Verb‐governed infinitival complementation in the recent history of English

1. Introduction

1.1 Clausal complementation in the recent history of English: An introduction

This monograph deals with morphological, syntactic, semantic/pragmatic and discursive variation between to- and bare-infinite (BI) verbal complementation in English. It investigates the reasons that lead certain verbs (specifically, dare, need and help) to opt for one solution or the other in the recent history of English. Non-finite complementation is a highly productive area of research (see Fanego 1996, 1998, 2004; Mair 2003: 329; Taeymans 2004a, b, 2006; Schlüter 2005; Cuyckens and De Smet 2007; Loureiro-Porto 2009, 2010; Lohmann 2011; van der Auwera and Diewald 2012: 131; Levshina 2018). While there are numerous studies examining to-infinitive vs BI complementation from a syntactic, semantic or cognitive perspective (see Section 1.2), this volume is an attempt to provide a comprehensive analytic analysis of the possible grammatical, structural, semantic/pragmatic, processing, period and identity-avoidance factors affecting to-infinitive/BI complement alternation.

Some historical background regarding the origin and rise of non-finite complementation is in order to put the analysis in context. Los (2005: 301) distinguishes three different infinitives in OE, based on distribution, structural and positional evidence:

(i) the bare infinitive; (ii) the to-infinitive as complement of the verb wesan ‘be’, as in the PE fossil he is to blame; and (iii) the to-infinitive as purpose adjunct, and as theme- or goal-argument of verbs other than wesan.

Los contends that the BI (type (i)) is a verbal head even though it occurs in positions reserved for case-marked nouns in the clause and therefore ←7 | 8→has nominal traits. With respect to the wesan-to-infinitive construction in type (ii), Los explains that the to-infinitive constituent behaves like an Adjective Phrase in terms of position, distribution and structure and “occurs as a predicate in a small clause”. In ME this infinitive undergoes re-analysis as an overtly passivised to be-infinitive. Type (iii) is restricted to control constructions and functions “as the non-finite counterpart of the subjunctive clause” (p. 301); this to-infinitive has clausal characteristics. The assumption that the to-complementiser derives from the homonymy preposition to is supported by Fanego (2004: 26), who observes that “cognitive research on complementation has generally argued that to-infinitival complements evoke aspects of imagery inherent in the source-path-goal image schema associated with prepositional to”.1 According to Verspoor (1999: 511), “to-infinitive expresses that the subject, figuratively speaking, moves towards the state of being expressed, is not there yet, but is projected towards it”. From a diachronic perspective, Fanego (2004: 27) points out that there are records that show that the source of the to-infinitive was a “directional adverb/preposition with the meaning ‘toward’”, but that “by Late Old English or Early Middle English (1100–1300) to had lost its prepositional character and had grammaticalised to an infinitive marker, so that it began to occur where previously only the bare infinitive was”. Los (2005: 303) suggests that the rise of the to-infinitive may have been influenced by its reanalysis as a non-finite subjunctive, together with the gradual loss of Object-Verb word order, and more importantly, by the obsolescence of finite subjunctive forms. As will be explained in Chapter 2, modal verbs, which typically select BI clauses, are also favoured as periphrastic markers of the subjunctive mood.

The outline of this chapter is as follows: This section provides a general description of non-finite complementation in English, specifically in relation to its morphological, syntactic and semantic/cognitive ←8 | 9→features. Section 1.2 deals with to-infinitive and BI clausal complementation and the determining features of each complementation alternative. Finally, Section 1.3 outlines and justifies the research questions that underlie this study.

1.1.1 Complementation: Morphological features

This volume presents an exploration of verb-governed non-finite clausal complementation in the history of English. Non-finite clauses are characterised by the absence of verbal inflection and characteristically fulfil a subordinate function in the syntactic constituents to which they belong. In fact, Huddleston and Pullum et al. (2002: 89) view non-finiteness as a ‘desententialization’ strategy since it involves the loss of properties that are associated with a clause standing alone as a full sentence. Unification and bonding are the minimal process, “at least pragmatically”, in clause combining and most often take place in subordination, showing “hierarchical downgrading and desententialisation” (Hopper and Traugott 2003: 178). As a consequence, one member of the complement structure is decategorialised and becomes syntactically non-canonical. In their cline of dependency, Hopper and Traugott place finite constructions on the leftmost end of the cline (lowest degree of dependency), and non-finiteness, expressed by clausal remnants such as infinitives and participles, on the right (lowest degree of grammatical integration).

