Prof. Dr. Joanna Błaszczak

Prof. Błaszczak erforscht, wie Wörter und sprachliche Strukturen in Texten nicht zufällig verteilt sind, sondern sich gegenseitig vorhersagbar machen. Sie untersucht linguistische Vorhersagbarkeit im Kontext — also wie das Auftreten eines Wortes die Wahrscheinlichkeit für andere Wörter in seiner Umgebung beeinflusst. Für Unternehmen und öffentliche Institutionen ist dieses Wissen relevant für die Verbesserung von Sprachverarbeitungssystemen, Textanalyse-Tools und automatischer Spracherkennung. Die Erkenntnisse ermöglichen präzisere Vorhersagemodelle in Bereichen wie Suchmaschinen, Chatbots, Dokumentenanalyse und maschinelle Sprachverarbeitung.

🔒 Das System hat 90 mögliche Industrie-Partner gefunden — Firmen, Scores und Begründungen sind nur für eingeloggte Nutzer:innen sichtbar. Anmelden

• Linguistische Vorhersagbarkeit im Kontext

  Quelle ↗

  Virtually never are linguistic expressions and features distributed over texts by pure chance – "Language is never, ever, ever, random" (Kilgarriff, 2005). Rather, the occurrence of a certain item (word form) guides expectations about the appearance of other items in its context and thus, makes these other items predictable at a certain probability. For theoretical linguistics, this raises the question where to locate such cooccurrence information: in grammatical principles, in constructions (structural patterns) represented as a whole or with variable slots, in syntax and/or in the lexicon. But cooccurrence patterns are also fascinating beyond pure linguistics, because they reveal pathways of writing and thinking in a language, and are very often characteristic of a given culture. This leads to the central research question for the project: How strong is the relation between regular cooccurrences in large text corpora on the one hand, and psycho-/neurolinguistic measures of collocation and selectional restrictions on the other?

• Das mehrsprachige Klassenzimmer

  2014

  DOI

• Kopulasätze mit den pronominalen Elementen "to/ėto" im Polnischen und Russischen

  2000

  ZAS Papers in Linguistics · DOI

  In this paper we investigate the structure of specificational sentences like [Raskol'nikov]NP 1 - ėto [ubìjca staruxi]NP2 'Raskolnikov - that is the murderer of the old lady' in Russian and Polish, which - depending on the type of NP1 and NP2 - correspond to English pseudo-cleft-constructions (What Raskolnikov is is the murderer of the old lady) and specificational sentences (The person I like most is my father), respectively. We propose that the Slavic constructions can be analysed similarly to their English counterparts: the first fragment contains a semantic variable, which is specified in the second fragment.
 
 We show that the pronouns "ėto" <Rus.> / "to" <Pol.>, which are obligatory in Slavic specificational sentences, have two functions. 1. the deictic function: "ėto/to" take an open proposition available in the discourse or reconstructed from it, and assign this open proposition to another proposition, which provides the value for the variable of the open proposition. 2. the operative function: "ėto/to" link two syntactically independent fragments, the first of which can be semantically interpreted as an indirect question comparable to the wh-clause in the English pseudo-clefts, and the second as an answer to this question.

• When verbs have bugs: lexical and syntactic processing costs of split particle verbs in sentence comprehension