# Plenary talks

• Bodo Winter

Workshop

Bodo Winter

Data visualization with ggplot2

Data visualization is an essential part of statistical analysis: we create graphs for ourselves in the process of exploring datasets, and we create graphs to communicate our results to others. The R package ggplot2 has become the lingua franca of data visualization, and this workshop is all about how to use it within efficient and reproducible work flows. The workshop will emphasize live coding, with participants getting hands-on experience in creating a range of plots for different kinds of linguistic data.

• Stefan Hartmann

Workshop

Stefan Hartmann

Regression modelling strategies for count data

Linguists love to count, but regression modelling strategies that are particularly well-suited to investigate count data are arguably still underused in linguistic research. While methods like simple linear regression and binomial logistic regression modelling have become part of the standard toolkit of linguists, some research questions require other modelling strategies. Methods like Poisson regression and negative binomial regression are well-suited for regression models with count data as outcome variable. Drawing on hands-on examples, this workshop provides a basic introduction to these methods using R. We will first explore the conceptual basics of these methods and then take a closer look at how we can implement them in R. Basic knowledge of other regression modelling strategies, and/or of R, is an advantage but not necessary for following the course. However, all participants should make sure to have R (https://www.r-project.org/) and RStudio (https://posit.co/download/rstudio-desktop/) installed in advance.

• Francesca Masini

Francesca Masini

No escape from multi-word expressions

Why should a morphologist bother about multi-word expressions (MWEs)? Morphology is about the structure and meaning of words and MWEs are not, strictly speaking, “words”. As a matter of fact, they are not “phrases” either, because they may have idiosyncratic properties (formal, semantic) and/or behave in a different way in terms of internal cohesion. Disorienting as it may be for those adhering to a modular view of grammar, this state-of-affairs is quite expected in constructionist approaches, where the notion of construction as a sign handles the morphology-syntax distinction in a gracious way.

It is not surprising, then, that MWEs did not have to struggle to become part of the picture in Construction Grammar. Quite to the contrary, they (specifically, idioms) have been central to the development of Construction Grammar since the beginning, and their relevance has been confirmed by the rise of Construction Morphology, which aims at integrating MWEs into a theory of morphology.

Despite these premises, morphologists adopting Construction Morphology as a framework have been playing little attention to the multiword level and its interaction with words and morphological processes, both within single linguistic systems and cross-linguistically. I will discuss why there is no escape from multi-word expressions if one wants to stick to the constructionist paradigm and what are the theoretical and methodological challenges we need to face to actually implement a unified view of linguistic structures.

• Steffen Höder

Steffen Höder

Here be dragons? Construction Grammar, phonology and the double articulation of language

Looking back over several decades of constructionist research, it is quite obvious that there has always been a strong urge not to meddle with phonology more than absolutely necessary. The main reason is that Construction Grammar has usually been defined as a theory dealing with the lexicon-grammar continuum, and the boundaries of grammar have been drawn along traditional lines, i.e. as encompassing syntax and morphology. In more recent years, however, things have started to move, and we now routinely discuss, for example, pragmatic and multimodal constructions as well.

In this talk, I will argue that if we claim (which we usually do) that the linguistic knowledge of speakers and communities can be modelled in its entirety in terms of constructional networks, then the exclusion of phonology from Construction Grammar is ultimately untenable:

1. Most obviously, lexically filled constructions have always been imagined to include a phonetic form, but exactly what this form consists of has largely remained obscure or simply been neglected, often leading to descriptions relying on conventional orthography rather than phonetic reality.
2. The idea of phonology being outside grammar follows from the notion, inherited from earlier structuralism, that language is composed of meaningful and merely distinctive elements (the so-called double articulation of language). However, this distinction is not as clear-cut as is often assumed, and it does not always coincide with the boundary between morphology and phonology. This boundary is challenged, among other things, by a wide range of grammatical phenomena that involve various kinds of phonological elements, from morphophonological alternations, other types of non-concatenative morphology, and prosodic clause-type marking to phonaesthemes, ideophones, and interjections.
3. Finally, from a usage-based perspective, Construction Grammar must be able to model regional, social, and functional variation within languages as well as multilingualism, especially because most speakers and communities worldwide are multilingual, and all languages exhibit some degree of intralingual variation. In such contexts, phonological variables are clearly both socio-pragmatically meaningful.

