The Mental Lexicon

The aim of this workshop is to bring together (historical) linguists, psycholinguists, dialectologists, language technologists, literary scholars, philosophers and historians in order to discuss how we can advance our understanding of the structure of the lexicon in time, space and cognition by linking different data sets to each other. The assumption is that linking can best be done using ontologies or thesauri. Ontologies have gained a lot of interest with the rise of the semantic web and linked open data. Many researchers are currently designing ontologies, and classifying words according to sets of cognitive synonyms (synsets), semantic fields or semantic frames. The relations of words within semantic networks varies, depending on the ontology that is chosen.

                If we succeed in connecting entities (words) from different networks and ontologies to each other, we will be able to reveal similarities and differences in networks. This will shed light on how the mental lexicon is structured. We will for instance find answers to research questions such as: What kind of concepts are connected with many synonyms, and is this the same in various ontologies or does this vary? How can we explain this? How do words in a particular language divide the semantic space in different semantic fields and how does this division vary cross-linguistically? Do words learned at an early age, or words with a high familiarity rating, show more or less synonyms than words learned later or than less familiar words? How do linguistic semantic fields based on synonyms, hyponyms and hyperonyms in, e.g., Wordnet, relate to the thematical ordering of word associations?

                A fundamental problem within lexical research, encountered in all disciplines, is how to describe the semantic relations between words. In Natural Language Processing these relations are often measured by semantic similarity or synonymy (ship, boat) and semantic relatedness (ship is related to sailing and mast). In the workshop we will endeavour to find definitions of the concepts similarity and relatedness that concord with those used within humanities research, philosophy and psychology. We will discuss how the performance of automatic measurements can be improved by using pre-existing knowledge resources as (synonym) thesauri or encyclopedias, composed by humanities researchers, and by adding new resources, such as historical data, data from word associations networks and familiarity ratings.   The final subject that will be discussed in the workshop is semantic search. In semantic search a search query aims not to find keywords, but concepts: synonyms or semantically related words. Humanities researchers, especially those concerned with historical texts, have an urgent need for a tool that supports semantic search, since we often do not know what words were used for certain concepts in the past. The workshop will consider how the network of words, meanings and ontologies can be utilized to improve semantic search in modern and historical texts and across data sets.