Semi-automatic construction of word-formation networks

Abstract

The article presents a semi-automatic method for the construction of word-formation networks focusing particularly on derivation. The proposed approach applies a sequential pattern mining technique to construct useful morphological features in an unsupervised manner. The features take the form of regular expressions and later they are used to feed a machine-learned ranking model. The network is constructed by applying the learned model to sort the lists of possible base words and selecting the most probable ones. This approach, besides relatively small training set and a lexicon, does not require any additional language resources such as a list of vowel and consonant alternations, part-of-speech tags etc. The proposed approach is evaluated on lexeme sets of four languages, namely Polish, Spanish, Czech, and French. The conducted experiments demonstrate the ability of the proposed method to construct linguistically adequate word-formation networks from small training sets. Furthermore, the performed feasibility study shows that the method can further benefit from the interaction with a human language expert within the active learning framework.

Swedish word family resource

“…Zeller, Šnajder & Padó, 2013;Šnajder, 2014). Derivational families exclude compounds and multi-word expressions, as do word formation networks (Lango, Žabokrtský & Ševčíková, 2021), even though the latter term suggests a broader view on word formation mechanisms. A range of different computational lexicons grouping vocabulary based on the principles of derivational families or word formation networks are described in Lango et al (2021), for instance, CELEX, 4 that exists for German (Baayen, Piepenbrock & Gulikers, 1996) and a few other languages; DerIvaTario for Italian (Talamo, Celata & Bertinetto, 2016); Derinet for Czech (Žabokrtský et al, 2016).…”

“…Hypothesis 2: (a) simpler words (i.e. words consisting of a minimal number of morphemes) within each family appear at earlier proficiency levels and are more frequent than more multimorphemic words in the family; and (b) relations between word family members are "ordered by morphemic complexity" through word formation mechanisms (cf Lango et al, 2021) which is reflected in receptive and productive data frequencies. Hypothesis 3: Extralinguistic aspects of language learning can be explored through word families, e.g.…”

“…Relations between word family members are "ordered by complexity"* through word formation mechanisms (cf. Lango et al, 2021) which is reflected in receptive versus productive data frequencies.…”

Universal Derivations 1.0, A Growing Collection of Harmonised Word-Formation Resources

“…DeriNet.ES is a DeriNet-like lexical database for Spanish which is based on a substantially revised lexeme set used originally in the Spanish Word-Formation Network (Lango et al, 2018). In DeriNet.ES, derivational relations were created using substitution rules covering Spanish affixation (Faryad, 2019).…”

“…The Polish Word-Formation Network is a DeriNet-like lexical network for Polish created by using pattern-mining techniques and a machine-learned ranking model (Lango et al, 2018). The network was enlarged with the derivational relations extracted from the Polish WordNet (Maziarz et al, 2016).…”