Multilingual...Friday?

Inspired by pne's recent question about "with", I've decided to rewrite my response in the form of a longer post, borrowing the format used by aadroma in his regular "Multilingual Monday" feature. Unfortunately, covering the subject in a manner both concise and accurate is hard to do without assuming some familiarity with certain grammatical concepts. Since I'm sure there will be some interested readers who aren't linguists, I've tried to put together this brief introduction.

Theta roles

To make cross-linguistic grammatical comparison easier, many linguists make use of something called theta roles. A theta role characterises the relationship of a participant to the action being expressed. All language need to express actions and they all need ways of distinguishing between different participants. But which roles are distinguished grammatically (optionally or otherwise) by different languages varies quite a bit.

For instance, a number of languages distinguish agents (participants which intentionally perform the action) from experiencers (participants which receive sensory impressions). English doesn't; you pluck a rose and you smell a rose. Perhaps in the second case you intended to smell it and perhaps you didn't, but the syntax is exactly the same regardless. Other roles are universal: Every language I've ever heard of has some way of indicating that rose in the example above is a patient (participant which undergoes a change of state as a result of the action), and more often than not indicating this is mandatory.

(I hope this brief explanation gives you some idea why linguists have created a whole parallel vocabulary instead of just saying "subject" and "object" like your grade-school language arts teacher, because I don't have time to go into more detail right now. Maybe later--with some mind-blowing examples from Basque!)

Linguistic typology

When it comes to morphology and syntax (or--as the hip kids say--"morphosyntax"), it's customary to arrange languages on a rough continuum from isolating to polysynthetic based largely on the ratio of morphemes to words. A morpheme is the smallest unit of meaning in a language. For example, the word bakers has three: The root bake, the agentive suffix -er, and the plural ending -s. Of these, only bake is a free morpheme--a morpheme that is also a stand-alone word. -er is a derivational morpheme, because it changes the word-class of bake (from verb to noun) and -s is an inflectional morpheme because it doesn't, but changes the meaning of the word all the same.

(The more linguistically-sophisticated among you are doubtless already clucking your tongues. Yes, this is a gross simplification of these concepts; thus the links. If you know me, you know how I love to point out the difficulty of answering the deceptively simple question "What is a word?", but there's no room for another digression in what is already a digression.)

An isolating language is one in which the ratio of morphemes to words is closest to one-to-one. Polysynthetic languages are on the opposite extreme; it's often said that a single word in one of them can correspond to an entire sentence in an isolating language. English is toward the isolating end--though not (as the example above demonstrates) on the extreme end with languages like Thai and Chinese. Osage, which you may have become familiar here with through previous entries is synthetic in its verb complex but not polysynthetic because it does not allow operations like object incorporation which are found in Navajo and Eskimo.

Any questions so far?

Use me, abuse me: Indicating the instrument

If you find yourself confused by some of the grammatical terminology in this entry, you may want to jump back and read the previous entry, which defines some of the concepts referred to here.

An instrumental role

One theta role which seems to be frequently marked is that of instrument (an inanimate participant that performs an action at the behest of the agent). A prototypical example would be knife in the following example:

I cut the bread with a knife.

Clearly, bread is the patient in this example: It undergoes a change of state due to the action of cutting. But what about the knife? It's what actually transforms the patient from uncut bread to cut bread, but it's inanimate, so it can't be an agent. Since it is I who has the intentionality in this example, I fulfils the agent role and knife is considered an instrument.

Now, obviously, not every language indicates the instrument by using a prepositional phrase, like English does. What are some of the other possibilities?

Derivative approaches

One solution is deriving a new verb that expresses "to cut with a knife" without the need to indicate an explicit instrument. English, which make liberal uses of verbing, often allows this. For instance:

He called me by phone. -> He phoned me.

But, in this case, the verb knife exists with a slightly different meaning, i.e.

I knifed him. <-> I stabbed him with a knife.

Applicatives represent a specific sort of "verbing" derivation, one that generally turns one sort of verb into another. Basically, an oblique argument (such as the object of a prepositional phrase) is "promoted" to the position of direct object. German forms applicatives productively with the prefix be-, e.g.:

Ich wohne in einer Eigentumswohnung. "I live in a condo." > Ich bewohne eine Eigentumswohnung. "I inhabit a condo."

Just as it was possible in this example to take the object in and make it the direct object of a verb, it is possible in some languages to lift "knife" from "with a knife" and do the same. For instance, Hausa:

an sṑke shì dà wuƙā "Somebody stabbed him with [a] knife." -> an sōkà masà wuƙā (lit.) "Somebody stabbed to.him [a] knife."

