āComplementation,ā in the words of Noonan (2007: 101), āis basically a matter of matching a particular complement type to a particular complement-taking predicate.ā In this sense complementation, even if the term was not used, was a prominent theme in traditional grammars of English, including Poutsma (1929 [1904]). In more recent work complementation has continued to be a central concern in generative studies, with Rosenbaum (1967) forming a basic framework for many later studies, in part because of the well-defined system of phrase structure rules offered in his book. Work in the generative framework was mostly based on intuitions and informant judgments, but the advent of large electronic corpora in the last two or three decades has given a new dimension to the field. The studies in this book take full advantage of the availability of large electronic corpora. In addition, the approach here draws on statistics in a more sophisticated way than is the case in most other current work on complementation. In particular, the use of multiple regression analysis promises new insights into the incidence of different types of sentential complements in recent English.
Two broad approaches can be identified in work on complementation that takes advantage of corpus evidence. The two points of departure follow naturally from the fact that complementation concerns a relation between a complement-taking predicate and a certain complement type. It is possible to proceed from the point of view of a particular complement-taking predicate or from the point of view of a particular complement type. A complement-taking predicate contains, or consists of, a lexical categoryāa verb, adjective, or nounāand the lexical constituent in question may be called a head, with the head selecting a complement type. The first point of view may then be termed the head-based approach, and it proceeds from identifying and examining the types of complements that a particular head selects. As for the second type of approach, it takes a particular complement type as its point of departure, and then examines the heads that select that particular complement type. A complement type is a phrasal category and as a phrasal category it may also be viewed as a pattern. The second point of view may then be termed the pattern-based approach.
In a pattern-based approach, a search string is chosen that does not specify the heads that select the construction, and the approach can yield information for instance on the verbs, adjectives and nouns that select a particular pattern. By contrast, a head-based approach recommends itself for instance when a particular head selects more than one type of complement such that the resulting head-complement constructions are fairly similar in meaning.
Most of the chapters of the present book are of the latter type, with only Chapter 3 diverging from this approach in that it takes up an analytical issue that arises in the study of adjectives co-occurring with infinitival complements of two types. In the other chapters the focus is on a head that selects two types of morphologically and morphosyntactically distinct sentential complements that are nevertheless so close to each other in meaning that they have often been treated under the same sense in major dictionaries. The objective is then to inquire into the different factors that affect the choice of complement on the basis of a selection of corpus data. A range of such factors are considered in the several chapters of this book, and advanced statistics is brought to bear on the salience of such factors. As far as the present authors are aware, this is the first book-length study of non-finite complements in English with such an orientation. The findings are based on the heads investigated, and may at first sight appear to be narrow in scope. However, the information gained here on factors affecting complement choice is useful beyond those heads in that it can be taken into account in subsequent work when other heads, quite possibly emerging from the application of the pattern-based approach, are examined in subsequent research. Ultimately, the head-based and pattern-based approaches can be combined and applied jointly in order to gain a full picture of the system of English predicate complementation.
The heads to be investigated are the three adjectives
accustomed,
afraid, and
prone. In the case of
afraid, the variation is between
to infinitives and what are in this book called
of -
ing complements. For initial illustrations, consider the sentences in (1aāb), both from the
Strathy Corpus of Canadian English.
- (1a)
I was afraid to hang up.
- (1b)
ā¦ Quebecers are not afraid of going it alone.
In the case of
accustomed and
prone, initial illustrations of the variation to be considered are given in (2aāb) and in (3aāb), respectively. (2aāb) are from the
Hansard Corpus and (3aāb) are from the Corpus of News on the Web (Davies 2013), henceforth referred to as the
NOW Corpus.
- (2a)
People in the Services are accustomed to disregard political prejudices. (Hansard 1952)
- (2b)
We are accustomed to fertilising the land on which we grow crops ā¦ (Hansard 1955)
- (3a)
McPhee is more prone to stress the agony of compositionā¦ (US, 2017)
- (3b)
Frankie seems prone to trying almost anything. (US, 2017)
An important basic assumption made in this volume is that all of the sentences in (1aāb), (2aāb), and (3aāb) involve a non-finite complement clause and that the complement clause, an infinitival clause in (1a), (2a) and (3a) and a gerundial clause in (1b), (2b), and (3b), has its own understood (or covert or implicit) subject. The idea that infinitival and gerundial constructions have covert subjects was made by major traditional grammarians, including Jespersen (1961 [1940]: 140), and it is generally made in current work. It is not shared by all linguists, but the present authors recognize that there is considerable evidence for understood subjects, in addition to an appeal to the tradition represented by Jespersen. One piece of independent evidence has to do with the generally accepted principle of binding theory that reflexives need to have a c-commanding antecedent (see for instance Radford 1997: 115). The well-formedness of a sentence such as Perjuring himself would not bother John then shows that the non-finite subject of the sentence has a covert subject, because otherwise the reflexive would lack a c-commanding antecedent. (The essence of the argument goes back to Postalās (1970) pioneering article, but today it is appropriate to present the argument in terms of binding theory; see Landau 2013: 115.)
The assumption of understood (covert) subjects, generally made in mainstream work today, also makes it possible to use a straightforward definition of the notion of control from Duffley (
2014), which explicitly presupposes a framework with covert (
or understood)
subjects:
Control has to do primarily with the question of what determines the identity of the unexpressed subject of non-finite verbal forms such as the infinitive or the gerund-participle in constructions such as Joseph tried to find a quiet place and Peter enjoyed going fishing in his boat. (Duffley 2014: 13; the term āgerund-participle,ā used by Duffley, corresponds to the traditional term āgerundā used...