uni-potsdam.de

Sie verwenden einen veralteten Browser mit Sicherheitsschwachstellen und können die Funktionen dieser Webseite nicht nutzen.

Hier erfahren Sie, wie einfach Sie Ihren Browser aktualisieren können.

Neural representations of lexical complexity

The lexicon is the product of the diachronic stratification of complex linguistic processes. Among these, morphological derivation offers a cross-linguistic relevant strategy, which, when synchronically productive, generates lexical items with new meanings, by combining existing morphemes. Yet, cross-linguistic neuroimaging literature suggested that derived words are accessed as whole forms in the bihemispheric fronto-temporal substrates supporting the semantic and pragmatic interpretation of non-combinatorial linguistic input, without requiring the left-lateralised perisylvian network specialized for grammatical analysis. In recent work using functional magnetic neuroimaging (fMRI), I re-examined these positions in the context of the combinatorially rich, root-based Italian lexicon, investigating how key linguistic coordinates, such as semantic transparency and affix productivity, affected the representational content of derived words in the language systems. The results revealed a dominant role of semantic transparency, modulated by affix productivity, in the neurocognitive encoding of derived forms. Only opaque forms showed cohort competition effects, consistent with earlier cross-linguistic evidence. However, behavioural and computational linguistic models of derivational complexity identified dissociable multivoxel representational patterns for opaquely and transparently derived words. The bilateral fronto-temporal activity associated with opaque words confirmed their non-compositional processing. At the same time, the left inferior frontal regions selectively reflected affix productivity in transparent forms, pointing to their decompositional representation. Combined univariate and novel multivoxel-pattern analyses, as performed here, are promising methods for detecting with an unprecedented detail the fine-grained representational properties of lexical complexity, which may vary across languages depending on their typological and historical characteristics.

Gehalten von

Dr. Francesca Carota

Veranstaltungsart

Gastvortrag

Sachgebiet

Forschungsschwerpunkt Kognitionswissenschaften
Linguistik

Universitäts-/ Fachbereich

Humanwissenschaftliche Fakultät

Termin

Beginn
09.02.2017, 11:00 Uhr
Ende
09.02.2017, 13:00 Uhr

Veranstalter

Potsdam Research Institute for Multilingualism

Ort

Universität Potsdam, Uni-Komplex Golm, Haus 5/ Raum 1.06
Karl-Liebknecht-Str. 24/25
14476 Golm
Lageplan

Kontakt

Judith Schlenter & Dr. Michaela Schmitz
Karl-Liebknecht-Str. 24-25
14476 Potsdam

Telefon 0331 977-2751