Chapter 1 - Introduction

Summary

This first chapter introduces the reader to the central research questions and hypotheses investigated in this book. In particular, it discusses the role that the morphologically and semantically related simple verbs (e.g. notice in the case of take notice of) may play for the semantic evolution of the CPs. It draws on previous literature, showing that not much has been done in the way of studying the role of the simple verb. Further, the chapter introduces other factors that may help predict or explain the evolution of the CPs. The chapter is rounded off by a summary of the book’s structure.

1.1 Setting the scene

English is well-known for its rich system of verbal constructions. Particular attention has been devoted to the emergence and spread of multi-word verbs, which consist of several parts (see e.g. Hawkins 2019; Berg 2014a; Claridge 2000). This book focuses on one type of multi-word verb, which is the composite predicate (CP). Two examples of such constructions, which consist of a semantically light verb like make or take followed by a nominal expression, are given in examples (1) and (2), which come from 19th-century British fiction.¹

(1)

She made mention of her early service in the family, and of her little pension.

(1859, NCF)

(2)

It was a regular road through the fields, and nothing very uncommon to see a young woman there, but I took notice of her because she looked white and scared.

(1859, NCF)

It is the aim of this book to study the semantic–syntactic evolution of such CPs in English.

Previous studies on CPs have mainly provided a synchronic snapshot either of the situation in present-day English (PDE) (see e.g. Mastrofini 2023; Mehl 2017; Ronan & Schneider 2015; Hoffmann et al. 2011) or of their distribution at one particular stage in the history of English, be it, for example, Old English (Akimoto & Brinton 1999), Middle English (Moralejo-Gárate 2003) or Early Modern English (Claridge 2000). By contrast, comprehensive overviews on the development of CPs over a time span of several centuries are much rarer (for an overview, see Traugott 1999; Brinton 2008). The present study aims to fill this gap: spanning British fiction from the 16th to the 20th century, it investigates the semantic–syntactic evolution of 24 different CPs, all of which have the composition illustrated in examples (1) and (2).²

The present study starts from the observation that each of the 24 CPs investigated has a morphologically related simpler verb in the paradigm that, in most cases, semantically overlaps with the CP (compare e.g. make mention of – mention and take delight in – delight in). It is precisely the role of this simple verb³ that has been grossly overlooked in previous studies (for two exceptions, see Kytö 1999; Berlage 2014) and that takes centre stage in this book.

Before I engage in introducing the three main research questions of this study, followed by an outline of the corresponding hypotheses, a few words on the genesis of this book are in order: this book was initially conceptualized as part of a variationist framework, but it will take the reader only a few pages to realize that it is no longer so. In fact, variation proper, in the sense of semantic interchangeability between the CP and the simple verb, is marginal at best (for more information, see Sections 1.2, 4.1, Chapters 6 and 7). It may possibly apply to such cases as make use of and use (for a fuller discussion of make use of, see Section 6.2), but a close look at hundreds of corpus examples reveals that the semantic scope of the simple verb in this, as in most other cases, is much broader than the semantic scope of the CP (see the references mentioned previously). This book therefore rests on a different assumption: the semantic scope and evolution of the simple verb may affect the semantic scope and evolution of the CP despite the fact that they are not semantically interchangeable in all the contexts in which they occur (in fact, they may not be interchangeable in many of them). If we envisage such two constructions as take leave of – leave or make impression on – impress, there will be some kind of competition between them (no matter how small the semantic overlap between them is, provided it is not zero). The main idea behind this book, therefore, is that the development of the CP is not entirely independent from that of the simple verb.

Although there is in reality not just one morphologically and/or semantically related simple verb for each CP, the present study deals with competition between a CP and exactly one morphologically related and, in many (but not all) cases, semantically related simple verb (for a closer look at the CPs included, see Section 4.1).⁴ The first and main research question to be addressed in this book thus reads:

1. 1) Can the semantic–syntactic evolution of the CPs (at least in part) be predicted on the basis of the semantic scope and evolution of morphologically corresponding simple verbs, provided that there is at least a minimal semantic overlap between them (i.e. mention in the case of make mention of and notice in the case of take notice of)?

