Saturday, January 10, 2009

Adjectives - who needs 'em?

Most languages have a class of words that express properties and behave differently from other words. These are called adjectives. In English, for example, words like "red" or "old" or "tall" behave differently from nouns or verbs. For example, you add -s to verbs in the present tense if their subject is 3rd person singular, like "he sings" or "she eats"; but you can't add -s to an adjective, so you say "he is red" rather than *"he reds". You can put "very" before an adjective ("very red"), but not usually before a noun (you can't say *"very food".) Verbs can't be placed between "the" and the noun (unless you add an ending like -ing or -ed), but adjectives can (you can say "the red car", but not "the move car").

It turns out, according to Dixon 2004, that practically every language - perhaps every language - has at least one separate class of words, definable purely on the grounds of their (morphosyntactic) behaviour rather than their meaning, that refer to properties. This class typically includes words expressing size, age, value, and colour, and sometimes more.

But often, a concept expressed using an adjective in one language is expressed only by a verb or a noun in another. For example, in Kwarandzyəy adjectives come between the noun and the plural marker:

ạdṛạ kədda yu
mountain small PL
"little mountains" (hills)

But there is no adjective "happy" in Kwarandzyəy; instead, you use a verb, yəfṛəħ "be happy, rejoice". And to say "the happy people", you say "the people who are happy/have rejoiced":

bạ γ i-ba-yəfṛəħ
person who they-PF-happy

Moreover, though they may always be distinguishable by some test, they usually tend to behave very much like another word class. In fact, Stassen 1997:30 (link goes to 2003) postulates that in every languages adjectives handle predication (saying "X is red", for example) in the same way as either verbs, nouns, or locations. For example, in English or Arabic, adjectives handle predication like nouns (you say "He is tall", just like "He is a footballer"); in Korean or Tamasheq, they do it like verbs; and some languages, like Japanese, have both verb-like and noun-like adjectives.

So clearly people can do without some adjectives, and clearly the behaviour of adjectives tends to be very similar to the behaviour of some other word class. Why not do without them altogether? It would be easy enough to construct a language where no morphological or syntactic tests could distinguish adjectives from verbs, or from nouns. So if practically every language does take the trouble to distinguish them, there must be some pretty powerful cognitive motivation for it - and some pretty powerful historical tendencies acting to separate adjectives from verbs and/or nouns. The question isn't directly relevant to my current work, but it's worth thinking about.


Mattitiahu said...

One language I've studied that has particularly ambiguous adjective classes is Coptic. "Adjectives" behave like nouns, and their usage is equally ambiguous as to which word is the actual modifier and the modified. Adjectival relations are expressed by a particle /ᵊn/ inserted between the two elements, one with an article and one without.

For example: /ou-nokʸ ᵊn polis/ 'a large one, viz. city' = a big city (= /ou-polis ᵊn ounokʸ/ 'a city, viz. big' = a big city) ('-' indicates prefixed article boundary)

The particle /ᵊn/ (which has many functions in Coptic) probably is expressing a genitival relationship here, probably literally expressing something in a Coptic speaker's mind 'a large thing of city' or 'a city of large'.

If this is so, very likely Coptic does not actually have adjectives as a distinct grammatical category, but rather the idea of adjectives are expressed by a nominal construction. (For more details, Bentley Layton's (2004) "A Coptic Grammar, 2nd Ed." Wiesbaden: Harrassowitz. §99ff. gives a good synchronic description of the phenomena of 'attributive nouns' as adjectives in Coptic.)

Oh, and really awesome blog you've got going here. Keep it up!

Anonymous said...

Yeah, welcome back Lameen.

Wagiman, like a number of far-northern Australian languages, has a separate class of word from nominals and verbs called coverbs, but they are pretty much verbs as we would normally term them; the 'verbs' being closer to auxiliaries.

There aren't any morphosyntactically definable adjectives, and this is the case for a lot of Australian languages. The meanings associated with adjectives are partly nominal and partly verbal. Colours in Wagiman, for instance, and morphosyntactically coverbs, yet take derivational morphology to form things that are closer to 'adjectives'. Interestingly, there are two distinct morphological frames that Wagiman coverbs can take to become adjectival, and the difference is dependent on whether the underlying coverb is a stative predicate, in which case it takes a prefix nu-, or an inchoative predicate, when it takes a circumfix ma- -yin. So a coverb like 'red' wirril-ma more often than not surfaces as nu-wirril-ma, and a coverb 'die' bort-ta derives the adjective ma-bort-ta-yin 'dead'.

I wouldn't say therefore that Wagiman behaves like 'normal' languages in having a morphosyntactically distinct adjectival word class. Instead I'd say Wagiman has two morphologically derived classes of 'adjectivally' things.

John Cowan said...

I love the way Mandarin adjectives work: they are of the verby type, and are differentiated from plain verbs thus: Any verb (or at least any intransitive verb, though it is unclear whether "intransitive" is a syntactic or merely a pragmatic category in Mandarin) can be used attributively if followed by the particle de, but adjective/stative verbs are permitted to omit the particle.

Lameen Souag الأمين سواق said...

