PhD 1: The inchoative construction in Spanish (Sven Van Hulle)
The first PhD deals with a new paradigm of inchoative auxiliaries in Spanish: Rompió a llorar (lit. ‘he broke to cry’). Inchoative constructions express the onset of an event, and consist of four slots: a subject, an auxiliary verb, a preposition, and an infinitive. As opposed to other Romance languages, the yet understudied construction has become strikingly productive in Spanish, where a wide gamut of verb classes fills the auxiliary verb slot, for instance: change of state verbs (e.g. romper), motion verbs (Se echó a reír, lit. ‘she threw herself to laugh’) or put verbs (Se mete a escribir, lit. ‘she puts herself to write’). A second source of productivity relates to the infinitive slot. We will track the historical development of the filler classes in both slots, and their interactions, against the background of the constructionalization of a more abstract inchoative pattern, and examine the parameters that determine the competition between the auxiliary verbs.
PhD 2: The minimizing construction in French & Dutch (Margot Van den Heede)
Minimizing constructions reinforce sentence negation. Such constructions typically contain noun-fillers originally referring to small objects or values, most often joined to negative elements such as Fr. (ne) … pas, Du. geen, and which combine with a limited set of verb fillers. Such constructions come in various variants, with indefinite singulars (Fr. Je n’y comprends pas un iota ‘I don’t understand not one iota’; Du. dat interesseert hem geen bal ‘That does not interest him a ball’), bare nouns such as jack shit, etc. Other constructions involve generalizing PPs such as au monde ‘in the world’ or even que + N (Je n’entrave que dalle ‘I don’t understand COMP dalle’). As intensifiers, they belong to one of the richest sources of language creativity and they are an important stage in Jespersen’s cycle through which languages renew their negative particles by further decategorializing nouns (for Old- and Middle French, e.g. Möhren 1980; Mosegaard-Hansen & Visconti 2009). Belgian Dutch (for Netherlandic Dutch, see i.a. Hoeksema 2009) and French data will be investigated, against the background of typological differences related to negation in general, taken from the literature on double negation and negative concord, negative polarity items and Jespersen’s cycle (cf. e.g. Giannakidou 2005).
PhD 3: The anti-causative construction in English, Dutch & French (Martin Godts)
The anticausative construction construes an event as happening ‘semi-autonomously’ (The door opened), in contrast to caused events (John opened the door). Each language has a set of possible strategies to express anticausative vs. caused events. English is noted for its frequent use of P-labile verbs (McMillion 2006), i.e. verbs that allow for both intransitive and transitive construals of an event in which the Patient, e.g. the chocolate, is either the first or second argument: The chocolate melted vs. We melted the chocolate. By contrast, Romance languages make use of a reflexive pattern (Fr. La porte s’ouvrait ‘The door opened-refl’ vs. Jean ouvrait la porte ‘Jean opened the door’) (Dobrovie-Sorin 2006). Another competing pattern is a periphrastic one, in which the alternation holds between a copula (Du. De deur ging open ‘The door went open’) vs. a caused resultative (Du. Jan deed de deur open ‘Jan did/made the door open’). Productivity will here be addressed at the level of competing constructions. Germanic vs. Romance strategies will be investigated against the background of typological differences in relation to alignment (ergativity vs. transitivity/ accusativity — Halliday 1967) and the encoding of resultativeness/ change-of-state meanings. English will be taken as a starting point, to be compared with Dutch and French.
PhD 4: The alternating Nom-Dat/Dat-Nom construction in Icelandic & German (Joren Somers)
This alternating Dat-Nom/Nom-Dat construction shows variation in terms of word order, i.e. between the dative and the nominative occurring pre- or postverbally, with both word orders being equally neutral:
| (1a) | Mir gefällt das Buch gut | (German) |
| (1b) | Mér fellur þessi bók vel í geð | (Icelandic) |
| ‘I like this book’ | ||
| (2a) | Das Buch gefällt mir gut | (German) |
| (2b) | Þessi bók fellur mér vel í geð | (Icelandic) |
| ‘This book pleases me’ | ||
| (3a) | Me dunkt dat | (Dutch) |
| (3b) | Dat dunkt me | (Dutch) |
This type of alternation is restricted to particular semantic verb classes (cf. Barðdal 2001a). Alternations of this type are documented at different times in the individual languages and it is hypothesized that those onset times and the further evolution of the pattern depend on changing productivity values of host verbs. We will zoom in on (i) changing productivity of language cognates and (ii) differences in word order between Dat-Nom and Nom-Dat constructions. Three Germanic languages will be studied: Icelandic, German and Dutch, which are three of four Germanic languages where the construction still exists (with Faroese being the fourth). The construction is most profound in Icelandic and German (Barðdal, Eythórsson & Dewey 2014), although its scope in Dutch remains to be investigated.
PhD 5: Productivity of sentence constructions: from corpus to sentence processing (Mariia Baltais)
Language speakers use a structured inventory of grammatical patterns, or constructions, which can be productive to varying degrees. On the one hand, productivity of a construction can be measured based on its usage as attested in language corpora. However, its extensibility (Barðdal 2008) goes beyond closed-ended corpora when speakers instantaneously apply grammatical patterns in sentence processing. Until now, it is not clear whether and how productivity hypothesized on the basis of language corpora matches productivity “at work” in the minds of native speakers measured on the basis of on-line and off-line language processing data. Additionally, previous research has not paid much attention to the influence of semantic differences or differences between individual speakers on productivity measures.
Therefore, the main goal of PhD 5 is to provide a better understanding of the relation between different aspects of productivity by conducting a series of psycholinguistic experiments. The project aims to answer two main research questions: – RQ1: To what extent are corpus measures of productivity predictive of processing data? – RQ2: How extensible are the constructions and how fast can speakers learn to extend them on-line through repeated exposure? To answer these questions, a series of sentence-rating and eye-tracking-during-reading experiments is planned. Four constructions from different but related languages will be examined (see PhD projects 1-4).
PhD 6: Individual variation in attitudes towards grammatical productivity / creativity in Dutch (Anouk Van Den Stock)
Corpus-based investigations of productivity tend to abstract away from user-related variables. This is problematic for two reasons. First, productivity is referred to as a constrained form of creativity (Goldberg 2019). Creativity, however, is a property of the speaker and not of the language (Zawada 2006), and therefore likely to be influenced by personal variables (Hoffmann 2018). Second, from a usage-based perspective, we can expect that every speaker’s unique experience with language results in individual variation in which speakers (i) will innovate within constructions, and (ii) will accept creativity within constructions.
In order to get a grip on why and how constructions can be/become more or less productive in language, it is therefore imperative to take into account individual differences between language users in their attitudes towards productivity. To test for such effects, this project will conduct several acceptability experiments in which Dutch native speakers are asked to rate both conventional and unconventional/creative instances of several constructions in Dutch. Materials will be selected on the basis of preliminary corpus investigations. Three types of user-related variables will be considered: (i) ‘prototypical’ sociolinguistic variables such as age, gender and education, (ii) personality traits such as Extraversion, Openness and Agreeableness – which will be measured through the BFI-2 questionnaire (Soto & John 2017), and (iii) cognition-related measures such as general intelligence, receptive vocabulary and print exposure.