In linguistics, a compound verb or complex predicate is a multi-word compound that acts as a single verb. One component of the compound is a light verb or vector, which carries any inflections, indicating tense, mood, or aspect, but provides only fine shades of meaning. The other, "primary", component is a verb or noun which carries most of the semantics of the compound, and determines its arguments. It is usually in either base or conjunctive participial form.
A compound verb is also called a "complex predicate" because the semantics, as formally modeled by a predicate, is determined by the primary verb, though both verbs appear in the surface form. Whether Noun+Verb (N+V) compounds are considered to be "compound verbs" is a matter of naming convention. Generally, the term complex predicate usually includes N+V compounds, whereas the term compound verb is usually reserved for V+V compounds. However, several authors also refer to N+V compounds as compound verbs.
Compound verbs are to be distinguished from serial verbs which typically signify a sequence of actions, and in which the verbs are relatively equal in semantic and grammatical weight.
Thus, there are two classes of complex predicates:
- V+V compounds: The true compound verb, where the first verb is a Light verb (LV), followed by a primary or Heavy verb. With a few exceptions all compound verbs alternate with their simple counterparts. That is, removing the light verb / vector does not affect grammaticality at all nor the meaning very much: निकला nikalā '(He) went out.' In a few languages both components of the compound verb can be finite forms: Kurukh kecc-ar ker-ar lit. "died-3pl went-3pl" '(They) died.'
- N+V compounds: A compound with Noun+verb, converting the noun into a verbal structure; the arguments and the semantics are determined by the N and the tense markers / inflections are carried by the V. This would include English stretched verb examples like take a walk or commit suicide. Often the Verbs participating N+V compounding are also those that participate as LVs in V+V compounds. The N+V compound appears in almost all languages, especially with LVs such as "do", "make" etc., and are sometimes not considered to be a true compound verb.
Languages with compound verbs
Compound verbs are very common in all the languages of India, though they are more frequent in the northern Indo-Aryan languages
than in Dravidian languages
. In addition to Hindi-Urdu
, they occur in other Indo-Iranian languages like Persian
, in Tibeto-Burman
languages like Limbu
. Also, they are very frequent in Altaic languages
, and in Central Asian Indo-European languages Kazakh
, and Kyrgyz
, and in northeast Caucasian languages like Tsez
Conventionally, the English language expresses fine distinctions as to the beginning, duration, completion, or repetition, of an action in the form of compound verbs, using auxiliaries
or other lexical
mechanisms. Examples here include was starting
, had lived
, had been seen
, etc. This usage reduced the need to create complex predicates.
Though V+V compound verbs are rare in English, one may illustrate the form with the example "start reading". In some interpretations, one may consider "start" as a light verb, which carries markers like tense. However, the main part of the meaning, as well as the arguments of "started reading", i.e. answers to questions such as who? (agent), or what was it that he "started reading" (object) are determined by the second, primary verb, "read". Note that "start" also modifies the meaning or the semantics, by focusing on the early part of the "reading". Also note that "start" carries plural/tense markers (they start|he starts reading), whereas "reading" appears in this fixed form, and does not change with tense, number, gender etc.
Whether gerundive forms like "start reading" are compound verbs is controversial in English; many linguists prefer to treat "reading" as a nominal in its gerundive form. However, the compound verb treatment may have some advantages, particularly when it comes to semantic analysis. For example, in X starts reading Y, the question what did X start is less revealing than what did x "start reading".
English has many examples of N+V compound predicates: see stretched verb.
Sometimes examples from English cited for compound verbs turn out to be serial verbs, e.g.: What did you go and do that for?; or your business might just up and leave.
Compound verbs are very common in Indo-Iranian languages
, such as Hindi-Urdu
, where as many as 20% of the verb forms in running text may be compounds.
For example, Hindi निकल गया nikal gayā, lit. "exit went", means 'went out', while निकल पड़ा nikal paRā, lit. "exit fell", means 'departed' or 'was blurted out'. In these examples निकल nikal is the primary verb, and गया gayā and पड़ा paRā are the light verbs.
Similarly, in both English start reading
and Japanese 読み始める yomihajimeru
-read" "start reading," the vector verbs start
and 始める hajimeru
"start" change according to tense, negation, and the like, while the main verbs reading
and 読み yomi
"reading" usually remain the same. An exception to this is the passive voice, in which both English and Japanese modify the main verb, i.e. start to be read
and 読まれ始める yomarehajimeru
-start" start to be read
Under the influence of a Quichua
substrate speakers living in the Ecuadorian altiplano
have innovated compound verbs in Spanish:
- De rabia puso rompiendo la olla, 'In anger (he/she) smashed the pot.' (Lit. from anger put breaking the pot)
- Botaremos matándote 'We will kill you.' (Cf. Quichua huañuchi-shpa shitashun, lit. kill-CP throw.1plFut, तेरे को मार डालेंगे )
Historical Processes and Grammaticalization
As languages change, the vector or light verb may retain its original meaning to different degrees of bleaching
, part of the process of grammaticalization
Thus, in the Hindi compound nikal paRā (exit+fall), paRā has almost none of its "fall" meaning, though some of the finality of the fall also is transferred as a perfective
aspect. On the other hand, the Japanese "meru" (start) retains a good deal of its independent word meaning even in the compound.
In the long run, it has been suggested that LVs that are particularly frequent, may become grammaticalized, so that they may now occur systematically with other verbal constituents, so that they become
an auxiliary verb (e.g. the English verb "be", as in "I am eating", or "had" in "they had finished"), or, after sound change, even a clitic (a shortened verb, as in "I'm"). In particular, some verb inflections (e.g. Latin future tense inflections) are thought to have arisen in this manner.