Languages with Few Parts of Speech

[Terra]

What (natural and constructed) language has the fewest parts of speech? How would one go about designing a language that has only a single PoS?

I tried looking on WALS for this, but couldn't find anything. Maybe my search terms were off. Furthermore, is there a WALS page for languages that use light verb constructions heavily? How about simply turning adjectives into verbs instead?

[Chagen]

I'm almost certain that a language with one part of speech is flat-out impossible.

You need verbs, at least. A language that has only nouns would have no way of expression actions (which the entire point of language), and if it has "nouns" that express actions, those are just verbs under a different name.

A language with only verbs is slightly more plausible, though I don't see how it can be analyzed as having no nouns if it has lexicalized verbs that refer to concrete things.

[Skomakar'n]

I tried to create a language with only verbs and pronouns and a few adverbials and stuff like that, but no nouns or adjectives (Vanga; it's in the C&C section), but apparently I failed. :p

[Whimemsz]

You might have to define this a little further. Plenty of languages, for example, lack a clearly distinct class of adjectives, treating them either as verbs or as nouns, but there may still be arguments for differentiating "adjectives" from other verbs or other nouns (I know Xephyr has been reading Dixon & Aikhenvald's volume on adjectives recently, he might be able to give more concrete examples of this). Do you only want languages where there is absolutely no distinction whatsoever there, or would you accept, say, a language where stative verbs which are semantically like adjectives might have available a smaller range of morphosyntactic options? There's been a lot written on whether Salishan (and some other NW Coast languages) distinguish nouns from verbs (it's fairly clear that they don't distinguish verb and noun roots at any rate: any root can theoretically function as a predicate or as a substantive), though from what I've read, I'd certainly say the two are distinct.

[Aurora Rossa]

I tried to create a language with only verbs and pronouns and a few adverbials and stuff like that, but no nouns or adjectives (Vanga; it's in the C&C section), but apparently I failed. :p

I have tried making a language without nouns (the old Socialese project) as well but ultimately found it rather impractical and inelegant. The language abolished nouns more or less by using verb roots with the meaning "to be X" and treating them as relative clauses. Hence the sentence "John drinks coffee" would become "he-who-is-John that-which-is-coffee he-drinks-it". That obviously leads to some long-winded and repetitious phrasing and it takes away the fun ability to incorporate nouns into verbs.

[Terra]

Do you only want languages where there is absolutely no distinction whatsoever there, or would you accept, say, a language where stative verbs which are semantically like adjectives might have available a smaller range of morphosyntactic options? There's been a lot written on whether Salishan (and some other NW Coast languages) distinguish nouns from verbs (it's fairly clear that they don't distinguish verb and noun roots at any rate: any root can theoretically function as a predicate or as a substantive), though from what I've read, I'd certainly say the two are distinct.

Either is fine.

Btw, what I had in mind when writing this was a conversation I had with a Chinese student: Instead of saying "The dog is red.", it's more like "The dog reds.". Such a construction applied uniformly could eliminate adjectives entirely.

[Skomakar'n]

I tried to create a language with only verbs and pronouns and a few adverbials and stuff like that, but no nouns or adjectives (Vanga; it's in the C&C section), but apparently I failed. :p

I have tried making a language without nouns (the old Socialese project) as well but ultimately found it rather impractical and inelegant. The language abolished nouns more or less by using verb roots with the meaning "to be X" and treating them as relative clauses. Hence the sentence "John drinks coffee" would become "he-who-is-John that-which-is-coffee he-drinks-it". That obviously leads to some long-winded and repetitious phrasing and it takes away the fun ability to incorporate nouns into verbs.

Yeah, at least if it's a long or very distinct affix, word or set of words to create those relative clauses. The little -(A)h and -(A)s endings of Vanga are obviously less prominent, and especially the -h one sometimes merges with the preceding consonant, simply devoicing or geminating it, giving it a different sound anyway, and it also has several allophones depending on the preceding sound even if it remains as -h.

