The volume Nodes and networks in Diachronic Construction Grammar edited by Lotte Sommerer and Elena Smirnova (2020) brings together ten contributions to the workshop on ‘Advances in Diachronic Construction Grammar’ held at the fiftieth meeting of the Societas Linguistica Europaea in September 2017 in Zürich. The aim of the workshop was to address a series of open questions pertaining to the still relatively young field of Diachronic Construction Grammar (henceforth DCxG).
The Introduction to the book by Smirnova & Sommerer is a fully fledged chapter which provides a deeply insightful and critical overview of open questions, inconsistently treated topics and unresolved issues relating to the nature of the node, the design of the constructional network, and the notion of change from a DCxG perspective. To pick out one example, one of the main controversies dealt with in depth in this introductory chapter – and indeed throughout the book – is the conceptual distinction between ‘constructionalization’ and ‘constructional change’, which has lain at the heart of many studies in DCxG since Traugott & Trousdale's (Reference Traugott and Trousdale2013) seminal book. In a nutshell, according to the definitions provided in Traugott & Trousdale (Reference Traugott and Trousdale2013), constructionalization involves the creation of a new node in the constructicon because of changes in the form and meaning of an existing construction or because of the emergence of an entirely new form–meaning pair in the network, whereas constructional changes encompass formal or meaning changes alone, and therefore do not involve the creation of a new constructional node. However, Smirnova & Sommerer convincingly point out several weaknesses in this distinction, at both the theoretical and empirical levels, based on original evidence of, among other things, the development of complex adpositions in German. For instance, they argue that Traugott & Trousdale's (Reference Traugott and Trousdale2013) definition of constructionalization paradoxically conflicts with de Saussure's concept of the sign. Although the concept of the construction is considered to be an extended version of the Saussurean sign (see Koutsoukos et al. Reference Koutsoukos, Van Goethem and De Smet2018), a definition of constructionalization which postulates that formal changes or semantic changes alone are not sufficient to determine the emergence of a new construction violates the basic concept of the Saussurean sign as a unified pairing of form and meaning. Moreover, Smirnova & Sommerer rightly identify a contradiction in Traugott & Trousdale's (Reference Traugott and Trousdale2013) use of the notion of ‘gradualness’ – also discussed in the chapter by Flach – when speaking of constructionalization, since change is by definition a gradual process, whereas the emergence of a new construction is conceptualized as an abrupt, instantaneous node creation. Thirdly, they argue convincingly that the concept of constructionalization cannot be applied in a unified way to the wide array of construction types, which are characterized by different degrees of abstraction, schematicity and complexity.
These theoretical issues give rise to empirical problems when change in particular constructions is investigated. In a case study on debonding of Germanic prefixoids (e.g. keihard ‘lit. boulder-hard; very hard’ vs kei te laat ‘lit. boulder too late; way too late’), Norde & Van Goethem (Reference Norde, Van Goethem and Booij2018) examine whether the bound and free uses of intensifying prefixoids should be considered as separate constructions, and demonstrate that the distinction between constructionalization and constructional change is hardly operationalizable. Traugott & Trousdale's (Reference Traugott and Trousdale2013) account of constructionalization is not very explicit as to how to qualify or quantify changes in form or meaning and raises the question of how many changes are needed for a form or meaning to be considered ‘new’, which causes several researchers to point to the analogy with the Sorites Paradox (How many grains of sand make a heap?) (cf. also the contribution by Flach). In addition, it may be argued that other types of change may be as relevant as formal or semantic changes when it comes to identifying the emergence of a new construction. In this context, Hilpert (Reference Hilpert2013) argues that the emergence of new constructions, called ‘constructional change’ in his account, manifests itself not only through form and meaning changes, but also through changes in frequency or distribution. Hilpert (Reference Hilpert2013: 7) states explicitly that ‘frequency changes … are no less indicative of constructional change than are developments in meaning or the phonological and morphosyntactic substance of a construction’. In the same vein, Norde & Van Goethem (Reference Norde, Van Goethem and Booij2018: 515) conclude their study on the debonding of Germanic prefixoids by stating that ‘[R]ather than concern ourselves with the question of whether or not the emergence of free uses of prefixoids is constructionalization, we feel it is more insightful to look at observable changes at different levels, considering as many factors as possible that can be operationalized quantitatively, including collocational properties and productivity.’