Matt: Thanks for the example - that gives me the excuse I needed to go and look at my Coptic grammar :) Judging by ancient Egyptian, I would have guessed that adjectives would be distinguished from nouns by showing gender and number agreement - do they?

Jangari: At first glance, it sounds like the morphologically defined class of stative verbs in Wagiman could reasonably be termed its adjective class. Are there any reasons why this is a bad idea?

JC: Yeah, Chinese grammar is cool. I would think the usage of hen "very" also offers some tests, but I'd have to look at a proper grammar instead of the beginners' texts I've used.

Anonymous said...

Some Berberologists claimed that the Amazigh language doesn't have adjectives at all. This claim was pretty much rebutted by Salem Chaker in his book "Articles de linguistique berbère".

In Amazigh we have hundreds of simple adjectives that are easily comparable with English. But many of them could be used also as nouns! And there are verbal adjectives in every dialect.

Tarifit and Tashelhit dialects reflect this:

In Tarifit we say:

Aryaz amẓẓyan (young man)
Iryazn imẓẓyanen (old men)
Tamghart tamẓẓyant (young woman)
Timgharin timẓẓyanin (young women)

But in many parts of Souss-Tashelhit we use verbal adjectives:

Argaz (i)mẓẓin : young man
Irgazn mẓẓinin : old men
Tamghart (i)mẓẓin : woman (who is) young
Timgharin mẓẓinin : women (who are) young

I reckon that this Tashelhit pattern has more to do with (older) Southern Berber (Tuareg). Tarifit has developed proper adjectives more than Tashelhit, but verbal adjectives are still present in Tarifit.

About constructing an adjectiveless language, I think it would be like travelling back in time. I can’t resist the idea that older languages were simpler, raw and less structured, in other words: primitive just like Man himself. Words could then be nouns, adjectives and verbs at the same time.

So this constructed adjectiveless language would surely evolve over time by creating “adjectives” and other word forms, if left free of regulations and standardization. I think that natural languages, as they evolve, tend to complicate themselves by creating categories and exceptions. Don’t you think?

Moubarik Belkasim

Anonymous said...

Any verb ([...]) can be used attributively if followed by the particle de

Absolutely anything can be used as an attribute if followed by the particle de. This is also how the complete lack of genetives and relative pronouns is handled, for example. Textbook sentence: "A Chinese teacher is a teacher who teaches Chinese" = Hànyǔ lǎoshī shi jiào Hànyǔ de lǎoshī = "Chinese teacher is teach Chinese de teacher", where de attaches to the whole phrase jiào Hànyǔ.

Mattitiahu said...

Adjectives in earlier Egyptian do indeed show declension, but it seems that only a few retain number and gender distinctions by the Coptic period:

ex. nfr 'good, beautiful'

nfr (m.s.) nfrt (f.s.)
nfrwy (m.du.) nfrty (f.du.)
nfrw ( nfrwt (

When the article became the main indicator of number and gender atop of its role in expressing the quality of definiteness, it seems that only a single form is retained (Middle Egy. nfr > Sahidic Copt. /noufe/), however there do seem to be seem to be some occasional adjectives that show seperate masculine, feminine, and common plural forms.

ex. sabe 'wise' (nominalized: wise person) /sabe/ /sabē/ /sabeˀewe/

These may be fossilized forms, but I'm not entirely sure at the moment. I only really dabble in Egyptian linguistics.

In any case, the syntax of the adjectival idea in Coptic at any rate appears to be for the most part that of a prepositional phrase, suggesting that the category of 'adjective' in Coptic is nominal in syntax, at least on the synchronic level. Coptic nouns themselves generally don't show singular or plural distinctions except as indicated by the article, but with some exceptions: usually in core vocabulary like 'father' sg. /jōt/ pl. /jote/ < Middle Egyptian 'jt' or 'dog' /ouhor/ (three different forms) /ouhore/, /ouhoˀore/, /ouhōre/ /ouhoˀor/

As for the status of adjectives in Classical Egyptian, the particle /ᵊn/ is not present, and potential difference in seperate nominal and adjectival forms by vocalism are obscured by the writing system, so that's difficult to say.

All that said, I'm not really much of an expert in Egyptian, and I've only been studying the various phases of it for about a year and a half now. I primarily work on Greek and Anatolian branches of Indo-European, but I do quite enjoy working on Afroasiatic languages as a hobby :)

I'll try to get a more straight answer for you on Egyptian adjectives tomorrow after I've had the chance to consult an actual Egyptian linguist on the matter.

Mattitiahu said...

