Event Date and Time
-
Location
Tydings Hall room 2108

Speakers: Amritha Mallikarjun & Rochelle Newman

Titles: Toddler's accommodation of accent: acoustic and experiential factors (Newman) The cocktail party effect in domestic dogs (Mallikarjun)

Abstracts:

Newman: Accents can pose multiple problems for a young learner:  First, phonetic and rhythmic variations can potentially make the mapping between accents difficult.  Second, speakers with foreign accents may be less consistent across their productions. Third, some accents may have sufficiently large acoustic differences that they can cause a perceived change in category identity resulting in overlap between intended categories. Recent work from our lab has revealed that: 1) toddlers’ (aged 18-20 months) ability to generalize across accents appears to be affected more by dialectal changes in vowel realizations than by changes in linguistic rhythm; 2) children aged 32 months can successfully accommodate accents that do not result in category overlap, but have more difficulty when accents have the potential to cause phonetic confusions, and 3) bilingual children are better able to generalize across unfamiliar regional dialects than monolingual children. We have also found that while children’s recognition of known words is impaired by the presence of an accent, their short-term memory is not. That is, assuming a word was identified correctly when it was spoken, the presence of an accent does not appear to further affect children’s ability to store that information for subsequent recall. These findings support a more nuanced view of the way that children both accommodate, and are affected by, variation in their language input. 

Mallikarjun: Like humans, canine companions often find themselves in noisy environments, and are expected to respond to human speech despite potential distractors. Such environments pose particular problems for young children, who have limited linguistic knowledge. Here, we examine whether dogs show similar difficulties. We found that dogs prefer their name to a stress-matched foil in quiet conditions, despite hearing it spoken by a novel talker. They continued to do so at signal-to-noise levels as low as 0 dB. We see better performance at name recognition in dogs that are trained to do tasks for humans, like service dogs, search-and-rescue dogs, and explosives detection dogs. The tasking dogs were of several different breeds, and their tasks were widely different from one another. This suggests that their superior performance may be due to generally more training and better attention.

HESP Seminar Series Presents talks by Amritha Mallikarjun and Rochelle Newman, pictured