Suprasegmentals

[din]

Hi

Whenever I read about sign language_{(if anyone has any good source on any sign language, please tell me about it -- that's how little I'm able to find)}, I am intrigued by the 'suprasegmental' features, such as facial expressions or mouthing which modify the signs. Another good example are the aspectual distinctions in, for example, ASL (see here).

I've come across suprasegmental features in spoken languages before, too. I believe Sami has suprasegmental palatalization. I'm not sure if it represents any kind of morphological operation, or if it's just a characteristic of a particular word (like word stress*).

_{* Yes, I am aware that English also uses stress in some words to derive nouns and adjectives from verbs, of which wiki has an overview here.}

I'm interested in these things, and would like to know if you know more examples of suprasegmental features in the world's spoken languages. I'm particularly interested in suprasegmental morphemes

Thanks in advance

[din]

Nothing at all? Is my question too unspecific? Because it was intentionally so: I just want to know what suprasegmentals exist, and if they have any specific function in a particular language, or if it's phonemic (like stress in English).

Somebody must have read something somewhere

[Radius Solis]

Tone, stress, pitch accent, and prosodic structuring are all suprasegmental, whether contrastive or not - and almost all languages have at least one of those things. So if you want to find out more, look those things up on wikipedia.

Occasionally other things can become suprasegmental, such as vowel frontness or height applying to the word level instead of the segment level; this is called "vowel harmony" if you want to look it up. I have heard of nasality being suprasegmental in some Amazonian languages, and of sibilant POA being suprasegmental in some Indic languages, but I can't recall which; this is occasionally called "consonant harmony". Advanced Tongue Root (ATR) is probably suprasegmental in at least some of the languages that contrast it.

For examples of suprasegmentals that are contrastive:
tone: Mandarin
stress: English
pitch accent: Japanese
prosody: Hawaiian
vowel harmony: Turkish

For examples of suprasegmentals used as morphemes to at least some extent:
tone: numerous West African languages
stress: English
pitch accent: I don't know
prosody: I suspect none exist
vowel frontness: German umlaut

HTH

[Melteor]

I would just like to add a few things.

Various languages have contrastive phonations e.g. the way "red rum" was spoken in the"Shining". The difference is that it's not always clear that phonations apply to vowels or consonants. Korean and various Central, south american and African languages have them

Various features associate with different features in different languages. As Radius Solis mentions, nasalization is occasionally contrastive. Otoh in Piraha it's shows only in the syllable immediately following a glottal consonant I.e. Rhinoglottophilia. In another language such as Cherokee, it helps show word boundaries by popping up in the final syllable of a word. Wichita would devoice such a vowel. French stresses final syllables when they're not schwa. Welsh has a stress and rising pitch accent that shows word boundaries.

Papiamentu has a stress and pitch accent system, but it applies to the whole intonation phrase (as I understand it) so it helps constrain clausal prosody.

So-called "micro-prosody" is poorly studied and to me, they seem a lot like "tropes". I've wanted to start a wiki with recordings of the various kinds of stereotyped accents encountered in English, and break it up with tags and recordings. I think ToBI is kind of too dense, and it faces several problems which include...A struggle to reconcile phonetic detail with phonological detail. A lot comes down to idiolectical variation, and nobody has the resources to study that. Even tones seem poorly documented. Intonation languages are described as having levels/registers or contours depending on whats in vogue tho that terminology is inappropriate for such languages which are not tonal i.e. Do not have phonemic tones. The relationship between stress, volume, timing and duration, and pitch on the other hand, are acknowledged only in cutting edge research e.g. with French.

French has been characterized as having a very simple prosodic system, but it's not very uniform, with a lot of variation in strategies between speakers for emphasizing things. Where the language is relatively consistent is how it uses prosody to segment speech, but that is challenging because then both stress and pitch seem to have various roles. To create a "grammar" for French (as ToBi users would call it) would ideally not just include pitch! but also lengthening effects of both vowels and consonants. So it is thorny. Here is a thesis I found. it pertains to the issue of disambiguating between the ridiculous syncretism available in French.

stress: English

I was surprised at how uncommon this actually is.In new verbs derived from nouns e.g. to google, it doesn't even show up.

[Radius Solis]

stress: English

I was surprised at how uncommon this actually is.In new verbs derived from nouns e.g. to google, it doesn't even show up.

It's uncommon and, at best, minimally productive. This is consistent with derivational morphemes in general. However, that list is far too short. I once posted a list of 37 minimal pairs, though I don't have it anymore. I'm pretty sure I've seen a longer list than just 37, too, somewhere on wikipedia; one of the replies on that link lists a source as having given 85.

Not all pairs work for all dialects, of course - but on the other hand, he whom you linked to is being silly about restricting his list to "true" minimal pairs where the "only" difference is stress. Because vowel reduction patterns are predictable from stress location, 1. it is not necessary to consider it in proving contrastive stress; 2. it could reasonably be defined as part of what the English stress system consists of, i.e. part of the thing being tested for; and 3. it's the location of stress that's contrastive, not its features: the contrast vanishes in monosyllables. Unstressed pronouns "I" and "we" do not contrast with stressed "eye" and "Wii". So the only benefit you gain by excluding pairs with stress-conditioned vowel alternations, over keeping those pairs, is to show something that's actually false anyway.

[Soap]

Well youd sound silly if you tried to stress the last syllable of Google, wouldnt you? But your point stands, of course, there are plenty of other newly coined words where the stress doesnt move. The moving stress pattern doesnt seem to have ever been the majority, though, so I dont think there's any reason to expect it to generalize.

Also some of the minimal pairs are coincidences: e.g. ENTR-ance vs en-TRANCE ... not cognates.

[TzirTzi]

Just to make a very minor but perhaps helpful point, the features discussed here are in generativist phonology more often called autosegmental than suprasegmental. The idea is that there is a tier of segments--phonemes--which is a linear ordered list of segments, each of which are defined by a series of feature settings ([+/-voice], [+/-back], etc.). If a feature behaves in some way independently of this linear ordering--it's behaviour is best described in terms of groups of segments, or it can be both before and after certain segments, or it tends not to be deleted when segments are deleted--it is understood as independent of the sequence of segments. That gets you supra-segmental - but the data actually shows that these features behave as linear-ordered bundles--segments--themselves, just on a different tier. So that's autosegmental--segmental, but independent of the main segmental tier.

Most of the features discussed can be either segmental or autosegmental depending on the language: tone can apply on a purely one-to-one basis to TBUs (segmental), or a one-to-many and/or many-to-one basis, with all the behaviours that go with that (autosegmental); likewise nasalisation (segmental in English, autosegmental in e.g. Guaraní (have a look at p4 of this article)), ATR, etc.. In theory I suppose any feature can be segmental or autosegmental--but I don't know of any examples of features other than those already mentioned in this thread behaving as autosegmental.