Is aspiration in English phonemic?

< Previous | Next >

Valentln

New Member
Spanish - Mexico.
Hello. I was reading this thread where someone found this source, but I'm not sure if I completely understand the subject.
In Spanish, just as in French, a 'voiced' alveolar plosive (d) has negative VOT, whereas in English, a 'voiced' alveolar plosive has zero VOT, which is actually voiceless, but without aspiration.
In the two romance languages, a 'voiceless' alveolar plosive (t) has zero VOT, whereas in English, a 'voiceless' alveolar plosive has positive VOT, so it is aspirated.

Many people say that in English, voicing is phonemical, but aspiration is not. I think it's the opposite.
If someone were to utter a bilabial plosive with zero VOT, a native English speaker would hear a 'b', and not a 'p', because it wouldn't be aspirated.
Also, I've heard that word-initial voiced plosives are partially or completely devoiced.
Do you think I could use a Spanish 't' (zero VOT) and English speakers would hear a /d/?
What sounds more natural to you, a plosive with negative VOT, or one with zero VOT?
I hope I made myself clear.
 
  • Nino83

    Senior Member
    Italian
    whereas in English, a 'voiced' alveolar plosive has zero VOT, which is actually voiceless, but without aspiration
    In the two romance languages, a 'voiceless' alveolar plosive (t) has zero VOT, whereas in English, a 'voiceless' alveolar plosive has positive VOT, so it is aspirated.
    I've listened to the samples in your second link and to my Italian ear the first two samples, with negative and zero VOT, sound voiced [b, d, g].
    Maybe I found the problem. I read here that unaspirated voiceless consonants have a positive VOT (15 ms for /t/ and 30 ms for /k/) while aspirated voiceless consonants have a longer VOT (70 ms for /t/ and 80 for /k/ in English).
    Zero VOT (like that on those samples) still sounds voiced to me.
    Do you think I could use a Spanish 't' (zero VOT) and English speakers would hear a /d/?
    It depends on the Spanish VOT.
    For example in this work the VOT of both Italian and English /t/ was measured: Italian 32.62 ms English 80.95 ms, so I think it is not probable that an English speaker would peprceive a /t/ with a VOT of 32 ms as a /d/.
     
    Last edited:

    gburtonio

    Senior Member
    UK, English
    Many people say that in English, voicing is phonemical, but aspiration is not. I think it's the opposite.
    If someone were to utter a bilabial plosive with zero VOT, a native English speaker would hear a 'b', and not a 'p', because it wouldn't be aspirated.
    Also, I've heard that word-initial voiced plosives are partially or completely devoiced.
    Do you think I could use a Spanish 't' (zero VOT) and English speakers would hear a /d/?
    What sounds more natural to you, a plosive with negative VOT, or one with zero VOT?
    I hope I made myself clear.
    Generally in phonology English consonant pairs are distinguished by energy, so the distinction is between 'fortis' (≈ voiceless) and 'lenis' (≈ voiceless).

    Experiments have been carried out whereby recordings of English fortis plosives have been manipulated so that the aspiration is removed from the recording, but the VOT maintained (i.e. the aspiration is replaced with silence). Native speakers still hear fortis sounds (i.e. 'voiceless') when they listen to the manipulated recordings, so it appears that it's not the aspiration that makes us hear a fortis plosive, but the positive VOT.

    Native speakers would be very likely to hear a Spanish /t/ as an English /d/. However, once you become familiar with a language where the VOT works differently, you're less likely to make the mistake.
     

