As I understand it, such a system is exceedingly rare (if not completely nonexistent) in natlangs. Sure, there are many languages in which nouns and verbs are very similar (Salishan comes to mind), and can act in very similar ways, but there seem to be no languages in which there is no difference whatsoever between nouns and verbs. The most common situation appears to be one in which nouns can appear as the head of the predicate, while verbs can head a noun phrase but are restricted in which modifiers they can take. (e.g. Makah verbs must be definite when heading a noun phrase; verbs in Lushootseed cannot take possessors in noun phrases; verbs in Fijian are obligatorily possessed in a noun phrase, and cannot take quantifiers.)[*] It's omnipredicative: basically, verbs or noun can indifferently function as predicate.
[*] For that matter, there's no real distinction between noun and verb. That is, words are translated either as nouns or verbs in English, but from a Simbri perspective, there's no good test for distinguishing the two.
Well, it depends. I like how François does it in his article on word classes in Hiw: carefully look at where each word can occur, then define word classes such that each word class occurs in a different set of environments. For instance, here’s how that works out for Hiw (in excerpt, I didn’t feel like copying out the whole table):"Nominal" and "verbal" morphology overlap quite a bit; but still wouldn't it make sense to describe them separately (noting when they overlap, of course). Or should I treat nouns as zero-derivations of verbs? Or the reverse, as the case may be?
Looking at this, we can immediately see the sort of questions you’re asking apply equally well to Hiw. Nouns can occur in predicates; does this indicate zero-derivation of nouns to verbs? No, of course not — it just means that the word class we’re calling ‘nouns’ can occur in a bunch of places, one of which happens to be the heads of predicates. Can we say that there is no distinction between nouns and verbs? No, we most certainly cannot — verbs are distinguished in that they can modify predicates (forming serial verb constructions), while nouns can’t. Adjectives can also head predicates; are adjectives verbs? No they aren’t — we can see that they have a wider distribution than normal verbs. Are adjectives a subclass of the verbs? Based on the table, that seems reasonable — we could say that adjectives are those verbs which can modify an argument phrase. Meanwhile, it seems that a small subset of roots have a wider distribution than any one class listed in that table — it turns out that they can occur anywhere that either ‘weak nouns’ or verbs can, making this a clear example of multicategoriality (or ‘zero-derivation’, if you prefer). And so on and so forth.François wrote:
verb adjective strong noun weak noun adverb head or argument phrase - - + - - modifier in argument phrase - + - + - head of TAM-inflected predicate + + + + - head of direct predicate + + + - - modifier in predicate phrase + + - - +
As you can see, I find that this sort of analysis helps enormously with questions such as yours. Should you describe nouns and verbs separately, or treat them as a single word class, or treat them as zero-derivation, or what? Well, it depends! Do what makes the description work out cleanest. If nouns and verbs really behave exactly the same in every respect, then there’s probably no point in even distinguishing them — it would be fine calling them just ‘bases’, or ‘roots’, or similar. On the other hand, if they share very many properties but are not identical, it may be better to do what François did with Hiw and revert to calling them ‘nouns’ and ‘verbs’, on the understanding that they have a far wider distribution than that of English ‘nouns’ and ‘verbs’.
I don’t really understand what you mean by this, or how such a system would work. All languages I know of have one syntactic constituent containing a noun, its modifiers, determiners, quantifiers etc., and another containing the verb, TAM inflection, adverbs etc. Generally, languages claimed to have no noun/verb distinction are said to have words which can head either noun phrases or verb phrases, rather than having no NP or VP at all.Syntactically, of course, there's little difference between noun phrases and verb phrases... So should I describe them as one? There are, still, more 'noun phrase-ish' constructions and 'noun-verbish' constructions after all...
My impression is that most current syntactic theories do that anyway, so you’re in good company if you take this approach!I mean, I could describe as a series of transformations from the 'saner' [VSO sentence] … But I feel that's cheating: it hides the complexity by pretending it's really an IE language underneath.
My own two cents: don’t bother with syntactic theories in a reference grammar! As you can see here, they don’t help much with describing the language itself, and only make it difficult for readers to understand what’s going on. (I personally loathe reading grammars which use syntactic theories — they just make it difficult for those of us who aren’t too familiar with syntax.)Drawing a syntactic tree is kind of difficult.
Would it help to describe it in terms of transformations, or generative grammar?
…
Or would a different theory suit the language better?
I’d advise taking a middle approach: utilise standard linguistic ideas to the fullest extent possible, but don’t force the language to follow a mould in which it doesn’t fit, and don’t feel bad if you have to abandon standard linguistics in one or two places. In general, try to describe the language using language-internal criteria, rather than trying to analyse it using some language-external convention which may or may not work.None of this, by the way, is unattested in natlangs. For Nahuatl (which I've drawn on quite a bit for inspiration), there seems to be two approaches:
[*] Jettisoning all conventional ideas about language, and using an ad-hoc terminology and specific nomenclature. Very accurate, but you don't understand a thing. Plus it makes Nahuatl look like a language spoken by extradimensional beings and makes it look a lot more complex than it really is.
[*] Hiding the weirdness, under the hood. In reference grammars Nahuatl is described as being just like Spanish, and the weird stuff is kept for specialists in linguistics papers. The approach as the advantage of actually working (you do understand how the language works!) but you know, writing a conlang grammar, I want to show the weird stuff!
Any ideas on what approach would work best?
A good example I’ve come across is that Hiw article I linked above. If you read that article, François doesn’t refer to other languages, and he doesn’t refer to linguistic theory as to what nouns and verbs ‘really’ are; instead he simply analyses Hiw words in terms of their position in Hiw sentences, and ends up with a lovely description of Hiw word classes which perfectly fits Hiw. Luckily, that analysis ends up corresponding pretty well to conventional terms like ‘noun’ and ‘verb’, but it also reveals details which would be missed by assuming those categories from the outset: for instance, Hiw has two noun classes rather than one, and both noun classes can head the verb phrase. By contrast, if he had started the analysis by assuming the existence of ‘nouns’ and ‘verbs’ with English-like properties, he most probably would have missed the former contrast altogether, and the latter would probably have ended up as something like ‘Hiw has no copula; nouns are apposed to give an equational sentence; the TAM markers normally applied to verbs can then be applied to the second noun’. Which, I mean, isn’t a wrong analysis as such, but it doesn’t really give the simplest description possible.
Another good example I know of is Döhler’s Komnzo grammar, describing a language which is mostly pretty normal… except for the horrific verbal morphology, where everything you know about morphological sanity goes out the window. Here is a diagram of a representative word (yfathwroth, ‘they hold him away’):
Clearly, such a structure is not amenable to analysis in the same way as, say, Spanish. Döhler instead ended up analysing the verb in terms of Komnzo itself: sometimes this ends up with standard linguistic terms being appropriate (e.g. -o is clearly an andative), while other times it is simplest to describe the language in terms of totally new categories (e.g. y- is the ‘third person masculine singular α’, and fath is an ‘extended stem’, which combine to form the ‘imperfective’ — a poorly-named category, in my opinion, for it is quite different to aspects more usually called ‘imperfective’). Again, I would call this a good analysis, since it describes Komnzo without reference to any external theory, and ends up giving a clear description of how the verbal morphology works (with the exception, as I said, of the unedifyingly-named ‘imperfective’).
Or places where word order is important-- you've already given one, namely focus.
