Belgian Journal of Linguistics - Volume 34, Issue 1, 2020

The German Constructicon Project (www.german-constructicon.de) aims at documenting grammatical constructions in contemporary standard German on the basis of annotated corpus examples, including relations between constructions and between constructions and evoked semantic frames. So far, the research focus has been mainly on the development and computational implementation of a constructicographic workflow (including a parsing pipeline) that allows for addressing any kind of constructions on varying levels of schematicity, idiomaticity, and abstractness. However, such an exemplar-driven procedure precludes us from systematically identifying constructional candidates. In this article, we scrutinize ways to operationalize and implement data-mining procedures to inductively identify construction candidates.

The present contribution summarizes findings on the understudied area of Italian valency coercion – i. e. the interaction of verbs and argument structure constructions in novel and creative ways – from four different studies. It highlights their innovative character, theoretical significance, and crosslinguistic implications for Construction Grammar. The paper suggests that valency coercion resolution involve different phenomena, such as distributional properties of constructions and compatibility between verb and construction. Sociolinguistic factors such as age and diatopic variables are also suggested to be relevant.

We develop the notion of pattern of coining found in some complete-inheritance models of Construction Grammar (Fillmore 1997; Kay 2013), which are processes used to coin new units based on analogy with an existing one. Unlike constructions, they cannot be considered systematically productive in synchrony. After providing measurement methods, we assess the productivity of three patterns (‧whelm, ‧licious and ‧holic). To do so, we carried out a statistical analysis using two web corpora. Unlike Kay, we show that the difference between constructions and patterns of coining is not so clear-cut, since patterns of coining may undergo constructionalization, and that qualitative aspects should be taken into account along with quantitative data when trying to assess the status of a word-formation pattern.

This squib discusses the question whether Construction Grammar can account for the assumption of universal grammatical categories (Bybee, Perkins, and Pagliuca 1994) that are prone to language change, e.g. tense. Most publications in Construction Grammar tackle individual constructions, such as the way-construction (Jackendoff 1990). But it remains unclear how grammatical categories as a universal phenomenon can be described in constructionist terms. We propose that there is a way to (a) describe grammatical categories, which per definition are encoded paradigmatically, as constructions themselves and (b) to thereby strengthen the assumption of a set of universal grammatical categories.

This squib applies and extends insights from (Diasystematic) Construction Grammar to the code-switching and loan-translation of English verbs (and verbal constructions) in US-German dialects. After presenting recent findings about the nature and interaction of language contact phenomena, I introduce the constructional principles guiding the analysis and the data sources. I then present a wide array of data and formulate hypotheses regarding the processes and motivations underlying each type, appealing to a constructional and usage-based view of the bilingual’s mental lexicon.

This squib briefly explores how contextualized embeddings – which are a type of compressed token-based semantic vectors – can be used as semantic retrieval and annotation tools for corpus-based research into constructions. Focusing on embeddings created by the Bidirectional Encoder Representations from Transformer model, also known as ‘BERT’, this squib demonstrates how contextualized embeddings can help counter two types of retrieval inefficiency scenarios that may arise with purely form-based corpus queries. In the first scenario, the formal query yields a large number of hits, which contain a reasonable number of relevant examples that can be labeled and used as input for a sense disambiguation classifier. In the second scenario, the contextualized embeddings of exemplary tokens are used to retrieve more relevant examples in a large, unlabeled dataset. As a case study, this squib focuses on the into-interest construction (e.g. I’m so into you).

This paper presents a frame-based constructional approach to argument structure satisfaction via unselected adjuncts, by focusing on one such case in Japanese. It points out an intriguing constructional phenomenon whereby causal adjunct clauses marked with node ‘because’, as used with main-clause predicates that evoke communication frames (such as Telling and Warning), serve to satisfy main-clause argument structure. The node clause precedes the main-clause speech act of telling/warning, and can be interpreted as a speech-act causal (Sweetser 1990). The node clause at the same time conveys the content of informing or warning, i.e., the core Frame Element message, which is absent as a main-clause complement. This analysis of argument structure satisfaction via unselected adjuncts provides evidence for a Frame Semantic approach to argument structure that incorporates Construction Grammar.