[chris_notts]

You might have to define this a little further. Plenty of languages, for example, lack a clearly distinct class of adjectives, treating them either as verbs or as nouns, but there may still be arguments for differentiating "adjectives" from other verbs or other nouns (I know Xephyr has been reading Dixon & Aikhenvald's volume on adjectives recently, he might be able to give more concrete examples of this).

I have that book. For what it's worth, Dixon seems to believe that all languages have a set of morphemes covering core adjectival notions that show some morphosyntactic differences from verbs and nouns. For example, he says:

It can sometimes be a tricky matter finding criteria to distinguish 'verb-like' adjectives from verbs, or 'noun-like' adjectives from nouns. I believe that in every language which is studied in detail, criteria can be found. Criteria are not always of the definitive 'yes-or-no' variety; as Alpher (1991) shows for the Australian language Yir-Yoront (see §5), a collection of statistical tendencies can combine to provide a satisfactory characterisation of the adjective class (as opposed to noun and verb classes).

The book is not a large scale typological survey, so I don't think it can be taken as proof of claims like the universality of adjectives as a distinct class. Instead, there are 13 descriptions of relevant phenomena in different languages. And some languages show stronger differentiation than others.

For example, in Wolof adjectives are mostly treated like verbs in predicative function. The main way in which they behave differently is in the order of definite marker / relativiser when they modify nouns. For example:

xale bu rafet bi
child REL pretty DEF
'the pretty child'

xale bi xam
child REL:DEF know
'the child who knows'

* xale bi rafet
* xale bu xam bi

The author of the chapter on Wolof, Fiona McLaughlin, says:

I conclude that although it is possible to posit an adjective class in Wolof based on the behaviour of relative clauses, as in Lao (see Enfield, ch. 14), the class of adjectives is best characterised as a sub-class of verbs.

This is just from flicking through, I'll have a more thorough look later.

[cromulant]

What (natural and constructed) language has the fewest parts of speech?

Languages with only one PoS, or two (substantives and grammatical markers) are an old conlanging trope. Many on the ZBB have gone there. Like this guy. Ithkuil is another (more famous and developed) example.

How would one go about designing a language that has only a single PoS?

I don't know; my interest in this idea was fleeting. I can easily imagine a language lacking any lexical distinction between word classes; that's easy. Tongan apparently does this. But to lack any syntactic or semantic distinction? How indeed.

I tried looking on WALS for this, but couldn't find anything. Maybe my search terms were off.

No, there's no feature on WALS for "Number of Parts of Speech." I'm very familiar with WALS; I'd know if it was there.

Furthermore, is there a WALS page for languages that use light verb constructions heavily?

Nope. Various examples may use light verbs incidentally, but they're not a focus anywhere, and there's no survey of their distribution.

How about simply turning adjectives into verbs instead?

For that, you're in luck.

[alice]

Don't forget:

Vance may be in a state of pulling our legs.

[----]

From what I've seen of Inuktitut, it's hardly anything except verbs and nouns. I haven't seen much though.

[Radius Solis]

Don't forget:

Vance may be in a state of pulling our legs.

He surely is, because there we are talking about languages that do not use words as substantives, and such a behavior cannot reasonably be expected to exist in natural language. This is not the same thing as a language lacking a noun-verb distinction!

When people talk about a language having no noun-verb distinction, like the Salishan and nearby languages Whim mentions, they do not mean its sentences lack substantive words. What they mean is that the roots in these languages have no lexical property of being or not being substantive. There's no division of content words into multiple classes, just one big set of roots, all of which can be used in a noun-like capacity or a verb-like capacity. However, at least for the Salishan ones, it is notable that words used in a noun-like capacity and those used in a verb-like capacity still take the same set of inflections: its "nouns" are obligatorily marked for things like tense, aspect, and person marking and can always stand alone as grammatical complete sentences.

[Chuma]

There will always be some sort of semantic categories, of course; some words represent actions, and some represent objects. (And some are in between; "alive"/"live", for example, can just as easily be a verb or an adjective.) But treating them as the same lexical category is quite possible.

