Linguistic Miscellany Thread

[Zju]

I don't know about [ii] and [uu], but there definitely is a difference between [i.i] and [u.u] on the one hand and [ji] and [wu] on the other. I agree that [j w] generally represent closer sounds, but then again their exact realisation and variation depend on the language in question.

Edit: Thinking of it, what is [uu] even supposed to represent? [u.u], [uː], [wu] and [uw] are one thing, but is [uu] even something different on its own?

[zompist]

FWIW J.C Catford agrees with Brad-- he says semivowels are just ultrashort vowels-- but my mouth disagrees. [w] has far more lip rounding than [u], so [uu]/[wu] are very distinct. There's less difference between [i] and [j], but the tongue is farther forward for [j].

(A reminder: [i] and [u] will be interpreted as BBcode, but we have a workaround for that-- the bk tag.)

[Ephraim]

You might as well try to say /uu/ and /wu/ are the same.

Well, of course they aren’t! However, in narrow transcription, [uu] and [wu] are the same, for precisely the same reason that [ii] and [ji] are the same.

I think it's true that there is no well-defined (and generally agreed-upon) language-independent difference in terms of tongue position and lip shape between the symbols [u] and [w], [i] and [j], [y] and [ɥ], and [ɯ] and between [ɰ]. The main difference is really that the first member of the pair is usually syllabic (unless it is not) and/or phonologically a part of the vowel system, and the latter member is (almost) never syllabic, and typically a part of the consonant system in some way.

However, I don't think it's true that [u] and [w] (etc.) would necessarily represent the same pronunciation even in quite a narrow phonetic transcription. The exact difference may depend on the language, though. There may be a difference, like [w] being shorter, slightly closer (or more open), somewhat less central (or more central), having different type or degree of lip rounding etc. The IPA-symbols are broad enough to cover quite a bit of variation, and I don't think any of the possible articulatory distinctions can be said to be "canonical". But also, the context matters quite a bit, and the difference in syllabicity may make quite a bit of difference in terms of timing, coarticulation effects, the timing of tone contours, among other things. This is really quite difficult to represent in a system where sounds are transcribed as discrete segments.

In a similar manner, I don't think there's a well-defined language independent articulatory difference between [kwa] and [kʷa], but they could be used to represent some phonetic difference (such as a difference in timing or type of lip rounding), which may even be phonemic (although I don't know of any language that makes this distinction).

[dɮ the phoneme]

Acknowledging that the precise phonetic difference between vowels and semivowels my be debated and/or language dependent, does anyone have thoughts on the phonological question that initially sparked this?

[Travis B.]

In a similar manner, I don't think there's a well-defined language independent articulatory difference between [kwa] and [kʷa], but they could be used to represent some phonetic difference (such as a difference in timing or type of lip rounding), which may even be phonemic (although I don't know of any language that makes this distinction).

I myself have noticed that /kw/ and /gw/ for me are [kʷw̥~kʷw] and [kʷw~ɡʷw] respectively for me, i.e. they have a clear semivowel component but the plosive component is also labialized, but this is such a phonetic detail I rarely bother to mark it in transcriptions.

[Estav]

When a phoneme has both syllabic and non-syllabic allophones that are determined positionally (e.g. /i u/ in Latin, etc.), what formalism is typically used to describe the phonological processes in play? We might naively write a rule like i > j /_V, but this predicts that e.g. /iiia/ would be realized as [jjja], when I suspect [jija] is the more realistic outcome. I'd like to use footing to account for this, but the problem is that syllabification must happen before footing, whereas this kind of allophonic variation necessarily happens before syllabification. I don't know much about Latin, but my impression is that sequences like /iiia/ are not very common, if attested at all. Are there any languages with similar positional variation of vowels and semivowels that do allow long strings like this, which might provide a test case for different formalisms?

José Ignacio Hualde has written some about analyses of Spanish that treat semivowels as allophones, but I forget what he says. In both Latin and Spanish, the distribution isn't actually 100% predictable from position.

[Richard W]

In a similar manner, I don't think there's a well-defined language independent articulatory difference between [kwa] and [kʷa], but they could be used to represent some phonetic difference (such as a difference in timing or type of lip rounding), which may even be phonemic (although I don't know of any language that makes this distinction).

Will the English (well, some English) phonemic distinction of [ʃʷ] and [ʃʷw] (as in Schweppes) do?