    Nino83

    Senior Member
    Italian
    In this study there are some data about the Spanish VOT for /p, t, k/, respectively 6.5 10.4 and 25.7 ms (compared to 58 70 and 80 ms of English).
    So maybe the Spanish /p, t/ (maybe not /k/) could be perceived as /b, d/ by monolingual English speakers.
    In this work it is said that the mean VOT for the Italian /p, t, k/ is 12 17 and 30 ms, so it is somewhat longer than in Spanish.
     

    gburtonio

    Senior Member
    UK, English
    Somewhat anecdotal, but I remember seeing a note written by a native English speaker of a word she'd heard in Greek, with its translation into English. The word she'd heard was either τίποτα [tipota] (= nothing) or μακάρι [makari] (= hopefully) – unfortunately I can't remember which, but it was one of the two. Whichever it was, what I noticed was that she's heard a voiced consonant, as she'd written (she didn't know Greek script), either 'dipota', or 'magari'. Greek has VOTs of 19 ms for [p], 27 ms for [t], and 49 ms for [k], according to:

    Raphael, L. J., Tobin, Y., Faber, A., Most, T., Kollia, B., & Milstein, D. (1995). Intermediate values of Voice Onset
    Time. In F. Bell-Berti, & L. Raphael (Eds.), Producing Speech: Contemporary issues: for Katherine Safford Harris (pp.
    117–127). New York: AIP Press
     

    berndf

    Moderator
    German (Germany)
    As a native speaker of a language where positive VOT is the distinctive feature of fortis stops as well, this story is very familiar. It happend regularly to me that I misheared Italian fortis stops as lenis (now less so as I live in a Romance language area and have adjusted to negative VOT as the distinctive property of lenis stops). Inversely, when I spell letter a letter combination like MBD (em-bee-dee) to an Italian colleague I can almost be sure he will write down MPT.
     

    Nino83

    Senior Member
    Italian
    So it's better to pronounce these consonants with aspiration in order to avoid ambiguities.
    I can almost be sure he will write down MPT.
    I think there is regional variation in Germany. For example, listening to these two pronunciations on forvo, the central one sounds like /'daavit/ while the Bavarian one sounds almost like /'taafit/ (even the intervocalic /v/ sounds like an /f/).
     
    Last edited:

    berndf

    Moderator
    German (Germany)
    So it's better to pronounce these consonants with aspiration in order to avoid ambiguities.
    Absolutely. Dialects that lack aspiration also lack the distinguishabity between fortis and lenis stops.
    I think there is regional variation in Germany. For example, listening to these two pronunciations on forvo, the central one sounds like /'daavit/ while the Bavarian one sounds almost like /'taafit/ (even the intervocalic /v/ sounds like an /f/).
    /f/ is not a stop and aspiration is not relevant. In many dialects of German [v] is still an invervocalic allophone of /f/ as it was in early Germanic when there was no /v/ phoneme. Bavarian has completely lost the notion of a voiced fricative, both phonemically and allophonically. I am an IT consultant and I have a customer in Munich. I always have to smile when I hear someone in the office there telling he wants to copy a file onto the surfer [søɐfɐ].:D
     
    Last edited:

    Nino83

    Senior Member
    Italian
    The voicelessness is even more pronounced in Austria. In this sample I hear /'tox/ for doch.
    I think this is the situation (VOT+ = positive; VOT- = negative; 0VOT = zero):
    Romance: /b, d, g/ = VOT-; /p, t, k/ = VOT+ (from 10 to 30ms)
    English, Dutch, Low German: /b, d, g/ = VOT- or 0VOT; /p, t, k/ = VOT+ (from 60 to 80ms)
    High German (expecially Bavarian): /b, d, g/ = from 0VOT to VOT+ (up to 30ms); /p, t, k/ = VOT+ (more than 60ms)

    In other words I hear the English /b, d, g/ (excluding some British accents in final position) voiced, while the same consonants seem voiceless in some German accents.
     

    berndf

    Moderator
    German (Germany)
    The voicing or non-voicing of lenis stops in those Germanic languages and dialect that rely on VOT+ for fortis stops is erratic. We simply don't care and don't pay attention. I am quite familiar with Eastern Austrian accents and the /d/ in that person's pronunciation of doch is not natural. The whole pronunciation is hyper-correct. What is particularly strange is that he combines VOT- with an energy level of the plosive release appropriate for fortis stops. Having said that, for a German ear it is still clearly a /d/ and not a /t/.
     
