Exploring the measurement of markedness and its relationship with other linguistic variables

Antonym pair members can be differentiated by each word’s markedness–that distinction attributable to the presence or absence of features at morphological or semantic levels. Morphologically marked words incorporate their unmarked counterpart with additional morphs (e.g., “unlucky” vs. “lucky”); properties used to determine semantically marked words (e.g., “short” vs. “long”) are less clearly defined. Despite extensive theoretical scrutiny, the lexical properties of markedness have received scant empirical study. The current paper employs an antonym sequencing approach to measure markedness: establishing markedness probabilities for individual words and evaluating their relationship with other lexical properties (e.g., length, frequency, valence). Regression analyses reveal that markedness probability is, as predicted, related to affixation and also strongly related to valence. Our results support the suggestion that antonym sequence is reflected in discourse, and further analysis demonstrates that markedness probabilities, derived from the antonym sequencing task, reflect the ordering of antonyms within natural language. In line with the Pollyanna Hypothesis, we argue that markedness is closely related to valence; language users demonstrate a tendency to present words evaluated positively ahead of those evaluated negatively if given the choice. Future research should consider the relationship of markedness and valence, and the influence of contextual information in determining which member of an antonym pair is marked or unmarked within discourse.