Recommendations

 

Verb phrase (VP)

This category is slightly more difficult to define, since there is disagreement over the extent of the verb phrase. In particular, should the verb phrase include only the words that are verbs, or should it also include the complements of the verb? In the examples given in this document, and in the sample texts in the appendices, we have chosen to include the complements, but it must be noted that this is an open issue, and we are in no way implying that this analysis is preferable to the alternative. The choice to be made at this level, i.e. the inclusion or exclusion of verbal complements in the Verb Phrase, is shown by the examples in 27 and 28, 27 showing the inclusion of the complement of the verb in the verb phrase and 28 excluding the complement:

(27)  He [VP took up [NP a clothes brush NP] VP]

(28)  He [VP took up VP] [NP a clothes brush NP]

An advantage in the type of analysis shown in 27 is that the relative levels of the constituents can be shown to a greater extent -- i.e. complements of the verb are included in the verb phrase, while adjuncts and peripheral adverbials are left at sentence level.

However, in a case where an adjunct occurs before the complement of the verb, the approaches used in 27 and 28 would cause problems, since either both the adjunct and the complement would be included as daughters of the verb phrase, or both would be daughters of the sentence, rather than keeping the complement as a daughter of the verb phrase and the adjunct as a sister of the verb phrase. These problems may be solved by an additional notation, but at some level, arbitrariness is inevitable.

Regardless of the choice made over the extent of the Verb Phrase, there arises a problem of discontinuous Verb Phrases. A complex verbal construction may be discontinuous, e.g. the auxiliary and the main verb are separated in inverted constructions in English, or the main verb is positioned at the end of the sentence in German and Dutch. Such discontinuity can be avoided by having different labels and constituents for the auxiliary verb and the main verb, resulting in an analysis as shown in the Dutch example 29 below:

(29)  [NP Ze NP] [AUX zullen AUX] [ADVP er ADVP] [VP [NP de VN-agenda [PP voor [NP het komende jaar NP] PP] NP] behandelen VP].

and in the English interrogative inverted example 30, using the so-called `dummy auxiliary' do:

(30)  [AUX Do AUX] [NP they NP] [VP confide [PP in you PP] VP]?

As with Noun Phrases, Verb Phrases can be identified by a constituency test. In strong constituency languages like English, the whole VP can be moved, but not part of it: compare 31 and 32:

(31)    Give in to blackmail, I never will

(32)  *Give in, I never will to blackmail

However, there are languages in which constituent tests do not work. These will typically be languages with flexible word order, such as Finnish. 33 is an example of a discontinuous VP (Vilkuna 1989: 26):

(33)  Maailmaa nähnyt hän on.

world-Part seen he is `He IS a widely-travelled person.'

For Finnish, then, evidence for a VP is less convincing than it is for English, and a dependency approach seems the more natural choice. (Covington (1990) provides a parsing strategy for variable word order languages and Covington (1991) for parsing discontinuous constituents, both using a dependency syntax approach.)

In Italian also, constituency tests cannot be applied. This can be shown through the distribution of VP-adverbs (e.g. completamente `completely', intenzionalmente `intentionally', attentamente `carefully') and S-adverbs (e.g. probabilmente `probably', certamente `certainly'). In English, these different classes of adverbs have a different distribution within the sentence. In contrast, in Italian, the distinct adverb classes cannot be distinguished on the basis of their distribution in the sentence. S-adverbs and VP-adverbs can occur in the same positions within the sentence, as illustrated in examples 34 to 37:

(34)  Attentamente/certamente, il bambino ascoltó la storia `Carefully/certainly, the child listened to the story'

(35)  Il bambino attentamente/certamente ascoltó la storia `The child carefully/certainly listened to the story'

(36)  Il bambino ascoltó attentamente/certamente la storia `The child listened carefully/certainly to the story'

(37)  Il bambino ascoltó la storia attentamente/certamente `The child listened to the story carefully/certainly'

Thus, in Italian as well as other languages, neither the position nor the syntactic context can help to decide whether an adverb is an S-adverb or a VP-adverb; this can only be stated by considering its semantic content and the way it relates to the content of the predicate or the sentence. This situation has consequences for the success of standard VP-tests.