The focus of the talk will be on how Construction Grammar – not least in its analysis of morphology – can benefit from studying such supposedly peripheral phenomena and how, in turn, such phenomena can be analysed using the same theoretical toolbox that we as constructionists are already familiar with.

• Muriel Norde

Muriel Norde

Creativity in language change: a challenge for Diachronic Construction Morphology

Diachronic Construction Grammar (DCxG), a constructionist approach to language change (Barðdal et al. 2015; Noël 2016; Sommerer & Smirnova 2020), is by now a well-established framework that is concerned with the emergence of and changes in constructions, i.e. form-meaning pairings of varying degrees of complexity and schematicity. A recent spin-off is Diachronic Construction Morphology (DCxM, cf. Norde & Trousdale 2023), which focuses on changes in morphological constructions, building on the (synchronic) framework of Construction Morphology (Booij 2010; Masini & Audring 2019).

Like CxM generally, DCxM is output-oriented, on the usage-based assumption that morphological patterns are shaped by usage (Diessel 2019: 19) and hence prone to change across the life-span, in what Kemmer & Barlow (2000: ix) have called a ‘feedback loop’. In bottom-up fashion, speakers generalise over similar sets of constructions, which is formalised in (D)CxM by means of schemas (Booij 2010: 51–93). Schemas consist of a formal pole and a semantic pole and the symbolic link (represented by a double arrow) between them. Examples of schemas are given in (1-3), illustrated with German hammer (Norde & Van Goethem 2018).

The adjective hammerhart ‘very hard’ in (1) can be represented either by a schema that is fully specific (1a) or by a fully schematic one that generalises over all simile constructions (e.g. blutrot ‘red as blood’ or riesengroß ‘big as a giant’). Because simile constructions invite an intensifying interpretation, the use of hammercame to be extended to adjectives where a direct comparison with a hammer no longer makes sense, e.g. hammergeil ‘very cool (?cool as a hammer)’. Such constructs show that hammer now also functions as a prefixoid (hence the angle brackets), which has a meaning different from the corresponding free noun. Accordingly, we now have, apart from the specific construction in (2a), a partially schematic schema (2b) with hammer as a fixed slot and a free adjective slot, which is related to the fully schematic schema (2c), which represents all prefixoid constructions where an adjective is intensified without a simile relation (e.g. saukalt ‘very cold (?cold as a pig)’ or scheißteuer ‘very expensive (?expensive as shit)’. In specific contexts, hammer was reinterpreted as an adjective, which resulted in a new, fully specific construction again, schematised in (3).

(1)              hammerhart ‘hard as a hammer; very hard’

b:               [[a]N,i[b]ADJ,J]k                            ⟷              [as SEMj as SEMi]]

(2)              hammergeil ‘very cool’

(3)              ein hammeres Ende ‘a cool ending‘

A development as sketched in (1-3) raises interesting questions related to creativity in morphology. Following Norde & Trousdale (submitted), I propose a model that draws on Sampson’s (2016) distinction between F-creativity (fixed creativity) and E-creativity (expanding creativity). The terms have been the subject of debate in the constructionist literature (Bergs & Kompa 2020; Cappelle 2020), but from a DCxM perspective the distinction could be useful, with a small adaption. In Norde & Trousdale (submitted) we argue that there are in fact two types of F-creativity, which we term F1 (for fully productive patterns such as (1b)) and F2, where speakers create a form that is not fully sanctioned by the schema, e.g. when the meaning of hammer has been extended from comparison to intensification more generally. While such new constructions are (initially) F2-creative, they may (when type frequency increases) lead to a new schema, as in (2b), which is E-creative, both at the semi-schematic level and at the fully schematic level. A new specific form (as the adjective in (3)) is also E-creative in this model.

