Referring expressions in direct and indirect speech in Czech, English, German, and Norwegian

Barbara Hemforth (1) , Kaja Borthen, , Barbara Schmiedtova, (2) , Bergljot Behrens, (3) , Cathrine Fabricius-Hansen, (3)

1 LLF UMR7110 - Laboratoire de Linguistique Formelle
2 Institut für Deutsch als Fremdsprachenphilologie
3 ILOS - Department of literature, area studies and European languages [Oslo]

This paper presents the results of a psycholinguistic experiment on specific indefinites. Whereas previous research related to specificity has focused on indefinite antecedents of anaphora, our experiment concerns the possibility of using referential terms to report on speech events that involve indefinites. According to Sæbø (2011) "A referential term can represent an indefinite in a speech report if (i) the speaker had the referent of the term in mind but (ii) such a term would not have made sense to the hearer". In our experiment we tested to what extent informants choose to use a referential term to report on a speech event that (most likely) involved an indefinite description. The experiment was conducted to answer two main questions: 1) What nominal form is preferred in each language in the direct speech examples?, and 2) What effect does the change from direct to indirect speech have? For the first question, we predict a strong preference for indefinite NPs assuming that participants perform embedded audience design as requested in the narration. For the second question, we assume that the number of definites and proper names will increase across languages in indirect speech. Our results support the hypothesis that indefinites are preferred in the direct speech examples, however with a few more exceptions than expected, especially in Czech. The exceptional choices of definites in the direct speech examples can be attributed to the informant's ability to adjust the context into one where the crucial referent is uniquely identifiable or familiar to the original hearer. The difference between Czech and the three Germanic languages can also be explained, since Czech does not have a definite article and therefore involves a demonstrative determiner (not a definite article as in the Germanic languages) as the "definite" NP alternative in our experiment. Given that demonstrative determiners underlie different constraints than definite articles it is as expected that Czech behaves differently from English, German and Norwegian. In the indirect speech contexts, Norwegian and Czech do not show the expected increase in use of definites. We propose that this may be due to a more complex definite morpho-syntax in Norwegian, and by the fact that the Czech demonstrative, according to Cvrček et al. (2010: 141-143), is almost only used in spoken language, only rarely in written texts. The differences in the use of demonstratives in direct and indirect speech in Czech thus reflect usage preferences in different modalities. English and German, though, show a significant increase in the use of definites and a decrease in the use of indefinites in indirect speech, compared to direct speech - which is in accordance with our hypothesis and in support of the referential report hypothesis of Sæbø (2011). The use of referential expressions is thus governed by general principles of audience design, which may become fairly complex in the framework of a narration, but also by fine-grained language specific usage preferences. References Cvrček, V. et al. 2010. Mluvnice Současné Čestiny. Praha: Karolinum. Sæbø, K.J. 2011. The Specifics of Reported Speech and Thought. Presentation at Discourse Representation, Comprehension and Production in a Cross-linguistic Perspective, CAS, Osla, June 6-8,2011

Mots clés: Experimental Pragmatics; Referential Expressions

Domaines: Linguistique

Barbara Hemforth, Kaja Borthen,, Barbara Schmiedtova,, Bergljot Behrens,, Cathrine Fabricius-Hansen,. Referring expressions in direct and indirect speech in Czech, English, German, and Norwegian. International Conference on Nordic and General linguistics, Apr 2012, Freiburg, Germany. ⟨halshs-00770485⟩