Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 17 October 2018

Construction Grammar: Introduction

Abstract and Keywords

This chapter discusses the history and principles of Construction Grammar. It explains that Construction Grammar has its roots in the Saussurean notion of the linguistic sign, and the chapter outlines some of the unifying principles of constructional approaches to the architecture of language. The chapter also provides a summary of the other contributions to the volume.

Keywords: Construction Grammar, linguistic sign

One of the central concepts of linguistics is the Saussurean notion of the linguistic sign as an arbitrary and conventional pairing of form (or sound pattern/signifiant) and meaning (or mental concept/signifé; cf., e.g., de Saussure [1916] 2006: 65–70). Under this view, the German sign Apfel and its Hungarian equivalent alma have the same underlying meaning ‘apple,’ but different associated conventional forms ([ˡɑpfl] and [ˡɑlmɑ]). Over seventy years after Saussure's death, several linguists then explicitly started to explore the idea that arbitrary form-meaning pairings might not only be a useful concept for describing words or morphemes but that perhaps all levels of grammatical description involve such conventionalized form-meaning pairings. This extended notion of the Saussurean sign has become known as a ‘construction’ (which includes morphemes, words, idioms, and abstract phrasal patterns) and the various linguistic approaches exploring this idea were labeled ‘Construction Grammar.’

Instead of assuming a clear-cut division of lexicon and syntax, Construction Grammarians thus consider all constructions to be part of a lexicon-syntax continuum (a ‘constructicon,’ Fillmore 1988; see also Jurafsky 1992 and Goldberg 2003: 223). Examples from this continuum are given in (1)–(4) (employing a fairly informal description of the form and meaning parts; (p. 2) for various different approaches to the representation of constructions, cf. the chapters in Part II):

  1. (1) word construction:

    apple [æpl]—‘apple’

  2. (2) idiom construction: e.g., X take Y for granted

    [X TAKE Y fә ɡɹɑːntιd]—‘X doesn’t value Y’

  3. (3) comparative construction: e.g., John is taller than you

    [X BE Adjcomparativeðәn Y]—‘X is more Adj than Y’

  4. (4) resultative construction: e.g., She rocks the baby to sleep1

    [X V Y Z]—‘X causes Y to become Z by V-ing’

The word in (1) is a classic pairing of form and meaning and thus qualifies as a construction. The meaning of the idiom in (2) is not completely compositional and must therefore be stored in a speaker's mental lexicon. Moreover, while (2) is an idiom that is partly substantive (i.e., whose phonological form is fixed in several places, i.e., [fә] and [ɡɹɑːntιd]), it also contains ‘slots’ in its subject and object position that can be filled by various elements (cf. Many people take their friends for granted. / She takes him for granted. / John and Jill take each other for granted. /…). Constructions that contain such slots are said to be schematic and the comparative construction in (3) can therefore be said to be more schematic than the idiom in (2), since the former only has one substantive element ([ðæn]) and several schematic slots (for the subject X, the form of BE, the comparative adjective and Y). Finally, the resultative construction in (4) is completely schematic, since it only contains slots for the cause X, the verb V, the affected complement Y, and the resulting state Z (and thus licenses such diverse structures as She kissed it better. / They wiped the table clean. / He colored his beard red. /…).

When tracing the history of the term ‘construction’ in linguistics, it is interesting to note that the analysis of syntactic structures as form-meaning pairings was commonplace in traditional grammars (and the use of constructions in this sense can be traced back at least as far as the twelfth-century Medieval Linguists; cf. Goldberg and Casenhiser 2006: 343). In fact, even the early Chomskyan transformational models (1957, 1965) employed the notion ‘constructions’ and included construction-specific rules. In his discussion of phenomena such as (2)–(4) in his seminal Aspects of the Theory of Syntax, Chomsky, for example, speaks of the ‘Verb-with-Complement’ (1965: 190) construction take for granted, the comparative constructions underlying John is more clever than Bill (1965: 178), or a causative construction akin to (4) (1965: 189). Yet, when the first major Construction Grammar publications appeared in the 1980s (in particular Fillmore 1985a, 1988; Fillmore, Kay, and O’Connor 1988; Lakoff 1987 or Wierzbicka 1988), these were all considered to be revolutionary and in stark opposition to the dominating syntactic theory at the time, the Chomskyan Principles-and-Parameters approaches (as exemplified by, e.g., Chomsky 1981).