I assume that the answer to this question is in the positive in the sense that the semantic scope and evolution of the simple verb correlates with the semantic scope and evolution of the CPs, provided there is a minimal semantic overlap between them (Hypothesis 1). This hypothesis aligns the present study to those diachronic studies which sympathize with the idea that competition between semantically related items can (at least in part) account for language change. Prominent examples in the history of English include such classic cases of variation as the dative alternation in give the boy the book vs give the book to the boy (see Zehentner 2019; 2022) and the genitive variation as in the girl’s book vs the book of the girl (see e.g. Wolk et al. 2013; Rosenbach 2002; In press). Although the present study is no such study on variation, it aims to show that the concept of competition, originally a psycholinguistic effect (see Berg 2014b; Dell & O’Seaghdha 1994), can be used to successfully explain which types of language change different CPs go through (see Section 9.1).

Further, this study is in line with the iconic assumption that one form (e.g. take notice of) is (in principle) reserved for one meaning, with an expression that is different in form (e.g. notice) also expressing a difference in meaning. More specifically, my study stands in the tradition of paradigmatic isomorphism – that is, the idea of a 1:1 match between form and meaning in the inventory of words (see e.g. Bolinger 1977; Haiman 1980; Wierzbicka 1988; Croft 2003: 105–10). This correlation between form and meaning motivates the following prediction: if the meanings of the simple verb are extended, the meanings of the CP should be reduced in order to avoid cases of synonymy.⁵ While it is perfectly clear that language change tolerates cases in which one form has two unrelated meanings (homonymy) or one meaning is covered by two forms (synonymy),⁶ we will here investigate if and to what extent the specialization of the CPs (those cases where they narrow down their meanings) can be accounted for by the principle of isomorphism.

In this study, we are not only concerned with the extent to which some CPs specialize and others do not, but we also want to explore how specialization shows (i.e. if there are different types of semantic specialization and what they are). For this purpose, we will look at changes in the modifier slot (e.g. restrictions to specific modifiers such as good in make good use of), changes in the determiner slot (e.g. restrictions to specific determiners such as no in make no mention of) and changes in the wider assertive and non-assertive context that these CPs occur in. On a theoretical level, it will be particularly interesting to see whether some CPs become polarity-sensitive as to the environment they occur in (i.e. become restricted e.g. to negative or other non-assertive contexts and therefore gradually turn into what we call negatively oriented polarity-sensitive items (NPIs)).⁷

Investigations into how different CPs specialize are linked to the question of whether semantic changes are accompanied by syntactic ones and, if they are, how these two types of changes compare to one another. Two examples in which the CPs make mention of and take notice of are both semantically restricted to negative contexts and syntactically restricted to specific types of determiners preceding the head nouns are given in (3) and (4). While public mention in make mention of is preceded by the determiner no, notice in take notice of co-occurs with the indefinite determiner any.

(3)

Tell me about the cartridges, I went on. I think you were wise to make no public mention of them.

(BNC)

(4)

Beales had not taken any notice of other customers until a man he recognized as MacQuillan came to the next table.

(BNC)

Although formal and semantic changes often go hand in hand (see the argumentation provided previously), we know at least since the extensive discussions held in the grammaticalization literature (see e.g. Heine & Kuteva 2002; Hopper & Traugott 2003: 33; Francis & Yuasa 2008) that they do not have to coincide. Rather, we may envisage three different scenarios.

• Scenario 1: Semantic and syntactic changes affecting a construction run in parallel. In this case, there is no need to disentangle the two processes because a high degree of syntactic fixation will be indicative of a high degree of semantic restrictedness and vice versa. In the literature, this view has become known as the ‘parallel reduction hypothesis’, arguing in favour of the ‘dynamic coevolution of meaning and form’ (Bybee et al. 1994: 20).

• Scenario 2: Semantic changes precede syntactic changes. In this case, a construction may have become semantically specialized, but this is not yet reflected in its syntactic structure. The idea behind this approach is that ‘form follows function’; that is, functional (semantic) changes precede formal changes. The most prominent proponents of this view are Givón (1991: 123), Haspelmath (1999: 1062) and Francis and Yuasa (2008). For a more detailed overview, the reader is referred to Dehé and Stathi (2016: 912). While this process has so far been attested for processes of grammaticalization (see, again, Dehé & Stathi 2016; Francis & Yuasa 2008; Heine & Kuteva 2002: 3), Berlage (2010) has shown that these processes are not genuine to cases of grammaticalization but may also be found in (selected cases of) lexicalization. The examples she cites (take prisoner, fall victim, play truant) can all be placed on a cline of lexicalization (Berlage 2010: 54–5).

