Firstly, all words have a minimum length constraint of at least two morae. So bal is fine, and yusay is fine, but *ni is not (though it’s fine as a clitic). Most often, words are two syllables long, though some particularly common ones are only one syllable.
Secondly, the verbal structure may be given by a (simplified) template of S=[(mute) V-(ASP) ADJ], where S= is a subject clitic, mute is the past tense marker, V is the verb, ASP is an aspectual suffix, and ADJ refers to a ‘noun adjunct’, a semi-incorporated nominal word used for light verb constructions. (e.g. In be=lhiise guqle ŋay ‘I find it’, lit. ‘I=look search it’, the noun adjunct is guqle ‘search’; we know it must be a noun adjunct rather than an argument because a sentence can have at most one direct object.)
Thirdly, there is extensive use of serial verbs. As is usual for serial verb constructions, all verbs are fully inflected, agreeing in tense and aspect.
Fourthly, the class of verbs is largeish but closed, with about 150–200 members. Other verbal meanings are expressed with either serial verbs or verb adjunct constructions:
Be=[waq fawetl]
1s=[do speech]
I speak (speak = do speech)
Ni=[siwi] [yusay] ŋay
2s=[take] [come] 3s
You bring it (bring = take come)
Ni=[wiilets] [qaathan] bal
2s=[contact] [fall] 1s
You push me (push = contact fall)
Ni=[paatli] ŋay [qisga tsilik]
2s=[make] 3s [break shatter]
You shatter it (lit. ‘you make it shatter’; shatter = break shatter)
(Note: for these glosses I’ve enclosed each pst+verb+adjunct in square brackets, as I did above.)
Now, all this is fine: each of the principles above is well-attested in natural languages. Even when they’re put together, the sentences don’t appear all that strange, though they’re a little bit verbose. However, the real problem is when these sentences are put into TAs other than the present imperfective. Recall that each verb is inflected independently, giving sentences like:
Ni=[mute siwi-i] [mute yusay-i] ŋay
2s=[PST take-PFV] [PST come-PFV] 3s
You brought it
Ni=[mute wiilets-i-ŋu] [mute qaathan-i-ŋu] bal
2s=[PST contact-PFV-TEL] [fall-PFV-TEL] 1s
You pushed me over
[Paatli-niiq] ŋay [qisga-niiq tsilik]
[make-IMP] 3s [break-IMP shatter]
Shatter it!
Now, maybe it’s just me, but I see a language where the word for brought is mute siwii mute yusayi to be stretching the bounds of credibility. As far as I’m aware, there is no known language with this level of verbosity — and as I said already, I consider such a language to be basically unspeakable.
Now, quite simply, my main problem is that there seems to be no obvious way of fixing this problem. I see a number of options, all of which are somewhere between undesirable and impossible:
- The most obvious solution would be to use what has been called ‘nuclear’, ‘complex’ or ‘shared’ serialisation, in which the prefixes and suffixes surround the whole SVC instead of applying to each verb — effectively turning it into a verbal compound. This would give sentences such as ni=[mute [wiilets qaathan]-i-ŋu] bal, which in terms of verbosity seems far more acceptable. However, this approach runs into problems with verbal adjuncts: since the adjunct is quite clearly outside the affixes, as seen above in constituents like qisga-niiq tsilik, this would make it impossible to use verbal adjuncts in SVCs — a constraint I see as extremely undesirable and unnatural.
- Along the same lines, a similar solution would be to restrict the marking to only one verb in the SVC, giving sentences like ni=[mute wiilets-i-ŋu] qaathan bal, which again has acceptable verbosity. However, I am not aware of any language which uses this as the only type of serialisation; thus this which would still leave the language with excessively long sentences (albeit not as many).
- Another possible solution would be to make the past tense marker a bit shorter, changing it from mute to something like mus. This does make the sentences slightly shorter, but not by much, and besides I quite like the current past tense marker.
- Potentially I could make the words themselves much shorter, preferring one-syllable rather than two-syllable words (recall the minimal word constraint allows CVV and CVC words). This would immediately make sentences much shorter — compare nimute wiiletsiŋu mute qaathaniŋu bal to something like nimus wiyeŋu mus qathiŋu bal — but it would require completely redoing the vocabulary (well, all 40 words of it, anyway), something I don’t look forward to doing.
- My preferred solution would be to require mute only once at the beginning of the SVC, so using nimute wiiletsiŋu qaathaniŋu bal rather than nimute wiiletsiŋu mute qaathaniŋu bal. This immediately makes the sentence less verbose, but I’m not sure whether this is plausible at all — at present, mute is a (phonologically independent) prefix to the verb, something which I know can exist, whereas this proposal would require mute to be an placed outside the verbal complex as an unbound particle, which is something I’m not sure can exist.
.