Comparatively speaking : a psycholinguistic study of optionality in grammar

GRAMMATICAL OPTIONALITY refers to the ability to realize the same meaning using more than one grammatical expression, as in the dative alternation, or the choice between active and passive encodings of a transitive event. Optionality has been studied extensively in the syntactic domain, and has provided numerous insights concerning the kinds of factors that influence production, the functional benefits of speaker choice, and the nature of syntactic development. In the present work, I inaugurate the psycholinguistic study of morphological optionality by considering an English alternation that occurs relatively commonly among adjectives in the comparative, namely the ability to inflect SYNTHETICALLY (e.g., angrier, prouder), or ANALYTICALLY (e.g., more angry, more proud). This investigation is motivated based on the pervasive character of morphological optionality cross-linguistically, and the fact that the English comparative has been unjustifiably used by a number of analysts as evidence in favor of a phenomenon known as BLOCKING. Blocking analyses view inflectional class as a categorical matter. I demonstrate that this has the effect of erroneously predicting that optionality should not occur. After a preliminary introduction in Chapter 1 to the issues and phenomena under investigation, I proceed in Chapter 2 to a series of experiments on comparative comprehension. Using both offline (Experiments 1 and 3) and online (Experiment 2) methods, I show that, for adjectives that participate in comparative optionality, listeners roundly prefer the analytic variant, most likely because it allows for earlier phrase-structure recognition, and avoids a temporary parsing ambiguity associated with the synthetic. Chapter 3 is a production study conducted jointly with L. Robert Slevc. In this work, we use an elicited production technique (Experiment 3) to establish that speakers tend to increase their rate of use of the analytic variant under conditions of increasing syntactic and semantic complexity. We argue that there is no evidence to suggest that this is done as a means of helping lessen addressee processing loads. The behavior is, however, consistent with the notion that speakers choose between competing variants as a means of facilitating their own production processes. In Chapter 4, I consider how the analytic and synthetic patterns of comparative inflection are acquired. Experiment 4 demonstrates that, contra the results of previous work on comparative development, both patterns are subject to overgeneralization errors (e.g., *dangerouser, *more fast, cf. faster). The results of Experiment 5 suggest that the discrepancy between my findings and those of other researchers is likely due to their failure to adequately control for response perseveration. I conclude in Chapter 5 by providing an extensive critique of various blocking analyses of the English comparative. As an alternative, I suggest a gradient, pattern-based approach that is consistent with the assumptions of WORD AND PARADIGM morphology. This analysis avoids becoming mired in difficulties that afflict more syntactocentric treatments of comparative inflection, and allows categorical inflectional outcomes and optionality to be modeled in the same unified framewor