auditory word recognition

A growing body of psycholinguistic research suggests that visual and auditory word recognition involve morphological decomposition: Individual morphemes are extracted and lexically accessed when participants are presented with multi-morphemic stimuli. This view is supported by the Morpheme Interference Effect (MIE), where responses to pseudowords that contain real morphemes are slower and less accurate than responses to pseudowords that contain invented morphemes. The MIE was previously demonstrated primarily for visually presented stimuli. Here, we examine whether individuals' sensitivity to morphological structure generalizes across modalities. Participants performed a lexical decision task on visually and auditorily presented Hebrew stimuli, including pseudowords derived from real or invented roots. The results show robust MIEs in both modalities. We further show that visual MIE is consistently stronger than auditory MIE, both at the group level and at the individual level. Finally, the data show a significant correlation between visual and auditory MIEs at the individual level. These findings suggest that the MIE reflects a general sensitivity to morphological structure, which varies considerably across individuals, but is largely consistent across modalities within individuals. Thus, we propose that the MIE captures an important aspect of language processing, rather than a property specific to visual word recognition.

Shafai-Bajestan, E. and R. H. Baayen
Classical linguistic, cognitive, and engineering models for speech recognition and human auditory comprehension posit representations for sounds and words that mediate between the acoustic signal and interpretation. Recent advances in automatic speech recognition have shown, using deep learning, that state-of-the-art performance is obtained without such units. We present a cognitive model of auditory comprehension based on wide rather than deep learning that was trained on 20 to 80 hours of TV news broadcasts. Just as deep network models, our model is an end-to-end system that does not make use of phonemes and phonological word form representations. Nevertheless, it performs well on the difficult task of single word identification (model accuracy 11.37%, Mozilla DeepSpeech: 4.45%). The architecture of the model is a simple two-layered wide neural network with weighted connections between acoustic frequency band features as inputs and lexical outcomes (pointers to semantic vectors) as outputs. Model performance shows hardly any degradation when trained on speech in noise rather than on clean speech. Performance was further enhanced by adding a second network to a standard wide network. The present word recognition module is designed to become part of a larger system modeling the comprehension of running speech. Index Terms: naive discriminative learning, auditory word recognition, wide neural networks, deep speech, frequency band summary features

A growing body of psycholinguistic research suggests that visual and auditory word recognition involve morphological decomposition: Individual morphemes are extracted and lexically accessed when participants are presented with multi-morphemic stimuli. This view is supported by the Morpheme Interference Effect (MIE), where responses to pseudowords that contain real morphemes are slower and less accurate than responses to pseudowords that contain invented morphemes. The MIE was previously demonstrated primarily for visually presented stimuli. Here, we examine whether individuals’ sensitivity to morphological structure generalizes across modalities. Participants performed a lexical decision task on visually and auditorily presented Hebrew stimuli, including pseudowords derived from real or invented roots. The results show robust MIEs in both modalities. We further show that visual MIE is consistently stronger than auditory MIE, both at the group level and at the individual level. Final...

Classical linguistic, cognitive, and engineering models for speech recognition and human auditory comprehension posit representations for sounds and words that mediate between the acoustic signal and interpretation. Recent advances in automatic speech recognition have shown, using deep learning, that state-of-the-art performance is obtained without such units. We present a cognitive model of auditory comprehension based on wide rather than deep learning that was trained on 20 to 80 hours of TV news broadcasts. Just as deep network models, our model is an end-to-end system that does not make use of phonemes and phonological word form representations. Nevertheless, it performs well on the difficult task of single word identification (model accuracy 11.37%, Mozilla DeepSpeech: 4.45%). The architecture of the model is a simple two-layered wide neural network with weighted connections between acoustic frequency band features as inputs and lexical outcomes (pointers to semantic vectors) a...