The reason for this lies in the fact that the Principles-and-Parameters models, which superseded the earlier Aspects approach,2 were all characterized by the claim (p. 3) that constructions are only epiphenomena, that is, merely a collection of structures that are the result of the interaction of universal principles and parameter settings (cf. Chomsky 1995: 129). Assuming the existence of an innate language faculty (called Universal Grammar), the focus of these Chomskyan Mainstream Generative Grammar approaches shifted from language-specific computational models of native speaker competence to an investigation of the mental principles and parameters underlying all human languages. Yet, while proponents of the Principles-and-Parameters theory claim to have achieved a high level of descriptive and explanatory adequacy (in the sense of Chomsky 1965: 24–26), Culicover and Jackendoff (2005: 25–37) draw attention to the fact that this alleged success is only achieved by ignoring a great number of linguistic phenomena. The standard approach of the Principles-and-Parameters paradigm is “to focus on the core system, putting aside phenomena that result from historical accident, dialect mixture, personal idiosyncrasies, and the like” (Chomsky 1995: 20). As it turns out, however, this results in a great many phenomena—all of which must be part of a speaker's mental representation of language—which cannot be accounted for by the Principles-and-Parameters theory. Besides, as Borsley (2006, 2007) points out, in order to achieve analyses that can capture the differences between clause types (as well as other abstract constructions) Principles-and-Parameters approaches heavily rely on invisible elements that head functional projections (such as vP, T, or C; cf. Chomsky 1995, 2000b) and which have specific properties and interpretations—an analysis “that is little different from a construction-based account that associates the same set of properties and interpretations directly” (Sag 2010: 488).

As the present volume will show, in contrast to this, Constructionist Grammar approaches provide a uniform analysis of more idiosyncratic ‘peripheral,’ as well as ‘core,’ linguistic features. They achieve this without recourse to transformations/derivations or the employment of empty elements. Instead, the mental grammar of speakers is claimed to consist of a network of schematic and substantive constructions (‘constructicon’) and it is the parallel activation of constructions that underlies a set of particular utterances (‘constructs’). This view of grammar as a mental network of constructions has recently received great empirical support by independent research on first and second language acquisition (see Diessel, this volume, and Ellis, this volume, respectively), psycholinguistics (see Bencini, this volume), as well as neurolinguistics (see Pulvermüller, Cappelle, Shtyrov, this volume). Moreover, constructional approaches have also provided important new insight into the diachronic evolution of languages (see Fried, this volume; Barðdal, this volume; Hilpert, this volume), as well as sociolinguistic (see Hollmann, this volume) and dialectal and discourse variation (see Östman and Trousdale, this volume).

Since Constructionist Grammar offers a psychologically plausible, generative theory of human language, it has spawned a large body of constructionist research. The present handbook now provides the first authoritative overview of the theory and its applications, as well as the various constructionist approaches. Each chapter is written by one of the leading researchers in the field, and the book therefore (p. 4) offers the reader the most exhaustive overview of Construction Grammar presently available. Next, we will briefly outline the handbook's overall structure and provide a short overview of the various chapters.

