Four experiments were conducted to explore the correlation between syllable number and visual complexity in the acquisition of novel words. In the first experiment, adult English speakers invented nonsense words as names for random polygons differing in visual complexity. Visually simple polygons received names containing fewer syllables than visually complex polygons did. In addition, analyses of English word-object pairings indicated that a significant correlation between syllable number and visual complexity exists in the English lexicon. In Experiments 2 and 3, adult English speakers matched monosyllabic novel words more often than trisyllabic novel words with visually simple objects, whereas trisyllabic matches were more common for visually complex objects. Experiment 4 replicated these findings with children, indicating that the assumption of a correlation between word and visual complexity exists during the period of intense vocabulary growth. Although the actual correlation between syllable number and visual complexity is small, other posited constraints on word meaning are also limited in strength. However, an increasing number of small, language-specific word-meaning correlations are being uncovered. Given the documented ability of speakers to detect and use these subtle correlations, we argue that a more fruitful approach to word-meaning acquisition would forgo the search for a few broad, powerful word-meaning constraints, and we attempt to uncover individually weak, but perhaps jointly powerful word-meaning correspondences.Throughout the centuries, introspections on the acquisition of word meanings have identified ostensive definition as the prototypical naming game. An adult directs a child's attention toward a certain region of the visual scene while simultaneously muttering some word or phrase. The child then infers that this sound is the linguistic symbol for the relevant feature of the visual scene, and encodes the associated pair into memory. If the child can correctly perform this inference once during every 1 or 2 waking hours between the 1st and 5th years of age, a normal kindergartener' s vocabulary will be attained (Carey, 1982).This apparently simple and intuitively workable naming game has received extreme criticism over the last three decades. Quine (1960) in particular argued that pointing toward an object while speaking a word or phrase does not imply an obvious inference about the meaning of the utterance. For instance, consider a situation in which a