Menzerath-Altmann Law for Word Length Motifs

Motifs are relatively new linguistic units which make possible an in-depth investigation of sequential properties of texts (for the general definition cf. Köhler, this volume, pp.89-90). They were studied in a handful of papers (Köhler 2006, 2008 a,b, this volume, pp.89-108; Köhler and Naumann 2008, 2009, 2010, Mačutek 2009, Sanada 2010, Milička, this volume, pp.133-145). Specifically, a word length motif is a continuous series of equal or increasing word lengths (measured here in the number of syllables, although there are also other options, like, e.g., morphemes).In the papers cited above it is supposed that motifs should have properties similar to those of their basic units, i.e., words in our case. Indeed, word frequency and motif frequency, as well as word length (measured in the number of syllables) and motif length (measured in the number of words) can be modelled by the same distributions (power laws, like, e.g., the Zipf-Mandelbrot distribution, and Poisson-like distributions, respectively; cf. Wimmer and Altmann 1999). Also the type-token relations for words and motifs display similar behaviour, differing only in parameters values, but not in models.We enlarge the list of analogous properties of motifs and their basic units, demonstrating (cf. Section 3.1) that for word length motifs also the Menzerath-Altmann law(cf. Cramer 2005; MA law hence forth) is valid. The MA law describes the relation between sizes of the construct, e.g., a word, and its constituents, e.g., syllables. It states that the larger the construct (the whole), the smaller its constituents (parts). In particular, for our data it holds the longer is the motif (in the number of words), the shorter the mean length of words(in the number of syllables) which constitute the motif. In addition, in Section 3.2 we show that for randomly generated texts the MA law is valid as well, but its parameters differ from those obtained from real texts.