In addition to the key issue of the nature of change in the constructional network, the introductory chapter tackles many other intriguing and complex issues, one of the most fundamental being the nature of (vertical and horizontal) constructional links, another so far insufficiently charted territory. In total, Smirnova & Sommerer identify seven theoretical questions that are central to the contributions in the book. These questions are the following (pp. 3–4):
1. How can node creation and node loss be implemented in the network model?
2. When is it warranted to postulate a new separate node in the network as a result of ‘constructionalization’ and when is it not (‘constructional change’ only)?
3. What kinds of connections exist between the nodes in the network (i.e. vertical and horizontal links)?
4. What is the theoretical status of ‘allostructions’, ‘homostructions’, ‘constructional families’ and ‘paradigms’?
5. How can the reconfiguration of node-external linking be modelled in the (existing) network model?
6. How do general cognitive abilities like analogical thinking, routinization/automatization, abstraction and categorization/schematization relate to the structure and reorganization of the constructional network?
7. (How) should the existing network model be enriched or reconceptualized in order to integrate aspects missing so far?
The nine contributions that follow the introductory chapter explicitly address one or more of these questions, and are logically structured around the following three topics: (1) The nodes: creation, change and loss; (2) The links: vertical and horizontal relations; and (3) Beyond existing models. While most contributions concentrate on the nature of vertical and horizontal network links (Gyselinck; Perek; Zehentner & Traugott; Percillier; Lorenz), two contributions focus on the creation, change and loss of the network nodes (Flach; Sommerer), and two other contributions discuss how the existing models should be enriched or reconceptualized (Diewald; Budts & Petré). In what follows, I briefly summarize these nine chapters.
The chapter by Susanne Flach tackles the terminological ambiguity of the concept of ‘constructionalization’, which refers either to the process that leads to the creation of a new construction or to the end point of this process, i.e. the node creation itself. On the basis of the diachronic development of the into-causative construction in English (e.g. They talked him into complying with the rules, p. 46) from a series of changes in the caused-motion construction (e.g. They talked him into compliance, p. 46), the author suggests that the term ‘constructionalization’ is only useful if it is reserved for the node-creation meaning, while the process leading to the creation of a new node would be better captured by the label ‘constructional emergence’. Furthermore, all changes in the network, whether or not related to the emergence of a new node, could be subsumed under ‘constructional change’. As such, the chapter advocates a view that foregrounds the dynamic connections in the network instead of narrowing down the focus to the emergence of new nodes.
The chapter by Lotte Sommerer is also an original contribution to the ongoing discussions, because – unlike the bulk of studies in DCxG – it does not concentrate on the emergence of a new construction, but on so-called ‘constructional loss’ or ‘constructional death’, i.e. the disappearance of a node in the network. More specifically, the author examines Old English constructions that feature co-occurrence of a possessive and a demonstrative (POSS DEM or DEM POSS), a construction type that is considered ungrammatical in Present-day English (e.g. *his that neighbor), and the driving forces leading to the disappearance of this constructional family. A multivariate mixed logistic regression model suggests that the constructions under discussion occur more often in translations from Latin than in original texts and that they decrease in frequency over time, especially between early and late Old English. The author argues that, ultimately, the loss is due to competition with a new productive NP construction that allows for only one determiner slot. Special attention is paid to the effect of frequency and cognitive factors (entrenchment and analogical thinking) on the network changes under consideration.
Emmeline Gyselinck concentrates on the Dutch intensifying fake reflexive resultative construction (e.g. Ik betaal me elke maand blauw, lit. ‘I pay myself every month blue’, ‘I pay a lot of money every month’, p. 108), a construction related to the fake reflexive resultative construction [SUBJ V REFL XP], but with the specific property that the postverbal phrase (XP) does not express the result of the verbal activity, but its intensity. This interesting case study presents itself as a well-suited candidate for closer examination of productivity shifts and internal reorganizations in the constructional network, among other reasons because intensifiers are known to be particularly prone to processes of renewal and innovation. Zooming in on the diachronic developments related to frequency, productivity, collocational behaviour and conventionalization in the intensifier slot, Gyselinck neatly reveals that the dynamicity of the network may involve diverse and even reverse reorganizational processes, from expansion and schematization (e.g. [SUBJ V REFL het vuur uit de sloffen (lit. ‘the fire out of the slippers’)]) to conventionalization and loss (e.g. [SUBJ V REFL wild (‘wild’)]).
