zompist bboard

How common is it for glottal stop and zero to indisputably contrast phonemically in word-initial position? I'm looking for a contrast that exists both in the lexicon and at the surface. For example, for German, it is argued that glottal stops are a consequence of stress.

On a related issue, should the term 'zero consonant' be taken to embrace glottal stops?

it's certainly attested. the classic example is hawaiian -- /alo/ 'front' vs. /ʔalo/ 'dodge'.

dunno how *common* it is, but UPSID lists a phonemic glottal stop in 48% of its languages, and i'd guess it contrasts initially with zero in a good portion of those. incidentally, UPSID reports a phonemic voiced glottal stop in nenets, which i'm adding to my gigabyte-large file called why_upsid_is_garbage.txt

Nortaneous wrote:which i'm adding to my gigabyte-large file called why_upsid_is_garbage.txt

For those of us who don't use it (often), why is it?

In Arabic there's a word-initial contrast, although it's been messed up in various dialects. The distinction is there when the word is preceded by another word as opposed to utterance-initial. Utterance-initially there is a glottal stop, but it can be elided (sometimes but not always along with the following vowel) in utterance-medial or final contexts.

Zju wrote:

Nortaneous wrote:which i'm adding to my gigabyte-large file called why_upsid_is_garbage.txt

For those of us who don't use it (often), why is it?

There's a large number of errors. Reporting the wrong phone due to scribal errors, or they just look at the symbol without even bothering to check to see what it's supposed to represent, or just flat-out wrong information that can't be reconciled (except maybe if they're using extremely odd sources that no one else agrees with). I thought I once even ran into a language that had a completely wrong inventory because they misspelled one language as a different one, but that might be me misremembering and it's not really *that* bad. Basically, UPSID can be used to help direct your searches into particular languages, but take absolutely nothing it says at face value.

Iirc languages can phonemically contrast zero-initial with a glottal stop, but that phonetically such a contrast won't be realized utterance-initially. There's no way (again, iirc) to distinguish between the initiation of voicing because of a preceding glottal stop and the initiation of voicing because it's the beginning of an utterance, unless a glottal stop has other effects like turning the vowel creaky. So if you're trying to hear the difference yourself, trying to say /a/ as an utterance and /?a/ as an utterance won't work because they're indistinguishable, you have to put them in context like with /sari a/ versus /sari ?a/.

That hypothesis strikes me as not so productive, because if we assume that all so-called utterance initial vowels really do have a glottal stop in front of them, it becomes a lot more difficult to explain languages which develop word initial consonants like /h, j, w, ŋ/ out of apparently nothing. I think the situation is actually that epenthetic word [or more pedantically, utterance]-initial glottal stop is quite common, but in other situations there is/was some kind of glide going on. There's no reason they can't open their glottis and then only afterward proceed with airflow.

vokzhen wrote:There's no way (again, iirc) to distinguish between the initiation of voicing because of a preceding glottal stop and the initiation of voicing because it's the beginning of an utterance, unless a glottal stop has other effects like turning the vowel creaky. So if you're trying to hear the difference yourself, trying to say /a/ as an utterance and /?a/ as an utterance won't work because they're indistinguishable, you have to put them in context like with /sari a/ versus /sari ?a/.

Do you mean "acoustically indistinguishable"? There is definitely an articulatory difference (though I suspect a three-way distinction with also [ɦa] would be more difficult).

vokzhen wrote:

Zju wrote:

Nortaneous wrote:which i'm adding to my gigabyte-large file called why_upsid_is_garbage.txt

For those of us who don't use it (often), why is it?

There's a large number of errors. Reporting the wrong phone due to scribal errors, or they just look at the symbol without even bothering to check to see what it's supposed to represent, or just flat-out wrong information that can't be reconciled (except maybe if they're using extremely odd sources that no one else agrees with). I thought I once even ran into a language that had a completely wrong inventory because they misspelled one language as a different one, but that might be me misremembering and it's not really *that* bad. Basically, UPSID can be used to help direct your searches into particular languages, but take absolutely nothing it says at face value.

yeah, there are loads of language-specific errors -- it's almost guaranteed that any given language inventory will have at least one error in it. (in the case of nenets, apparently the glottal stops are word-final allophones of other consonants, and the 'voiced glottal stop' is an artifact of different consonants that reduce word-finally to the glottal stop behaving differently in sandhi.) and it's a very limited resource that relies on outdated papers ('nenets' is properly two languages, and the most recent paper it cites for it is from 1975) and probably won't ever be updated again. the only reason to use it over PHOIBLE is that PHOIBLE b. doesn't have search.

