What We Talk about When We Talk about Paradigms: Representing Latin Word Formation

Word and Paradigm approaches to derivational morphology have been increasingly discussed in the last three decades (since Bauer 1997, but also Jackendoff 1975). However, notwithstanding the many examples of the evidence of paradigmatic derivation described in Štekauer 2014, it is still not clear whether it is possible to envisage larger, all-embracing paradigms. These would be essential to both deeply account for and predict recurrent patterns in a language. In our work, we are employing data contained in the Word Formation Latin (WFL) lexicon (Litta et al. 2016) to determine what need to be the main requirements for the physical representation of a derivational paradigm. The method employed to build the WFL lexicon has raised a number of problems (Budassi & Litta 2017): derivational families are represented via tree-graphs, and word formation rules are modelled as directed one-to-many input-output relations between lexemes. On the other hand, the WFL database contains relevant information on all derivational patterns in Latin, which makes it a precious source of data for our purposes. We decided to represent the paradigm in the shape of a table, highlighting the importance of the “cell”. In addition, we adopted aspects of Construction Morphology (CxM, Booij 2010) in order to add information to the cell. The system is thus enhanced with both morphological and semantic information. In this paper, we are going to explain which concepts borrowed from CxM suit the description of the function of a paradigm cell, and which ones we need to forgo in order to account for Latin's variety of development throughout time and space.