Hmm... quick update, I was formalizing my responses here for a post on my own blog, and I noticed something with Egy. √nfr and its reflexes in Coptic that I failed to notice while writing my last point, here's what I've been writing:

As for the status of adjectives as a seperate category in Classical Egyptian, the Coptic particle /ᵊn/ is not present, and potential difference in seperate nominal and adjectival forms by vocalism are obscured by the writing system, so it is difficult to say on the evidence of Classical Egyptian alone. If Egy. nfr and the verbal, nominal, and adjectival forms preserve distinct fossilized forms in Sahidic (inf. ⲛⲟⲩϥ͞ⲣ /noufṛ/ stat. ⲛⲟϥ͞ⲣ/nofṛ/ pred.adj. ⲛⲉϥ͞ⲣ– /nefṛ/ n.f. ⲛⲟϥⲣⲉ/nofre/, ⲛⲟⲃⲣⲉ/nobre/ adj. ⲛⲟⲩϥⲉ /noufe/) provides any evidence however, it is suggestive that the Older Egyptian root and vowel pattern morphology had seperate morphological patterns for nouns and adjectives based off a semantic root. This, of course, is to be expected from comparative Afroasiatic. The 'adjective' form /noufe/ however according to Lambdin (1983: 252) is a rare form and is most often used in compounds.

I'm planning on returning to the topic of adjectival usage in Egyptian at a later date, but for now I'm going to stick with analyzing the Coptic evidence synchronically. My insufficient knowledge of Late Egyptian and Demotic are presently handicapping me in making a better diachronic analysis of this phenomenon.

Lameen Souag الأمين سواق said...

Moubarik: yes, all Berber languages have adjectives, whether verb-like or noun-like or both. Siwi definitely prefers noun-like ones (eg azəggaγ "red"), but also has a number of verbs expressing properties (eg yəmm°əm, təmm°əm, etc. "be sweet").

Matt: thanks for the further comments.

Anonymous said...

I'm kinda surprised by Dixon's counter-claim that adjectives are a universal class. I skimmed through the link to google books, and though I'm not entirely convinced, it was interesting to know.

On the subject of Egyptian, the Coptic attributive particle does occasionally appear in Middle Egyptian, and is fully productive in Late Egyptian and esp. Demotic as the language becomes more isolating. Coptic has both nominal and relativized verbal "adjectives" in a way similar to Hausa, which also uses an attributive particle before a noun. Not all that unusual really. Old/Middle Egyptian adjectives were almost all derived from nouns and verbs, e.g. you can have a nisbe construction with the addition of -i to a noun in masc. forms, like Hebrew and Arabic.

Anonymous said...

By the way, the sabe forms are nouns that can be used attributively. They just happen along with a some other nouns to have retained the older gender/number marking.

Anonymous said...

Well, Italian have very little difference between nouns and ahjectives, even though these terms are used in the egrammers.

In Italian you can find expressions like "macchina lavatrice" - washing mashine. The word "lavatrice" is here used as an adjective.

The same is true about Finnish.

John Cowan said...

Lameen, you're quite right to point to "hen", though (as with most things Mandarin) it's not a matter of syntactic rules but of what it means: is it 'very' (in which case the verb is an adjective) or just a throat-clearing noise (in which case, not an adjective).

David: Absolutely. The coolness of Chinese grammar is its minimalism and its tendency to understand things by common sense rather than rules. "You could say it that way, but usually we wouldn't" is the equivalent of an ungrammaticality judgment in Mandarin.

Anonymous said...

or just a throat-clearing noise (in which case, not an adjective).

The way I was implicitly taught it is that it's the copula that's used with adjectives, while shi is the copula used with nouns... however, it's 4:20 at night here, so don't take my word for it.

(In case anyone is interested, hen can still mean "very", while in Very Classical Chinese shi was a demonstrative pronoun.)

John Cowan said...

David, I was speaking of how you distinguish adjectives, aka stative verbs, from just plain verbs. You're right that shi tells you which words are nouns as opposed to adjectives or verbs.

Languagehat said...

I've posted about this here, and people seem to be strongly disputing the idea that all languages have adjectives. How strong is Dixon's evidence for this?

Panu said...

The same is true about Finnish.

Uh, I don't know how much you can call it "the same", but it is true that in Finnish adjectives are declined in the same categories, with the same endings etc. as nouns, and grammars for us native speakers generally operate with the class of "nominit" (from Latin "nomina", plural of "nomen", cognate with English "noun"), which includes nouns ("substantiivit") and adjectives ("adjektiivit").

IMHO though, there is one important difference: you can turn adjectives into nouns, but you definitely cannot use every noun as an attributive adjective. Take the noun tekijä ("doer, maker, active person", nomen agentis of the verb tehdä, to do). You can create a compund word (tekijämies) where the second part only is declined (tekijämiehen, tekijämiestä, tekijämiehet...), but what you definitely cannot do is to decline tekijä as an attributive adjective: *tekijän miehen, *tekijöitä miehiä etc.

joe said...

In Newar (the native language of the Kathmandu valley in Nepal), the only way to regularly distinguish verbs from adjectives is by the morphology on the habitual form. So, say, khwauM 'cold (of food etc.)' is essentially distinguished from a habitual verb by the fact that -auM isn't a good ending for a verb. Apart from that (as far as I can tell) these adjectives can take verbal morphology (eg. tense, as in khwauMye 'It will be cold') and verbs are frequently used adjectivally (when combined with an attributive suffix, which adjectives also require when used attributively). The 'very' test doesn't really work particularly well because the Newar word for 'very' encompasses meanings such as 'a lot', 'frequently', etc.