We can think of the part of speech as the "favourite" grammatical function of the word. Verbs usually have a predicative function, nouns a nominal function, and adjectives an attributive function. But they can easily be changed, when you want them to have a different function:

Code: Select all

            predicative     nominal         attributive
verb        sings           singing one     singing
noun        is a dog        dog             which is a dog
adj         is big          big one         big

So it's clearly possible to treat all those words the same way, if you like. The most common choice seems to be to treat all of them as verbs, and the second most common to treat them as nouns. I've found it easier to think of them basically as adjectives, for various reasons. (Adverbs are easily merged with adjectives, so that's hardly worth mentioning.)

So in my conlang, you can think of all content words as being adjectives by default. I express all the grammar with separate form words - articles, copulas, conjunctions etc. If you add an article to a content word, it becomes like a noun, and if you add a copula, it becomes like a verb. So you can say
"the big dog is singing"
"the dog singing is big"
"the singing big is dog"
They all work the same way. It's not particularly difficult, if you're used to English.

One problem can be that you need a lot of forms, since every content word has to be able to take all the inflections of all the parts of speech. My solution is to let the form words take care of all the inflections. Another solution is to have very few mandatory inflections - skip things like number, definiteness, tense, etc.

If you think having separate form words is cheating, you can always use affixes instead. For an agglutinating language, it shouldn't be all to hard to have affixes for all your type-changing needs, and having lots of forms should be easier than for a more fusional language.

There is one thing to keep in mind, that many conlangers forget, including the guy who made Esperanto: A root can often define just one verb, but several nouns. So "teach" can define both "teacher", "student", "lesson", etc. You need to make sure you know which one your noun form is.
I use case markers in a way to mark each of them - absolutive for "lesson", ergative for "teacher", dative for "student", locative for "school" etc. See this thread for more on that.

[Gojera]

The last thing I read on the lack of a noun-verb distinction was something talking about Austronesian languages, in which a language was argued to have not made a clear distinction among them.

But even if this was the case, it argued that a more general universal is true: that every language makes a distinction between predicates and their arguments. This may often line up with the noun-verb distinction, but not always.

[Trailsend]

I'm almost certain that a language with one part of speech is flat-out impossible.

You need verbs, at least. A language that has only nouns would have no way of expression actions (which the entire point of language), and if it has "nouns" that express actions, those are just verbs under a different name.

Those who say it cannot be done should not interrupt those doing it.

[Radius Solis]

But even if this was the case, it argued that a more general universal is true: that every language makes a distinction between predicates and their arguments. This may often line up with the noun-verb distinction, but not always.

Yes. All languages have predicates and arguments, presumably. But what sort of words you find taking those roles is variable, and in some languages, all full words can easily take either role, including those that represent e.g. pronouns, prepositions, or personal names.

Also:

I need to make a correction to something I said above, having re-read the relevant section of LNNA last night: in Salishan languages there is one type of morphology that can appear only with some roots and not others: possessor affixes. Apparently only one kind, but there is at least that. The linguists arguing in favor of still finding a noun-verb distinction in these languages are basing their argument primarily on 1. that fact and 2. what may or may not be a correct analysis of a certain prefix, which has been analyzed alternatively as a nominalizer (by them) or as an aspect (by everyone else. Notably, the same morpheme is uncontroversially considered to mark continuative aspect when it occurs in the predicate, and is part of a paradigm of mutually exclusive aspect markers.). And even those linguists are quick to point out that there is no difference in syntactic behavior between any full word and any other.

[Whimemsz]

I'd also suggest that at the very least all languages (that I know of, anyway!) also have at least one class of particles/interjections/whatever. In highly synthetic languages, these are generally the words that take no inflections (in Ojibwe this includes adverbials, interjections, discourse particles, and so on--but note that these groups have different syntactic behaviors and so can probably be analyzed as multiple POSs). I'm not so sure what the distinction morphosyntactically might be in very isolating languages, but I'd be a bit surprised were there a language with no such distinction.