Abstract
We study the distribution of parts-of-speech sequences in Ukrainian texts by Ivan Franko. The defined units demonstrate frequency behaviour resembling to a significant extent that of ordinary words, therefore, we refer to them as ‘part-of-speech words’ (PoSW). It is shown that Zipf’s law holds for them, with an exponent changed for ranks in the domain near the h-point. Lengths of these units are modelled using discrete and continuous distributions. For frequency spectra of PoSWs, a continuous model based on the non-additive modifications of ordinary exponentials is applied together with two discrete distributions. All the proposed models yield a good fit with the determination coefficients R2 > 0.99. We expect that further studies of the proposed PoSW units both in Ukrainian and other languages can reveal new features of texts on the sentence and supra-sentence levels.
Notes
1. Literally Stonemasons or Quarriers, translated as The Pioneers by Percival Cundy (see Manning, Citation1948, p. 104).