Russian prefixed verbs as constructional schemas

This study tests the morphological gradience theory on Russian prefixed verbs. With the help of a specially designed experiment, in which participants were asked to evaluate the semantic transparency of a prefixed nonse verb given in minimal context, as well as to semanticise it by suggesting an existing Russian verb with the same prefix, we offer evidence that these verbs can be analysed as constructional schemas and that the degree of their morphological decomposition depends upon the different levels of activation of their sequential and lexical links. We prove that speakers of Russian are very sensitive to the etymological connection between verb prefixes and the prepositions they are related to. Thus, prefix-stem constructions with prefixes that correspond to prepositions are more likely to be morphologically decomposed, while prefix-stem constructions with prefixes that do not relate to prepositions tend to be regarded as single lexical units. Moreover, the general, highly abstract semantics of Russian prefix-stem constructions, especially of those that retain their ‘prepositional’ meaning, is undoubtedly accessible to language users, which is confirmed by the fact that the interpretability of these constructions is affected by priming.

Neural Modeling for Named Entities and Morphology (NEMO2)

Named Entity Recognition (NER) is a fundamental NLP task, commonly formulated as classification over a sequence of tokens. Morphologically rich languages (MRLs) pose a challenge to this basic formulation, as the boundaries of named entities do not necessarily coincide with token boundaries, rather, they respect morphological boundaries. To address NER in MRLs we then need to answer two fundamental questions, namely, what are the basic units to be labeled, and how can these units be detected and classified in realistic settings (i.e., where no gold morphology is available). We empirically investigate these questions on a novel NER benchmark, with parallel token- level and morpheme-level NER annotations, which we develop for Modern Hebrew, a morphologically rich-and-ambiguous language. Our results show that explicitly modeling morphological boundaries leads to improved NER performance, and that a novel hybrid architecture, in which NER precedes and prunes morphological decomposition, greatly outperforms the standard pipeline, where morphological decomposition strictly precedes NER, setting a new performance bar for both Hebrew NER and Hebrew morphological decomposition tasks.

Finding a 'flower' in a 'peanut' is as easy as in a 'garden': towards a lemma-based model of bilingual word recognition

The present study investigates how morphological information is processed and represented in the bilingual lexicon. We employed a masked cross-language morphological priming paradigm to examine morphological decomposition and semantic transparency in bilingual lexical processing. A robust and reliable morphological priming effect was observed for both transparent compounds and opaque compounds, though there was a strong trend for more facilitation in the former than the latter. To account for these results, we propose a lemma-based bilingual model specifying the activation/competition between lemmas during cross-language activation at the morphological level. Our novel findings advance the understanding of interplay between morphology and bilingualism.