The many guises of productivity: a case-study of Spanish inchoative constructions

Abstract This study investigates the productivity of a family of inchoative constructions in (Peninsular) Spanish ( empezar a + inf., echar a + inf.) and the way it correlates with semantic openness. Productivity is defined as lexical openness of a slot (here the infinitive) of a construction. A range of productivity measures (e.g., type/token ratio, hapax/token ratio, and other parameters related to the frequency of the most frequent types) are discussed and compared. Semantic openness is operationalized on the basis of distributional semantics, yielding two measures viz. semantic range and semantic sparsity . The correlations between productivity measures and semantics are studied by means of a Principal Components Analysis. It is shown that the so-called ‘anti-productivity’ measures (related to entrenchment due to high token frequency) on the one hand, and the productivity measures (including the hapax-based ones, traditionally associated with extensibility) on the other hand, strongly correlate and constitute opposites along the same dimension of lexical openness. In addition, the semantic measures strongly correlate with productivity, which suggests that often semantic and lexical openness go hand in hand in this dataset. As to the behavior of the individual auxiliaries (= micro-constructions), they appear to be rather extreme, showing very productive and less productive profiles.