Following this introduction, Part I. Principles and Methods starts off with Adele Goldberg's contribution (‘Constructionist Approaches’), which highlights the fundamental assumptions shared by all constructionist approaches. As Goldberg points out, there are four main tenets that distinguish Construction Grammar approaches from Chomskyan Mainstream Generative Grammar:

  • analysis of phrasal and clausal phenomena as grammatical constructions

  • the focus on surface form and the resulting eschewment of transformations or derivations

  • the claim that constructions form a network

  • and the acknowledgment and careful analysis of crosslinguistic variability and the role of domain-general cognitive processes in forming crosslinguistic generalizations.

The next three chapters then illustrate that despite these common assumptions, the question as to what counts as a construction is far from clear. While all Construction Grammarians would normally agree that any idiosyncratic pairing of morphological, syntactic, lexical, semantic, pragmatic, or discourse-functional properties (such as the idiom X take Y for granted) must be considered a construction (cf. Croft and Cruse 2004: 263), other issues such as the role of frequency or the status of semiregular phenomena remain disputed: in ‘The Limits of (Construction) Grammar,’ Paul Kay, for example, advocates the most conservative view, which only considers those linguistic phenomena as constructions that a speaker needs to know to “produce and understand all possible utterances of a language and no more.” Consequently, he makes a distinction between constructions as fully productive processes that are part of a speaker's grammar, on the one hand (such as the All-cleft construction), and semiregular processes such as the ‘A as NP’-pattern (as exemplified by dumb as an ox, dead as a doornail, or green as grass), on the other, which he sees as mere patterns of coining (and which are part of metagrammar, i.e. “a compendium of useful statements about the grammar”).

In contrast to this, Joan L. Bybee (‘Usage-based Theory and Exemplar Representations of Constructions’) outlines a view in which the mental grammar of speakers is shaped by the repeated exposure to specific utterances and in which domain-general cognitive processes such as categorization and cross-modal association play a crucial role in the entrenchment of constructions. Under this view, there would be no distinction between grammar and metagrammar, but all linguistic knowledge is instead viewed as emergent (Hopper 1987; Kemmer and Barlow 2000; Ellis and Larsen Freeman 2006a) and constantly changing. Moreover, specific phonological linguistic usage-events are said to be stored together with rich semantic and pragmatic information (including, e.g., positive, as well as negative, contexts yielding so-called semantic prosody effects; cf. Sinclair 1996b). In these approaches, (type and token) frequency plays a major role and anything that has been encountered often (p. 5) enough to be accessed as a whole is considered a construction, even if it exhibits no idiosyncrasy of form and meaning (and therefore could also be assembled on-line as a construct that is fully licensed by the combination of other constructions).

Frequency also plays an important role in Ray Jackendoff's Parallel Architecture approach (Constructions in the Parallel Architecture). In the Parallel Architecture model phonology, syntax and semantics are described as independent generative components within a speaker's mental grammar. Moreover, the small-scale interface rules between the three generative components, which Jackendoff calls ‘lexical items’ (i.e., pairings of phonological, (morpho)syntactic, and semantic structures), are synonymous with what most other constructionist approaches would call constructions. On top of that, similar to usage-based approaches, Jackendoff contends that it is frequency that decides which lexical items/constructions become listed in a speaker's long-term memory. Yet, in contrast to most Construction Grammar approaches, lexical items in the Parallel Architecture do not always comprise Saussurean signs (i.e., pairings of form and meaning) but also include ‘defective’ lexical items, that is, abstract syntactic principles without meaning (such as the VP-construction [VPV NP]) or semantic principles without syntactic form, such as reference transfer in I have read Shakespeare (Shakespeare = ‘books/plays by the author Shakespeare’).

The final chapter in this part, by Stefan Th. Gries (‘Data in Construction Grammar’), then moves away from these theoretical issues but nevertheless addresses a fundamental issue of Construction Grammar: What kinds of data are used in constructionist approaches and what inferences for their theories do Construction Grammarians make from these data types? As Gries shows, Construction Grammar researchers use a wide variety of different data types from introspection through corpus evidence to experiments. However, his chapter is not merely a valuable survey of previous research. Gries also meticulously outlines the parameters along which data types can be classified (with respect to the naturalness of the data, the type of stimuli used, and type of response collected) and how this impacts the kind of inference that can be made for Construction Grammar models of speakers’ mental grammars.

