/l/ recognition

[Taernsietr]

So I've been analyzing this for a while now, and I think it's time for input.

I've always been bothered by [l] (and [lˠ]) appearing in certain places in phonetic transcriptions, in both English and Portuguese, since I don't quite recognize said phonemes in everyday speech. For example, [Rio de Janeiro] Portuguese /l/ appears as [l] before back and central vowels, though many people speak it as [lˠ]. Before /i/ it might appear as [l lˠ ʎ]. But before any consonant it's almost always transcribed as [lˠ], whereas I always hear [w], maybe [u̯~ʊ̯].

For example, a friend of mine has a velarized /l/ in most positions where I have a plain /l/:
lado 'side' -> [lˠadʊ] vs. [ladʊ]
My mother has a palatal where I have a velarized /l/ before /i/:
linha 'line' -> [ʎĩɲa] vs. [lˠĩɲa]
... but we both have coda /l/ as [w]:
jornal 'newspaper' -> [ʒohnaw]
... which is the same as final /u/:
cacau 'cocoa' -> [kɐkaw]

Now, I know these may vary between dialects, but I've seen this [l] description for most Brazillian dialects, and it seems specially funny for my dialect.


In English, coda /l/ sometimes seems like [w] to me as well, but more rarely than what I've described in BP (well, I mostly have contact with English through TV series and whatnot, so there you have it). For example, I have [ɪnɚnaʃonəlˠ] for 'international', and it's generally what I hear. But I have [feɪ̯ɾəw] for 'fatal', and I think I've never heard this word, for example, with final [l] or [lˠ].

What say you?

[linguoboy]

In English, coda /l/ sometimes seems like [w] to me as well, but more rarely than what I've described in BP (well, I mostly have contact with English through TV series and whatnot, so there you have it). For example, I have [ɪnɚnaʃonəlˠ] for 'international', and it's generally what I hear. But I have [feɪ̯ɾəw] for 'fatal', and I think I've never heard this word, for example, with final [l] or [lˠ].

Full on [w] for coda /l/ is stigmatised for most American English speakers, as it's strongly associated with (a) AAVE, (b) non-native (esp. Asian) varieties, and (c) defective juvenile speech. This is most true of vocalisation of syllabic /l/, but it also applies to cases like fatal or milk.

[Khvaragh]

Not exactly coda, but in a coda consonant cluster: isn't this what happened with <talk>? Something like [tʰal̩k] > [tʰaɫ̩k] > [tʰɔk~tʰɔ:k]? AFAIK, this is standard, I don't know anyone who preserves the [l].

[finlay]

In English, coda /l/ sometimes seems like [w] to me as well, but more rarely than what I've described in BP (well, I mostly have contact with English through TV series and whatnot, so there you have it). For example, I have [ɪnɚnaʃonəlˠ] for 'international', and it's generally what I hear. But I have [feɪ̯ɾəw] for 'fatal', and I think I've never heard this word, for example, with final [l] or [lˠ].

Full on [w] for coda /l/ is stigmatised for most American English speakers, as it's strongly associated with (a) AAVE, (b) non-native (esp. Asian) varieties, and (c) defective juvenile speech. This is most true of vocalisation of syllabic /l/, but it also applies to cases like fatal or milk.

I get the impression that L-vocalisation is more common in the UK. Still stigmatised though.

[----]

The coda realization of /l/ for me (and a few other people on this board I believe) isn't too different from /w/ phonologically, it's almost always [ʟ~ɰ] which is only differed from [w] with labialization, which can occur anyway in my speech as a result of rounded vowels in the nucleus; i.e. <mule> is [mjuʊ̯], *not* the expected [mjuʟ].

[Nortaneous]

I suspect the picture in English is a bit more complicated by the fact that (at least IMI) /w/ is commonly pharyngealized or uvularized or something along those general lines.

[Grimalkin]

I get the impression that L-vocalisation is more common in the UK. Still stigmatised though.

This might be a slight tangent, but, L-vocalisation is near universal among younger speakers where I live (Essex), and across much of southern england. I've seen vocalised L transcribed as [w], [ʊ] and even [o]. My impression is that it's more consonant-like, i.e. [w], at the end of a word, and more vowel-like before another consonant - like in bald [bɔʊd].

[Astraios]

bald [bɔʊd]

This word has too many different pronunciations. My friend from London pronounces it as [bo:d].

[----]

I have it as [bɑɫd̥]. The only time my /l/'s vocalize is word-finally and before coronals.

bald [bɔʊd]

This word has too many different pronunciations.

Any time liquids are involved in English there's like 100 different pronunciations.

[finlay]

I have it as [bɑɫd̥]. The only time my /l/'s vocalize is word-finally and before coronals.

yet somehow not in this word

[Taernsietr]

I suspect the picture in English is a bit more complicated by the fact that (at least IMI) /w/ is commonly pharyngealized or uvularized or something along those general lines.

Might as well be the case, however, the main distinction I'm concerned with is the loss of the lateral characteristic of the sound. In Portuguese, if my ears do not fail me, there is also uvular/pharyngealization to some extent in some people's speech.

