Knowledge Bank

In a crowded room, listeners easily organize and understand speech arising from various sources. This process involves auditory organization, specifically requiring listeners to organize spoken words coming from different locations. To investigate this, Freggens and Pitt (in preparation) created a cross-ear integration paradigm to measure the organization of speech split in location. Participants hear words either intact (span) or separated (/s/ on the left and "ban" on the right) and are told to only report what they hear on one side. If integrated across ears, participants should report the word with /s/ attached to the voiceless stop /p t/ ("span") and, if not, report a word with the voiced stop /b d/ "ban" (span – /s/ = ban). Freggens and Pitt found that even when parts of the word were separated by the maximum distance (180 degrees), participants always integrated the word pieces ("span"). Experiments 1 (N=25) and 2 (N=23) were designed to test the generalization of this phenomenon to other word clusters and other scenes, using spelling change as an indication of integration. Experiment 1 used words starting with the clusters /sp st/ to replicate the findings of Freggens and Pitt as well as these clusters /sl sm sn sw shr fl fr/ to generalize results. I found overwhelming integration for all clusters tested, suggesting default integration for speech. Experiment 2 was conducted to investigate whether default integration occurs with more complex speech scenes. Participants performed the same task but heard two simultaneous words (a target word and a competing word) instead of one split word. The first phoneme of the competing word served the same function as the isolated fricative from Experiment 1, and spelling-change was used to indicate integration. Unlike in Experiment 1, I found a very low integration rate for all clusters in the experiment. This means that the competing word on the unattended side interrupted the default integration across locations found in Experiment 1. In conclusion, Experiment 1 supports the default integration mechanism for speech organization, although Experiment 2 extends the literature by showing that this integration mechanism for speech organization is limited only to situations where the speech sound has nothing else to attach with.