A fundamental assumption regarding spoken language is that the relationship between sound and meaning is essentially arbitrary. The present investigation questioned this "arbitrariness" assumption by examining the influence of potential non-arbitrary mappings between sound and meaning on word learning in adults. Native English-speaking monolinguals learned meanings for Japanese words in a vocabulary-learning task. Spoken Japanese words were paired with English meanings that: (1) matched the actual meaning of the Japanese word (e.g., "hayai" paired with "fast"); (2) were antonyms for the actual meaning (e.g., "hayai" paired with "slow"); or (3) were randomly selected from the set of antonyms (e.g., "hayai" paired with "blunt"). The results showed that participants learned the actual English equivalents and antonyms for Japanese words more accurately and responded faster than when learning randomly paired meanings. These findings suggest that natural languages contain non-arbitrary links between sound structure and meaning and further, that learners are sensitive to these non-arbitrary relationships within spoken language. (Contains 1 figure.)