Iirc languages can phonemically contrast zero-initial with a glottal stop, but that phonetically such a contrast won't be realized utterance-initially. There's no way (again, iirc) to distinguish between the initiation of voicing because of a preceding glottal stop and the initiation of voicing because it's the beginning of an utterance, unless a glottal stop has other effects like turning the vowel creaky. So if you're trying to hear the difference yourself, trying to say /a/ as an utterance and /?a/ as an utterance won't work because they're indistinguishable, you have to put them in context like with /sari a/ versus /sari ?a/.

what, it'd be possible for a language to insert ɦ (or h, or a homorganic glide as in mandarin(?)) before word-initial vowels

thetha wrote:That hypothesis strikes me as not so productive, because if we assume that all so-called utterance initial vowels really do have a glottal stop in front of them, it becomes a lot more difficult to explain languages which develop word initial consonants like /h, j, w, ŋ/ out of apparently nothing.

palauan ŋ-excrescence was apparently morphological, but that still leaves nganasan for ŋ

Word initial glottal stop appears to be default for vowel-initial words in German but IIRC in Gaelic it's /j/. Maybe these can be generalised further. Maybe it is a general thing in the Goidelic languages? Dutch maybe does like in German?

Nortaneous wrote:

thetha wrote:That hypothesis strikes me as not so productive, because if we assume that all so-called utterance initial vowels really do have a glottal stop in front of them, it becomes a lot more difficult to explain languages which develop word initial consonants like /h, j, w, ŋ/ out of apparently nothing.

palauan ŋ-excrescence was apparently morphological, but that still leaves nganasan for ŋ

I could have sworn that there were even more examples for that one. I know Nenets also had the same change; e.g. in the word "nenets"

thetha wrote:I could have sworn that there were even more examples for that one. I know Nenets also had the same change; e.g. in the word "nenets"

The full list of documented outcomes of Proto Samoyedic vowel initial words is ŋ- in Nganasan, ŋ- before back vowels and ɲ- before front vowels in Nenets (Tundra and Forest) and Yurats, Ø- before back vowels and n- before front vowels in Forest Enets, and Ø- (vocalic onset) everywhere else.

vokzhen wrote:Iirc languages can phonemically contrast zero-initial with a glottal stop, but that phonetically such a contrast won't be realized utterance-initially. There's no way (again, iirc) to distinguish between the initiation of voicing because of a preceding glottal stop and the initiation of voicing because it's the beginning of an utterance, unless a glottal stop has other effects like turning the vowel creaky. So if you're trying to hear the difference yourself, trying to say /a/ as an utterance and /?a/ as an utterance won't work because they're indistinguishable, you have to put them in context like with /sari a/ versus /sari ?a/.

The first point is a useful one, because it may be relevant for writing, especially if words are normally sounded out. The ease with which glottals stop consonants and independent vowels substitute for one another in Brahmic scripts had struck me, and I was wondering if it were just that there were no cases of contrast in the relevant languages.

Initial unaspirated stops have been reported to have an effect on vowels in Thai, and the initial glottal stop, often dismissed as unreal, has been recorded as having the same effect.

There may be another case where an initial contrast is possible. There are the initial clusters [ʔj] and [ʔw] which exist in various Tai dialects, and contrast with [j] and [w]. These contrasts appear in the native Indic writing systems.

vokzhen wrote: Iirc languages can phonemically contrast zero-initial with a glottal stop, but that phonetically such a contrast won't be realized utterance-initially. There's no way (again, iirc) to distinguish between the initiation of voicing because of a preceding glottal stop and the initiation of voicing because it's the beginning of an utterance, unless a glottal stop has other effects like turning the vowel creaky. So if you're trying to hear the difference yourself, trying to say /a/ as an utterance and /?a/ as an utterance won't work because they're indistinguishable, you have to put them in context like with /sari a/ versus /sari ?a/.

I'm not convinced. I can hear the difference just fine without context.

gestaltist wrote:I'm not convinced. I can hear the difference just fine without context.

I agree; vokzhen seems mistaken.

Richard W wrote:There are the initial clusters [ʔj] and [ʔw] which exist in various Tai dialects, and contrast with [j] and [w].

Okinawan also has this contrast. The glottal stop isn't otherwise contrastive though.

kanejam wrote:

gestaltist wrote:I'm not convinced. I can hear the difference just fine without context.

I agree; vokzhen seems mistaken.

Me, too.