By 24 months, children rapidly infer the meanings of words from sentence context: for instance, inferring from “She eats the dax” that “dax” refers to a food item. However, prior work has typically presented these informative sentence contexts in the presence of the target referent; it is less clear whether sentence context on one exposure will influence learners’ referential hypotheses on subsequent exposures. Here, we present results suggesting that even adults have some difficulty using a word’s prior sentence context to constrain its subsequent hypothesized referents, and 3-year-olds fail to make these cross-exposure inferences at all. Finally, we examine an alternative avenue for sentence context to guide word learning: adults successfully use informative sentence contexts to disambiguate between previously encountered referents. Ongoing work examines this question in children. These results suggest linguistic context is most useful for forming in-the-moment hypotheses but may, in some contexts, also facilitate learning across exposures.

Previous research suggests that a verb’s meaning is learned partly through the aggregated profile of syntactic frames associated with it. For example, “turn” occurs with transitive and intransitive frames in causative alternation (“He turned the car”/“The car turned”), indicating it is a causal verb. Some evidence demonstrates that young children combine multiple frames to map verbs to appropriate events. However, previous work always presented these frames together, in a single dialogue. What remains unknown is how verb learning occurs when the frames are separated, uttered in different referential contexts, as is likely in children’s everyday life. In a series of cross-situational word-learning experiments, we show that adult learners update their hypothesis about a novel verb's meaning when they encounter the verb again in a new frame, integrating their previous hypothesis about the verb’s meaning with the new frame. These results shed light on the cross-situational mechanisms of syntactic bootstrapping.

Children and adults use linguistic context to learn new words. For instance, in “She wears the dax,” “dax” likely refers to clothing. We asked whether learners retain these constraints across ambiguous word-learning exposures. Adults (n=139) learned 12 words. On Exposure 1, novel nouns were presented as the object of either Restrictive (e.g., “wears”) or Non-restrictive (“finds”) verbs. Participants selected which of two compatible referents the noun referred to. On Exposure 2, participants heard each noun again and chose between a distractor referent and the unselected referent from Exposure 1—now the only referent compatible with the previous Restrictive verb. If adults retain selectional restrictions, then Restrictive verbs should increase selection of the compatible referent. However, we found no difference between Restrictive (M=.74, SD=.26) and Non-restrictive (M=.76, SD=.24) conditions, p=.24. This suggests adults use verbs’ selectional restrictions to identify referents in the moment but do not retain these restrictions across exposures.

Previous research suggests that a verb’s meaning is learned partly through the aggregated profile of syntactic frames associated with it. For example, “turn” occurs with transitive and intransitive frames in causative alternation (“He turned the car”/“The car turned”), indicating it is a causal verb. Some evidence demonstrates that young children combine multiple frames to map verbs to appropriate events. However, previous work always presented these frames together, in a single dialogue. What remains unknown is how verb learning occurs when the frames are separated, uttered in different referential contexts, as is likely in children’s everyday life. In a series of cross-situational word-learning experiments, we show that both adults and three-year-olds generalize verb meanings across different syntactic frames in a cross-situational learning task. These results shed light on the cross-situational mechanisms of syntactic bootstrapping.

Infants’ earliest words are learned by observation of the referent world, but substantial research suggests such learning is highly error-prone. However, recent work suggests that even learners’ incorrect guesses may fall within the correct meaning’s semantic neighborhood—enabling learners to converge on the correct meaning across exposures. Here, we evaluate the semantic similarity of adults’ hypothesized word meanings in a cross-situational word-learning task. We find evidence for a weak semantic neighborhood effect: incorrect guesses are judged as similar to correct meanings (Study 1). However, this effect is not associated with successful word-learning. While learners tend to provide similar, internally consistent guesses across exposures, their accurate guesses are not similar to their previous guesses (Study 2). Moreover, incorrect guesses similar to the target do not increase accuracy on the subsequent exposure (Study 3). These results suggest early word-learning is driven by cues available in-the-moment, not by gradual exploration of semantic space.

Early word-learning opportunities are often highly ambiguous, with this problem being especially difficult for verbs. While a verb's syntax can help to identify the referent event from the environment, learners still need to contend with temporal and spatial misalignment between verbs and their referent events. Although children are shown to use syntax to infer verb meaning when there is initially no co-occurring referent event, it remains unclear what role syntax plays in verb learning across exposures in tandem with referential information. With three adult word-learning experiments, we showed that while syntax independently informed verb meaning in the absence of referents, it did not additionally constrain subsequent mappings when a referent was present. These results reveal both the power of syntax in cross-situational verb-learning–persisting across exposures –and its limitations–failing to supersede co-present referents.

A central challenge of cross-situational word learning is retaining word-referent mappings across exposures. We evaluate Memory-Bound Pursuit (MBP), a hypothesis-testing model of cross-situational word-learning which aims to account for learners’ memory constraints via a single parameter targeting the number of words that can be learned concurrently. Here, we show that by varying this parameter with age, MBP can capture both children’s and adults’ cross-situational word-learning success under varying levels of ambiguity. We also present new experimental findings supporting novel predictions made by MBP about the retention of word-referent mappings across intervening exposures. These findings suggest that MBP provides a strong baseline model of cross-situational word learning, capturing both developmental trends and experimental evidence of memory limitations for word learning.