Zompist Bboard Again

Right time to resurrect this thread with some further posts on some random bits of things I've uncovered about the language since this thread was last written on. I've got three main ideas that I would like to discuss here. In this post I'm planning to discuss new discoveries as to relative segment frequency, on the basis of the two relay texts I have so far done (admittedly not the best corpus one could hope for, but still has some relevance all the same). In the next two posts I will discuss new revelations regarding the stress system and for the final post I will do a proper dive into valency, which I realise I didn't actually cover properly before.

OK, so based on the two relay texts I have done, the AkanaWiki Frequentizer outputs the following graphs.







Firstly, the obvious and somewhat intentional results - the schwa and the glottal stop are by far the most common sounds in their respective categories and globally. The full vowels other than /e/ are the next most common sounds, with /i/ pipping /a/ I suspect mainly due to affectation from palatal consonants (ditto for why /u/ is as frequent as it is) as well as its frequent use in the 3rd person agent/possessor prefix ‘i-. By contrast the least frequent segments are in large part those that I have intended to be so, namely the labials as well as those with assumed historical origins as clusters/diphthongs

The vowels really show little patterning that seems indicative outside of the off-cuve relative frequencies of /u/ and /e/. If we take /e/ as having originated from a historic *ai diphthong, then its relative infrequency is partly explained, but then the relative frequency of /u/ may be further explained by hypothesising that the original corresponding *au merged with /u/ from other sources at some point.

There is more to say with the consonants. Firstly, after the glottal stop (whose relative frequency in all its uses is pretty much self-evident to anyone who's ever seen an Asta text) the next most common consonants are /n/, /w/, /t/ and /x/, which is very much to be expected given their use as singular noun class prefixes. The next consonant being /y/ is also not a surprise either, being found both as a plural noun class prefix as well as in the perfective suffix -yə. Similar arguments can be made for the frequency of /s/, however it also crops up frequently in consonant clusters as an assimilation from /x/. Finally /m/ and /r/ are found as first and second person prefixes respectively, and the latter also in the causative infix, so their relative frequency is in part assured.

After those two we enter the realm of consonants whose relative infrequency is either the product of their mostly lexical distribution or due to the properties of the texts being analysed. The former is undoubtedly the case with the labiovelars, as they only crop up in inflections as the product of the progressive with particular stems, as also the bilabial stop /p/, which doesn't even occur stem-finally. On the other hand, with cases such as /tr/ versus /ŋ/, I suspect the textual choice plays a bigger role - a cursory glance at my lexicon seems to suggest that /ŋ/ is slightly more frequent overall, particularly being somewhat common as a final consonant in verb stems (with the caveat that it does not always surface as such in inflections), and in natural speech one would assume that the irrealis infix -ŋə- would occur slightly more frequently than the applicative infix -atr-, the only inflectional affixes which either consonant occurs in. In the texts we have however, the irrealis is used barey at all, whereas the applicative crops up comparatively frequently. There's also the matter of the absence of otherwise fairly basic lexical items that use the velar nasal, such as taŋa "house" which compounds this.

There's good stuff here! I vaguely recall seeing this thread before the most recent post, but back then I had basically no interest in polysynthesis. Now that I'm working on my own such conlang, I can appreciate this much more than I did.

Frislander wrote: ↑Fri Dec 14, 2018 6:38 am Type II incorporation comes from the re-addition of a new argument. With no other marking this is usually interpreted as the possessor of the incorporated noun, however the applicative can be used to raise other roles.

yiyan miyessiyənesin
yiyan mə-y-essə-yə-nesin
child.PL 1s-Ipl-stab-PRF-puppy
I killed the children's puppy/puppies

miyessatriyənesin yiyan
mə-y-essə<atr>-yə-nesin yiyan
1s-Ipl-stab<APPL>-PRF-puppy children
I killed puppies for the children

This is a confidence boost since I'm currently writing a post about "genitive incorporation", which sounds like the same thing as your possessor argument here. I hadn't considered applicative meanings with incorporation, either, so this is new & shiny for me! I don't know that I'll do anything with it myself, but it's good to think about.

A very belated "nice work" to you!

Vardelm wrote: ↑Fri Sep 25, 2020 12:55 pmThis is a confidence boost since I'm currently writing a post about "genitive incorporation", which sounds like the same thing as your possessor argument here. I hadn't considered applicative meanings with incorporation, either, so this is new & shiny for me! I don't know that I'll do anything with it myself, but it's good to think about.

I feel like we read different posts, or at least I can't understand your reaction here...

Frislander is combining regular incorporation of the direct object with regular applicatives. The suffix -tri- marks dative applicative. The weird part is that when a verb has an incorporated direct object with no such applicative marker, that is the default, then a space is left for another real direct object ('children') that gets interpreted as possessor of the incorporated noun ('puppy/puppies').

But when I read your post about "genitive incorporation" and "applicative meanings with incorporation", at first glance it seems you're talking about incorporating things that aren't direct objects, like incorporating a dative (1SG.NOM gave-ID-children the gifts.ACC, where -ID- = marker that the incorporated noun is dative) or illative (1SG.NOM went-ID-mountain the last year)...? Which sounds interesting, but this would be a different thing.