Compare the English sentence "I slice a knife into the bread". But, once again, the similar English phrase ?"I cut a knife into the bread" is jarringly unidiomatic.

Another possibility is deriving a compound verb, knife-cut (Cf. hand-cut, e.g. I hand-cut the fries <-> I cut the fries by hand.) But, again, the freedom of English isn't quite enough for this usage to sound idiomatic. That might not be the case in other languages. (For a slightly different type of compound, see the Chickasaw example, below.)

Getting analytical

The idiomatic choice in English (and in other Standard Average European languages, like Spanish, German, and even Welsh) is to use free morphemes to construct a prepositional phrase, i.e. with a knife, con un cuchillo, mit einem Messer, etc.

Chinese favours a similar solution, although--strictly speaking--it doesn't use prepositions but "co-verbs".

我用刀把麵包切 wǒ yòng dāo bǎ miànbāo qiè "I cut the bread with a knife."

用 yòng, corresponding here to "with", can also be a full verb meaning "use" (e.g. 教我怎样用刀叉 jiāo wǒ zěnyang yòng dāochā "Teach me how to use a knife and fork"). Thus, the sentence could also be translated as "I, using a knife, seize [another co-verb] bread, cut". This represents what is often called a serial verb construction.

Theoretically, it would be possible to use word order alone to express the instrumental role. All you would need would be a predicate which takes at least three arguments--one for the agent, one for the patient, and one for the instrument. (Not necessarily in that order.) English does have verbs like this, e.g. give with slots for agent, theme, and patient, but nothing I can think of which allows one of the arguments to be an instrument. I'm sure the constructed language Lojban would allow such a predicate, but I couldn't tell you what it would be.

Inflection on the noun

Agglutinative" endings can be difficult to distinguish from independent words. I imagine much of the reason why the Korean instrumental affix (으)로 /(u)lo/ is considered the former is that is has allomorphs based on the form of the stem. (The (으) /u/ appears after some consonant stems in order to break up awkward clusters.)

나는 빵을 나이프로 베었어요 /na nun ppang ul naiphulo pay.ess.e.yo/ "I cut the bread with a knife"

Turkish ile is right on the border between postposition and suffix. It can appear as a separate word (e.g. bıçak ile "with a knife") or it can attach to the noun and follow vowel harmony, just like any other inflectional suffixes (e.g. bıçakla; cf. bıçakta "on a knife").

Fusional endings, however, are less ambiguously inflectional because the same ending simultaneously expresses multiple categories such as number and gender in addition to case. For instance, Czech nožem "with a knife". For other nouns, this ending could appear as -ou, -í, or -ím in the singular and -y, -i, -ami, -ími, etc. in the plural. (By contrast, Turkish would simply stick on the plural suffix -lar before the ending, i.e. bıçaklarla "with (a/the) knives".)

Inflection on the verb

Osage, on the other hand, takes quite a different tack: It adds an instrumental inflection to the verb rather than the object. This is the locative prefix i-. Compare:

wacue hpaase "I cut the bread"
wacue mąhį ihpaase "I cut the bread with a knife."

As I've mentioned elsewhere, Osage is highly synthetic in its verb complex, but not elsewhere. There's no marking on either wacue "bread" or mąhį "knife" to indicate which is patient and which is instrument. Generally, direct objects precede other complements in Osage, but the semantics would allow one to reorder the sentence if the speaker wanted to emphasise mąhį. The semantics (a knife can cut bread, but bread can't cut a knife!) would make the meaning clear in any case.

In general, Chickasaw uses a similar expedient in the form of the instrumental prefix isht, e.g.

tali' is(h)sabo'watok "S/he hit me with a rock". (Cf. bo'li "to beat, to pound".)

However, in the specific case under scrutiny, it uses a participial construction. That is, the verb bashtabli "to cut with a knife" is composed of the participle of bashli "to cut, to saw" and the root verb tabli "to cut, to pull apart". Thus, the statement paskaã bashtablilitok "I cut the bread with a knife" could be more literally translated as "Sawing, I cut the bread".

Getting complex

The second Chickasaw example is right on the border of another possibility, that of using a subordinate clause of some sort. This is the sort of construction required by the polysynthetic Wakashan language Kyuquot Nootka (a relative of Kwakiutl).

č̓ičƛa·qƛ 'uḥw̓ał x̣uta·y "He will cut it using a knife."

The English above is, obviously, only a rough equivalent. The literal meaning of 'uḥw̓ał is "it-use".