Word length distribution in Brazilian‐Portuguese texts
Arne Ziegler · Journal of Quantitative Linguistics · 1996
In most languages it is a characteristic feature that lexical parts contain words of different length. Thus, the distribution of word length can be used to characterize texts and language itself. The present analysis tries to give a first contribution to the word length distribution of a variety of Portuguese, namely Brazilian Portuguese. It tries to find out the frequencies of different word lengths in Brazilian‐Portuguese press texts and whether the 1‐displaced mixed Poisson distribution can be considered as one suitable model of word length frequencies for Romance languages. The sources of all processed texts of the survey were two Brazilian newspapers, which take a stand on topics like Brazilian literature, language, culture and history. The examination defines specific criteria of measuring and counting word lengths in texts and modifies those criteria according to the characteristic features of the Portuguese language and concerning the Portuguese orthographical conventions. As one result the examination shows the fitting of the theoretical to the empirical distributions of the processed texts. The analysis provisionally tries to confirm the assumption of a non‐accidental distribution of word lengths in Brazilian Portuguese texts, on the one hand, and the hypothesis that texts of an individual language can follow different models, on the other.