Ser wrote: ↑Fri Sep 25, 2020 2:17 pm But when I read your post about "genitive incorporation" and "applicative meanings with incorporation", at first glance it seems you're talking about incorporating things that aren't direct objects, like incorporating a dative (1SG.NOM gave-ID-children the gifts.ACC, where -ID- = marker that the incorporated noun is dative) or illative (1SG.NOM went-ID-mountain the last year)...? Which sounds interesting, but this would be a different thing.

My "genitive incorporation" will end up looking much like the first gloss I quoted. I think mine is starting from a different place to get there, though.
The main point was I appreciate seeing that construction.

Right, time to discuss stress more fully. When I last mentioned stress, in a passing comment in the first post, I said this:

Stress is basically word-initial. Some speakers show shift to the post-initial in cases when the initial has a schwa and the post-initial a non-schwa vowel, but this isn't universal.

I've now come to the conclusion that this is very much an incomplete description that deserves a lot more discussion. This will involve some discussion of the historical background which led to the development of the current Asta vowel system, as this makes a lot of the complexities found in the system much more transparent.

So, to begin, what is the actual stress system of Asta? Well essentially it is a mobile stress system with the quality of the vowel nuclei being the main determining factor. The primary stress will always fall within the root (note, not the phonological word as the above description implies), and always upon a full (i.e. non-schwa) vowel within that root. Which of the "full" vowels is selected however is not always entirely predictable, at least on the surface. In particular, while the default seems to be the first full vowel, there are a number of cases where the primary stress instead seems to fall on the second, e.g. -uŋwá "closed/shut", -iyáwə "awake/alert" and ne‘é‘ "ant". The vowel /e/ in particular seems be be strongly attractive to stress, with vanishingly few occurrences in unstressed syllables.

What's going on here? Let's start from the beginning. Proto-Asta, as indicated elsewhere, is believed to have had a three-vowel system *a *i *u, with a length contrast and two diphthongs *ai *au. In this system stress was once again mobile but weight-sensitive - the primary stress was on the initial syllable of the root unless a longer vowel or a diphthong followed it in the next syllable, in which case the stress moved rightward to that syllable in the case of a root-initial short vowel and may optionally have moved in that direction in the case of a root-initial long vowel (this explains a seemingly irregular post-initial stress in a handful of roots such as -aŋé "loud").

In the development of this system to modern Asta, the main change was that unstressed short vowels were reduced to schwa. This happened after rounding/palatalisation had been transferred to preceding consonants. Furthermore the long vowels merged with their short counterparts, further obscuring the weight-based motivaion for the system, and the diphthongs did likewise. *ai in particular underwent a noticeable a stress-based split to /e/ in syllables where it carried stress as /i/ in other cases, with the differences with long *ii in unstressed syllables being recoverable from the presence/absence of palatalisation of a preceding consonant. The comparative status of the *au diphthong is somewhat less clear as the merger with /u/ seems very near complete, though an intermediate stage of *o in stressed syllables may have broken into modern /wa/ after velars.

Afterwards resulting schwas underwent the affectation processes that still operate in the language today - rounding before labiovelars, fronting before palatals and total assimilation over the glottal stop. This latter process produced the vanishingly few instances of unstressed /e/ that we have, including the ne‘é‘ "ant" example given above. In particular, it will be noted that thee affected schwas are still tret like schwas for the purpose of stress assignment.

I will note that these processes have not as of yet generated any pairs of words which differ solely on the basis of stress, and I will continue not to mark primary stress in examples, nevertheless the potential is there for a more robust stress contrast to develop.

Now after that groundwork on primary stress location it is time to turn to the larger topic of feet and rhythm. Asta seems to prefer trochees over iambs, as the default root-initial stress location would appear to indicate, but is tolerant of both. In particular, the addition of prefixes frequently entails a shift from trochaic to iambic rhythm, as they do not appear to have affected the primary stress location (with the possible exception of roots with two long vowels/diphthongs, which may have been post-intially stressed without a prefix but initially stressed with one. If so the is no synchronic evidence for this alternation however). The reverse shift is also possible with roots that have post-initial primary stress but a prefix with a full vowel (e.g. min- "first person plural), which resultss a trochaic rhythm with a secondary stress location before the primary stress (see below). Additionally, the iambic rhythm appears comparatively unstable, as it is possible for a word to switch from iambic to trochaic rhythm but not the other way round (see below).

The secondary stress locations will thus be determined by this rhythm, with trochaic rhythm placing secondary stress on successive odd syllables and iambs on successive even ones. Schwas (effected or not) will tend to push secondary stresses rightwards another syllable, combining with the preceding foot to create a dactylic foot in trochaic rhythm and the following foot to create an antidactylic foot in iambic. In the case of iambic rhythm this only occurs when the preceding vowel is also schwa, otherwise the secondary stress is instead pulled leftwards and the rhythm switches to trochaic. If that still results in a secondary stress on a schwa, this is typically allowed, but the schwa will tend to be pronounced more open in the direction of /a/. In the rare cases where we get more than one schwa prior to the primary stress location a degenerate foot will result.