[Richard W]

Acknowledging that the precise phonetic difference between vowels and semivowels my be debated and/or language dependent, does anyone have thoughts on the phonological question that initially sparked this?

It reminds me of the semantics of the programming assignments 'x := y; y := x;' - does this swap the values (as in Statemate) or make them the same (as in Ada)?

If we apply the rule regressively, then iiia > iija > jija.

[Travis B.]

In a similar manner, I don't think there's a well-defined language independent articulatory difference between [kwa] and [kʷa], but they could be used to represent some phonetic difference (such as a difference in timing or type of lip rounding), which may even be phonemic (although I don't know of any language that makes this distinction).

Will the English (well, some English) phonemic distinction of [ʃʷ] and [ʃʷw] (as in Schweppes) do?

Interestingly enough I don't labialize /ʃ/ even in /ʃw/ as in Schweppes.

[Ephraim]

I myself have noticed that /kw/ and /gw/ for me are [kʷw̥~kʷw] and [kʷw~ɡʷw] respectively for me, i.e. they have a clear semivowel component but the plosive component is also labialized, but this is such a phonetic detail I rarely bother to mark it in transcriptions.

That makes sense. A fun exercise might be to actually try to pronounce [kwa] without any sort of lip rounding on the plosive, i.e. to begin the transition from unrounded to rounded lips only after the release of the plosive. This is certainly possible to do, but you get a sort of [ɰ]-like off-glide which may be quite noticeable, something like [kɰwa].

In a similar manner, I don't think there's a well-defined language independent articulatory difference between [kwa] and [kʷa], but they could be used to represent some phonetic difference (such as a difference in timing or type of lip rounding), which may even be phonemic (although I don't know of any language that makes this distinction).

Will the English (well, some English) phonemic distinction of [ʃʷ] and [ʃʷw] (as in Schweppes) do?

That’s an interesting contrast, but I was actually thinking about labialized velars specifically. The reason for this is that [w] is itself velar. After the release of the stop of [kʷa], (typically) the tongue has to transition from the close back position of the velar stop to the open position of the vowel, while the lips transition from the rounded to the unrounded shape. This creates a [w]-like off-glide. This is not necessarily the case for non-velar labialized consonants since the tongue does not have to pass through the velar position.

It would be interesting to know if there is a similar contrast in a language where labialized consonants are also velarized, though.

[Otto Kretschmer]

Any good samples of languages going from isolating to agglutinatove?

[Moose-tache]

Any good samples of languages going from isolating to agglutinatove?

I don't think there are any attested languages that went from clearly isolating to clearly agglutinative within the historical record. But many agglutinative languages show evidence of once being isolating, usually by having bound morphemes that are suspiciously similar to independent words. Tibetan is like this, as is Choctaw, Sumerian, Turkish, and some would argue PIE.

[Travis B.]

Any good samples of languages going from isolating to agglutinatove?

I don't think there are any attested languages that went from clearly isolating to clearly agglutinative within the historical record. But many agglutinative languages show evidence of once being isolating, usually by having bound morphemes that are suspiciously similar to independent words. Tibetan is like this, as is Choctaw, Sumerian, Turkish, and some would argue PIE.

Old French was certainly not isolating, but within the historical record we have seen nominative pronouns reduced to affixes in French.

[Kuchigakatai]

When a phoneme has both syllabic and non-syllabic allophones that are determined positionally (e.g. /i u/ in Latin, etc.), what formalism is typically used to describe the phonological processes in play? We might naively write a rule like i > j /_V, but this predicts that e.g. /iiia/ would be realized as [jjja], when I suspect [jija] is the more realistic outcome. I'd like to use footing to account for this, but the problem is that syllabification must happen before footing, whereas this kind of allophonic variation necessarily happens before syllabification. I don't know much about Latin, but my impression is that sequences like /iiia/ are not very common, if attested at all. Are there any languages with similar positional variation of vowels and semivowels that do allow long strings like this, which might provide a test case for different formalisms?

Ioana Chitoran in The Phonology of Romanian (2001) analyzes Romanian as having /i, ii, iii/ (e.g. copiii for the last one). Maybe you'll find something in that book useful, but beware its framework is Optimality Theory... she doesn't use rules but picks well-formed surface forms from a generator that works in rather mysterious ways.

