Direct speech annotation for ELTeC-fra

Sampling

For each of the 100 novels in ELTeC-fra, 10 sentences have been randomly sampled, for a total of 1000 sentences or around 18600 words.

Annotation

Each sentence has been annotated with one of four speech categories:

• n = narrator speech, including indirect character speech
• c = direct character speech
• x = mixed speech, including character and narrator speech
• u = undecidable / other, e.g. free indirect speech or thought as well as letters.

No annotation guidelines, one annotator, no inter-annotator agreement checks.

Visualization

• Overall percentages of the four categories in the sample
• Percentages for the four categories by decade.
• Sentence length by speech category.