The linguistic organization of grammatical text complexity: comparing the empirical adequacy of theory-based models

Although there is a long tradition of research analyzing the grammatical complexity of texts (in both linguistics and applied linguistics), there is surprisingly little consensus on the nature of complexity. Many studies have disregarded syntactic (and structural) distinctions in their analyses of grammatical text complexity, treating it instead as if it were a single unified construct. However, other corpus-based studies indicate that different grammatical complexity features pattern in fundamentally different ways. The present study employs methods that are informed by structural equation modeling to test the goodness-of-fit of four models that can be motivated from previous research and linguistic theory: a model treating all complexity features as a single dimension, a model distinguishing among three major structural types of complexity features, a model distinguishing among three major syntactic functions of complexity features, and a model distinguishing among nine combinations of structural type and syntactic functions. The findings show that text complexity is clearly a multi-dimensional construct. Both structural and syntactic distinctions are important. Syntactic distinctions are actually more important than structural distinctions, although the combination of the two best accounts for the ways in which complexity features pattern in texts from different registers.