    Last edited:

    Nino83

    Senior Member
    Italian
    It seems he pronounces all the "d" in that way (another example Deutschland). What do you mean with hyper-correct? Is the Eastern Austrian pronunciation of "d", you're used to, more or less "voiced" than that of the speaker on forvo?
    (I agree on the fact that Germans don't care. The voicing of /b, d, g/, if I'm not wrong, is only allophonic because it's aspiration that makes the difference. Similarly to the Shanghainese language where non aspirated stops are pronounced [b, d, g], like in Middle Chinese, but that's not important because tone and aspiration make the difference in that case, not voicing)
     

    Valentln

    New Member
    Spanish - Mexico.
    I've listened to the samples in your second link and to my Italian ear the first two samples, with negative and zero VOT, sound voiced [b, d, g].
    Maybe I found the problem. I read here that unaspirated voiceless consonants have a positive VOT (15 ms for /t/ and 30 ms for /k/) while aspirated voiceless consonants have a longer VOT (70 ms for /t/ and 80 for /k/ in English).
    Zero VOT (like that on those samples) still sounds voiced to me.

    It depends on the Spanish VOT.
    For example in this work the VOT of both Italian and English /t/ was measured: Italian 32.62 ms English 80.95 ms, so I think it is not probable that an English speaker would peprceive a /t/ with a VOT of 32 ms as a /d/.
    In this study there are some data about the Spanish VOT for /p, t, k/, respectively 6.5 10.4 and 25.7 ms (compared to 58 70 and 80 ms of English).
    So maybe the Spanish /p, t/ (maybe not /k/) could be perceived as /b, d/ by monolingual English speakers.
    In this work it is said that the mean VOT for the Italian /p, t, k/ is 12 17 and 30 ms, so it is somewhat longer than in Spanish.
    I found on the article you shared what I stated before: "In English, "voicing" can successfully separate /b, d, ɡ/ from /p, t, k/ when stops are at word-medial positions, but this is not always true for word-initial stops. Strictly speaking, word-initial voiced stops /b, d, ɡ/ are only partially voiced, and sometimes are even voiceless". On the same article says the following: Simple unaspirated voiceless stops, sometimes called "tenius" stops, have a voice onset time at or near zero, meaning that the voicing of a following sonorant (such as a vowel) begins at or near to when the stop is released. (An offset of 15 ms or less on [t] and 30 ms or less on [k] is inaudible, and counts as tenuis).

    I read here that "word-initial phonemes /b/, /t/ and /g/ can be realized by a stop with negative, zero or positive VOT, but these values are usually in the range of -20 to +20 ms and so are perceptually equivalent to zero VOT". I guess the right term for this kind of stops is "tenius stops", and not "plosives with zero VOT", because most of the time they don't really have null VOT; actually, they can have negative or positive VOT, however, it's too short and it can't be perceived.

    If all this information is right, it means that Spanish and even Italian /p, t, k/ can be used as word-initial English lenis stops /b, d, g/.

    I listened again to these samples and I played them to three other Spanish speakers, and we all agree that the second one (zero VOT) sounds voiceless. This could be because our fortis plosives are generally tenius. Furthermore, our word-initial voiced stops have a negative VOT ranging from -40 to -90 ms or less.
    I wonder if more Italians hear zero VOT plosives as voiced, or if those samples are actually voiced, but Spanish speakers can't perceive it.