• Scenario 3: Syntactic changes precede semantic ones. As far as I am aware of, this type of change has not been attested to in the literature so far. The idea is that a construction may become syntactically frozen while it is still semantically transparent. As in Scenario 2, the semantic and syntactic status of the construction must not be confused because there is no direct correlation.

In this monograph, I want to figure out which of the three scenarios holds for the CPs under investigation. By analysing those CPs which undergo semantic specialization/restrictedness, we will see whether they all pattern alike or whether there are differences between them in the sense that different CPs can be assigned to different scenarios (see Section 8.2). Our second research question thus is:

1. 2) Do semantic and syntactic changes (affecting the constructions) run in parallel, or is one type of change faster than the other?

My hypothesis is thereby in line with Francis and Yuasa (2008) and Berlage (2010): mismatches between form and meaning are not genuine to cases of grammaticalization but also apply to the history of CPs. In line with the sparse evidence provided for cases of lexicalization, I expect the CPs to be affected by semantic changes earlier than by syntactic ones.

If we want to classify the CPs and their morphologically related simple verbs as being either analytic or non-analytic (relative to each other), the CP should represent the analytic end of the scale. This decision is based on the assumption that only the analytic construction reserves one form for one meaning: the light verb has the function of aspectual marking (see e.g. Renský 1964: 295; Prince 1972; Brinton 1996: 198–200; Vogt 2002; Trousdale 2008) and the nominal expression carries the main semantic content of the predication (see e.g. Huddleston et al. 2002: 290). In my terminology, I thus follow Traugott (1999: 247), who also refers to the CPs as ‘analytic’ constructions.

On the basis of this typological distinction, it will now be interesting to see how the two forms develop in language history. It will be one of the aims of this study to closely examine if the CPs investigated really become more frequent during the Late Modern English period and all the way up to contemporary English or whether the trend towards more analyticity that started in the Middle English period (for the expansion of CPs, see Kytö 1999; Claridge 2000: 94) is not supported by the CPs investigated in this study. If, in the present study, the frequencies of the simple verbs can be shown to be higher than those of the CPs in 20th-century English, such a finding runs counter to the fact that PDE is replete with light verb constructions (see e.g. Hawkins 2019; Berg 2014a). If, diachronically, the frequencies of the CPs are either decreasing as we move from the 19th to the 20th century or remain at a consistently low level, such a development questions the widely stated belief that English has consistently increased its degree of analyticity (see e.g. von Schlegel 1846; Sapir 1921; see also Hawkins 1986). Our third research question thus reads:

1. 3) Is the evolution of the CPs and their corresponding simple verbs in line with the general trend of the English language to become ever more analytic?

In line with what Szmrecsanyi (2012) and Neubauer (2024) have found out for cases of inflectional and derivational morphology, respectively, I assume that the development of the CPs defies the trend of the English language to continuously increase in its degree of analyticity.

In the remainder of this chapter, I will shed light on those factors that may influence the evolution of the CPs. The focus will first and foremost be on the simple verb and the question of how the semantics of the simple verb compare to those of the CPs. This will be illustrated with the help of four different pairs of CPs and simple verbs (see Section 1.2). Next, we will look into potential other factors that may impinge on the evolution of the CPs. Although they will not be discussed in as much detail as the role of the simple verb, they become relevant when it comes to the question of why the development of some CPs can be better predicted than that of others (see Section 1.3). The chapter is concluded by an outline of how this book is structured (see Section 1.4).

1.2 The role of the simple verb

In the literature, the CP and the simple verb have often been claimed to differ only slightly in terms of their meanings (see e.g. the overview in Brinton & Akimoto 1999: 4). With respect to CPs involving make, Brinton and Akimoto (1999: 4) equate the examples make a claim – claim and make a choice of – choose. The Oxford English Dictionary (OED online) supports this view by stating for the light verb take, ‘Often take forms with the object a phrase which is a periphrastic equivalent of the cognate verb: e.g. to take a leap is equivalent to to leap, to take a look to to look, to take one’s departure to to depart.’ Moralejo-Gárate (2003: 150) talks about CPs and their ‘semantically equivalent simple verb’ but hastens to add that CPs and simple verbs have ‘different structural, semantic and textual possibilities’, thereby relativizing the notion of semantic equivalence.

