Byzantine Construction Grammar: The Learnability and Acquisition of Medieval Greek Syntax, and the Prefabricated Style of Byzantine Literature

The recent publication of a textbook for the teaching of Greek (Miscellanea Laurentiana) has prompted a reconsideration of how linguistic knowledge is acquired and represented. Drawing on lexical priming, generative syntax, and construction grammar, this paper argues that Byzantine textbooks for learning Greek functioned as repositories of exempla and rules that collectively depict a linguistic system in which each lexical item contains stored syntactic information (‘features’) governing and constraining syntactic derivations. Within this framework, linguistic competence emerges as a network of stronger and weaker primings (Hoey 2005), whereby words and syntactic constructions are learned and stored in the brain as clusters of ‘lossy memory traces of formal patterns and their associated messages-in-context within their hyper-dimensional conceptual space’ (Goldberg 2019: 73)—where ‘lossy’ indicates that mnemonic traces are not complete and precise copies of all the utterances or words we have heard, but rather partial, simplified, and functionally effective abstractions. From this premise, this paper advances three hypotheses on Byzantine literary language. (1) The syntax—largely absent from Byzantine manuals, which focus overwhelmingly on vocabulary—was implicitly learned as one of the inherent, or (sociolinguistically) attributed properties of lexical items (Searle 1958). (2) Texts composed by authors trained in this grammatical tradition display a mixed texture of fixed bound phrases (depending on memory, sometimes syntactically peculiar or semantically noncompositional) and flexible bound phrases (generated anew in context, exhibiting regular syntax and semantics). (3) Mapping the conceptual associations, collocations, and bound phrases across Byzantine texts can illuminate their compositional methods, intertextuality, originality, and sociolinguistic dynamics. Using the case study of the database currently being developed by the ERC CoG MELA project, this paper contributes empirical evidence to these three working hypotheses and proposes a cognitive-linguistic model of Byzantine literary production grounded in learnability and constructional patterning.