    Thanks to all for helping me out. :)
     

    Nino83

    Senior Member
    Italian
    In this work there is the (Canadian) French VOT (of monolingual speakers) for /p, t, k/, which is, respectively, 18 23 and 32 ms.
    So we have:
    Spanish: 6.5 10.4 25.7
    Italian: 12 17 30
    French: 18 23 32
    Greek: 19 27 49
    I wonder if more Italians hear zero VOT plosives as voiced, or if those samples are actually voiced, but Spanish speakers can't perceive it.
    Maybe because the consonants with a zero VOT are near to the limit between voiced and voiceless. In Italian, French and Greek the limit is higher (more than 10ms).
    Anyway as far as I remember, I've never mistaken an English initial /b, d, g/ as a /p, t, k/ (while it occurs with German lenis consonants, which can have a positive VOT).
     

    berndf

    Moderator
    German (Germany)
    I think it is about time to draw conclusions from this for the question of this thread. Perceptions of foreigners may be an indication how consistent and relevant certain features of English lenis and fortis stops are but in the end it can't answer the question. For that it only matters what triggers native speaker's perception as either lenis or fortis.
     

    Forero

    Senior Member
    Hello. I was reading this thread where someone found this source, but I'm not sure if I completely understand the subject.
    In Spanish, just as in French, a 'voiced' alveolar plosive (d) has negative VOT, whereas in English, a 'voiced' alveolar plosive has zero VOT, which is actually voiceless, but without aspiration.
    In the two romance languages, a 'voiceless' alveolar plosive (t) has zero VOT, whereas in English, a 'voiceless' alveolar plosive has positive VOT, so it is aspirated.

    Many people say that in English, voicing is phonemical, but aspiration is not. I think it's the opposite.
    If someone were to utter a bilabial plosive with zero VOT, a native English speaker would hear a 'b', and not a 'p', because it wouldn't be aspirated.
    I don't. A Spanish p sounds like a Spanish p, not like a b. And the p in speak still sounds like a p, not a b, even though it is not aspirated like the p in peak.
    Also, I've heard that word-initial voiced plosives are partially or completely devoiced.
    I don't believe this.
    Do you think I could use a Spanish 't' (zero VOT) and English speakers would hear a /d/?
    A Spanish t does not sound like an English d to me. Neither does the t in stake, though it is not aspirated like the t in take.
    What sounds more natural to you, a plosive with negative VOT, or one with zero VOT?
    I hope I made myself clear.
    In the English I speak, a voiced initial consonant always has a negative VOT.

    By the way, aspiration as I understand it is an audible sound that occurs at the release of some consonants and is independent of voicing. There are unvoiced consonants aspirated in varying degrees and voiced consonants aspirated in varying degrees.

    Aspiration is not phonemic in (my variety of) English in the sense of separating one consonant phoneme from another, but it does help to distinguish one word or phrase from another because it helps in locating morpheme boundaries (e.g. "Pike's peak" vs. "Pike speak") and stressed vs. unstressed syllables (more aspiration in stressed syllables).
     

    berndf

    Moderator
    German (Germany)
    In the English I speak, a voiced initial consonant always has a negative VOT.
    It's possible. But according to all studies I have read you would be a member of a tiny minority.
    By the way, aspiration as I understand it is an audible sound that occurs at the release of some consonants and is independent of voicing. There are unvoiced consonants aspirated in varying degrees and voiced consonants aspirated in varying degrees.
    Voiced aspiration exists but not in English.

    Also, I've heard that word-initial voiced plosives are partially or completely devoiced.
    I don't believe this.
    Take the first two (sorted by rating) recordings on but in Forvo. They are both pronounced by AmE speakers and that sound fairly representative to me. To the level of accuracy I could them with a smartphone waveform editor app, VOT is in both cases around +15ms (give or take 5), which is percepually zero.
     
    Last edited:

    merquiades

    Senior Member
    English (USA Northeast)
    I don't. A Spanish p sounds like a Spanish p, not like a b. And the p in speak still sounds like a p, not a b, even though it is not aspirated like the p in peak.I don't believe this.A Spanish t does not sound like an English d to me. Neither does the t in stake, though it is not aspirated like the t in take.In the English I speak, a voiced initial consonant always has a negative VOT.

    By the way, aspiration as I understand it is an audible sound that occurs at the release of some consonants and is independent of voicing. There are unvoiced consonants aspirated in varying degrees and voiced consonants aspirated in varying degrees.