Much more suspicious of an assumed semantic identity between the simple verb and the periphrastic construction is Kytö (1999: 168), who states that it ‘is not always the case in PDE that two alternatives – the simple verb (e.g. to differ) and the corresponding verb + deverbal noun construction (make a difference) – amount to one and the same (or nearly the same) meaning’. The same point is raised in Quirk et al. (1985: 751), who claim that there are cases in which the CP ‘clearly does not have the same meaning as the verb alone’. The examples cited are, for example, make love (to) and take trouble (over). At least equally critical of an assumed identity in meaning is Claridge (2000: 78), who states on the basis of her Early Modern English data that ‘the simplex and the combination are by no means identical’.

I will here critically assess the idea of an assumed semantic equivalence between the CP and a morphologically related simple verb, showing that some constructions differ only minimally in meaning while others have hardly any semantic overlap. In order to illustrate the semantic spectrum of similarity and dissimilarity between the CP and the simple verb, the introduction offers four pairs of constructions. Only in those cases where there is at least some degree of semantic overlap can we expect to find competition between the simple verb and the CP (for the concept of competition, see Section 1.1).⁸

We will start with the two constructions take possession of and possess. While the simple verb possess is stative, take possession of is inchoative; that is, it describes the starting point of an event (see Hiltunen 1999: 137; for the inchoative function of take a V constructions more generally, see Shahrokny-Prehn & Höche 2011: 242).⁹ Transferring Wierzbicka’s discussion of the more restricted take a V construction (1982: 795) to take possession of, we can argue that the CP is likely to involve ‘an initial momentary movement’, while possess does not. Dixon (2005: 474) adds that the take a V construction implies physical effort on the part of the subject, which is precisely what we see when someone grabs something in order to get it under control. Given these semantic differences, it is not surprising that the two constructions do not alternate. Example (5) illustrates this very clearly: taking possession of something is not the same as possessing it; rather, the state of possession is a result of having taken possession of something. It is for this reason that the simple verb possess cannot take the place of take possession of.

(5)

She could not think of it as her house any longer, it belonged to Nina and Joe, they had taken possession of it and suddenly, Eline realized that it didn’t matter to her at all.

(BNC wrid.1)

Two other pairs of constructions that are clearly semantically different are take offence at – offend and make love to – love. Starting with the former, we see that the periphrastic construction take offence at is passive in meaning – that is, it corresponds to the passive construction to be offended by rather than to the active verb to offend (see Algeo 1995: 206; Matsumoto 1999: 71). For make love to as in (6), it is clear that the periphrastic construction refers to the act of sexual intercourse in PDE, exclusively. This means that the meanings of make love to and love are clearly distinguished today (see also Quirk et al. 1985: 751). In earlier examples, however, the meaning of the CP can best be paraphrased as ‘to court’ or ‘to pay amorous attention to’ (see OED online), as can be seen in example (7).

(6)

Edward made love to her that night…

(BNC wrid.1)

(7)

It is an old saying, ‘Praise the child, and you make love to the mother’.

(1829, quoted from OED)

We should notice that the ‘court’ sense, which seems somewhat archaic to us today, is still found in texts up and until the 20th century (see OED online). This means that, for several centuries, we have a situation in which both readings are, in principle, possible, depending on the context. What is crucial for our comparison of the simple verb and the CP is the fact that the simple verb to love cannot be reduced to the physical sense covered by make love to (even the ‘court’ sense implies a physical expression of one’s love) but means, much more generally, ‘to have or feel love towards/to entertain a great affection/fondness for’ (OED online). It follows that make love to and love cannot be considered semantic competitors of each other.

