Impact of Word Predictability on Metrical Alignment in English Verse
Michael Hugh Kelly · 2024
Predictable, less informative words are reduced in duration in conversational speech. This study extends this pattern to the rhythmic structure of English verse. Study 1 shows that monosyllabic content and function words are more likely to appear in weak positions in Shakespeare’s verse when they are relatively predictable (e.g., high in frequency, high in conditional probability given the following word, and/or high in conditional probability given the previous word). Similar patterns appear in Milton’s verse in study 2, indicating that the patterns generalize beyond works like Shakespeare’s which were typically composed for live spoken performance. Study 3 confirms that the results in Studies 1 and 2 are not due to a small set of idiosyncratic words as they appear also in an analysis of word types in Shakespeare and Milton, in the process reinforcing that, similar to spoken prosody, the conditional probability of a word given the following word is the most influential of the predictability variables. These patterns in written verse provide one case where word predictability effects cannot readily be interpreted through ease of articulation, a common explanation for their presence in speech. Exemplar-based lexical models are discussed as an alternative, including further predictions about verse structure if correct. This study aligns with other research showing that the rhythmic structure of verse follows the same principles as spoken prosody, but extends the parallels to contextually-driven variation. It also adds to other research suggesting that, across domains, information value in a message is encoded by variables associated with prominence.