Congreso, coloquio o simposio
On the occasion of the Grammar and Corpora Conference, the Language Productivity Work Consortium organizes a workshop on productivity of syntactic constructions.
When speakers produce or interpret language structures, they rely on a structured inventory of grammatical rules or patterns. Some of these are highly productive: they have a broad domain of application and are readily available to coin new expressions.
This phenomenon has long been observed in morphology. For instance, speakers of Dutch can readily apply the morphological rule Verb+baar to create new adjectives meaning ‘that can be Verb-ed’, such as in een twitter·baar stuk tekst ‘a twitterable text chunk‘. By contrast, other rules such as Verb+(e)lijk, as in ondraag·lijk ‘unbearable’ are not productive (Booij 2002). As a consequence, *twitter·lijk is completely out. But also syntactic rules and constructions can be productive to varying degrees, since they can be applied to a range of words (which fill one or more slots), including neologisms. For instance, the transitive construction X+Verb+Y (e.g. to eat an apple), which accepts many verbs, is far more productive than the nominative-genitive construction in Modern German, which is restricted to only a handful of verbs (Barðdal 2008: 150), one of which is gedenken (Wir gedenken der Opfer ‘We remember the victims’).
Productivity is an abstract property of linguistic structures that forms part of the implicit knowledge speakers have about a language. Not only does it play a fundamental role in synchronic language description, it is also a crucial concept in language change (i.a. Hilpert 2013, Traugott & Trousdale 2013, Perek 2016) and language acquisition (Tomasello 2003; Hartsuiker & Bernolet 2017). Up until the present day, however, the phenomenon of productivity is poorly understood, especially with regard to syntactic constructions.
The workshop aims to address the following questions:
1. What are the different guises of productivity?
Productivity is a theoretical construct that can only be observed indirectly, i.e. when it is “at work”. Traditionally, corpus linguists have inferred productivity from the sum of utterances produced by the speakers of a language, i.e. from language usage (e.g. Baayen 1991, Zeldes 2012). Yet, how do these usage data match with language intuitions as measured via acceptability ratings? And how are they processed in on-line adult speech, both in production and comprehension? And what about speaker-related differences?
2. How can we measure (aspects of) productivity?
Some of these conceptual problems boil down to more technical issues about the interrelations between corpus-based metrics (such as type/token ratio, hapax/token ratio, hapax/type ratio, etc.) and how they capture distinct dimensions of productivity (i.a. Baayen 1992, 2001, 2009; Barðdal 2008; Zeldes 2012; Van Wettere 2021). Diachronically, an intriguing question is that of how type frequency relates to (changing) (token) frequency of each of the types and of the construction as a whole (cf. Feltgen 2020). Another important question is how extensibility is affected by strong entrenchment of a very small number of highly token-frequent fillers, which might be considered a factor of anti-productivity (cf. Barðdal 2008: 49, referring i.a. to Bybee 2001: 118–126).
3. What are the differences between morphological and syntactic productivity?
4. How does semantics affect productivity?
What is the role of semantic similarity to already attested fillers (Barðdal 2008, Suttle & Goldberg 2011)? And what about the impact of semantic variability (cf. Goldberg 2019), including the semantic range and density of the filler spectrum?
Other topics might also be proposed, as long as they are relevant for a better understanding of productivity.
A more detailed description can be found at our webpage.
Call for Papers:
We invite submissions for 20-minute oral presentations (plus 10 minutes for discussion). The conference language is English.
Abstracts should be fully anonymous and clearly state the research question(s), approach, method, data, and (expected) results. They should not exceed 500 words, excluding data, figures, and references. All submissions will be reviewed anonymously by at least two reviewers.
Peter Lauwers, Universidad de Gante
Jóhanna Barðdal, Universidad de Gante
Timothy Colleman, Universidad de Gante
Renata Enghels, Universidad de Gante
Rob Hartsuiker, Universidad de Gante
Miriam Taverniers, Universidad de Gante
Ludovic De Cuypere, Universidad de Gante
Anne-Sophie Ghyselen, Universidad de Gante
Universidad de Gante