While the previous sets of constructions are clearly distinguished in terms of their meanings, the meaning difference seems to be minimal in the last set mentioned here, which is make provision for and provide for. The close semantic similarity between make provision for and provide for is obvious from the definition of provide for given in the OED online. Here, the CP is paraphrased in terms of the prepositional verb provide for. This suggests a substantial semantic overlap. Despite this close similarity, I want to argue that make provision for and provide for should not be considered synonyms of one another. For one, we know that languages generally avoid cases of true synonymy (see e.g. Croft 2003: 105; Johnson 1755: v; the discussion in Miller & Charles 1991: 2; Clark 1987: 4; see also Bergs 2005: 214); for another, periphrastic constructions involving light verbs like make and take have often been claimed to show aspectual differences when compared to the simple verb (cf. e.g. Renský 1964: 295; Prince 1972; Brinton 1996: 198–200; Vogt 2002; Huddleston et al. 2002; Trousdale 2008). Accordingly, the interpretation of the CP make provision for should be along the following lines: the activity of providing for something is turned into a (quasi-)accomplishment and thereby receives a temporal interpretation (cf. e.g. Brinton 1996: 200; see also Berlage 2014: 221) – while make provision for consequently denotes a temporal event, provide for relates to an activity that is unbounded.

In Berlage (2014: 221), I have argued that subtle semantic differences (as they exist, for example, in the cases of make provision for – provide for) may be overridden by discourse-functional constraints. It has thus been shown that CPs are more prone to modification patterns than the corresponding simple verbs are (see e.g. Poutsma 1926: 395–6; Jespersen 1942: 117; Live 1973: 34; Nickel 1968: 15–6; Brinton 1996: 194; Kytö 1999: 179; Bergs 2005: 214). Nickel (1968: 15) explains this in the sense that noun modification is ‘easier to manipulate and more versatile’ than verb modification. If the noun inside the periphrastic construction is modified, however, we may arrive at a situation where the context rules out variation between the CP and the simple verb. This is illustrated in examples (8) and (9). While example (8) is perfectly acceptable, example (9) is ungrammatical.

(8)

Have a good laugh at them, Jones.

(Greene; quoted from Stein 1991: 20)

(9)

*Laugh well at them, Jones.

(Stein 1991: 20; see also Berlage 2014: 221)

Diachronically, such a preference of the periphrastic construction for modification patterns may (gradually) lead to clearer semantic distinctions between the CP and the simple verb in the sense that only the CP occurs in combination with (specific) adjectival modifiers (see e.g. Brinton & Akimoto 1999: 17; Matsumoto 1999: 78; Tanabe 1999: 113–7; Huddleston et al. 2002: 291). This is illustrated for the case of make use of – use in examples (10) and (11), the latter of which should be strongly dispreferred because it sounds unidiomatic.

(10)

God grant me another chance, and I’ll make better use of it!

(BNC wrid.1)

(11)

^?God grant me another chance, and I’ll use it in a better way!

(example (10) rephrased)

Examples (10) and (11) suggest that even in those cases where variation is possible in principle (because there is a sufficient semantic overlap between the CP and the simple verb), there will be many contexts that do not allow for it. A more detailed list of such contexts is provided in Hoffmann et al. (2011). To this list belong all those cases where there is no adjective–adverb correspondence (illustrated in examples (8) and (9)). In addition, the list contains nominal modifiers as in (12), which do not comply with the idea of isomorphism between the CP and the simple verb, and examples in which multiple premodifying adjectives as in (13) cannot be turned into a coordinated adverb phrase accompanying the simple verb.

(12)

… Intach in the meanwhile proposed to Calcutta Tramways that a tram car be made available for tourists to take a heritage tram ride to view some of the city’s attractions.

(The Statesman, 15 February 2002; quoted from Hoffmann et al. 2011: 266)

(13)

Take a good hard look at the registered madrasas, which have been provided unlimited state funds without accountability.

(The Statesman, 10 February 2002; quoted from Hoffmann et al. 2011: 265)

Against the background presented, we may conclude that there is variation between the simple verb and the CP, but there are many syntactic and semantic contexts in which the CP and the simple verb do not alternate. I would therefore like to argue that the decision to treat the simple verb and the CP as fully semantically equivalent is only justified from a purely onomasiological perspective (see Mehl 2017) or a context-free definition of variation in syntax but seems oversimplified when discussed against a context-bound definition of ‘sameness’ (see Rosenbach 2002: 22–3; 2019). In the present study, I will assume that there is some semantic overlap between most sets of simple verb and CP (e.g. take notice of and notice) but that there are also substantial differences between them (with some sets not allowing for variation at all, as e.g. in the cases of take possession of – possess and take offence at – offend).

