Profile-Based Text Classification for Children with Dyslexia

2. Linguistic Complexity And Reading Difficulty

Text readability is closely related to and even determined by the linguistic complexity of a text in the sense that the readability of a text increases as linguistic complexity decreases and vice versa. Therefore, linguistic complexity is a central notion when dealing with text classification.

Defining linguistic complexity is currently one of the most hotly debated notions in linguistics. In a first quantitative description of linguistic complexity, Blache (2011) identifies the types of constructions that are considered complex and thus difficult to process. She differentiates between local complexity, which refers to structural complexity, difficulty, which involves processing aspects and cognitive load, and global complexity, which refers to the language as a system rather than the complexity of a given realization (see Miestamo (2008) for a similar classification). Of the two levels, local complexity is considered measurable and has drawn considerable attention in the literature. Local complexity therefore includes phonological complexity (e.g. size of phonemic inventory, incidence of marked phonemes, phonotactic restrictions, maximum complexity of consonant clusters), morphological complexity (e.g. extent of allomorphy use and morphophonemic processes), syntactic complexity (e.g. level of clausal embedding and recursion), semantic and lexical complexity (e.g. extensive occurrence of homonymy and polysemy, type/token ratios), pragmatic complexity (e.g. degree of pragmatic inferencing) (see Szmrecsanyi and B. Kortmann (2012) for a review).