In the same vein as the contribution by Gyselinck, the chapter by Florent Perek also concentrates on changes in schematicity and productivity in order to capture the nature of constructional change. More specifically, Perek delves into the relationship between the two concepts and argues that, despite their close connection, schematicity and productivity should be distinguished as separate properties of a given construction that do not necessarily work in tandem. To do justice to this relationship, a distinction should be made between the schematicity of the lexical slots, i.e. generalization of a construction's lexical fillers, on the one hand, and that of the meaning of the construction as a whole, on the other. The author argues that only the former is directly related to productivity. This is illustrated by a closer look at the recent change in the English way-construction (e.g. She managed to talk her way out of the ticket, p. 152).
The chapter by Eva Zehentner and Elizabeth Closs Traugott concentrates on the emergence of the benefactive alternation constructions (e.g. John baked Mary a cake / John baked a cake for Mary, p. 169) in Early Modern English, which can be seen as a particular subset of the dative ditransitive constructions. Special emphasis is laid on the interaction and horizontal network links between benefactive and ditransitive constructions at different levels of abstraction. The empirical analysis shows that horizontal connections are key to accounting for the development of a division of labour between the constructions under discussion, and specifically for the crystallization of the preposition for in the benefactive prepositional allostruction, as well as for the generalization of an underspecified benefactive alternation schema (or ‘constructeme’). Ultimately, the analysis demonstrates the benefits of postulating horizontal links between constructions in addition to vertical ones in order to gain a better understanding of the changes under consideration.
Michael Percillier's chapter investigates the changes in the network of secondary predicate constructions with as, for, into and to (e.g. crown someone to/as king, p. 213) that took place in the Middle English period, and especially the transition from the predominant use of the variant with the preposition to to the Present-day English pattern featuring a preference for prepositional as. As in the previous contribution, the theoretical focus is placed on the horizontal (either polysemic or allostructional) connections between the network nodes. In addition, by investigating the influence of Anglo-Norman, which introduced new secondary predicate constructions with French-based verbs via replication in Middle English, the chapter shows how external factors, in particular language contact, can be modelled into the DCxG framework. Such constructions which are introduced through language contact and function as homonymic counterparts to the pre-existing native constructions because of formal similarity but divergent semantic properties are labelled as ‘homostructions’.
The central claim of the study by David Lorenz is that the formation of new schemas goes hand in hand with changes in both vertical (instantiation and inheritance) and horizontal (associative) connections in the network. More specifically, the study concentrates on the emergence of new horizontal links on the basis of a quantitative corpus study of to-contraction (want to > wanna, going to > gonna, got to > gotta) in American English since the nineteenth century. While various contracted forms first emerge merely as phonetic reductions and pronunciation variants of the full forms, the contracted forms gonna, gotta and wanna are ultimately emancipated from their source forms by an increasing frequency that is accompanied by increasingly similar usage patterns. This constructional split from the full forms results in an emerging pattern of contractions with increasing strength of association, which the author accounts for as the emergence of a ‘metaconstruction’, i.e. ‘an associative link that defines the analogical status of the contractions relative to the full forms’ (p. 267).
Gabriele Diewald's chapter pleads for a revalorization of the concept of ‘paradigm’ in Construction Grammar. Although the key role of paradigms is widely recognized in the context of grammaticalization and morphological theories, the author regrets that the notion has largely been lost in the field of CxG. Drawing on existing approaches such as the second order schemas of Booij (Reference Booij, Hippisley and Stump2016) and evidence from changes in German and English modality, tense, determiner and number paradigms, Diewald suggests incorporating the notion of paradigm as a ‘hyperconstruction’ into constructional models. In this approach, a grammatical paradigm such as tense should be conceived of as a complex node in which an ordered hierarchical system of individual constructions is embedded and whose meaning is the sum total of these embedded relationships. A paradigm node should then be embedded into other constructions wherever its function is needed.