When we look at diachronic studies addressing the co-existence of the CP and the simple verb, we find only a few qualitative studies on related verb pairs (see e.g. the examples of have knowledge of – know, make/take/have a promise – promise and take a labor – labor in 15th-century letters by Tanabe 1999: 108–13). Even more sparse are quantitative studies on the topic. To my knowledge, the only quantitative study is provided by Kytö (1999: 198), who conducts two case studies on the Early Modern English section of the Helsinki Corpus. Here, the author compares a) the CPs make/have use to the simple verb use and b) the CPs take/have care to the simple verb care in terms of their relative frequencies of occurrence. She shows that the gradual decrease of the simple verb is paralleled by an increase in the use of the CP, suggesting, on a more general level, a connection between the frequencies of the simple verb and those of the periphrastic construction. In her conclusions, Kytö (1999: 200) points out that the CP gains ground ‘at the cost of simple verb uses’, suggesting that their frequencies inversely correlate: as the frequencies of the CP increase, those of the simple verb decrease.

I will here take the correlation between the evolution of the simple verb and the CP as the starting point for the investigations presented in the present study. Based on Kytö’s results, I will assume that the semantic scope and evolution of the simple verb can be used to predict the semantic scope and evolution of the CP.

1.3 Other factors

While this study mainly focuses on the effect that the semantic scope and evolution of the simple verbs has on the evolution of the CPs, it would be too simplistic to assume that the semantic evolution of the CPs is governed by one factor alone. It is the purpose of the present section to highlight those other factors which may potentially impinge on the evolution of the CPs. In the following, I am going to outline those five factors that play a role in the present study.

The first such factor is other verbal alternatives to the CP. Naturally, semantic competition between verbal expressions is not limited to morphologically related structures. Rather, a CP like make fun of will not only compete with the extremely rare verb to fun but also with a morphologically unrelated verb like to ridicule. Although the scope of this study does not allow us to take into consideration all semantically related verbal expressions when we discuss the evolution of the CPs, other verbal competitors will come into play where we see that the semantic evolution of the CPs cannot be successfully predicted on the basis of the semantic scope and evolution of the simple verb alone. At this point, we will take into consideration both simple verb competitors to the CP and other non-morphologically related periphrastic constructions (e.g. pay attention to in the case of take heed of). While this procedure does not provide us with a comprehensive picture of all the constructions that compete with the CPs in semantic space, it is perfectly clear that non-morphological semantic competitors of the CPs may influence their specialization processes. More specifically, I assume that the relative strength of other verbal alternatives depends on the semantic generality and evolution of the morphologically related simple verb in the sense that it differs for simple verbs that are or become relatively broad in terms of their semantic scope (e.g. notice) and those that are or become more narrow (e.g. apologise/apologize for). The role of other potential competitors will be elaborated on in Section 9.1.

Another factor related to the verb phrase is the frequencies of CPs that contain the light verb make as compared to those that contain the light verb take (see Section 7.2). This factor is distinct from competition between the CP and other verbs and raises the question of whether differences in the frequencies of the CPs involving make and take correlate with higher or lower specialization/restrictedness rates of the corresponding CPs. This question will be addressed in Section 7.2.

A non-verbal factor that may play a role in the evolution of the CPs is the frequencies of the head noun (which occurs inside the complement of the CP) outside the CPs (see Section 7.3). The idea behind such an investigation is that those head nouns which are very frequent elsewhere (e.g. the noun use) may lead to higher frequencies of the CP compared to head nouns that are used infrequently elsewhere (e.g. the noun heed). If this is the case, the frequencies of the head nouns in contexts outside the CPs will serve as a predictor of the frequencies of the CPs, which again indicate higher or lower degrees of semantic restrictedness and, possibly, semantic specialization (for the operationalization of semantic restrictedness and specialization, see Section 4.2). Section 7.3 tests this hypothesis against empirical data.

