The "passive" set-up is described as subjects being instructed to ignore the sounds and concentrate on the silent film. That sounds foolproof.
In this study the "passive listening" group were asked to "simply listen":
Speech segmentation by statistical learning depends on attention
Juan M. Toro, Scott Sinnett, Salvador Soto-Faraco
Abstract "We addressed the hypothesis that word segmentation based on statistical regularities occurs without the need of attention. Participants were presented with a stream of artificial speech in which the only cue to extract the words was the presence of statistical regularities between syllables. Half of the participants were asked to passively listen to the speech stream, while the other half were asked to perform a concurrent task. In Experiment 1, the concurrent task was performed on a separate auditory stream (noises), in Experiment 2 it was performed on a visual stream (pictures), and in Experiment 3 it was performed on pitch changes in the speech stream itself. Invariably, passive listening to the speech stream led to successful word extraction (as measured by a recognition test presented after the exposure phase), whereas diverted attention led to a dramatic impairment in word segmentation performance. These findings demonstrate that when attentional resources are depleted, word segmentation based on statistical regularities is seriously compromised. q 2005 Elsevier B.V. All rights reserved. Keywords: Language; Attention; Speech Segmentation
Saffran et al. (1997) showed that statistical learning took place while participants performed an incidental task completely unrelated to the speech input. Their concurrent task, however, could hardly allow for a strict control of attentional load because it consisted of self-paced free drawing. Our results from Experiment 2 (750 ms SOA) replicate Safran’s (1997) findings, as statistical learning was found in an incidental situation, when the speech stream was combined with a visual distracter task. However, when the concurrent task was more demanding (500 ms SOA) or within the same sensory modality as the speech stream, performance on the segmentation task dropped to chance levels. Thus, the present results clearly show that at least some attentional resources must be available and directed to the speech stream in order to segment it. This conclusion is along the same line as other results in the field of language perception, specifically, on word reading (Rees et al., 1999) and audiovisual speech integration (Alsius, Navarra, Campbell, & Soto-Faraco, in press).
Thus, our data add to these recent demonstrations and call for some discretion in the interpretation of previous results indicating segmentation under incidental situations. In conclusion, the present results show that even if speech segmentation based on statistical learning can occur without the need of explicitly instructing the listener to focus on the speech stream, it is clear that some degree of attention is needed to attain word extraction successfully. Another question is, for instance, whether certain features of speech are in fact a very salient stimulus for the human, thus, potentially recruiting whatever attentional resources left available that are not being used in other processes, or even capturing attention already allocated to less salient stimuli. This is especially important in the context of language development...'