Part II. Constructionist Approaches then moves on to particular frameworks in which the notion of constructions plays a central role. Historically, the first approach that explicitly labeled itself as a ‘Construction Grammar’ was Berkeley Construction Grammar (BCG). In fact, when the first papers appeared in this framework in the late 1980s to early to mid-1990s (Fillmore 1985a, 1988; Fillmore, Kay, and O’Connor 1988; Michaelis 1994; Michaelis and Lambrecht 1996), they simply called their approach Construction Grammar (with the attribute ‘Berkeley’ only added recently to distinguish it from other constructionist approaches that emerged later on). In the present chapter, one of the founding fathers of BCG, Charles J. Fillmore, outlines what motivated him and Paul Kay to pursue an analysis of grammar in which peripheral/‘noncore,’ as well as core grammatical patterns were described and explained by the same mechanisms. The chapter introduces the ‘boxes within boxes’ representation and the attribute-value-matrix (p. 6) (AVM) notation of BCG and also discusses issues such as unification, constituency, unary constructions (‘pumping’), valence, complementation, co-instantiation, and null instantiation. Moreover, Fillmore presents BCG analyses of various constructions such as the Subject-Predicate construction, Head-Complement construction, Inversion construction, or Left-Isolation construction.

While work within BCG is still being carried out and important constructionist work is published in (revised versions) of the framework (cf. the work of Mirjam Fried, Jan-Ola Östman, Toshio Ohori, and others; e.g., Fried and Östman 2004a; Ohori 2005; Fried 2008, 2009b), neither Fillmore nor Kay continue to pursue this approach. These days Fillmore's work focuses on issues of frame semantics, that is, the conceptual structure of lexical meaning (Petruck 1996), and the technological implementation of frame semantic and syntactic information in the FrameNet database <http://framenet.icsi.berkeley.edu/〉;3 (cf., e.g., Fillmore, Johnson, and Petruck 2003; Boas 2006, 2009b). Yet even though his work on frame semantics is not explicitly couched in any particular Construction Grammar framework, it is nevertheless highly relevant for constructionist analyses (cf., e.g., Iwata 2008; Boas 2009a; Hasegawa et al. 2010; Fillmore, Lee-Goldman, and Rhodes 2012).

Paul Kay, on the other hand, has started to collaborate with Ivan Sag and Laura A. Michaelis on an approach called Sign-Based Construction Grammar (SBCG). As Michaelis points out, SBCG evolved out of ideas from BCG and construction-based Head-Driven Phrase-Structure Grammar (HPSG) (Ginzburg and Sag 2000). Like BCG and HPSG, SBCG employs AVMs or feature structures to model linguistic phenomena. However, unlike BCG but similar to HPSG, the feature structures of SBCG are typed, that is, arranged in a hierarchical inheritance classification (Carpenter 1992). SBCG is a highly formalized theory in which constructions are defined as type constraints on local trees (so-called ‘constructs,’ note how this technical definition of construct and construction differs from the informal one given above; cf. Michaelis 2009, this volume; and Boas and Sag 2012 for details). In her contribution, Michaelis tracks the historical connection between SBCG and BCG and also shows in which aspects the former differs from the latter. Other issues covered in her chapter include the strict locality constraint of SBCG, the avoidance of overgeneralization, and inheritance, as well as the treatment of inflectional and derivation processes.