As concerns the head nouns of the CPs, the question emerges of whether their phonological weight (i.e. the number of syllables the nouns contain) may impinge on the frequencies with which the CPs occur in the sense that CPs with longer head nouns (as in e.g. make allowance for or make apology for) are particularly infrequent (see Zipf’s First Law; Zipf 1949). While the frequencies of the head nouns are interpreted as a predictor of the semantic status of the CPs, the phonological weight of the head nouns is taken as an explanation in those contexts where the semantic status and evolution of the CPs cannot be predicted on the basis of the semantic scope and evolution of the simple verbs (see Section 7.1.2).

Another factor explored in those cases where the predictions of ‘specialization’ or ‘no specialization’ are not borne out by the data is genre effects. Since long and phonologically bulky nouns in general tend to be more at home in more elaborate genres than in less elaborate genres (see e.g. Haiman 1983; Rohdenburg 2012: 139), I will investigate if CPs with long head nouns such as allowance or apology (in make allowance for and make apology for) differ as to their preferences for more or less formal text types. By comparing the occurrence of the same CPs in different genres, we can test whether increases or decreases in the CPs over time come down to genre effects or whether they represent genuine changes over time.

To summarize, the present study primarily concentrates on how the semantic scope and evolution of the morphologically related simple verbs affects the specialization of the CPs (see Hypothesis 1); in two smaller studies, I test whether the frequencies with which the light verbs make and take occur inside the CPs and the head nouns selected are correlated with the specialization/restrictedness of the CPs (see Sections 7.2 and 7.3). The role of non-morphologically but semantically related verbs, the phonological weight of the head nouns and genre effects are discussed as (post hoc) explanations where the evolution of the CPs runs counter to the hypothesis of a correlation between the semantic scope and evolution of the simple verbs and that of the CPs (see Chapter 6). The role that other verbal alternatives play when it comes to the evolution of the CPs is, however, taken up again in the model of competition unfolded in Section 9.1. While their influence is only suggested in post hoc explanations in Chapter 7, it becomes clear that the model can in principle predict the relative strength of different verbal alternatives to the CP.

1.4 Outline

My book is organized into ten chapters. All of the chapters presenting empirical results deal either with the semantic scope and development of the CPs/simple verbs (see Chapters 5–7) or their syntactic evolution (Section 8.1) or both (Section 8.2). Since this study primarily concentrates on processes of semantic specialization and, secondly, on the interaction of semantic and syntactic changes, semantic changes are generally portrayed before syntactic ones (compare the ordering of Chapters 6–8). This has consequences for the organization of Chapters 2 and 3 in the sense that previous research on semantic changes is depicted before research on syntactic changes and that the semantic composition of the CPs is reported on prior to their structural composition. The more detailed organization of the book is as follows:

The introduction is followed by a chapter on the evolution of composite predicates in the history of English. Here, I will summarize what previous research has to tell a) on the grammaticalization and lexicalization of CPs in English (Section 2.1), b) on semantic changes of CPs, involving both the concepts of idiomatization and semantic specialization (Section 2.2), and c) on syntactic changes in CPs, with a particular focus on the decategorialization of the noun phrases (NPs) inside the CPs (Section 2.3). While CPs have variously been discussed as constructions that undergo grammaticalization, lexicalization or both (see Section 2.1) and while their changing NP status has also been explored (Section 2.3), there is surprisingly little research on their semantic specialization. The present study seeks to fill this gap.

Chapter 3 describes the CPs under investigation in the present study, both in terms of their semantic and in terms of their structural composition (Sections 3.1 and 3.2). Further, it introduces the simple and prepositional verbs which are morphologically and, in many cases, semantically related to the CPs (Section 3.3). Finally, Section 3.4 unfolds the three central hypotheses under scrutiny (which have been briefly introduced in Section 1.1).

Chapter 4 is devoted to the methodology applied in this book. To start with, Section 4.1 presents the mechanisms that underlie the selection procedure of the 24 CPs analysed. Section 4.2 is then concerned with the operationalization of the variables in the hypotheses. Further, it explains how the different types of specialization can be read off from changes in the modifier slot (Section 4.3.1), the determiner slot (Section 4.3.2) and the wider assertive and non-assertive contexts that a construction occurs in (Section 4.3.3). In Section 4.4, I will finally outline which electronic corpora are used in this study, which time periods are investigated and which statistical tests are applied.