The coda realization of /l/ for me (and a few other people on this board I believe) isn't too different from /w/ phonologically, it's almost always [ʟ~ɰ] which is only differed from [w] with labialization, which can occur anyway in my speech as a result of rounded vowels in the nucleus; i.e. <mule> is [mjuʊ̯], *not* the expected [mjuʟ].

Guess another word I spell with lateral /l/

On a slight change of course, what traits languages that have this /l/ -> [w] variation tend to have?

[Nortaneous]

L-vocalization implies ɫ, and ɫ implies ʎ. (English doesn't have /ʎ/, but my guess is that it picked it up from French.)

Are there any languages that have both l-vocalization and ʎ-vocalization (yeísmo)? Also, why does <yeísmo> have the accent when the stress is penultimate?

[TaylorS]

I have something like a lateralized [ɰ] after back vowels, otherwise it's plain [ɫ], though I am not a good sample for studying allophones of /l/ since I had speech therapy to say it properly. I believe Travis turns coda /l/ into [ʊ]

[TaylorS]

Also, why does <yeísmo> have the accent when the stress is penultimate?

to indicate a hiatus rather than a diphthong. I would pronounce it as [je.ˈiz.mo].

[----]

I have it as [bɑɫd̥]. The only time my /l/'s vocalize is word-finally and before coronals.

yet somehow not in this word

My fault, should have been more specific. When it's word-final and not in a consonant cluster, except when the other consonant is a coronal, it's vocalized. So <ball> would be something like [bɒː] but in <bald> it's still an L type thing. However, in <Balkan>, it would be [bɒːkɨn]

[tron cat]

In English, coda /l/ sometimes seems like [w] to me as well, but more rarely than what I've described in BP (well, I mostly have contact with English through TV series and whatnot, so there you have it). For example, I have [ɪnɚnaʃonəlˠ] for 'international', and it's generally what I hear. But I have [feɪ̯ɾəw] for 'fatal', and I think I've never heard this word, for example, with final [l] or [lˠ].

Full on [w] for coda /l/ is stigmatised for most American English speakers, as it's strongly associated with (a) AAVE, (b) non-native (esp. Asian) varieties, and (c) defective juvenile speech. This is most true of vocalisation of syllabic /l/, but it also applies to cases like fatal or milk.

Also New Zealand accents, but it's stigmatised here too, maybe not as much.

[finlay]

I have it as [bɑɫd̥]. The only time my /l/'s vocalize is word-finally and before coronals.

yet somehow not in this word

My fault, should have been more specific. When it's word-final and not in a consonant cluster, except when the other consonant is a coronal, it's vocalized. So <ball> would be something like [bɒː] but in <bald> it's still an L type thing. However, in <Balkan>, it would be [bɒːkɨn]

This doesn't make any sense, though. By your second sentence of this paragraph, you are saying that L vocalises word-finally and in clusters with coronals, hence [bɒː], [bɒːd], [bɒɫkɨn]. I think the key word you need to take out there is "not" – I would rephrase it as "It's vocalised when it's word-final or in a consonant cluster, except when the other consonant is a coronal".

Could also stem from you not understanding the word 'coronal', but I doubt that.

[----]

I have it as [bɑɫd̥]. The only time my /l/'s vocalize is word-finally and before coronals.

yet somehow not in this word

My fault, should have been more specific. When it's word-final and not in a consonant cluster, except when the other consonant is a coronal, it's vocalized. So <ball> would be something like [bɒː] but in <bald> it's still an L type thing. However, in <Balkan>, it would be [bɒːkɨn]

This doesn't make any sense, though. By your second sentence of this paragraph, you are saying that L vocalises word-finally and in clusters with coronals, hence [bɒː], [bɒːd], [bɒɫkɨn]. I think the key word you need to take out there is "not" – I would rephrase it as "It's vocalised when it's word-final or in a consonant cluster, except when the other consonant is a coronal".

Could also stem from you not understanding the word 'coronal', but I doubt that.

Oh gosh, you're right; I mixed up two of the terms. Replace every instance of coronal in there with dorsal. I don't know how I messed that up

[Travis B.]

I have something like a lateralized [ɰ] after back vowels, otherwise it's plain [ɫ], though I am not a good sample for studying allophones of /l/ since I had speech therapy to say it properly. I believe Travis turns coda /l/ into [ʊ]

What I have varies wildly in practice, actually, because it is controlled by both the closeness and roundness of what it follows, except if /l/ precedes /w/, where then the two are assimilated together as [wː].

For coda /l/ not before /w/, it seems that I have four different realizations that can actually show up, them being [ɤ̯], [o̯], [ɯ̞̯], and [ʊ̯]. Rounding always matches that which /l/ follows, and height is largely determined by the height of what it precedes and overall carefulness; in my everyday speech open-mid and opener consonants tend towards [ɤ̯] and [o̯] while close-mid and closer consonants tend towards [ɯ̞̯] and [ʊ̯], but in more careful speech even more open vowels can take [ɯ̞̯] and [ʊ̯] rather than [ɤ̯] and [o̯].