Two other frameworks that are also highly formalized and based on the unification of AVMs are Fluid Construction Grammar (FCG) and Embodied Construction Grammar (ECG). Besides their formal nature, these two approaches also share the fact that they are specifically aimed at computational implementation. In his contribution, Luc Steels draws attention to the fact that FCG is a formalism that allows Construction Grammar researchers to formulate their findings in a precise manner and to test the implications of their theories for language parsing, production, and learning. Moreover, FCG has been fully implemented and is freely available from the project's homepage <http://www.fcg-net.org/>. The FCG chapter in this volume provides an overview of the formalism concerning its fundamental primitive data structures and operations, as well as its computational (p. 7) background to parsing and production. On top of that, Steels also discusses issues of design patterns (i.e., the methods and techniques adopted for the implementation of complex linguistic phenomena).

ECG is another computational implementation of Construction Grammar, but as Benjamin Bergen and Nancy Chang point out, the driving question of this framework is how language is used in actual physical and social contexts. Thus, ECG is an attempt to computationally model the cognitive and neural mechanisms that underlie human linguistic behavior. One of the main tenets of ECG is that mental simulation plays a crucial role in processing language. Chang and Bergen survey this role of mental simulation in processing and outline how language can be seen as an interface to simulation. In addition, they illustrate the role of constructional knowledge for mental simulation and show how constructions are represented in ECG. Finally, based on psycholinguistic evidence, they also outline an ECG-based model of language comprehension.

In contrast to these formal approaches, the remaining approaches in this part explicitly eschew formalization and instead adopt a strong usage-based perspective. The first of these, Langacker's Cognitive Grammar (CG), developed independently of Construction Grammar, but as Cristiano Broccias notes also shares many assumptions with other constructionist approaches. The basic premise of CG is that language is grounded in embodied human experience and language-independent cognitive processes such as association, automatization, schematization, and categorization. Moreover, since CG assumes grammar to be inherently meaningful, constructions as form-meaning pairings have a central place in the framework. In CG, constructions are defined as bipolar structures that pair a semantic pole with a phonological one. Consequently, unlike other approaches, CG does not include syntactic information in the form pole. In the chapter for this handbook, Broccias outlines the role of constructions in CG, but also addresses topics such as semantic and grammatical roles in CG, the status and analysis of grammatical classes, as well as the view that meaning arises through the dynamic process of conceptualization.

Radical Construction Grammar (RCG), on the other hand, emerged as a result of William Croft's extensive typological research. Arguing that all grammatical categories are language-specific and construction-specific, RCG assumes no formal syntactic structure other than the part-whole structure of constructions and the grammatical roles that occur in constructions. RCG is thus a nonreductionist theory that takes constructions to be holistic gestalt phenomena (in which part-part relationships play no role). Moreover, RCG advocates an exemplar semantics model of the syntax-semantics mapping in which specific situation types are organized in a multidimensional conceptual space. Formal construction types are then said to have a frequency distribution over that conceptual space. Other central topics in Croft's chapter include the central role of distributional analysis in RCG, the internal formal structure of constructions, as well as the analysis of typological as well as language-internal variation, and the status of grammatical generalizations.

(p. 8) Finally, Cognitive Construction Grammar (CCG) aims at providing a psychologically plausible account of language by investigating the general cognitive principles that serve to structure the network of language-specific constructions. As Hans C. Boas notes, one result of this approach is, for example, the attempt to explain how the form of constructions is motivated by their meaning. In his chapter, Boas traces the foundations of CCG by outlining Adele Goldberg's seminal work on Argument Structure constructions. On top of this, he discusses constructions that have important discourse-relevant properties (such as the Nominal Extraposition construction or the Deprofiled Object construction), and the role of phonological factors for licensing constructions (cf., e.g., the role of prosody in the Nominal Extraposition construction or rhetorical questions). Besides this, Boas also looks at the interaction of multiple constructions, the role of networks and inheritance hierarchies, as well as frequency and productivity, from a CCG perspective.

