This volume revolves around semantic and, above all, syntactic analyses of complement constructions – many of them based on a broad set of data. The book is organized into three parts, the second of which accounts for nearly half of the book: (part I) ‘Semantic description of constructions’, (part II) ‘Variation and change in complementation patterns’ and (part III) ‘The emergence of new patterns’. Except for Höglund and Rohdenburg, who (also) look at complements of an adjective and a noun, respectively, all other contributors are concerned with verb complements (some including object control): marked infinitives (Duffley, Egan, Höglund, Saarimäki, Rohdenburg, Slomanson and Hietaranta), bare infinitives (Egan and Rohdenburg), gerunds (Saarimäki), prepositional gerunds (Duffley, Rickman & Kaunisto, Höglund and Rohdenburg), finite clauses (Saarimäki and Rohdenburg) and nominal complements (Rostila, Rohdenburg, Callies and Hoffmann).
In the introduction, Mark Kaunisto takes the opportunity to foreshadow the debates and issues addressed by the authors. A good half of them refer to the following tendencies (see p. 3): (i) the Great Complement Shift, a more than 500-year-old drift in English, which ‘includes the spread in the use of gerundial non-finite complement clauses at the expense of infinitival complements’ (see also Vosberg Reference Vosberg2006); (ii) Rohdenburg's Complexity Principle according to which ‘features contributing to the complexity of the syntactic structure of the sentence … are likely to result in the use of grammatically more explicit complement patterns’; and (iii) Rudanko's Choice Principle which involves ‘the role of agentivity in connection with the selection of complements’.
In part I, from a merely synchronic perspective, Patrick J. Duffley discusses the question of why some causative verbs take a to-infinitival object complement in Present-day English (PdE) as in ‘She convinced me to stay’ while semantically comparable constructions have a gerundial one introduced by the preposition into (‘She talked me into staying’). In a quantitative analysis of the Corpus of Contemporary American English (COCA), he is able to show that the verbs get, force, convince and persuade are practically confined to the marked infinitive. The verb coerce is rare, though twice as common with the prepositional construction as with the infinitive. Talk and trick can only take the prepositional gerund with talk being more frequent than trick in such a construction. In the (mainly British) Bank of English corpus, Duffley also finds that both convince and persuade are very rare with the prepositional construction while persuade is four times as common with it as convince. In line with Rudanko (Reference Rudanko2006), he argues that it is the literal meaning of the preposition into that emphasises the process of gradually moving forward and leading to a containment while the infinitive marker – derived from the preposition to – is clearly result-oriented. Therefore, the verbs talk and trick and others like ease, which all ‘denote a gradual … transition into a new state’ (p. 21), can hardly be found with the to-infinitival pattern. Finally, the question is raised as to whether or to what extent the Constructional Grammar approach can account for the findings. What might have been worth looking at as well is the following construction involving the opposite meaning of the preposition into: ‘She talked me out of staying.’
Taking both a synchronic and diachronic approach, Thomas Egan's empirical and multifactorial analysis of the British National Corpus (BNC) reveals that the lexical rivalry between the two permissive verbs allow and let is mainly influenced by two syntactic and one semantic factor: the passive voice in the superordinate expression almost exclusively triggers the verb allow as in ‘I was (not) allowed to pass’ rather than let, which is accounted for by the prototypical meaning of the verb let. In the active, however, there is a tendency for negated matrix verbs (thus denoting prohibition) to choose the verb let exemplified in ‘He didn't let me pass’. In the (more frequent) cases of positive polarity (denoting permission) in the active voice, the distribution of the two verbs in question depends on the semantic type: barrier-removal of the type in ‘The border guard allowed me to pass (after having raised the barrier for me)’ shows a strong affinity for allow while let is clearly favoured in the sense of non-imposition of a barrier as represented by ‘The border guard let me pass (with the barrier already being raised)’. In the rare cases of passive let, the following observations in the COCA and the Corpus of Historical American English (COHA) are made: first, contrary to causative make, the marked infinitival complement (as in ‘I was let to pass’) is about to disappear. Second, however, the to-infinitive tends to be protected (against the rising predominance of the bare infinitive) in fictional genres (this result is statistically not significant), cases of negative polarity and the case of positive polarity, by non-imposition of a barrier rather than barrier-removal. This result is explained in terms of the meaning of to as a target-oriented (infinitive) marker.
It should be noted at this point that cognitive factors such as the Complexity Principle can account for the choice of the more explicit complement variant (here the marked infinitive rather than the unmarked one) in cases of increased cognitive complexity (here passivation or negation) and are often found to be much stronger than any semantic tendencies.