    Aspiration is not phonemic in (my variety of) English in the sense of separating one consonant phoneme from another, but it does help to distinguish one word or phrase from another because it helps in locating morpheme boundaries (e.g. "Pike's peak" vs. "Pike speak") and stressed vs. unstressed syllables (more aspiration in stressed syllables).
    I would agree with you here.
    Aspiration is not so important in my northeast variety. Some people don't even do it. Voiced consonants are clearly voiced (but I can't calculate when VOT would occur)
    I have heard some German natives mix p/b, and say Paveria. They also don't understand me on occasion, but there could be numerous reasons for that!

    Note: Spanish d is a voiced interdental consonant so it wouldn't correspond to English d or t, or Italian either.
     

    Forero

    Senior Member
    It's possible. But according to all studies I have read you would be a member of a tiny minority.

    Voiced aspiration exists but not in English.


    Take the first two (sorted by rating) recordings on but in Forvo. They are both pronounced by AmE speakers and that sound fairly representative to me. To the level of accuracy I could them with a smartphone waveform editor app, VOT is in both cases around +15ms (give or take 5), which is percepually zero.
    I don't think 15ms is perceptually zero.

    The first but, by SeanMauch, sounds normal to me, definitely with a voiced b.

    The second but, by anakat, sounds "affected". The b is almost, but not quite, a Spanish p, and the t is on its way to becoming [ts].

    It is possible the audio file is lossily compressed and does not do justice to anakat's pronunciation.

    Voiced aspiration does exist in English, but it is not phonemic, except to help distinguish stressed syllables from unstressed ones.
     

    Forero

    Senior Member
    If not, you would have heard it as an aspirated /p/.
    Why?
    Yet the VOT is positive. The perception of voicing does not always agree with the physical fact of voicing.
    I am not convinced.

    I think for us to get on the same page, we may need to have a means of synchronizing the audio in question with a video showing both the time domain and the frequency domain. It would help to be able to adjust the sample length for the Fourier transform (because of the infamous uncertainty principle) and to see a simultaneous laryngograph to confirm.

    Otherwise you will continue to say you have evidence of something and I will have to continue to say I don't see it but in fact doubt it.

    And, until I see evidence to the contrary, I will continue to say that positive VOT does not imply an audible escape of air (which is what aspiration means to me).

    (As it is, my browser will only consent to playing what I click on after five or six tries, and Forvo presents its own "sophisticated" obstacles to data access. This is frustrating.)
     

    berndf

    Moderator
    German (Germany)
    I've got me doubts as well now because I hear the voicing, too. I will come back to you when I am in front of a computer. Smartphone apps have their limitations.
     

    berndf

    Moderator
    German (Germany)
    :oops:
    I had misread the waveform graph. I had measured the time from closure to voice onset and not from release to voice onset. The latter is of course negative.
     

    Forero

    Senior Member
    It's possible. But according to all studies I have read you would be a member of a tiny minority.
    I don't think so, but note that I speak Southern AmE, not some other variety, and I have noticed from speakers of other varieties a tendency to pronounce, for example, seventeen and seventy without the distinction I expect between the stressed -teen and the unstressed -ty, which either makes them sound too much alike, both with aspirated ts, or makes seventy sound like "7 D", with a hard (not tapped) d sound.

    In my dialect, seventy does not lose its t, as twenty often does, and that t is always either unvoiced or tapped (or both), and seventeen always has an aspirated, unvoiced t, but with less aspiration than for "7 T".
     

    berndf

    Moderator
    German (Germany)
    I don't think so, but note that I speak Southern AmE, not some other variety, and I have noticed from speakers of other varieties a tendency to pronounce, for example, seventeen and seventy without the distinction I expect between the stressed -teen and the unstressed -ty, which either makes them sound too much alike, both with aspirated ts, or makes seventy sound like "7 D", with a hard (not tapped) d sound.