Chapter 5 depicts the evolution of the simple verbs – both in terms of their semantic coverage and in terms of their frequencies of occurrence. While Section 5.1 provides a synchronic snapshot of the simple verbs in PDE (relying on information provided by the OED online), Section 5.2 gives us the frequencies of the verbs in 19th- and 20th-century British fiction. In Section 5.3, I predict two different types of development for the CPs: depending on the semantic scope and evolution of the simple verbs, the CPs either undergo specialization or do not. These predictions are tested against empirical data in Chapters 6 and 7.

Chapter 6 deals with what I call ‘Type I-CPs’ – that is, those CPs whose simple verbs have an extensive semantic coverage or are extending their semantic scope. For these CPs, the prediction is that they undergo specialization or are semantically restricted throughout. In order to find out whether this prediction holds exclusively for CPs that have morphologically related simple verbs with a semantic overlap or can be extended to those CPs which do not have a semantic overlap, I will test this prediction against the relevant CPs in Section 6.1. On the basis of these findings, I will then proceed to a closer (and more specific) investigation of processes of semantic specialization that can be read off from changes in the modifier slot (Section 6.2), the determiner slot (Section 6.3) and the wider assertive and non-assertive contexts that the CPs occur in (Section 6.4). This analysis will reveal which CPs become polarity sensitive in the course of time (i.e. become restricted e.g. to negative or other non-assertive contexts). Section 6.5, eventually, shows which and how many CPs confirm Hypothesis 1 (see Sections 1.1 and 3.4). In addition, this section summarizes the different types of specialization that occur and asks whether, on the basis of these results, we can make any predictions as to which CP shows which type of specialization.

Chapter 7 is then devoted to what I call ‘Type II-CPs’ – that is, CPs whose simple verbs do not have an extensive semantic coverage or do not extend their semantic scope. The prediction is that these CPs do not undergo specialization. As in Section 6.1, a first analysis compares the effects of CPs which have simple verbs with a semantic overlap to those without. The more specific analysis of these CPs is then divided into two groups – depending on whether they contain the light verb take (Section 7.1.1) or make (Section 7.1.2). For those cases that do not fit in with the prediction, I will explore which potential other factors may be responsible for the development of the CPs. To these belong other verbal competitors in the paradigm, the phonological weight of the head nouns and genre effects (Sections 7.1.1 and 7.1.2). Still other factors that may impinge on the semantic status (and, possibly, the evolution) of the CPs are the frequencies with which CPs occur that involve the light verbs make as compared to those that involve the light verb take (Section 7.2) and the frequencies of the head nouns outside the CPs (Section 7.3). In Section 7.4, I conclude which and how many CPs confirm the hypothesis of ‘no specialization’ and highlight those factors that successfully predict this outcome.

While the empirical analyses provided in Chapters 6 and 7 have focussed on semantic changes in the CPs, Chapter 8 is concerned both with the syntactic fixation of the CPs and with a comparison of the semantic and syntactic changes that the CPs go through. In Section 8.1, I will place all CPs on a synchronic cline that indicates their degree of syntactic fixation, ranging from constructions that are (completely) syntactically fixed to those that are syntactically flexible. I will here explore the question of whether there is a correlation between the degree to which a construction has become semantically specialized/restricted and its degree of syntactic fixation. In Section 8.2, I will then provide empirical data that answer the question of whether semantic and syntactic changes run in parallel or whether they proceed at different rates in the sense that we get form–meaning mismatches as in cases of grammaticalization (see Francis & Yuasa 2008).

Chapter 9 finally provides a theoretical discussion based on the empirical analyses presented in this study. I will here deal with each of the three hypotheses (see Section 3.4) in turn: while Section 9.1 focuses on the concept of competition and the extent to which competition can account for language change in the history of the CPs (providing a model of semantic competition between the CP and the simple verb), Section 9.2 asks if the principle of iconicity can account for the fact that mismatches between formal and semantic changes are eventually resolved. Section 9.3, finally, deals with the question of how (potentially) low frequencies of the CPs compare to the prevalent tendency of the English language to become ever more analytic and which theoretical conclusions can be drawn from this.

Chapter 10 rounds the book off by summarizing the main findings of this study and pointing out avenues for future research. Specific attention will be devoted to the relevance of the functional principles of competition, iconicity and economy of expression for language change.