DFG project

PI: Prof. Dr. Nikolaus P. Himmelmann

Most theoretical and typological work in linguistics proceeds on the (usually implicit) assumption that differences between spoken and written language can safely be ignored for the purposes at hand such as modelling word order alternations or typologizing ditransitive constructions. A major consequence of this assumption is the fact that typological or theoretical work making use of data from languages without a writing tradition – the large majority of the world’s languages – is often exclusively based on written transcripts of spoken language. The current proposal asks whether this practice and the assumptions underlying it are sound. To answer this question, two interrelated issues need further investigation. 1. What exactly happens in transcription; that is, what decisions do native speakers and re-searchers (have to) make when representing spoken language in writing? To what extent do these decisions provide evidence for grammatical structures? 2. When does the fact that data from unwritten languages are essentially spoken language data actually matter for grammatical analysis? Is it possible to provide a principled delimitation of the set of phenomena where typical features of spoken language are clearly relevant for grammatical analyses, and ones where spoken language features can safely be ignored? The project concerns morphosyntactic analyses where attention to the specific features of spoken (as opposed to written) language is generally not part of standard analytical procedures. This includes morphosyntactic topics such as constituent structure, word classes and grammatical relations, i.e. topics primarily related to morphosyntactic form, not function. In contrast, it is clear, and widely acknowledged, that for topics such as deictics, it makes a major difference whether or not one includes features of spoken language such as prosody, gesture or gaze in the analysis. Importantly, the current investigation does not presuppose that the differences between written and spoken language – and the concomitant differences in generating primary data – are relevant for all types of typological and theoretical enquiry. Rather, it is the goal of the project to determine when and where exactly the difference matters, and in which ways the process of producing primary data (transcription) itself may generate important data for such enquiries.

Dutch Research Council (NWO)

PI: Dr. Eva van Lier

In linguistic typology, human communication is regarded as a key factor shaping the diversity of language structures world-wide. Yet, many typologists do not support their explanations with data from actual language processing. On the other hand, psycholinguists have looked at only a tiny proportion of all human languages, relying on an (implicit) assumption of linguistic universality. This study connects the fields of linguistic typology and psycholinguistics, focusing on alternating verbs. These are verbs whose arguments can be coded in more than one way, requiring speakers to choose between grammatical constructions. According to usage-based linguistic theory, such verb-argument constructions are not formed by abstract, general rules operating on the entire verbal lexicon, but rather emerge from speakers' experience with specific verbs in communicative context. In typology, however, verb-specific grammatical structures have mostly been treated as exceptions: until recently, whole languages were classified with a single value of a particular structural variable, e.g. as having a nominative-accusative case system. Yet, psycholinguistic priming experiments show that speakers access and use verb-specific grammatical knowledge in language processing. This project combines typological and psycholinguistic methods. First, we investigate cross-linguistic diversity in the grammatical behavior of alternating verbs, using spoken language corpora. Second, together with language specialists, we carry out experimental case studies on Kamang (Indonesia) and Chechen (Russia). These two languages have argument-coding properties that are almost non-existent in European languages. This project sheds light on the psychological reality, nature, and role of verb-specific grammatical knowledge, beyond the very few languages and constructions that have been studied experimentally to date. Thus, it shows how language processing and linguistic diversity are connected in verb-argument construction alternations world-wide.