Exploring crosslinguistic and genre variation in syntactic complexity: insights from a multilingual parallel corpus

Abstract Syntactic complexity varies not only between languages but also across textual genres, reflecting structural, stylistic, and communicative factors. This study explores the interplay of these two dimensions by analyzing six syntactic complexity metrics (SCMs) across 17 languages and four genres in a large multilingual corpus. We examine how SCMs correlate across languages and genres and assess which metrics best predict language versus genre. Correlation analysis reveals two robust metric clusters – clausal and phrasal. Inter-metric correlations are high in legal texts and movie subtitles, but substantially lower in parliamentary proceedings and fiction, likely due to greater internal variation in syntactic complexity within these genres. Using random forest classification, we show that SCMs strongly predict genre, with noun phrase metrics ranking highest. In contrast, mean dependency distance (mdd) and its standard deviation are most predictive of language. Misclassification tendencies confirm the affinity of English with Romance languages, previously observed in lexical studies. By emphasizing not only central tendencies (mean values) but also the dispersion of metrics within texts, this study adds methodological nuance to the measurement of syntactic complexity. It also confirms the need to account for both language and genre in crosslinguistic syntactic complexity research.