Beskrivning
A growing body of evidence indicates that listeners constantly make probability estimates of the speech signal yet to unfold (Kuperberg & Jaeger, 2016; Kutas et al., 2011). Listeners appear to take advantage of a multitude of linguistic cues to predict upcoming features and structures (Hjortdal et al., 2022; Klein et al., 2012; Mitsugi, 2020; Roll et al., 2017; Roll et al., 2015; Salverda et al., 2014; Söderström et al., 2016).Classifiers are function words or morphemes which impose a classification on the nominal lexicon of a language, based on some shared semantic feature(s) (Aikhenvald, 2000; Seifart, 2010). It may therefore be hypothesized that classifiers, when preceding a noun, could aid the listener’s language processing by pre-activating a set of possible or likely referents. A series of studies investigating classifiers in Japanese and Mandarin Chinese suggest that listeners use classifiers predictively (Deng et al., 2022; Frankowsky et al., 2022; Huettig et al., 2010; Klein et al., 2012; Mitsugi, 2020). To our knowledge, only classifiers in East Asian languages have been investigated.
In the present study, we investigated the predictive function of classifiers in Baniwa [bwi], an Arawak language of Northwest Amazonia with approximately 3-4000 speakers (Aikhenvald, 2007). Baniwa has a set of around 50 classifier morphemes, obligatorily marked as suffixes on lower numerals. Numerals precede nouns in the noun phrase, thus making the Baniwa system suitable for studying the possible predictive function of classifiers. Classifiers in Baniwa mainly encode physical shape. Further, Baniwa also has a generic classifier, -da, which contrasts with the other classifiers in being compatible with a much larger and more diverse set of nouns.
We investigated whether classifiers in Baniwa facilitate the processing of following words either by pre-activating a set of noun candidates or certain semantic features. For this, we designed a response time study with a forced-choice component. Participants were presented with a verbal stimulus and two images on a screen, and were instructed to select the image that represented the verbal stimulus. The verbal stimulus consisted of a noun phrase including three components: the numeral ‘one’, a classifier (which is obligatorily marked on the numeral), and a noun. One of the presented images represented the noun in the noun phrase while the other represented either a competitor taking a different classifier (mismatch competitor) or a competitor sharing the same classifier with the noun in the noun phrase (classifier competitor). We hypothesised that if speakers use classifiers to make estimates about the identity of upcoming nouns, they would respond faster to target words in the mismatch condition compared to the classifier competitor condition.
Further, we hypothesised that classifiers which are more constraining regarding which nouns can follow result in faster response times. Classifiers were divided into two groups based on their (hypothesized) usefulness as predictors of upcoming nouns: shape and generic. Mandarin Chinese listeners have been reported to respond faster to non-generic classifiers compared to the generic classifier, gè (Frankowsky et al., 2022) and, in eye-tracking studies, fixate faster on those (Klein et al., 2012).
During the seminar, we will present preliminary results from the response time experiment carried out in Brazil in September and October this year.
| Period | 2022 nov. 17 |
|---|---|
| Vid | Allmän språkvetenskap |
| Omfattning | Lokal |