Stress Test for BERT and Deep Models: Predicting Words from Italian Poetry

TL;DR

This study evaluates BERT's ability to predict words in Italian poetry, revealing its sensitivity to noncanonical structures, word frequency, and compositional meaning, with a focus on subword units for out-of-vocabulary words.

Contribution

It provides a detailed analysis of BERT's performance on Italian poetic sentences, highlighting challenges posed by linguistic complexity and nonliteral language.

Findings

BERT is highly sensitive to noncanonical sentence structures.

Performance is affected by word frequency and local compositional effects.

Subword units help address out-of-vocabulary word prediction issues.

Abstract

In this paper we present a set of experiments carried out with BERT on a number of Italian sentences taken from poetry domain. The experiments are organized on the hypothesis of a very high level of difficulty in predictability at the three levels of linguistic complexity that we intend to monitor: lexical, syntactic and semantic level. To test this hypothesis we ran the Italian version of BERT with 80 sentences for a total of 900 tokens mostly extracted from Italian poetry of the first half of last century. Then we alternated canonical and noncanonical versions of the same sentence before processing them with the same DL model. We used then sentences from the newswire domain containing similar syntactic structures. The results show that the DL model is highly sensitive to presence of noncanonical structures. However, DLs are also very sensitive to word frequency and to local non literal meaning compositional effect. This is also apparent by the preference for predicting function vs content words, collocates vs infrequent word phrases. In the paper, we focused our attention on the use of subword units done by BERT for out of vocabulary words.