Part III. Constructicon: From Morphemes to Clauses and Beyond then illustrates how constructionist approaches can be used for the analysis of all types of (morpho)syntactic phenomena from the lexicon-syntax cline. Importantly, the chapters in this part are not committed to a particular Construction Grammar framework but are instead couched in a general constructionist terminology in order to make their findings and results maximally accessible to all readers. Part III starts off at the bottom end of the lexicon-syntax cline with a discussion of the smallest types of constructions. In Morphology in Construction Grammar, Geert Booij presents a whole range of arguments for a Construction Grammar approach to morphology. First, he shows how specific lexical items such as playable or debatable interact with their corresponding abstract construction. As Booij emphasizes, a constructicon with both these types of constructions arranged in a default inheritance network not only licenses the specific lexical conventions of a language but also accounts for the generative power of grammar. Then he discusses the fact that psychological evidence supports the view that morphological constructions have holistic properties. In addition to word-formation, Booij also covers inflectional morphology and discusses the relationship between morphological and syntactic constructions.

One step further up on the lexis–syntax cline, Stefanie Wulff then touches upon important issues in the constructionist analysis of Words and Idioms. Wulff surveys the main arguments from Fillmore, Kay, and O’Connor's (1988) classic study on idioms and illustrates what lead Fillmore et al. to the conclusion that these phenomena are best analyzed as constructions. Besides presenting Fillmore et al.'s classification of idioms along the categories of ‘familiar vs. unfamiliar pieces’ and ‘familiarly vs. unfamiliarly arranged,’ she also discusses the interpretation of idioms from a usage-based Construction Grammar point of view. In particular, she addresses the problem of how the degree of semantic and syntactic irregularity of different constructions can be measured quantitatively. As Wulff points out, the notion of collocates might provide an important clue to the answer of this question: she argues that the compositionality of an idiom construction (e.g., beg DET question) crucially depends on how semantically similar its collocates are (p. 9) compared to those of the simple word constructions that the idiom is formally associated with (here beg and question). In particular, drawing on corpus data, she proposes a quantitative measure (weighted R) that numerically expresses this degree of semantic similarity and that can be considered a numerical measure of compositionality.

In a similar vein, Anatol Stefanowitsch presents Collostructional Analysis, a family of quantitative corpus-linguistic methods that allow researchers to express the strength of the relationship between word constructions and the grammatical structures they occur in. After familiarizing the reader with the three major variants of collostructional analysis, namely ‘simple collexeme analysis,’ ‘distinctive collexeme analysis,’ and ‘co-varying collexeme analysis,’ Stefanowitsch gives several case studies and shows how the results from these varying collostructional measures can enlighten constructionist analyses of lexical and grammatical constructions. Finally, he addresses important methodological issues of collostructional analysis such as the use of inferential statistics, the cognitive mechanisms assumed, as well as the choice of statistical tests.

Next, moving toward the more syntactic end of the lexis-syntax cline, Thomas Hoffmann looks at Abstract Phrasal and Clausal Constructions. As the title suggests, this chapter outlines how constructionist approaches can describe and model even the most abstract of syntactic structures. On top of this, however, Hoffmann emphasizes that such analyses also further our understanding of the phenomena in question. First, he turns to abstract clausal constructions and discusses the arrangement of declarative, interrogative, imperative, and relative clauses in default inheritance networks and points out differences between those Construction Grammar frameworks that take a usage-based approach and those that do not. Then he moves on to filler-gap constructions and illustrates how these can be analyzed within a Construction Grammar approach and how these constructions interact in English with preposition placement (pied-piping To whom should I talk? vs. stranding Who should I talk to?). Finally, he presents an analysis of the English comparative correlative construction (e.g., the more you read, the smarter you’ll get) and provides empirical evidence for a specific intonational signature of this construction.

