An interdisciplinary workshop on understanding and explaining syntactic variation
Hosted by the Radboud University Nijmegen, November 15-17, 2012
Joan Bresnan (Stanford University)
Adele Goldberg (Princeton University)
Sali Tagliamonte (University of Toronto)
Antal van den Bosch (Radboud University Nijmegen)
Syntactic variation concerns the alternation between constructional alternatives such as He gave the boy the book and He gave the book to the boy. Syntactic variation research investigates the factors which determine why one of these alternatives is preferred over the other in specific linguistic and situational contexts.
Syntactic alternations are investigated in very different schools of linguistics and psychology, which each produce valuable results and predictions. Many of these findings, however, are unknown to colleagues in other fields, because they are deemed theoretically uninteresting, or because they are based on evidence which is inaccessible to non-initiated colleagues. Also, the best models and the most accurate predictions so far have been the result of a methodological interdisciplinarity which does not widely exist yet (discounting some rare exceptions like Joan Bresnan’s Spoken Syntax Lab at Stanford). This workshop therefore convenes the best people in the field to collaborate across disciplinary borders.
No matter how disjointed the field of present-day syntax, these are extremely exciting times for syntacticians. The past decade has witnessed revolutionary theoretical innovation in combination with a rapidly increasing quest for more (and more diverse) quantitative data and analytical methods.
This changing attitude is perhaps most visible in the fact that the generative community is becoming increasingly data-conscious. After four decades of insistence on rule-based analyses on the basis of subjective data, many generative theoreticians are beginning to embrace experimental methods of data collection, such as Magnitude Estimation (Bard, Robertson, and Sorace 1996; Featherston 2007). This has made generative theories testable, and generative research more cumulative and incremental.
Generative grammar is also overcoming its penchant for mono-factorial explanations – whereby one factor is claimed to motor constructional preference. The increasingly influential belief in much syntactic work that syntactic choice is probabilistically and multi-factorially determined (viz. by more than one non-categorical factor or constraint) has lead to the emergence of Optimality Theory (OT), which proposes that the observed forms of language arise from the interaction between competing constraints. The introduction of bidirectional OT at the turn of the millennium was very much pioneered by Dutch linguists such as Helen de Hoop (Nijmegen) and her students. Bidirectional OT has lead to important new insights into, amongst others, case marking (De Swart 2001), but De Hoops controversial claim that the ungrammatical use of object pronouns in subject positions in Dutch (Van Bergen et al. 2011) is a syntactic optimization (instead of an anomaly) also builds on bidirectional OT.
Sociolinguistics has from its inception on insisted on multi-factorial explanations, albeit with a view to studying how society impacts syntactic choices rather than uncovering their cognitive architecture. Until recently the sociolinguistic enterprise was almost exclusively concerned with uncovering the demographic and situational correlates of syntactic preferences, which often lead to theoretically underspecified analyses of limited cognitive relevance. The past decade, however, has witnessed a theoretical and methodological renaissance, which was pioneered in many ways by Sali Tagliamonte, the most prolific and influential sociosyntactician. In her seminal (2012) book, Tagliamonte significantly widens the field of variationist sociosyntax to include theory building which goes well beyond the original generative focus. Tagliamonte and Baayen (submitted, available on her website) illustrates a second of Tagliamonte’s prior concerns, the quest for the optimum tool for sociosyntactic analysis.
One of the theories to which sociolinguists are increasingly attracted is Exemplar Theory. While generative theory is rule-based to the extent that it conceives of grammar as a set of rules which combine stored items into larger structures, exemplar theorists hold that memory-stored lumps of previous language experiences – sounds, words, idioms, up to complete structures – are used as a reference point for the generation of new output. Rather than compositionally created out of a closed set of stored words, “grammar arises as a set of analogical generalizations over stored chunks of previously experienced language – lexical phrases or constructions – which are used to build new expressions analogically” (Bresnan and Hay 2008: 256).
