that this is common usage would be appropriate. As Nohat said, approximants are often slightly (and sometimes not so slightly) fricated, e.g. in emphatic usage. Should I give English as an example of a language with a voiced palatal fricative, just because when I say "Yes, indeed!" with the right emphasis, the 'y' is slightly fricated? That's a comment for the approximant article. I was hoping that you would correct the comment on Spanish yourself, but since you won't, I will. kwami 18:14, 2005 Jun 8 (UTC)

I'm sorry if I came across as being overly critical, but your somewhat high handed attitude is quite taxing on one's patience considering that you're making a lot of edits that are downright reckless and to some extent even erroneous. I appreciate your input, but please try to tone down your own POV and try to balance your views with established use. Most importantly, though, show some proper respect when discussing. Right now you're resorting to quite petty and low-level rhetorical feints and quite obvious incivility. This degrades you as much as it does me, so please try to stick to factual discussion.

Peter ^Isotalo 20:43, Jun 8, 2005 (UTC)

Sorry for being rude, Peter. I would have worded things differently if I'd waited a day. I think this has taxed both of our patience. kwami 21:36, 2005 Jun 8 (UTC)

[edit] It doesn't sound right

The sound sample doesn't sound right. My name is pronounced like this, and according to Dutch_language#Consonants, it is the Voiced velar fricative. However, this sound sample sounds clearly softer than the way we pronounce it in Dutch. Where does this sound sample come from? Gerrit ^CUTEDH 20:57, 17 September 2005 (UTC)

P.S. Please note that I'm not a linguist, so I can't really follow the discussion above. I don't know the technical words, I don't know what a voiced velar fricative is, I just know that the sound I hear is not like my name, but it still is linked in the Dutch language article as if it is! Gerrit ^CUTEDH 21:01, 17 September 2005 (UTC)

Footnote #3 on the Dutch consonant chart might explain the discrepancy: for most Dutch speakers, 'g' is pronounced [x], not [ɣ], at least when it's the first sound in a word. (For some speakers it's only [ɣ] between vowels.) Why don't you listen to the sound at Voiceless velar fricative and tell us if that sounds better? Also, before the vowel [e], the sound might be closer to a Voiceless palatal fricative; the velar might only occur near "back" vowels like [a, o, u]. If these sounds are still off, then we might have some work to do. kwami 22:08, 17 September 2005 (UTC)

[edit] Greek letter gamma?

Current version says the IPA symbol "ɣ" is the "Greek letter gamma". In contrast, the Unicode code charts have this character as U+0263 LATIN SMALL LETTER GAMMA [1], distinguished from "γ" U+03B3 GREEK SMALL LETTER GAMMA [2]. Would it be more correct to say the IPA symbol is a variant of the Greek gamma? Stebulus 04:56, 13 August 2006 (UTC)

No responses, I guess. I say it's a variant. Updating page accordingly. Stebulus 00:32, 12 October 2006 (UTC)

Yeah, that's what I would call it. Ƶ§œš¹ ^{[aɪm ˈfɻɛ̃ⁿdˡi]} 02:04, 12 October 2006 (UTC)