While the previous chapters in this part focused more on the relationship between lexical and syntactic constructions, Jaakko Leino instead concentrates on the role of Information Structure in constructionist approaches. Since information structure influences by which construction a particular meaning is expressed and why speakers therefore choose one construction over alternative ones in specific situations, it is an element of sentence grammar that has attracted considerable interest within the Construction Grammar community. In his contribution, Leino surveys the central notions of information structure (presupposition and assertion, identifiability and activation, topic and focus) and how these factors can be incorporated into a Construction Grammar view of mental grammar.

Part IV. Acquisition and Cognition shifts the focus to the psycholinguistic and neurolinguistic underpinnings of Construction Grammar, with a particular (p. 10) focus on how children and adults acquire constructional knowledge at different stages of development. The first chapter, by Holger Diessel, is concerned with first language (L1) acquisition. Diessel contrasts the generative approach to L1 acquisition with the usage-based approach, the latter of which has been strongly influenced by—and has in turn influenced—the more general theory of constructional knowledge. Drawing on work by Michael Tomasello, Diessel considers the lexical specificity of early pivot schemas that children use, and their relationship to the verb-argument constructions that typify adult language. This involves a consideration of how the more specific knowledge displayed by children in pivot schemas generalizes to the more schematic constructional knowledge associated with adult grammars. The evidence suggests that children “learn… a network of related constructions in which the same event is construed from different perspectives so that speakers can choose the construction that is most appropriate to realize their communicative intention in a particular situation.” It appears that children's knowledge of complex sentences involving finite and nonfinite complements develops in a way that is similar to that of their knowledge of verb-argument constructions.

The usage-based approach features heavily in Nick Ellis's contribution on second language (L2) acquisition. Observing that L2 acquisition involves “processes of construction and reconstruction,” Ellis shows how the frequency, form, and function of constructions interact as learners acquire another language. Learners of English, for instance, typically “first acquire the most frequent, prototypical, and generic exemplar of the verb island” (such as give in the Double Object construction); this process appears to hold true for acquisition of other constructional types. Crosslinguistic transfer effects are also considered, and Ellis draws attention to potential bias effects of knowledge of constructions in L1 on developing acquisition of L2 constructions. The contribution concludes with a summary of possible future directions for research into constructional approaches to L2 acquisition.

After these two acquisition chapters, Giulia Bencini turns to psycholinguistic issues of language production. She first outlines the psycholinguistic “consensus model” for production, which sets out the necessary explananda for a successful theory of language production, as well as the overall architecture of the linguistic system. Then she surveys the empirical evidence for and against the two-stage model of language production, which assumes separate levels for functional (semantic and syntactic) processing, as well as for positional processing (linearization and phonological realization). In addition, she gives an overview of experiments that provide support for “the existence of lexically independent structure building operations in language production in addition to lexical representations.” As she argues, such lexically independent structural processes receive a straightforward interpretation as abstract constructions in a Construction Grammar framework. Finally, she also discusses the hypothesis that the same processes underlie understanding and producing language and adduces cross-modal priming evidence in favor of this claim.

(p. 11) The final contribution to this part, by Friedemann Pulvermüller, Bert Cappelle, and Yury Shtyrov, moves the discussion into the realms of neuroscience, and the “neurophysiological plausibility” of some of the claims of Construction Grammar with regard to syntactic structures. As the authors show, evidence from neuroscience has highly important repercussions for linguistic theory building in general. They also argue that the constructionist enterprise receives considerable empirical support from neurolinguistic studies. Specifically, the authors attempt to illustrate some of the ways in which the relationship between syntactic processing and constructional knowledge may be elucidated by drawing on “brain theories of grammar.” The research reported on in this chapter suggests that “[r]ecurrent word sequences and more abstract constructions generalizing over… specific sequences are also stored in the brain,” that the brain responds differently to grammatical and ungrammatical sequences, and that human language may legitimately be considered “as an extension of other (and older) neurocognitive features,” which supports many of the claims made by proponents of Construction Grammar. In contrast to some constructionist approaches, they also contend that neurological evidence indicates that it makes sense to postulate flexible constructional templates (i.e., syntactic linkage operations between meaningful units), as distinct from lexical construction storage.