Jouni Rostila's theoretical discussion delves into the question of whether verbs like nibble/chew, feed or live and rely or depend complemented by an oblique object involving the preposition on can be assumed to form one single type of argument construction in PdE. In contrast to Goldberg's (Reference Goldberg2014) prototypical ‘rely on’ construction featuring verbs of (concrete or abstract) ingestion with the function of gaining sustenance, he argues against overgeneralization and suggests splitting both examples up into two different constructions: while ‘She nibbled on a roll’ focuses on incremental progress, ‘She relied on him’ emphasises prospectivity. Rostila supports his idea by the fact that equivalent German constructions require different prepositions here like the ones in ‘Sie knabberte an einem Brötchen’ or ‘Sie verließ sich auf ihn’. He argues that Goldberg's hypothesis is too general because her construction would not be able to generate examples of the type ‘They decided on their course of action’. For cases like these, he therefore introduces a third type of construction which also expresses prospectivity, but without the notion of (gaining) sustenance.
In part II, like Duffley, Paul Rickman & Mark Kaunisto look into the so-called into -ing complements as in ‘He tricked her into paying for dinner’, a very productive syntactic pattern in PdE. What might have fostered the popularity of the construction is the influence of the Great Complement Shift. Comparing two periods (1995–8 and 2010–12), the authors now analyse the construction in the newspaper language of New Zealand English (NZE). (It is well known that ‘newspaper prose is still first-rate material for linguists interested in ongoing change’, Mair & Hundt Reference Hundt and Mair1999: 236.) In line with previous findings for American English (AmE), the empirical study shows that the overall frequency of occurrence has been slightly decreasing between the two periods. Moreover, the number of different matrix verb types has been decreasing significantly. Additionally, the semantic characteristics of the matrix verbs found in the corpus involve different forms of pressure: intimidation/surprise, physical force, (verbal) persuasion, deception and guidance/inspiration. Finally, the analysis has revealed 19 innovative uses of matrix verbs previously undocumented: half of them show morphological complexity, and two of them are attributed to local NZE lexical inventiveness. However, the authors did not find any influence of te reo Mãori. What might have been of interest is a comparison with British English (BrE), which is suggested by the data and methods section but absent from the results.
Another gerundial complement option governed by a preposition is analysed by Mikko Höglund, who traces the evolution of different structural variants after the adjective ashamed in AmE between 1810 and 2009. In particular, he aims to look at the question of how and under which circumstances gerundial complements introduced by of have evolved (compare the example in ‘He was ashamed of answering the question’ with ‘… ashamed to answer …’). Unfortunately, the inclusion of noun phrases (NPs), finite and zero complements distorts the statistics of the development of gerundial complement options in relation to infinitival ones. Although absolute numbers are missing, what can indirectly be inferred from the two diagrams charting the frequencies of both non-finite alternatives is that, with a few exceptions, the share of gerundial variants in both types of non-finite complements rises steadily – albeit not very dramatically – from about 7 per cent in the 1810s to about 20 per cent in the 2000s. Moreover, it is found that, in present-day AmE, complement clauses involving a copula clearly attract the of -ing pattern while extractions seem to favour the infinitive. The former of these two results is attributed to the Choice Principle while the latter, though not statistically significant, supports the Extraction Principle. Additionally, the data reveal that verbs such as say or admit (followed by a that-clause) after ashamed are quite frequent with the infinitive. This suggests that, like regret, the adjective ashamed is here complemented by a speech act verb in performative function, which generally tends to be expressed by the well-known infinitive (see Vosberg Reference Vosberg2006: 187–9) in such a formulaic (and fossilised?) expression.
While Höglund's data imply some evidence of the Great Complement Shift, Veera Saarimäki is concerned with a counterexample of matrix expressions subject to this evolutionary tendency. In her empirical and diachronic study, she investigates the following types of sentential complements after the verb propose in Late Modern and present-day BrE, largely including fictional texts from the years between 1780 and the 1990s: ‘She proposed ((to) him) to go there’, ‘She proposed (his) going there’ and ‘She proposed (to him) that he should go there’. Again, it is shown that Late Modern English was a period of significant change and distinct variation and that, basically, finite complements have long tended to be replaced by non-finite ones (see Vosberg Reference Vosberg2006: 273). In particular, while the verb as a whole has become relatively infrequent in the course of the twentieth century, the infinitival complement of propose increased at the expense of both the gerund and the finite clause, then peaked at the turn of the twentieth century and has stabilised since then. With the meanings ‘intend/purpose’ and ‘recommend/suggest’, propose is not only semantically complex but also involves an intricate control situation involving subject, object and unspecified control in non-finite complements. Still, no significant correlation between the two factors is found in the data. It would be interesting, however, to analyse to what extent horror aequi (as in ‘He was proposing to go / going there’ or in ‘He was about to propose to go / going there’) plays a role in the distribution of the two non-finite complement forms.