Some examples of all this with previous examples from this thread:

(muˈwi)(ˌsatriyə)(ˌtixarə)
(ˈmexwəŋən)(ˌnennən)(ˌnennu)(ˌtyi)
(ˌmintə)(ˈtayən)(ˌte)
(miˈnyes)(ˌsessuwə)
(miˈyi)(ˌtasə)(ˌpan)
(ˈmeŋe)(ˌŋixwi)(ˌnyen)

Updates!

First, some "patch notes" for what's changed about the language since I last posted.

The number of adverbial suffixes has been stripped down, with most being expressed by other means (a couple of which will be discussed below). The following suffixes remain: -pan 'distributive'; -tyuxə 'excessive'; -ri‘i 'frustrative, narrow-scope negation'; -run '(into) fire'; -nyen '(into) water'; -rex ' andative'; -yex 'venitive'; and finally there is a new suffix -te‘ex `perlative'.

minəŋəsnute‘ex xiyəssiyəx
min-<ŋə>əsnu-te‘ex xiyəssi-yəx
1p-<IRR>sail-PER sea-ADV
We should sail across the sea

Secondly, the form of the iterative is now altered. In particular, the reduplicated root is placed after other suffixes on the verb (including adverbial suffixes) and incorporated nouns, and prefixed with a -p-. It may also co-occur with other aspect forms.

wuyuri‘ipuyu
w-uyu-ri‘i-p-ROOT
Is-sleep-FRUS-ITER
He kept trying (and failing) to sleep

mətiriyəpirə
mə-t-irə-yə-p-ROOT
1s-Vs-do-PRF-ITER
I did it again

Thirdly, the generic nominalisation -‘ə now reduces to simply -ə after consonant stems

‘ətrinə
‘-ətrin-‘ə
IV-fall_through-NOM
diarrhoea

wəxentə
w-əxent-‘ə
Is-be_born-NOM
baby, infant

Fourthly, body part nouns may drop the noun-class prefix, especially in fixed verbal expressions.

namməninin
n-ammən-ini-n
IIs-flat-face-PAT.NOM
owl

mexuntawəxpuntə
mex-untə-awəx-p-ROOT
1>2-hold-mind-ITER
I've kept thinking about you

This leads onto a new development - "pseudo-modals".

As is probably clear from previous posts, Asta doesn't really do "auxilliary" verbs at all, not having any true participle forms or verbs that take direct complements. Some of the modal meanings that would be carried by a modal auxilliary in English may be carried carried by the irrealis infix <ŋə> or the frustrative -ru‘u

yuriŋəratrə
yur-<ŋə>irə<atr>
2>1-<IRR>do<APPL>
You could do (it) for me

mexəŋənne‘
mex-<ŋə>ənnet
1>2-<IRR>show
I need to/would/should show you

muwaxaxuŋwəri‘i
mə-w-VCC-axən-wə-ri‘i
1s-Is-PROG-hear-PROG-FRUS
I can't heard him

Commonly, though, Asta makes use of an entirely different construction which bears some resemblance to auxilliary verbs, but have an entirely different origin. A number of body part nouns in combination with the adverbial suffix and replacing the noun class prefix with personal/possessive prefixes carry modal meanings that in English would use an auxilliary verb and/or complement clause. The most common ones include: -awisəx 'think that' from xawəx 'mind'; -a‘yusəx 'want to, need to' from xa‘yux 'heart'; -ə‘trayəx 'fear to, am afraid to' from xə‘tra 'liver'; -ammiyəx 'can, be able to' from xammə 'hand, arm'; and -inyuyəx 'ought to, be obliged to' from xinyu 'back'. Note that these are external to the verb phrase and therefore the verb being modified still retains full verbal agreement, nor is there any restriction on the position of the pseudo-modal element.

mawisəx məxistyi xitrənə.
m-xawəx-yəx mə-xistyi
1s-mind-ADV 1s-leg IIIs-break-NOM
[I think my leg is broken

rətənnənneŋwə tiya ra‘yusəx
rə-t-VCC-ənnen-wə t-iya r-xa‘yux-yəs
2s-Vs-PROG-see-PROG Vs-PROX 2s-heard-ADV
You need/want to see this

Note that as these are historically possessed nouns, third persons use the invariant ‘- from the 3rd person ergative/possessive ‘i-, rather than showing noun-class agreement prefixes.

munasən natə ‘ə‘trayəx
mə-nasən n-atə ‘i-ə‘tra-yəx
1s-dog IIs-swim 3erg-liver-ADV
My dog is afraid of swimming

pa‘ ye‘ə‘ yu‘xwesə ‘ammiyəx
pa‘ y-e‘ənt y-wu‘xwesə ‘i-xammə-yəx
NEG Ip-be_initiated PL-foreigner 3erg-hand-ADV
Foreigners can't be initiated