MatrixDoc_ClausalComplements

Clausal Complements

In most languages, verbs like think, believe, "know" (and many others, e.g. fear, annoy) can take predicates as arguments. For example, in

I know that Kim runs every day.

that Kim runs every day

is the object argument of "know" and is a predicate (verby rather than nouny) structure. Similarly, in

That Kim runs every day annoys Sandy

that Kim runs every day

is the subject argument of "annoy".

These arguments, both subjectival and objectival,

• are called sentential,

or clausal, complements of the verb.

(NOTE: there can be different views on what is the head of that Kim runs every day, e.g. Pollard&Sag analyze the verb as a head while an alternative analysis employed in the Grammar Matrix is that it is the complementizer), but it can be called a predicate regardless).

The current version of the questionnaire only covers objectival clausal complements. Objectival complement clauses may take the same object position as noun complements (e.g. come before the verb in OV languages) or, in many languages they are extraposed to the end of the sentence and after the verb, even if the language is generally OV.

Each complementation strategy that you specify must correspond to a clausal verb type that you can define in the Lexicon section.

There are several ways in which languages usually mark clausal complements</b>. Many languages employ more than one way.

1. The complement clause is marked by a complementizer word, usually at the edge of the clause.

I believe that Kim is sleeping

In the above example, that is a complementizer.

If you would like some verbs to go only with one complementizer but not the other, add a FORM feature to your complementizer (e.g. FORM that).

2. The complement clause is marked by some special morphology on the verb.

The most common example of this is nominalization. The verb takes a special affix and starts behaving like a noun to a degree. In nominative-accusative languages, it is common that the nominalized verb's subject then needs to take a non-nominative case, like genitive.

For example, in Modern Standard Turkish (MST), like in many Turkic languages, the verb can be nominalized with a special affix, take some person and number inflectional affixes (but not the TAM markers; and the finite verb's inflectional paradigm and the nominalized verb's inflectional paradigm are generally different in MST), then take noun case markers after that. The verb's subject takes genitive case:

Ali-nin gecen aksam nehr-in kenar-in-da kos-tug-un-u gor-dum

• Ali-GEN past evening river-GEN shore-3.SG-LOC run-FN-3.SG-ACC see-PAST.1.SG

I saw that Ali was running along the river the other evening (Kornfilt).

There are different kinds of nominalization. In particular, some nominalized verbs can be modified by adverbs (like regular verbs):

Kim instantly eating the pizza did not surprise anyone.

while others can be modified by adjectives, like nouns:

Kim's instant eating of the pizza did not surprise anyone.

In some languages (e.g. Japanese), the nominalized verb that can be modified by adverbs needs to keep the case frame of an indicative clause, however, that is not the case in e.g. Turkic languages, where the nominalized verbs with genitive subjects can be modified by adverbs (Asarina and Hartman). In this questionnaire, you have an option to implement "high" or "low" nominalization; that means the verb will be nominalized "higher" or "lower" in the tree. In the case of "high" nominalization, there will be a verby constituent which will then be turned into something nouny; in contrast, with "low" nominalization, the verb is turned into something nouny and then a nouny constituent is formed with its arguments.

You will need to make a technical decision about how the semantics of high nominalization will be represented in your grammar. You can choose to either include a nominalization relation in the MRS representation of the embedded constituent or assume that it is semantically empty. In the former case, the complement-taking verb's ARG2 will be the ARG0 of the nominalization relation. In the latter case, the ARG2 will instead point directly at the embedded verb.

NOTE: A nominalized clause is a type of non-finite clauses which is used very commonly in clausal complements in world languages; other non-finite clauses include subjunctive mood and infinitive form. In general, those also fall in the category where the verb has some special morphology on it when it acts as part of a clausal complement. However, the subjunctive tends to occur with causative constructions and object raising, and the infinitive---in constructions where the subject is shared by the matrix and the embedded clause. Neither subject nor object raising is currently supported by the questionnaire, though you may use the FORM feature to indicate that the embedded clause is some form and then develop your grammar further by hand.

The below example may be an example of subjunctive complement clause that is covered by the current version of the questionnaire, assuming the complementizer+particle is analyzed simply as a single special complementizer.

In Russian, the subjunctive inflection looks the same as the past tense. Note that the example below is also is an example of a complement clause having a complementizer which attaches before the embedded clause. In Russian, a special particle "by" is needed in addition to the complementizer to ensure the subjunctive reading.

Ja ne dumaju chto=by Boris eto sdela-l I NEG think COMP=SBJ Boris that do.PRF-3SG.SBJ I don't think that Boris could do this. (Modified slightly from Noonan by a native speaker, to make it more felicitous out of context).

Specify nominalization strategies on the Nominalized Clauses subpage; FORM on Other Features subpage; MOOD in Tense and Aspect.

Specify the morphological rules in the Morphology section.

3. The complement clause's word order is different from the matrix clause. For example, in German, in which the word order is normally V2, clausal complements exhibit OV order.

Es ist war dass er schalu ist It is true that he cunning is It is true that he is cunning (Noonan).

References

Kornfilt, Jaklin. "On the syntax and morphology of clausal complements and adjuncts in the Turkic languages. Aspects of typology and universals" 1 (2001): 63.

Asarina, Alya, and Jeremy Hartman. "Genitive subject licensing in Uyghur subordinate clauses." WAFL VII (2011).

Noonan, Michael. "Complementation". Language typology and syntactic description, vol. 2: Complex constructions, ed. by Timothy Shopen, 42-140. (1985).