Animacy in verschillende teksttypes

Abstract

Although it is well known that in canonical transitive sentences the subject is animate and the object is lower in animacy than the subject, corpus research does not always corroborate these findings. Probably the most influential corpus study that investigated the use of animacy in transitive sentences is the study of Dahl and Fraurud (Dahl and Fraurud, 1996) on Swedish. They report a preference for constructions with animate subjects and inanimate objects (47,7%). This preference follows the markedness principles of Comrie with a subject that is higher in animacy than the object. Remarkably though, the percentage of constructions that violated this markedness principle was higher than the percentage of unmarked structures. The question arises what causes this pattern. Therefore we replicated the study of Dahl en Fraurud. In our comparative corpus study we used four different corpora that were either newspaper articles or literature texts intended for adults or children. This made it possible to investigate how the use of animacy in transitive relations is affected by genre or target group. Furthermore, the data were analyzed with respect to different animacy classifications: (i) a lexical analysis, in which only lexical items directly representing humans were qualified as animate; (ii) a referential analysis, in which lexical items that represented (a group of) human beings were also qualified as animate (e.g. committee). Results showed an animate subject-inanimate object preference, which was independent of the animacy classification. The lexical analysis revealed an influence of genre (literature vs. newspapers) as well as target group (adults vs. children). In contrast to the findings of Dahl and Fraurud, who used a corpus with mainly textbook chapters, newspaper and scientific articles, our literature corpora showed higher occurrences of sentences with two animate NPs than with two inanimate NPs. For newspapers, however, a similar pattern was found as in Dahl and Fraurud’s study with more sentences containing two inanimate NPs than two animate NPs. For the referent analysis, however, the pattern for the newspaper corpora was reversed: both genres showed a higher incidence of sentences with two animate NPs. With the animacy classification based on the referent instead of the lexical item similar patterns for all corpora were observed. Differences between genre and target group were leveled out and could no longer be observed. Since the pattern revealed high incidences of unmarked structures as well as relative high incidences for sentences with two animate arguments we concluded that the universal preference for animate subjects is confirmed. Furthermore, we argue that the referent analysis is the analysis that best reflects the processing of the reader or listener.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 31 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.