    In my dialect, seventy does not lose its t, as twenty often does, and that t is always either unvoiced or tapped (or both), and seventeen always has an aspirated, unvoiced t, but with less aspiration than for "7 T".
    My reply was to this
    In the English I speak, a voiced initial consonant always has a negative VOT.
    statement (highlighting mine). In most varieties of English, aspiration is so ubiquitous in initial fortis stops that voicing of initial lenis stops is effectively optional.

    But I admit, my knowledge of those subtleties is limited with regard to AmE. I have primarily studied BrE and in the realisation of stops there are significant differences between American an British speakers. Most BrE speakers consistently aspirate their fortis stops in all positions (ignoring peculiarities like t-glotalisation), except in clusters sp-, st- and sk- (non-aspiration in these clusters is a feature common to all Germanic languages).
     

    Forero

    Senior Member
    My reply was to this

    statement (highlighting mine). In most varieties of English, aspiration is so ubiquitous in initial fortis stops that voicing of initial lenis stops is effectively optional.

    But I admit, my knowledge of those subtleties is limited with regard to AmE.
    I don't believe the speakers of AmE who always voice initial /b/, /d/, and /g/ are a tiny minority. It is indeed voicing, not aspiration, that makes a b different from a p and a d different from a t in English minimal pairs ("parrot's beak" v. "parrot speak", "too" v. "do", etc.), and I suspect it is a tiny minority who don't voice /b/, /d/, and /g/, if indeed it is any native speaker at all.

    And I think this
    Good, because I heard only voiced consonants in those two samples. :thumbsup:
    statement confirms what I am saying, especially since I hear the b in the second sample as "almost, but not quite, a Spanish p".

    In contrast, the "b" (Pinyin representation) sound in Mandarin Chinese is never aspirated but sounds voiceless to me at the beginning of an utterance (e.g. in the sentence bu yao [don't want / no]).

    This is exactly what the literature says about it too. The creators of Pinyin chose b not because they hear it differently than Wade and Giles but because it simplifies the spelling and is more likely to elicit an understandable sound from the average user of the Roman alphabet.

    In fact, I would be more likely to believe that the p in speak is aspirated, ever so slightly, than that the b in but is ever unvoiced.
     
    Last edited:

    Nino83

    Senior Member
    Italian
    As I said, Germans and Chinese (and Koreans) seem not to have voiced stops in initial position. On the other hand intervocalic lenis consonants can be voiced (for example in Shenzhen I often hear [ɖʐ] instead of [ʈʂ]).
     
    Last edited:

    Forero

    Senior Member
    I've listened to the samples in your second link and to my Italian ear the first two samples, with negative and zero VOT, sound voiced [b, d, g].
    To my ear, the "bz" and "az" samples (bilabial and alveolar with "zero" VOT) are unvoiced, with the bilabial one like an exaggerated Spanish p, but the velar one is voiced.
    In this sample I hear /'tox/ for doch.
    It sounds like /'dox/ to me.
    It seems he pronounces all the "d" in that way (another example Deutschland).
    Also voiced.
    these values are usually in the range of -20 to +20 ms and so are perceptually equivalent to zero VOT". I guess the right term for this kind of stops is "tenius stops", and not "plosives with zero VOT", because most of the time they don't really have null VOT; actually, they can have negative or positive VOT, however, it's too short and it can't be perceived.
    I don't know where that author gets the -20 to +20 ms figure for "perceptually equivalent to zero VOT", but I am fairly sure it does not reflect the ability of the average human ear.

    Before this thread, I was even starting to think the definitions had changed (closure being something other than physical closure, voicing being something other than audible laryngeal vibration, or aspiration being something other than audible air turbulence).

    Now I believe it is the assumption that VOTs in a wide range like -20 to +20 ms are zero that must be at the center of all the peculiar statements I've read recently about what distinguishes consonants in English, and possibly even in German, and the nature of aspiration and voicing.

