Texts belonging to belles-lettres style are heterogeneous from the point of view of statistics. This is due to the fact that these texts contain the author’s speech and direct speech, both of them possessing formal distinctive features. The analysis of their syntactic structure is based on a corpus of 50,000 words taken from direct speech fragments.
The average sentence length in the direct speech (x̄=4.7±3.2words) is several times lower than in the author’s speech (x̄=12.3±9.9words). Very short sentences, consisting of 1–8 words (90.9 %) predominate; half of these are sentences having from 1 to 4 words. Most of them are simple sentences (79.9 %), their average length (x̄=3.6±2.2words) being much shorter than the average length of complex sentences (x̄=9.1±4.0words). Among the latter prevail complex sentences (35.5 %); asyndetically joined clauses (28.8 %) and compound sentences (26.8 %) are less dominant. Their average length is stereotypical: syndetically linked clauses consist of 8.2 words, asyndetically – 7.3 words.
Sentence length in the direct speech depends very much on the author’s individuality. Balancing sentence length in the author’s speech and in direct speech helps to ensure the stylistic uniformity of a text.