From Lemmas to Dependencies: What Signals Drive Light Verbs Classification?

TL;DR

This paper investigates Turkish light verb constructions (LVCs), a challenging class of Turkish MWEs where idiomatic predicate meaning can align with literal verb–argument structure. It tests whether signals for LVC classification come from lemma identity, morphosyntactic configuration, or full surface context by comparing lemma-based, grammar-based, and full-input models using a UD-derived supervision and a controlled diagnostic set. The main contributions show that coarse morphosyntactic cues alone do not robustly identify LVCs, lemma identity provides stronger signals but requires careful calibration, and lemma normalization introduces distribution shifts that complicate 'lemma-only' representations. These findings underscore the need for targeted, split-wise evaluation in Turkish MWE processing and suggest that lemma-only representations are not a single, fixed input form.

Abstract

Light verb constructions (LVCs) are a challenging class of verbal multiword expressions, especially in Turkish, where rich morphology and productive complex predicates create minimal contrasts between idiomatic predicate meanings and literal verb--argument uses. This paper asks what signals drive LVC classification by systematically restricting model inputs. Using UD-derived supervision, we compare lemma-driven baselines (lemma TF--IDF + Logistic Regression; BERTurk trained on lemma sequences), a grammar-only Logistic Regression over UD morphosyntax (UPOS/DEPREL/MORPH), and a full-input BERTurk baseline. We evaluate on a controlled diagnostic set with Random negatives, lexical controls (NLVC), and LVC positives, reporting split-wise performance to expose decision-boundary behavior. Results show that coarse morphosyntax alone is insufficient for robust LVC detection under controlled contrasts, while lexical identity supports LVC judgments but is sensitive to calibration and normalization choices. Overall, Our findings motivate targeted evaluation of Turkish MWEs and show that ``lemma-only'' is not a single, well-defined representation, but one that depends critically on how normalization is operationalized.