Rhythm in Prose and the Serial Correlation of Sentence Lengths: a Joyce Cary Case Study
Alan Parry Roberts · Literary and Linguistic Computing · 1996
The study of rhythm in prose does not seem to have examined regularities over any span longer than the few syllables corresponding to a poetic foot. Here the question is posed: can we detect in prose works a tendency to vary sentence–length in a regular (periodic) way? As a case study, extracts from three Joyce Cary novels were chosen, each told in the first person, with the narrators distinctly different characters. The sequence of sentence–lengths in each was examined for any tendency to periodicity (called here a‘beat’), by calculating the serial (‘auto’) correlation coefficient for lags from one to five. To be confident that a discovered beat was unlikely to be a chance event, the largest coefficient found was compared with the largest given by a sample random in its order of sentences. Each of the 100 ‘texts’ in the latter was formed by randomly shuffling the sentences of the original text into a new order, and then calculating its correlation coefficients. The usual serial coefficient was found, and also a ‘mean–side’ coefficient which, though it discards most of the information in the text's sentence–length distribution, proved more sensitive. The results showed a definite tendency to beats, the length of a sentence correlating with that of the sentence four before (two texts) or three before (one text). With the question about detectable periodicity thus answered, other questions arise for example, whether this is characteristic of Cary, or of an entire sub–culture, orI? One could of course extend the study of periodicity by looking for regularities in the appearance of other features of literary style, opening up the (highly speculative) possibility that something like an author's ‘characteristic spectrum’ might emerge.