Retour vers liste

Détail de la contribution

Auteur: Gertraud FENK-OCZLON

Relationships between Semantic Complexity, Structural Complexity and Markedness: Frequency Matters

Abstract/Résumé: In view of the fundamental role of frequency in information processing and cognition this paper focuses on frequency effects on semantic complexity and structural complexity. It discusses, moreover, the relationship between the theoretical construct markedness and the complexity of linguistic forms. Starting from several definitions of semantic complexity it will be shown that any definition of semantic complexity implies strong connections to frequency or to frequency-based explanations. For example: Semantically less complex forms are “less specific”, having “less semantic features” but a “greater extension” (Andersen 2001). And semantically less complex entities are, according to Mayerthaler (1981), more frequent. A definition of polysemy and homophony as central facets of semantic complexity (Fenk-Oczlon & Fenk 2008) recalls the well-known association (Zipf 1949) between number of different meanings and high token frequency. In most of these conceptualizations higher semantic complexity is reflected in higher structural complexity (e.g. Lehmann 1974, Mayerthaler 1981). It will be argued that the correspondences between semantic complexity and structural complexity can easily be attributed to the general principle 'the more frequent or the more predictable a sign, the shorter it is' (Zipf 1949). In an understanding of semantic complexity as ambiguity (polysemy, homonymy), the relationship between semantic complexity and length of form is the exact opposite: There are strong associations between a sign’s number of meanings and its shortness. Again frequency matters: High frequency reduces the length of signs but renders them more irregular, less transparent and also more ambiguous. On the other hand, high frequency of use not only reduces structural complexity but also the respective cognitive costs (Fenk-Oczlon 2001). Concerning the relationships between structural complexity, semantic complexity and markedness I shall present arguments and empirical data from previous papers (Fenk-Oczlon 1991, 2001) that argue, similar to Haspelmath (2006), that the term markedness could be easily replaced by frequency. It will be shown that in cases where (semantic) markedness and frequency diverge, as e.g. in the Russian aspect-system, frequency proved to be the better predictor variable. Frequency is, moreover, a tangible variable whereas markedness is a theoretical construct. Interrelationships between the variables semantic complexity, structural complexity, frequency, and predictability will be discussed within the framework of the Constant Information Flow hypothesis (Fenk & Fenk-Oczlon 1993, Fenk-Oczlon 2001) and the closely related Uniform Information Density model (Jaeger 2010).