This study seeks to gain a better insight into the origin and expansion of the construction <va a ser que sí/no> (lit. goes to be that yes/no) in Peninsular Spanish. We argue that this construction derives from the use of the periphrastic future construction <ir a ‘go to’ + inf> in a pseudo-cleft sentence whose subject is a deictic element or an element that conveys the speaker’s attitudinal assessment of the propositional content expressed in the attribute, a complement que-clause. The etymological structure evolves through a process that formally implies the suppression of the explicit subject and the fusion of the components va a ser que leading to the conventionalization of refutative and assertive values. To demonstrate this directionality, we examine recent stages of change and develop syntactic and semantic-pragmatic arguments grounded in a data-based approach.

This article explores the possibility of identifying mentally stored constructions in writing process data, that is, data that reproduce the process through which a text is written. The unit that serves as a basis for the identification of constructions is the burst of writing, which corresponds to a chunk of text produced between two pauses. Bursts are examined in L1 French and L2 English keylogging data from the Process Corpus of English in Education and their potential constructional status is considered.

This paper suggests two possible ways in which cognitively-oriented constructionist approaches (Cognitive Construction Grammar, Radical Construction Grammar, and Embodied Construction Grammar) could enhance the explanatory power of constructions. First, the anatomy of a construction should spell out how the morphosyntactic realizations of arguments are specifically mapped onto their inherent semantico-pragmatic properties, while also including detailed information concerning illocutionary force, information structure, register, politeness, etc. Second, it is argued that coercion should be best understood as a continuum allowing for varying degrees of (in-)compatibility between the verb and the construction taken as a whole. Moreover, parameterization and linguistic cueing prove useful to handle the dynamic interaction of the morphosyntactic, semantico-pragmatic, and discourse-functional hallmarks of constructions, including those which invite metonymic inferencing.

When communicating across closely related languages or varieties (e.g. in interdialectal communication or in regions such as Mainland Scandinavia), speakers have to learn how to decode words that show partial phonological differences from the equivalents in their L1. Although contact situations like these are rather common, interlingual decoding has scarcely been addressed in the CxG literature. As a contribution to this field of research, the paper discusses how (a particular stage in) emerging receptive multilingualism can be modelled from a CxG perspective. Specifically, it deals with the idea that repeated interlingual decoding generates partially schematic cross-linguistic constructions mirroring the speaker’s knowledge about sound correspondences, as suggested by Diasystematic Construction Grammar (Höder 2019).

In Construction Grammar theory, constructions are generally described as form-meaning pairings. It will be argued here that the formal specifications of some abstract constructions are so vague that the notion of form needs to be discussed rather critically. We aim to demonstrate how, in the English predicative and intransitive-motion constructions, the slots of more general constructions can be seen as being specified indirectly through sets of mini-constructions.

One of the hallmarks of scientific theories is their falsifiability, i.e. the fact that they make predictions that can objectively be proven wrong. Thus, it is paramount that researchers, including linguists, are able to state what kind of evidence would lead them to abandon their scientific theory. Yet, researchers just like all other human beings are susceptible to confirmation bias, i.e. the fact that they only seek evidence that supports their existing views. In this squib, I will raise the question whether Construction Grammar can become a falsifiable theory.

We provide a practical step-by-step methodology of how to build a full-scale constructicon resource for a natural language, sharing our experience from the nearly completed project of the Russian Constructicon, an open-access searchable database of over 2,200 Russian constructions (https://site.uit.no/russian-constructicon/). The constructions are organized in families, clusters, and networks based on their semantic and syntactic properties, illustrated with corpus examples, and tagged for the CEFR level of language proficiency. The resource is designed for both researchers and L2 learners of Russian and offers the largest electronic database of constructions built for any language. We explain what makes the Russian Constructicon different from other constructicons, report on the major stages of our work, and share the methods used to systematically expand the inventory of constructions. Our objective is to encourage colleagues to build constructicon resources for additional natural languages, thus taking Construction Grammar to a new quantitative and qualitative level, facilitating cross-linguistic comparison.

