• Linguistic complexity in 20th-century American stage and screen musicals; A comparison of 1930-1949 Broadway, Hollywood and Adaptation song lyrics’ lexical density

This essay concentrates on differences in linguistic complexity among three categories of musicals: Broadway-originating, Hollywood-originating, or Adapted from stage to screen. Concretely, motivated by previously, mostly qualitatively observed variations in the nature of artistic pursuits according to their widespread, the study inspects whether discrepancies or resemblances exist, in terms of lexical density (type-token ratios), between song lyrics of 1930 to 1949 American musicals, mainly according to their popularity and their Broadway, Hollywood or Adaptation provenances. In that regard, remarkable incentives for the investigation are also the particular interest of the 20th century’s theatre and cinema businesses as revolutionised by sound technology, the lack of formal linguistic perspectives in the bibliography comparing them, and ongoing debates on the connections between lexical complexity and community size.

Hypothesising that, linked to their smaller audience, Broadway lyrics might display higher complexity levels than Hollywood ones (with Adaptations in-between), a representative corpus of 188 tunes is compiled, treated and statistically measured to allow a meaningful quantitative analysis that, ultimately, can confirm or deny such initial assumptions. Striving for historical accuracy and analytical significance, the definition of the sample texts is stratified, whereas the inferential statistical tests realised are predominantly non-parametric.

Among them, those focusing on the interplay between musical categories (x) and lexical density (y) assert the negligibility of any of their interactions (e.g. Kruskal Wallis p- value of 0.1019 > 0.05), whereas the ones addressing potential links between popularity indicators (predictors) and type-token ratios (outcome) also evidence the weakness and consequent insignificance of these relations (e.g. Spearman correlation general p- value of 0.5101 > 0.05).

The project’s implications, however, lie beyond the impossibility of rejecting the null hypothesis of such interrelations’ triviality: most notably, for example, doubts are cast on the extent to which formerly qualitatively perceived divergences between the scrutinised categories are quantitatively patent in linguistic metrics, as well as on the predictive strength of popularity, seen as parallel to community size, for intra- and cross-categorical lexical density variance.