When looking for the referents of novel nouns, adults and young children are sensitive to cross-situational statistics (Yu and Smith, 2007; Smith and Yu, 2008). In addition, the linguistic context that a word appears in has been shown to act as a powerful attention mechanism for guiding sentence processing and word learning (Landau and Gleitman, 1985; Altmann and Kamide, 1999; Kako and Trueswell, 2000). Koehne and Crocker (2010, 2011) investigate the interaction between cross-situational evidence and guidance from the sentential context in an adult language learning scenario. Their studies reveal that these learning mechanisms interact in a complex manner: they can be used in a complementary way when context helps reduce referential uncertainty; they influence word learning about equally strongly when cross-situational and contextual evidence are in conflict; and contextual cues block aspects of cross-situational learning when both mechanisms are independently applicable. To address this complex pattern of findings, we present a probabilistic computational model of word learning which extends a previous cross-situational model (Fazly et al., 2010) with an attention mechanism based on sentential cues. Our model uses a framework that seamlessly combines the two sources of evidence in order to study their emerging pattern of interaction during the process of word learning. Simulations of the experiments of (Koehne and Crocker, 2010, 2011) reveal an overall pattern of results that are in line with their findings. Importantly, we demonstrate that our model does not need to explicitly assign priority to either source of evidence in order to produce these results: learning patterns emerge as a result of a probabilistic interaction between the two clue types. Moreover, using a computational model allows us to examine the developmental trajectory of the differential roles of cross-situational and sentential cues in word learning.
12 Figures and Tables
FIGURE 1 | Internal representation of the example item Mommy ate broccoli in a visual context.
Table 1 | Lmer models and p-values from MCMC sampling for learning prob, Exp. 1 probsReferential Uncertainty + (1|sub) + (1|item).
FIGURE 2 | A sample input scene from experiments of Koehne and Crocker (2010, 2011), paired with a spoken sentence Si gadis bermamema si worel (the woman will eat the broccoli ).
Table 2 | Lmer models and p-values from MCMC sampling for chosen meanings in conditions (high-frequency meaning choices, low-frequency meaning choices),TestType 1, Exp. 2. ChosensVerb Type + (1|sub) + (1|item), family = binomial (link =“logit”).
FIGURE 3 | Processed version of the training item shown in Figure 2.
FIGURE 4 | WordNet hypernym hierarchies for broccoli (a food item) and for skirt (a clothing item), as well as their corresponding meanings in our lexicon, extracted from the hypernym hierarchies.
FIGURE 5 | Koehne and Crocker (2010), Experiment 2: the left panel depicts the proportion of nouns learned in each condition in the original experiment, and the right panel shows the mean selection probability of the target object in our simulations.
FIGURE 7 | Simulation results for K&C 2010-Experiment 2 on noisy pre-training data sets of different size.
FIGURE 8 | Koehne and Crocker (2011), Experiment 2,TestType 1.
FIGURE 9 | Koehne and Crocker (2011), Experiment 2,TestType 2.
FIGURE 10 | Simulation results for K&C 2011-Experiment 2,TestType 1 on a noisy pre-training data sets of different size.
FIGURE 12 | Simulation results for K&C 2011-Experiment 2,TestType 2 on a noisy pre-training data sets of different size.
Download Full PDF Version (Non-Commercial Use)