Analogies in Complex Verb Meaning Shifts: the Effect of Affect in Semantic Similarity Models

Maximilian Köper,Sabine Schulte Im Walde

doi:10.18653/v1/n18-2024

Abstract

We present a computational model to detect and distinguish analogies in meaning shifts between German base and complex verbs. In contrast to corpus-based studies, a novel dataset demonstrates that “regular” shifts represent the smallest class. Classification experiments relying on a standard similarity model successfully distinguish between four types of shifts, with verb classes boosting the performance, and affective features for abstractness, emotion and sentiment representing the most salient indicators.

Highlights

IntroductionGerman particle verbs are complex verb structures such as anstrahlen ‘to beam/smile at’ that combine a prefix particle (an) with a base verb (strahlen ‘to beam’)
German particle verbs are complex verb structures such as anstrahlen ‘to beam/smile at’ that combine a prefix particle with a base verb
While we focus on German particle verbs, we expect our explorations to be applicable to other types of meaning shifts or languages

Summary

Introduction

German particle verbs are complex verb structures such as anstrahlen ‘to beam/smile at’ that combine a prefix particle (an) with a base verb (strahlen ‘to beam’). They are highly ambiguous, and they often trigger meaning shifts of the base verbs (Springorum et al, 2013; Koper and Schulte im Walde, 2016). The goal of this work is to provide a computational model of meaning shifts for German particle verbs. We show that (a) there are variants of (ir)regular meaning shifts that go beyond what was found in corpus-based explorations; (b) generalisation via classification boosts the strengths of salient verb features; and (c) affective features (i.e., abstractness, emotion and sentiment) play the predominant role in similarity models of meaning shifts. Proceedings of NAACL-HLT 2018, pages 150–156 New Orleans, Louisiana, June 1 - 6, 2018. c 2018 Association for Computational Linguistics

A Collection of BV–PV Analogies

Representations of BV–PV Analogies

Basic Distributional Similarity Model

Generalisation Models

Affect Models

Experiments on BV–PV Analogies

Findings

Conclusion