Günter Rohdenburg reports on quite a number of omissible clause markers in English verb and noun complements: ± that in finite complements (see the examples (1) and (2) below), ± to in infinitival complements (3), ± should in mandative finite complements (4), ± from complemented by gerunds (5), (6) and ± of/about followed by wh-clauses (7). In large-scale corpus analyses of an enormous and mainly present-day British (and American) newspaper database, the Complexity Principle is comprehensively corroborated in the case of the following complexity factors: passivisation (1b), (3c), (5c), (7b), transitivity (6), negation (4), NP complexity (1), clauses other than (easy-to-process) there-existentials (2b) and discontinuous structures involving intervening elements (5a). The reported findings go far beyond those presented in the tables. So, incidentally, the following other factors are examined: horror aequi (3a,b) and national variety (2a), (4,) (5), (6). Here, the horror aequi Principle is clearly confirmed. Moreover, AmE is found to be more explicit in the structure represented by (5) and less explicit in (2), (4) and (6) than BrE so that ‘AmE appears to be more sensitive to a complexity contrast than BrE’ (p. 143). In (4), the subjunctive is obligatory in AmE. In a similar vein, the marked infinitive is mandatory in the passive of assist (3c) nowadays. In (7), what is further advanced in the establishment of the prepositions than how. Interestingly enough, in (6), we are dealing with one of the rare cases of a cataphoric complexity factor.
(1) (a) She told him / the man (that) they / the pictures were nice.
(b) He was told (that) they / the pictures ….
(2) (a) This increases the risk (that) the UK will break apart.
(b) There is the risk (that) the UK ….
(3) (a) She was assisting him (to) do his homework
(b) She tried to assist him (to) do ….
(c) He was assisted to do ….
(4) She asked that the letter (should) (not) be forwarded.
(5) (a) These fears might put off many investors from adopting such schemes.
(b) These fears might put many investors off (from) adopting ….
(c) Many investors are put off (from) adopting …
(6) People will hold off (from) buying (new cars).
(7) (a) This reminded her (of/about) what she did. / … how miserable she felt.
(b) She was reminded (of/about) what … / how ….
Even more fascinating than the conclusions drawn above seems to be the following question put forward by the author himself at the end of his contribution: where are the limits of the Complexity Principle, and to what extent is it compatible with Hawkins’ (Reference Hawkins1999) Domain Minimization Principle?
Focusing on the verbs graduate, impact and shop, Marcus Callies also explores the extent to which certain function words, prepositions in particular, are about to become redundant in present-day BrE and AmE. Such a development turns the intransitive verb use into a transitive one:
(8) (a) He graduated (last year) (from/at) Harvard University.
(b) He graduated (last year) (with/as) MA at Harvard.
(9) The crisis impacted (heavily) (on/upon) the economy.
(10) We were shopping (in/at) IKEA.
What Callies finds, basically, is that, in BrE, the direct object is now the preferred variant in constructions such as (9) (current web data) but still very infrequent in (8a) and (10). In AmE, the transitive use is also favoured in (9) and still very rare in (8b) and (10). Surprisingly, patterns like (8b) show 80 per cent for the transitive structure in the BNC (and only 7 per cent in COCA 1990–94). In all cases except (8b), however, AmE is spearheading the trend towards the loss of the prepositions. In accordance with the Complexity Principle, adverbial insertions between the verb and the prepositional complement invariably trigger the use of the preposition.
In part III, taking a diachronic perspective on present-day Singapore newspaper English since 1951, Sebastian Hoffmann observes the reverse trend towards adding a preposition after a typically transitive verb. Contrary to the structure given in ‘They were still awaiting for details’, which already peaked in 1981 and was most likely retained from earlier uses of the input variety BrE, the type in ‘We entered into an apartment’ (with a literal meaning of enter) is shown to be a very recent (2001) innovation. Instances such as in ‘The government should adapt and leverage on social media’ also emerged in 2001, reached their preliminary peak in 2011 and are a pan-ESL (English as a Second Language) phenomenon also frequent in other Southeast Asian as well as African countries. The construction in ‘He mentions about the fact that …’, however, is frequent in South or Southeast Asia but rare in Africa. The prepositional variant in ‘Customers requested for fresh produce’ peaked in 2001 and can doubtless be regarded as the manifestation of a nativisation process. The decline after 2001 might be explained by prescriptive internal style guides of newspapers. The fact that all these prepositional alternatives are practically unknown to the native varieties of English (e.g. BrE or AmE) but did emerge in many ESL varieties is possibly due to analogy (e.g. with the noun request + for) or the need for explicitness, clarity or transparency.