The final part of the handbook is concerned with V. Language Variation and Change. Mirjam Fried's contribution considers the application of principles of Construction Grammar to language change. In addition to the presentation of a particular change in an Old Czech morphological construction, Fried addresses more general issues for a constructional approach to grammaticalization phenomena, and considers some of the ways in which constructions may change ‘internally,’ as well as changes that relate to a construction's distribution within larger constructions. The research reported in this chapter draws attention to the gradual nature of constructional change, the micro-steps involved at different constructional levels, and the importance of context.

Jóhanna Barðdal's contribution takes a constructional approach to the difficult issue of syntactic reconstruction. Barðdal points out that historical-comparative reconstruction is usually concerned with lexical, morphological, or phonological reconstruction, and doubt has been cast on the validity of syntactic reconstruction. She argues that a constructional approach to language, which takes form-meaning pairings as grammatical primitives, is an “optimal theoretical framework for reconstructing syntax.” Drawing on evidence from Argument Structure constructions in a range of languages, Barðdal illustrates that a constructional approach overcomes some of the alleged difficulties with syntactic reconstruction (e.g., that syntactic cognates are lacking, or that syntactic change lacks regularity).

As the chapters in other parts by Stefanowitsch and Gries show, there is a close relationship between constructional approaches to language and corpus linguistics. Martin Hilpert's chapter shows how corpus linguistics has significantly affected our understanding of constructional change. Hilpert focuses on three different kinds of constructional change: on changes in a construction's frequency, in its (p. 12) form, and in its function. Frequency changes involve change in text frequency and productivity; form changes include morphophonemic change, as well as change in argument structure; functional changes may involve metaphoric and metonymic change, as well as changes involving analogical extension. In all cases, Hilpert shows that data retrieved from corpora can reveal fine-grained detail regarding the gradual nature of constructional change, and facilitate the quantitative analysis that typifies much work on linguistic variation and change.

Ongoing change in language may be manifest in synchronic variation, and synchronic variation in different dialects and discourses is the subject of the contribution by Jan-OlaÖstman and Graeme Trousdale. In this chapter, the modeling of inherent variability in cognitive linguistics in general and in Construction Grammar in particular is illustrated by way of three case studies that present data that demonstrates structural variation, contextual variation, and variation in frequency of use associated with different regional varieties of the same language. Furthermore, the authors show how the usage-based nature of much research in Construction Grammar may be applied to emergent variation in discourse structures, particularly in dialogic contexts.

In a discussion of the relationship between Construction Grammar and sociolinguistics, Willem Hollmann illustrates some of the ways in which both sociolinguistic theory and theories of constructions in language may be synthesized. Hollmann draws on Croft's notion of ‘social-cognitive linguistics,’ that is, on the interactional nature of language in use, on Gries's work on correlations between particle placement in English verb-particle constructions and spoken vs. written language, on Grondelaers's work on correlations between variation in Dutch er-presentatives, regional and register differences, and finally on his own work with Anna Siewierska on grammatical and morphophonological variation in Lancashire English constructions. Throughout the article, Hollmann stresses the need for greater dialogue between Construction Grammarians and sociolinguistics in order to achieve a psychologically and socially plausible account of linguistic variation.

In sum, this handbook presents a comprehensive account of current work on Construction Grammar (its theoretical foundations, and its applications to and relationship with other kinds of linguistic enquiry). We are grateful to the contributors for their detailed, informed, and authoritative accounts of research into constructional approaches to human language.

Notes:

(2.) For an overview of the evolution of Chomskyan Mainstream Generative Grammar, cf. Culicover and Jackendoff 2005; Jackendoff 2002a.

(3.) A FrameNet analysis of Japanese can be found at <http://jfn.st.hc.keio.ac.jp/people.html> (cf. Ohara 2009, 2011).