    Of course I am still saying all this without access to sonogram data. Can anyone tell me the actual VOTs for the samples I have referred to in this post?
     

    berndf

    Moderator
    German (Germany)
    I don't know where that author gets the -20 to +20 ms figure for "perceptually equivalent to zero VOT", but I am fairly sure it does not reflect the ability of the average human ear.
    This is standard textbook opinion. From my own experience it seems plausible.
     

    berndf

    Moderator
    German (Germany)
    This could mean that for English speakers the difference between voiced and unvoiced is more similar to German than to Romance (i.e aspiration is quite important).
    In BrE it certainly is. In AmE apparently a bit less.

    I am not quite sure if fortis/lenis distinction can entirely be reduced to VOT. If we take the Chinese b, Romance speakers hear p and German speakers hear p. With English speakers it seems to be more complicated. American speakers generally hear p, British speakers often report they hear something that for them is neither a p nor b but a sound that has properties of both.
     

    Valentln

    New Member
    Spanish - Mexico.
    I measured the VOT of my fortis Spanish plosives (I used 5 words for each consonant, all of them were in word-initial position and adjacent to a stressed syllable with one of the five Spanish vowels). Their mean VOT was as follows: 9ms for /p/, 15ms for /t/ and 30ms for /k/. These values are significantly different from the Spanish values that are shown in this study (6.5, 10.4, 25.7), and they're also closer to the Italian values that Nino83 posted (12, 17 and 30).
    I'd like to mention that my /t/ and my /d/ are either apico-alveolar or lamino-alveolar, never interdental.

    I analyzed the 'zero VOT' samples that caused controversy. They're almost identical to my Spanish plosives, the bilabial plosive has a short-lag VOT of 10ms, the alveolar plosive has a VOT of 18ms and the velar plosive one of 29ms.

    I was clearly wrong when I said that Spanish has fortis plosives with zero VOT. It has plosives with positive VOT, often referred to just as 'short-lag VOT' because their VOT doesn't exceed 30ms (I'm not very convinced that –30ms/30ms is not perceptible). English normally has partially voiced lenis plosives, meaning that the voicing occurs at a short time before the closure of the consonant, just as in the word 'but' pronounced by Anakat, which has a negative VOT of –9ms. Both languages have plosives with short-lag VOT, but Spanish always has fortis plosives with positive short-lag VOT, not negative, that's why Anakat's pronunciation sounds 'affected', almost like a Spanish /p/, but not identical.

    Concerning the pronunciation of the word 'speak', or for that matter, the three clusters /sp/, /st/ and /sk/, I'm not sure if they sound like the fortis Spanish plosives. I used Praat to see how long was the VOT of the word 'speak', and I found this: the pronunciation of Clc30 has a short-lag VOT of 6ms, the pronunciation given by Rdbedsole has a short-lag VOT of 22ms. Both are American. I recorded myself pronouncing the same word and I produced a short-lag VOT of 16ms, which is longer than all the Spanish words I measured, but it's still within the parameter of fortis Spanish stops shown in this study.

    I read this article, and it shows that Forero's definitely not part of a minority. In the South of the US, people tend to pronounce word-initial lenis plosives with negative VOT. 740 of 951 (78%) tokens in the study were prevoiced, the median being –92ms (just like SeanMauch's 'but' (-169ms)). The other 22% were produced with positive short-lag (M=13.2ms). This means I could use my Spanish /p/ or /t/ (9ms and 15ms) and they could sound like lenis English plosives, but they would still sound sort of weird very often, so I'd rather use prevoiced lenis consonants to avoid misunderstandings. However, I'm pretty sure I've heard many Californians use short-lag (positive) plosives in function words like 'don't' or 'do'. Please listen to how this guy says 'dude'.

    Oh, and the word 'doch' by Wien has a short-lag of 16ms, very similar to my Spanish /t/.

    You can download all the audios I analyzed along with Praat screenshots of their respective VOTs.
     
    Last edited:
    < Previous | Next >
    Top