Finally, the chapter by Sara Budts and Peter Petré has the unambiguous objective of ‘[P]utting connections centre stage in diachronic Construction Grammar’ (see the title of the contribution, p. 318). The authors argue that the constructicon is more complex than a basically two-dimensional network because syntagmatic and paradigmatic relations are claimed to exist between constructions in both the vertical and the horizontal connections of the network. To support this view, evidence is provided from two quantitative corpus studies in which the emergence of a new construction is related to shifts in syntagmatic and similarity-driven associations with other constructions, which result in changes in paradigmatic affiliations. The first case study investigates the emergence of the [be going to inf] construction and the second examines the periphrastic do construction in close relation to English modal auxiliary constructions. By means of these case studies, the authors demonstrate both the intrinsic interrelatedness between a multiplicity of dynamic connections in the network and their key role for a full understanding of constructional change.
I warmly congratulate all the authors included in this volume on having contributed to this inspiring work with high-quality chapters that cover a wide range of linguistic phenomena and address challenging theoretical questions grounded in rich empirical data. I also praise the editors for their admirable in-depth and critical introduction to the body of burning questions that DCxG currently has to address. I highly recommend this book to anyone interested in Construction Grammar and/or diachronic linguistics: it is a perceptive and thought-provoking contribution to the field that not only openly poses problematic questions and highlights challenges for future research, but also presents fruitful new suggestions to the DCxG community. By way of example, the distinction made by Sommerer between ‘constructionalization novo loco’ (i.e. the addition of new nodes) and ‘constructionalization in situ’ (i.e. the local substitution of existing nodes) or that between ‘constructionalization’ (i.e. node creation) and ‘constructional emergence’ (i.e. the process leading to node creation) introduced by Flach, may potentially enable us to solve terminological ambiguities and to further refine concepts that are central to DCxG. Moreover, the final chapter by Budts & Petré, which goes beyond the traditional two-dimensional network models and proposes a dynamic connectionist model of language and change, most certainly deserves further exploration.
I can point to only three ‘imperfections’, which I would prefer to formulate positively as an invitation to the entire DCxG community to continue the valuable work presented in this book.
First, as a researcher in comparative linguistics, I feel I must point out the nearly exclusive focus on English: seven of the volume's nine chapters deal with English constructions, albeit in different historical stages of English; only the contributions by Gyselinck (Dutch) and Diewald (German) include other languages. This main focus on English is unfortunately typical of a long tradition of studies in CxG. For my part, I would see this volume as an opportunity to examine the validity of the newly provided theoretical proposals and empirical research designs for as many other languages as possible. If we aim to establish a theoretical model of DCxG that strives for psychological plausibility (p. 24), it should have universal power and should not be restricted to a single language.
Second, in spite of possible shortcomings, I sincerely believe that the merits of Traugott & Trousdale's (Reference Traugott and Trousdale2013) work, which has provided us with the first monograph-length and unified account of DCxG and has been a rich source of inspiration for many researchers in the field since its publication, cannot be sufficiently stressed. To conclude the book by offering a right of reply in the form of a commentary chapter to the criticism expressed towards concepts such as ‘constructionalization’ and ‘constructional change’ could therefore have been the icing on the cake.
Third, all the chapters in the book contribute explicitly to the questions raised in the Introduction by providing innovative accounts based on new empirical evidence. Several chapters concur in highlighting the importance of both vertical and horizontal connections in the network and the crucial role they play in constructional change. Nevertheless, as also pointed out in the conclusion of Lorenz's chapter, although the analyses are consistent and the newly introduced concepts seem useful, the accounts are still quite divergent and many terms are (re)introduced (‘sister nodes’, ‘constructeme’, ‘allostructions’, ‘homostructions’, ‘metaconstructions’, ‘hyperconstructions’, etc.). It remains to be seen in future research if these accounts and terminology can finally result in a (more) unified approach to DCxG.
That said, I am confident that the many inspiring insights and suggested responses to open questions that this volume provides are a significant stepping stone towards new common ground for future research in DCxG and that the volume has brought us ‘closer to a flexible but more constrained, albeit cognitively plausible network model of change’ (p. 36), thus fulfilling one of the aims initially set out for the book.