To elaborate on what Estav alluded to about Latin/Spanish, Latin /w/ isn't completely predictable: solvit 's/he released' (solv-ō, solv-is, solv-it...) vs. soluit 's/he was used to' (solu-ō, solu-istī, solu-it). The stems are /sɔlw-/ vs. /sɔlu-/... /j/ does much better in predictability, if you stick to native words, but Greek borrowings screw the system up with the likes of īātrīa [iː.aː.ˈtriː.a] 'medical treatment' < ἰατρεία and Iōnia [i.ˈoː.ni.a] 'Ionia (region in Asia Minor) < Ἰωνία [i.ɔː.ˈni.aː], and even then there's the annoying native name Gāius [ˈgaː.i.ʊs] < Old Latin Gāvios (related to gaudeō 'I rejoyce'), which for some reason evolved the way it did.

Spanish /j/ seems completely predictable to me, but only as long as you keep it apart from /ʝ/, and there are reasons for why you may want an analysis that treats [i j ʝ~ɟʝ] as the same /i/ phoneme, notably the way in which Spanish speakers tend to pronounce other languages, e.g. English "you" as [ɟʝu], etc. Nevertheless, the likes of yo niego [ˈnjeɣo] 'I deny' vs. enyesar 'to plaster, put in a cast' (morphologically en-yesar, from yeso 'chalk, plaster') are problematic, unless you go for /nj/ vs. /nʝ/, or rely on access to morphology (as you should in this case, IMO). /w/ does much better in purely phonological predictability, but only as long as you give up the notion of stress phonemically attached to syllables and start stressing single vowels instead, otherwise the likes of continúo [kon.ti.ˈnu.o] 'I continue' vs. continuó [kon.ti.ˈnwo] 's/he continued' are problematic: easily handled as /kontinúo kontinuó/ (no need for /w/ in the latter), but otherwise you have the ambiguous /kontiˈnuo/, unless you "phonemicize" syllable boundaries (a dubious notion) as /kon.ti.ˈnu.o kon.ti.ˈnuo/, or simply admit a distinct /w/: /kontiˈnuo kontiˈnwo/.

In spite of all the above, I think the attempt at trying to arrive at more predictability, and a reduced number of phonemes, without access to morphology and syllable boundaries at the phonemic level, rather than less predictability and more phonemes, is generally worth one's time, but it admittedly can easily come with issues unsolvable without morphology or more phonemes.

[Otto Kretschmer]

Dunno if I hadn't asked this before - did Primtitive Irish have a strong initial stress causing reductions of unstressed syllabes as did French and Germanic languages?

[Moose-tache]

Primitive Irish is poorly attested and does not show phonetic/phonemic stress, so I'm not sure what you're expecting. Primitive Irish does have some finals that are absent in the modern language, like nominative s, so maybe there wasn't initial stress. Or maybe it just hadn't had enough time to do its destructive business. Who knows?

[hwhatting]

Or maybe it just hadn't had enough time to do its destructive business.

That's the general assumption I know - i.e., Primitive Irish had initial stress, and that led to the syncopes and apocopes that we know from Old Irish, but it hadn't yet succeeded in causing them when the older Ogham inscriptions were written..

[Travis B.]

Or maybe it just hadn't had enough time to do its destructive business.

That's the general assumption I know - i.e., Primitive Irish had initial stress, and that led to the syncopes and apocopes that we know from Old Irish, but it hadn't yet succeeded in causing them when the older Ogham inscriptions were written..

From all appearances, the changes from Primitive Irish to Old Irish were greater than those from, say, Old Norse to modern Danish (and that is clearly the most progressive of the North Germanic languages) yet occurred in only a few centuries rather than 1000+ years.

[malloc]

This may sound silly, but I am still somewhat confused about how the perfective aspect works. If the perfective marks verbs as completed and considered in their entirety, does that mean it generally implies past occurrence even in languages without grammatical tense marking? What about future-tense verbs and imperatives? The verb has not yet occurred, making it incomplete, yet it refers to an event that will eventually complete.

[zompist]

This may sound silly, but I am still somewhat confused about how the perfective aspect works. If the perfective marks verbs as completed and considered in their entirety, does that mean it generally implies past occurrence even in languages without grammatical tense marking? What about future-tense verbs and imperatives? The verb has not yet occurred, making it incomplete, yet it refers to an event that will eventually complete.

First... it's not uncommon for an imperfective/perfective distinction to be made only in the past, as in French or Spanish.

Second... there are languages where the distinction works in all three tenses. You've been speaking one all your life, in fact. So to answer your question, think about what it means to say "I will read the book" vs. "I will be reading."