This view on creativity poses two challenges. The first one is that change is incremental, so that the line between F- and E-creativity is not easy to draw (Ungerer & Hartmann 2023: 44), although the emergence of Zipfian distributions (e.g. Feltgen 2020) might hint at E-creativity. There are also cases where F2-creativity does not lead to E-creativity, namely when a pattern becomes less constrained, e.g. when German un-X-bar, which originally requires a verbal base (e.g. undenkbar ‘unthinkable’) is found with an adjectival base (unkaputtbar ‘unbreakable, lit. un-broken-able’) (Hartmann 2014: 182). When such adjectival bases become increasingly common, selection restrictions on an existing schema are lifted, but a new schema, pairing new formal and semantic properties, does not emerge. The second challenge concerns the relation between creativity and extravagance (see for instance the papers in Eitelmann & Haumann 2022a). Where Eitelmann & Haumann (2022b: 5) see E-creativity as coming close to extravagance, (Hoffmann 2020) argues that some cases of E-creativity, e.g. the development of English to be going to as a future auxiliary, are not deliberate. To address these challenges, I will test the model across a number of case studies.

• Kristel Van Goethem

Kristel Van Goethem

Dutch compound constructions in learner language: cross-linguistic influence and exposure effects

Several studies have demonstrated that Dutch has a stronger tendency towards compounding than French (e.g., Du. zoutwaterzwembad vs Fr. piscine d’eau salée ‘saltwater pool’) (Van Goethem 2009; Van Goethem & Amiot 2019), especially when adopting a restrictive approach of compounding in which the presence of prepositions and/or internal inflection in multi-word expressions is considered evidence for their syntactic formation (Fradin 2009). The example above illustrates that Dutch compounding differs from French in another important aspect: while Germanic compounding is by definition right-headed, French has a general tendency towards left-hand headed compounds and phrases.

Nevertheless, little attention has been paid so far to the impact of such cross-linguistic differences on the use of compounds in learner language, even though word-formation awareness has been proven crucial for learners’ proficiency and creativity (Balteiro 2011).

In this presentation, we investigate the impact of these cross-linguistic differences on the acquisition of Dutch compounds by French-speaking learners in the context of multilingual Belgium. Moreover, we explore the impact of additional target-language input through CLIL programs (Content and Language Integrated Learning) (Hiligsmann et al. 2017, Van Mensel & Hiligsmann 2023). The corpus data are drawn from the MulTINCo database (Meunier et al. 2023) and we will focus on the acquisition of Dutch nominal compounds (Hendrikx & Van Goethem, forthc.).

The results are described and interpreted from the theoretical perspective of Diasystematic Construction Grammar(DCxG) (among others Höder et al. 2021), which conceptualizes the linguistic competence of multilingual speakers as one integrated network of constructions, containing language-specific idioconstructions and shared diaconstructions

• Livio Gaeta

Livio Gaeta

Up and down the Constructicon: Paradigmatic and syntagmatic aspects of Construction Morphology

Construction Morphology generally represents derivational relations in hierarchical terms, crucially relying on the mechanism of Default Inheritance. This can be concretely visualized by means of vertical relations connecting more general patterns or schemas lying high up in the Constructicon and more specific constructions placed down. Constructions ordered in such vertical relations are usually piled up in terms of specificity and/or reach of the schema, while these two criteria are not really elaborated in the current literature. In this talk, the attempt will be made at refining this view, especially focusing on possible mechanisms and forces guiding processes of top-down and/or bottom/up generalization – such as the challenging and elusive concept of markedness – which are held to influence the productivity of word-formation patterns within the Constructicon.