13. Mai 2025
Hagit Borer
Queen Mary University of London, Department of Linguistics
Reflecting on the Generative Word
Gemeinsame Veranstaltung mit dem Theoretical Linguistics Colloquium (dort findet am 12. Mai ein weiterer Vortrag statt: https://sites.google.com/view/totlvienna/2025-05-borer)
Abstract: An important aim within Generative Grammar has always been to formulate syntactic operations with maximal formal simplicity and generality. Given this aim, modelling the properties of words in general, and complex words in particular, remains a challenge. On the one hand, there’s much to suggest that words are independently listed with their unpredictable properties, such as phonological realization, meaning, argument array, syntactic category, and syntactic insertion context. Equally powerful, however, is the evidence that complex words have a hierarchical structure which is syntax-like, comprising, at the very least, headedness, complementation, and locality constraints. The tension between these sets of properties, and its implication for syntax has not only informed major developments within GG, but has also given rise to several major theoretical rifts, cantered directly on disagreements on the nature of words and their interaction with Syntax (Generative Semantics, Construction Grammar, Lexical Functional Grammar, to name a few).
By the end of the twentieth century most generative approaches were ‘lexicalist’ insofar as they were postulating a powerful lexicon, in which listed words controlled at least the following:
1. lexical semantic information, resulting in argument array and event properties
2. syntactic features and syntactic insertion frames (to the extent not derived from (1))
3. non-syntactic hierarchy building combinatorial operations of words and word parts (WF)
4. word (and sub-word) phonological properties
At the very same time, however, the early 90’s saw the emergence of diverse theoretical perspectives, which challenge the strict division between words and syntax, and which seek to postulate a single structure-building operation (call it Merge) involved in constructing both words and phrases (and possibly phonological constituents as well). In the past three decades, these perspectives have gained considerable force, to become a major contender to the lexicalist perspective. The primary challenge, to such approaches, was, and remains, the modelling of idiosyncratic word properties, including at the very least unpredictable phonological realization, non-compositional meaning, and argument array.
Beyond that primary challenge, there lies a bigger one: What are Words? Does the intuition that they ‘exist’ in a sense that a composed syntactic phrase does not, actually correspond to a theoretical or cognitive construct? Does the fact that our word use is largely drawn from a stored reservoir, but our phrasal use is largely formed on the fly, theoretically significant, and if so, how can it be modelled within a single Merge model?
In this presentation, I will discuss in greater detail all these developments within the history of GG and offer some speculations on what ‘words’ might be.