In addition to pioneering Exemplar Theory in syntax, Joan Bresnan’s Spoken Syntax Laboratory at Stanford University continues to be a major centre for collaborative work on syntactic variation using multiple sources of evidence. Bresnan interacts with linguists, psychologists, and computer scientists to develop and test probabilistic models of how grammar varies in speaker groups across space (Bresnan and Hay 2008) and time (Wolk et al.: submitted). In this sense she is exemplary for an interdisciplinary interaction which is still rare at this moment. In The Netherlands, Grondelaers’ work on syntactic variation in existential constructions – the presence of absence of existential er “there” in sentences such as In de wiskunde zijn (er) problemen “In mathematics there are problems” – exemplifies a similar interdisciplinarity in that it builds on both laboratory evidence and corpus-based regression analysis (see especially Grondelaers et al. 2008, and Grondelaers et al. 2009).
In functionally oriented schools of syntactic thought (such as Cognitive Linguistics), exemplar theory surfaces in Construction Grammar, an analytic model which abandons the distinction between lexicon and syntax, because both words and constructions are believed to be memory-stored and meaningful. Construction Grammar has developed into an increasingly popular school of thought (elaborated in such seminal volumes as Goldberg 2006) with a number of associated quantitative methodologies (Stefanowitsch and Gries 2003). In The Netherlands, Construction Grammar is represented in the work of Arie Verhagen (2005).
In the last decennium, two promising new data sources have enriched the study of syntactic variation, although they are not used on a wide scale yet. Language technology, to begin with, has evolved sufficiently to test and compare some of the predictions of exemplar and rule-based theory. Researchers in The Netherlands have spearheaded many of the new functionalities. Walter Daelemans and Antal van den Bosch (University of Nijmegen), for instance, have developed memory-based learning (see Daelemans and Van den Bosch 2005 for the theoretical background; the learning algorithm itself is implemented in the Tilburg Memory Based Learner or TiMBL). Building on TiMBL, Theijssen et al. (submitted) have shown that memory-based learning makes predictions about the dative alternation (I gave him the book vs. I gave the book to him) which are no less accurate than manual and automatic multi-factorial (regression) analyses.
Neurolinguistic data such as event-related brain potentials (ERP) are even sparser in theoretical linguistic work, but since they were introduced in linguistics at the beginning of the Eighties, ERP-data are increasingly used as dependent variables of the semantic or syntactic conditioning of grammatical variation. Every content word elicits a measurable N400 component (so called because it peaks at 400 ms. after word onset) with an amplitude which increases when a word is harder to relate to its sentence-semantic context (glass in “I eat my soup with a glass” would elicit an N400-effect), while the P600-effect, by contrast, reflects syntactic violations. Building on ERP-data, Hagoort (2009) has addressed the “neural infrastructure for parsing and encoding”, but ERP-data are also used at Hagoort’s Donders Institute at the University of Nijmegen to study the correlation between contextual probability and information-structural focus status (Wang et al. 2010).
There a number of reasons for the fact that theoretical linguists so rarely interact with empiricists/experimentalists in psycholinguistics, neurolinguistics and language technology. All the disciplines involved, to begin with, have their own publication and funding channels with “rules of the game” one has to abide by in order to get published or funded in them. Experimental psychology, for instance, is not keen on theoretical elaboration in journal articles (having always “overplayed the role of research methods at the expense of theory building” (Brysbaert & Rastle 2009: 332)). And since language technology is one of the few linguistic paradigms which can attract attention and funding from industrial partners, language technologists cannot always be bothered to test the predictions of their colleagues in theoretical syntax. At the same time, neurolinguists and language technologists often complain about the shortage of theoretical predictions which are sufficiently accurate for neuro-imaging or computer modelling. The projected workshop is therefore organized with the specific ambition to provide a forum in which syntactic theorists and empiricists can acquaint themselves with each others’ work.
The confrontation between the different disciplines in which (variationist) syntax research is conducted increasingly reveals theoretical and methodological concerns in many multi-factorial analyses. Construction grammarians and psycholinguists, for instance, have largely ignored the sociolinguistic (demographic and situational) conditioning of the variation they study (see Heylen et al. 2008 for the constructionist neglect of sociolinguistic factors). Sociosyntacticians, conversely, seem to be satisfied when the syntactic variable they study can be shown to pattern with a number of language-structural and (especially) social factors, but they are not primarily interested in the origin and nature of the language-structural factors, and they rarely bother whether, and to what extent factors interact, and how much variation they account for.