The contribution by Peter Slomanson also focuses on South Asia and, strangely enough, identifies parallel developments between the contact language Sri Lankan Malay (SLM) and late Old English (OE) in the establishment of to-infinitival complement clauses. It is argued that, similar to the OE inflected infinitive (e.g. to writenne), the SLM infinitival construction is based on a prepositional phrase (PP). Drawing on Los (Reference Los2005), who points out that the English inflected infinitive, though already purely verbal in OE, originated from a PP containing the preposition to with an irrealis mood and forward-looking meaning and a deverbal noun inflected for dative case in proto-Germanic, Slomanson now finds that the entirely verbal SLM marked infinitive involves a semantically irrealis infinitival prefix (mә-) and the infinitival suffix (nang), which is also used as a dative/allative clitic in NPs. Additionally, like OE to, SLM mә- cannot occur in the same clause as a modal. Due to contact with Tamil as spoken by Sri Lankan Muslims and contrary to what is generally assumed in language contact research, an isolating and analytic language such as Malay has turned into a synthetic one in a relatively short time. The following two conclusions can thus be drawn from these findings: first, languages can be substantially influenced in their grammatical system even by contact with typologically different languages. Second, well-attested languages (e.g. OE) can help to reconstruct the historical development of weakly attested languages (e.g. SLM) lacking diachronic corpora.
Like Slomanson, Pertti Hietaranta is concerned with the evolution of infinitival complementation in a non-Indo-European language: alongside the traditional and unmarked Finnish rakastan puhumista (‘I love talking’), the pattern rakastan puhua (‘I love to talk’) seems to have emerged quite recently. The author identifies essentially three interdependent factors influencing the development in this respect: (A) Internal: the infinitival construction has existed (as an already fully productive pattern) with certain matrix verbs (such as fear) for a long time and is now spreading to other verbs like love. (B) External: English is the dominant foreign language in educational and social or cultural life in Finland, and, as a lingua franca, it is becoming more and more important in the media (e.g. music and films) and modern communication (e.g. the internet). Additionally, Finnish translations of English books might want to highlight the foreign identity of the source text and thus imitate the English infinitival pattern. (C) Sociological/philosophical: the new and marked structure appears to be mainly used by younger speakers as an identifier in subculture communication. The question raised here refers to the influence of peer behaviour and language fashion or ideology: to what extent are speakers of one language willing to adopt features of another?
In conclusion, most of the contributions provide valuable theoretical insights into a variable field of grammar through which some of the most prominent (and in part probably universal) tendencies and principles that have been explored so far become visible. In particular, the Great Complement Shift is corroborated by Rickman & Kaunisto and Höglund, while a counterexample is found by Saarimäki. Furthermore, the Complexity Principle is substantiated across the board by Rohdenburg and Callies, and the Choice Principle is confirmed by Höglund. In addition to these three, the following two tendencies are addressed by Rohdenburg and Höglund, respectively: Rohdenburg's horror aequi Principle (‘the tendency to avoid the coincidence of two (near-)adjacent marked infinitives’, p. 137) and the Extraction Principle (the tendency to prefer the infinitive to the gerund if an element is fronted by extraction out of the complement clause; see p. 93 and Vosberg Reference Vosberg2006: 72).
Certainly, it would have been desirable to include even more factors of grammatical variation like the following: historical stages, especially (negative-implicative verbs in) Early Modern English (see Fanego Reference Fanego1996a: 57); style, especially the degree of formality (in addition to the careful suggestions in the results presented by Egan; see ‘more formal, more form’ in Rohdenburg Reference Rohdenburg1996: 152); micro-semantics, especially can/could + non-assertive contexts (see Fanego Reference Fanego1996b: 44–5, 58–9; Rudanko Reference Rudanko2000: 119ff.); phonology, especially prosody (see the Principle of Rhythmic Alternation in Schlüter Reference Schlüter2005: section 5.4); frequency and the degree of familiarity (see Krug Reference Krug1998; Rudanko Reference Rudanko2000: 149) and information structure, especially givenness in AcI constructions vs that-clauses (see Noël Reference Noël1997: 282; Reference Noël1998: 1061; Reference Noël, Rohdenburg and Mondorf2003: 366, 370).
