A Quantitative and Typological Approach to Correlating Linguistic Complexity

The equal complexity hypothesis states that "all human languages are equally complex" (Bane, 2008). Menzerath's law is well-known for explaining the phenomenon of selfregulation in phonology: "the more sounds in a syllable the smaller their relative length" (Altmann, 1980). Altmann, who made the mathematical formula of this law (Forns and Ferrer-i-Cancho, 2009), assumed that it can be applied to morphology as well "the longer the word the shorter its morphemes" (Altmann, 1980) and proved that the clause length depends on sentence length (Teupenhayn and Altmann, 1984). Some previous works on morphological complexity (Bane, 2008; Juola, 1998) asserted that morphology is a good starting point for complexity computation for its clearness, compared to other more ambiguous domains such as semantics. The best-known method of calculating morphological complexity is to take the numbers of linguistic constituents into account (Bane, 2008; Moscoso del Prado, 2011), with different mathematical formula to be applied to these figures. The following two paradigms are commonly employed: i) information theory (Fenk et al., 2006; Moscoso del Prado et al., 2004; Pellegrino et al., 2011) ii) Kolmogorov complexity (Bane, 2008; Juola, 1998). The main goal of our work is to explore interactions between phonological and morphological modules by means of crossing parameters of these two linguistic levels. This paper provides preliminary results obtained from a corpus-based cross-language study.