This book looks into the variation and change in the use of non-finite sentential complementation patterns in English. The study of complementation in general is a subject which has received a great deal of attention in recent years, and the availability of large electronic corpora has added a new perspective to the investigation of the different factors that play a role in the choice between alternative complement patterns. The book aims to contribute to this growing body of knowledge.
To understand the nature of complementation, it is helpful to consider the statement by Huddleston and Pullum (2002, 219) that the âmost important property of complements in clause structure is that they require the presence of an appropriate verb that licenses them.â Of course, other types of lexical heads besides verbs may also license or select complements, but the close relation expressed by the term licensing is at the heart of the concept of complementation. The relation can also be expressed by using the term âmatchingâ: â[c]omplementation is basically a matter of matching a particular complement type to a particular complement-taking predicateâ (Noonan 1985, 90).
In work on the matching issue it is possible to identify two different points of departure. An investigator can identify a particular syntactic pattern and then examine the class or classes of heads, or complement-taking verbs, adjectives or nouns, that license that particular syntactic pattern. This may be called the pattern-based approach, and a recent example of how it may yield linguistic generalizations is Kim and Davies (2016). They placed a focus on verbs that select what may be termed the transitive into -ing pattern, as in They talked John into taking part, where a matrix verb selects an NP and a following -ing clause. An investigator adopting a pattern as a point of departure would typically carry out corpus searches with a search string that identifies the syntactic type of the constituent selecting the pattern as a lexical head but would not specify the specific lexical head in question, instead spelling out the complement as specifically as possible. This type of work may for instance yield generalizations about semantic classes of verbs that select a particular syntactic pattern.
The other point of departure in work on the matching issue, by contrast, may be called the head-based approach. Using this approach, an investigator spells out the specific complement-taking verb, adjective, or noun as the key part of the search string but leaves the nature of the complement or complements selected by the head open, at least to some extent. The head-based approach is particularly appropriate when a head selects two or more types of complement that are fairly close to each other semantically. For instance, the adjective accustomed has selected both to infinitival and what may be termed to -ing complements in recent English, as in accustomed to dine alone and accustomed to dining alone, and an investigator interested in variation between the two types of complements will want to include at least part of the sentential complement in the search string used, in addition to of course including accustomed in the search string. A head-based approach can then be expected to yield insights into subtle differences of meaning between the types of complement selected by the head in question or into other factors influencing their use, and the insights can then be tested in later work by taking other heads into account.
The present authors recognize the value of the pattern-based approach. They also recognize that the two approaches can be used to complement each other. However, in this book they adopt the head-based approach as a point of departure. The reason is that in each chapter they have a focus on predicates that select two different types of non-finite complements that are or may be close to each other in meaning. The head-based approach can be expected to yield information on the meaning and use of each variant in such cases. For instance, in Chap. 2 the authors investigate variation between what may be termed at -ing and on -ing complements of the matrix verb work in recent American and British English as illustrated in (1aâb), from the Corpus of Contemporary American English (COCA):
- (1)a. But heâs a very hard worker, and he works at getting better and better. (COCA, 2000, NEWS).b. While training in Australia, Bennett logged thousands of miles on her bike and worked on getting stronger to drop her time in the rune [âŚ] (COCA, 2012, NEWS)
In sentence (1a) the matrix verb work selects an at -ing and in (1b) an on -ing complement. To facilitate discussion, it is also helpful to use the term âconstructionâ here, to denote a pairing of a form with a meaning (cf. Goldberg 1995, 1), and to recognize that constructions can exist at different levels. Thus in (1aâb) the larger construction of interest is of the type work at/on -ing, including the complement at/on -ing, which itself constitutes a construction, as does the -ing clause part, which is a gerund. (See the comments on âdifferent levels of constructional assemblyâ in De Smet 2013, 34â35.) Recognizing that constructions can be identified at different levels of syntactic structure makes it possible to zero in on the semantic contribution of each level to the meaning of the sentences in question and on the use of each variant, and this is done in Chap. 2. This focus on meaning and on subtle differences in meaning between two constructions that appear similar at first sight is in the spirit of what has been termed Bolingerâs Generalization, to the effect that a âdifference in syntactic form always spells a difference in meaningâ (Bolinger 1968, 127). Analogous considerations hold for other pairs of constructions in later chapters.
One basic question relating to the analysis of infinitival and gerundial complements made in this study should be addressed at the outset. In some recent work the assumption is made that âwhat you see is what you getâ (for instance, see Goldberg 2003, 219, 2006, 10). However, for their part the present authors recognize a role for understood constituents. That is, it is assumed here that even though infinitival and gerundial clauses often do not have overt subjects, there is a role in their analysis for covert (or understood, implicit) subjects. In support it is possible to invoke Jespersen:
An appeal to Jespersen is in the nature of an ap...Very often a gerund stands alone without any subject, but as in other nexuses (nexus-substantives, infinitives etc.) the connexion of a subject with the verbal idea is always implied. (Jespersen [1940] 1961, 140)