Japanese mimetics, and its psychomimes (e.g. gakkuri ‘disappointed’), in particular, are usually accompanied in speech with bodily movements, including gestures and postures. I have already argued that certain patterns in co-speech gestures and postures that accompanied psychomimes showed a relatively high rate of concord across speakers (Kanetani 2019). Taking the co-speech bodily movements as metonymic representations of embodied metaphors of emotion, this paper suggests that these kinetic features may be stored as part of the speaker’s knowledge of the words and argue that Japanese psychomimes are multimodal lexical constructions. I also show how such multimodal constructions are represented in the mind and how they are expressed in actual use. In particular, I describe and examine two-dimensional form-meaning pairings (based on Kita 1997) and show that one of the two dimensions may be selectively expressed in a given context.

In this paper we use corpora of four monolingual German-speaking children at 2 years of age to analyze the effect of input on the activation of chunks and frame-and-slot patterns. For this purpose, we first investigate to what extent chunks and patterns can be traced back to the direct input compared to input which is not part of the immediate discourse situation. Second, we take mean length of utterance (MLU) into account to see how the level of proficiency influences the amount of priming in each child. Results indicate that children with a lower MLU rely more on priming than children who are more proficient. This conclusion is consistent with the usage-based assumption that children’s linguistic development starts with a strongly item-based reproduction of input patterns that gradually gives rise to increasingly creative and productive uses of constructions.

Syntactic amalgams are innovative phrasal constructions that combine otherwise incompatible subparts of other constructions (Lambrecht 1988; Brenier and Michaelis 2005). We describe pleonastic formations like flavorize in English and ψηλαφ-ίζ(ω) [psilafízo] ‘palpate’ in Modern Greek as functional amalgams at the word level. We examine these formations through the lens of (function-oriented) Sign-Based Construction Grammar (Sag 2012), arguing that once we see derivational morphemes as signs, and sign combination as construction-driven rather than head-driven, we can describe such words as coercive combinations that serve a variety of semiotic functions.

Despite recent advances in Construction Pragmatics, a systematic way for delimiting coded pragmatic information has yet to be offered. This squib provides a step in establishing such an account by assessing what kind of pragmatic information speakers generalize from various usage-events. Drawing on findings from Conversation Analysis, I propose a distinction between pragmatic functions as speakers’ actions, and interactional patterns as discourse-information sequences. A synchronic examination of the Hebrew multifunctional discourse marker ′at/a yode′a/′at (′know.prs.m/f.sg′) demonstrates the consistent use of the construction in an interactional pattern across numerous usage-events. A qualitative diachronic analysis of ′at/a yode′a/′at suggests that speakers may associate forms with interactional patterns rather than with functions. This preliminary evidence provides support for the generalization of interactional patterns.

This squib provides a theoretical discussion on the use of the terms semantics and pragmatics in Construction Grammar. In the literature, the difference between semantics and pragmatics is often conceptualized either in terms of conventionality or in terms of truth-conditionality (Huang 2014, 299). It will be shown that, even though constructionists claim that there is no semantics–pragmatics distinction, both these underlying concepts are central to the study of constructions. Therefore, the aim is twofold. First, in keeping with Cappelle (2017), it will be argued that constructionists should make more explicit the distinction between the two types of (encoded) meaning. Second, it will be shown that constructionists need to be more terminologically consistent and agree on how to use the terms semantics and pragmatics. Following Depraetere (2019), I will argue that the terms semantics and pragmatics are most explanatory when defined in truth-conditional terms. In this way, finer-grained understanding of the meaning of constructions can be achieved.