1.1.2 Complementation: Syntactic features

Hopper and Traugott (2003: 175) point out that “ordinary discourse does not consist of isolated, context-free utterances but of linked discourse units”. ‘Clausal complementation’ refers to the process of embedding a clause so that it becomes a complement within another syntactic unit (Hamawand 2002: 1). Complement clauses are embedded clauses in the complement position of another clause or phrase, and are “regularly introduced by a complementizer, a morpheme or a paired morpheme, whose function is to identify the structure as a complement” (Hamawand 2002: 1). Hamawand (2002: 87; 2003: 66) claims that “complementisers iconically encode different degrees of distance ←9 | 10→between the referents of the main and complement clauses”, and analyses them as complex categories comprising “a central prototypical meaning from which other meanings are derived” (2003: 70). As Matthews (2007: 42) explains, the term ‘complementiser’ dates from the mid-1960s, and was known in the beginning as ‘marker’ specific to a ‘predicate complement’. Under this definition, Matthews (2007: 42) includes that, as in example (4), morphemic markers such as -ing (5), ‘POSS’ (6) and to (7):

As pointed out in Hopper and Traugott (2003: 175), the “development of markers of clause linkage” and complementisers, such as those in (4)–(7), has been attributed to the total or partial grammaticalisation of lexical items that led to the integration of initially separate clauses within larger main units,2 where the embedded clause fulfils a syntactic function (Hamawand 2002: 1).

Hamawand (2002) analyses clausal complementation as a type of subordination: the process of linking individual clauses so that they adopt a different, dependent syntactic status. He uses the term ‘superordinate’ clause to refer to the main clause and ‘subordinate’ for a complement clause. Subordinate clauses are characterised by syntactic incompleteness because interpretation of a subordinate clause often relies on inference from the main clause or from a broader context ←10 | 11→(Huddleston and Pullum et al. 2002: 89). Matthews (1981: 174), for example, explains that in (8a) the incomplete clause to get him to the airport, which lacks a subject, can be transformationally related to the complete non-finite clause (for) it to get him to the airport, in (9a), or the finite clause in (9b). Clauses such as that found in (8a) are classed as ‘reduced units’: they are “incomplete (…) in that at least one potential element of a main clause, or at least one of the potential categories of its predicator, is excluded” (Matthews 1981: 174).

The matrix verb of the main clause functions as the ‘licenser’ of different types of complementation (Matthews 2007: 77). Terms such as ‘valency’ or ‘valence’ (Tesnière 1969: 238), ‘complementation’ (Quirk et al. 1985: 65–67) and ‘argument structure’ (Fromkin 2000: 685) are used variously to refer to verbs that require complements (e.g. love), verbs that may take complements but do not have to (e.g. sing), verbs that require a complement and may take an additional locative (e.g. throw), and verbs that only require a locative (e.g. put). In other words, verbs may be intransitive or transitive with one or two complements. Among transitive verbs, some permit or require clausal constructions comprising a subordinate finite or non-finite verb (Matthews 2007: 85).3 ←11 | 12→Matthews (2007: 92) explains that complements are defined relationally as parts of a construction that are either required or ‘licensed’ by specific lexical units, in line with Huddleston and Pullum et al. (2002: 219), who state that “the most important property of complements in clause structure is that they require the presence of an appropriate verb that licenses them”. According to Matthews (1981: 148), that the complement clause and the main clause enter into an exocentric construction relationship, where none of the constituents can substitute for a whole, or into a ‘directive’ relationship, in Hockett’s (1958: 191) terminology, in which the verb is a ‘director’ and the complement is the ‘axis’. Modifiers, by contrast, are dependent but not specifically ‘licensed’. Matthews (1981: 147) adapts Bloomfield’s (1933: 194) notion of ‘endocentricity’ in his definition of modification, applied to constructions where at least one of the elements may be substituted for the whole.

Hamawand (2002) identifies three word classes that license subordinate or complement clauses: nouns, as shown in (11) below, adjectives, as in (12), and verbs, as in (13).4 Huddleston and Pullum et al. (2002: 1176) add prepositions to the list, as illustrated in (14) – I will not consider examples such as (14) as complement clauses but as an adjunct, because the main verb is intransitive and does not license a complement clause.

(11)At the meeting there was a desire to change section 3 of the agreement. (Rudanko 2000: 7)

(12)He was determined to commit himself. (Rudanko 2000: 90)

(13)John forced me to perform menial tasks. (Rudanko 2000: 3)

My analysis does not adopt this catenative interpretation and treats to hear from them soon in (18) as a complement clause functioning as the object of the main clause. For the purposes of comparison, however, ←13 | 14→in what follows I will summarise the relevant literature on catenative constructions and comment on any overlaps with my own research. In Huddleston and Pullum et al. (2002: 1177), ‘catenative’ refers to constructions that take non-finite internal complements where the main and subordinate verbs are chained or concatenated, as in (19). In (19), the notional subjects of the first three catenative verbs (intend, try and persuade) co-refer with she, whereas him is the subject of the last catenative verb help. Two types of catenative constructions are present, simple and complex, depending on the absence or presence of a Noun Phrase (NP) intervening between the main clause and the subordinate clause that functions as the complement of one of the clauses.