The Christiansen model of word segmentation [1] is a connectionist framework for modeling how infants combine multiple cues in learning and processing language. Most studies applying this model assume idealized input with adult-like representations of phonemes and features, with little or no degradation of the input signal. From these studies, it is difficult to tell if the model is robust to non-idealized, noisy input, which may correspond more closely to an infant languagelearnerÂ’s experience.
This study tests the robustness of the Christiansen model by providing input from a minimally-trained phone recognizer on infant-directed speech. Some degradation of performance is observed, but the model still performs above chance. This finding represents a first step in developing more realistic input representations for models of child language acquisition.