Note that my [o̯] and [ʊ̯] are generally not equal in quality to [o] and [ʊ], as [o̯] is less rounded than [o] as it can only take its roundness from the less-rounded [ɒ], and [ʊ̯] is almost invariably backer than [ʊ], especially as it can never undergo any centralization.

Similarly, I invariably vocalize syllabic /l/, from /ʊl/ and /əl/, as either [ɯ̞] or [ʊ] depending on the roundness of what it follows (i.e. whether it follows /w/, as that is the only rounded consonant I have) in careful speech or when stressed, while in everyday speech when not unstressed I often open my careful realization of [ɯ̞] to [ɤ], particularly finally. Note that my [ʊ] here is not the same as my [ʊ] from /ʊ/, as that is almost always at least a bit centralized, and often very much so, whereas this [ʊ] always remains fully backed.

Unstressed intervocalic /l/ is also vocalized in everyday speech in a manner similar to coda /l/, except that the vowels on both sides seem to have some part in determining the closeness and roundness of the resulting glide, and I have yet to figure out the exact rules here, except that the closer vowel on either side seems to have a part in determining its closeness.

All other cases of /l/ not accounted for above except geminate /l/*, whether from /ll/ or from any other source, will typically vocalize in everyday speech to [ɰ], e.g. prevocalically (especially as part of consonant clusters) and in stressed/careful intervocalic environments. However, in careful speech any of these environments may not be vocalized but rather be realized unvocalized as [ʟ̞]. Note that even in careful speech /l/ in prevocalic consonant clusters is very likely to be vocalized, while prevocalic /l/ not in clusters may be left unvocalized.

Conversely, geminate /l/* is never vocalized, but always is realized as [ʟ̞ː]. Note that aside from /ll/ it particularly comes from consonant cluster reduction, e.g. that of /dl/.

(I should note that this is not idiosyncratic at all, but rather that my dialect lacks unvocalized coda and syllabic /l/ altogether, and whose preservation of unvocalized /l/ even in very careful speech is limited to prevocalic, intervocalic, and geminate environments environments, aside from individuals who have sporadically borrowed different /l/s from other dialects.)

* Note that syllabic /l/ followed by an onset/intervocalic /l/ in the next syllable does not count as a geminate /l/, as that is just how /ʊl/ and /əl/ before a vowel are typically realized, even when that vowel belongs to another morpheme or even another word, where essentially an extra /l/ is added before the vowel synchronically without forming a geminate.

[Travis B.]

In English, coda /l/ sometimes seems like [w] to me as well, but more rarely than what I've described in BP (well, I mostly have contact with English through TV series and whatnot, so there you have it). For example, I have [ɪnɚnaʃonəlˠ] for 'international', and it's generally what I hear. But I have [feɪ̯ɾəw] for 'fatal', and I think I've never heard this word, for example, with final [l] or [lˠ].

Full on [w] for coda /l/ is stigmatised for most American English speakers, as it's strongly associated with (a) AAVE, (b) non-native (esp. Asian) varieties, and (c) defective juvenile speech. This is most true of vocalisation of syllabic /l/, but it also applies to cases like fatal or milk.

So why the hell do I keep on hearing such vocalized coda pronunciations of /l/ all the fucking time from North American English-speakers (and I don't mean from just back in Wisconsin, but from all over the place in the northern part of the eastern US, all the way out to here in Maryland), and I mean individuals who practically speak General American too and whose speech really is not stigmatized one bit?

Fucking hell, I only infrequently actually hear coda [ɫ] from anyone I come into contact with in real life, much the less the syllabic [ɫ̩]...

[äreo]

My syllable-final /l/ is pretty much always a lightly rounded [o], and no one really seems to notice in Florida or in Texas, and in fact most kids I hear have some degree of l-vocalization as well. It seems much commoner in the States than you might think.

[Travis B.]

Now that I think of this, this is likely a regional thing, actually. While coda/syllabic /l/ vocalization seems extremely common and unmarked in the northern part of the eastern US, when I have heard speech from people speaking GA-ish varieties in more westerly parts of North America, coda/syllabic /l/ vocalization has often been entirely absent. (I have noticed this because the lack of it sounds very overly careful and hence marked to my ears, almost as if one is trying to speak GA as a non-native speaker, whereas its presence is so unmarked to my ears that I normally do not notice it unless I am listening for it.)

(Heh - when I try to speak with [ɫ] for nonsyllabic /l/ and [ɫ̩] for syllabic /l/, to me it really sounds as if am a foreigner trying to speak English...)

[Ser]

Now that I think of this, this is likely a regional thing, actually.)

Maybe, [ɫ] and [ɫ̩] seem to be the rule around here...

[Nortaneous]

I think it might just be that some people with l-vocalization have an unrounded, usually pharyngealized/uvularized/whatever vowel, which is obviously going to be harder to distinguish from [5].

[Travis B.]

I think it might just be that some people with l-vocalization have an unrounded, usually pharyngealized/uvularized/whatever vowel, which is obviously going to be harder to distinguish from [5].

My thoughts seem to be similar here as well...