Bard, Ellen G., Dan Robertson & Antonella Sorace. 1996. Magnitude estimation of linguistic acceptability. Language 72, 32-68.
Bresnan, Joan and Jennifer Hay. 2008. Gradient grammar: An effect of animacy on the syntax of give in New Zealand and American English. Lingua 118, 245–259.
Brysbaert, M. & K. Rastle. 2009. Historical and Conceptual Issues in Psychology. London: Pearson Education.
Daelemans, Walter and Antal van den Bosch. 2005. Memory-Based Language Processing. Cambridge: Cambridge University Press.
De Swart, Peter. 2011. Sense and simplicity: Bidirectionality in differential case marking. In Anton Benz and Jason Mattausch (eds.), Bidirectional Optimality Theory, 125-150. Amsterdam and Philadelphia: Benjamins.
Featherston, Sam. 2007. Data in generative grammar: the stick and the carrot. Theoretical Linguistics 33, 269–318.
Goldberg, Adele. 2006. Constructions at Work: the nature of generalization in language. Oxford: Oxford University Press.
Grondelaers, Stefan, Dirk Speelman and Dirk Geeraerts. 2008. National variation in the use of er “there”. Regional and diachronic constraints on cognitive explanations. In Gitte Kristiansen and René Dirven (reds.), Cognitive Sociolinguistics: Language Variation, Cultural Models, Social Systems, 153-204. Berlin: Mouton de Gruyter.
Grondelaers, Stefan. 2009. Woordvolgorde in presentatieve zinnen en de theoretische basis van multifactoriële grammatica. Nederlandse Taalkunde 14, 282-299.
Hagoort, Peter. 2009. Reflections on the neurobiology of syntax. In D. Bickerton, & E. Szathmáry (Eds.), Biological foundations and origin of syntax, 279-296. Cambridge, MA: MIT Press.
Stefanowitsch, Anatol and Stefan Th. Gries. 2003. Collostructions: Investigating the interaction between words and constructions. International Journal of Corpus Linguistics 8, 209-43.
Tagliamonte, Sali. 2012. Variationist Sociolinguistics. Change, Observation, Interpretation. Wiley-Blackwell.
Theijssen, Daphne, Louis ten Bosch, Lou Boves, Bert Cranen and Hans van Halteren. Submitted. Choosing alternatives: Using Bayesian Networks and memory-based learning to study the dative alternation.
Van Bergen, Geertje, Wessel Stoop, Jorig Vogels and Helen de Hoop. 2011. Leve Hun! Waarom hun nog steeds hun zeggen. Nederlandse Taalkunde 16, 2–29.
Verhagen, Arie. 2005. Constructions of Intersubjectivity. Discourse, Syntax, and Cognition. Oxford: Oxford University Press.
Wang, L., Bastiaansen, M. C. M., Yang, Y., & Hagoort, P. 2011. The influence of information structure on the depth of semantic processing: How focus and pitch accent determine the size of the N400 effect. Neuropsychologia, 49, 813-820.
Wolk, Christoph, Joan Bresnan, Anette Rosenbach, and Benedikt Szmrecsanyi. Submitted. Dative and genitive variability in Late Modern English: Exploring cross-constructional variation and change.
Please send an anonymous abstract (500 words without references) which pertains to (one of) the topics discussed in the Background and Specific Topic sections above to S.Grondelaers@let.ru.nl. Precedence shall be given to contributions with an interdisciplinary and (theoretically or empirically) innovative focus.
Please specify your name, affiliation and contact details in the message body and append your anonymous abstract to the message. Abstracts are due on September 30, 2012 and will be reviewed by the workshop organizers.
Stefan Grondelaers (Radboud University Nijmegen)
Roeland van Hout (Radboud University Nijmegen)
September 30, 2012 Abstracts due
October 10, 2012 Notification of acceptance
November 1, 2012 Final abstracts due for inclusion in workshop programme
November 15-17, 2012 Workshop
From the invited and submitted contributions, 15 papers will be selected for inclusion in an edited volume. The workshop organizers are currently negotiating a volume in the series Studies in Language Variation with Benjamins.