Language, memory and neuronal mechanisms are reciprocally connected
in the human brain. Cortical mechanisms can be investigated
using a specific event related potential (ERP), called mismatch negativity
(MMN). MMN is greatly dependent on phonological feature
specific information. However, there are limited MMN studies using
‘‘natural’’, finely controlled speech stimuli.
The aims were (1) to develop natural-sounding English word and
pseudoword speech stimuli, using a new method for strictly controlling
feature-specific information. (2) To determine the robustness of
MMN responses to these stimuli in healthy young adults (n = 17).
Experimental stimuli included one word and one pseudoword:
peace and ‘‘peash’’. The researchers recorded (1) multiple repetitions
of these stimuli uttered by a male native speaker of (Hiberno) English
and (2) selected stimuli from these repetitions, whose vowels matched
in fundamental frequency (F0), mean energy and overall duration. To
create the final stimuli, cross splicing was used. These stimuli were
presented in two, reverse single contrast paradigms (each stimulus
acted as a ‘‘standard’’ (probability 90%) in one contrast and as a
‘‘deviant’’ (probability 10%) in the reverse contrast).
Carefully controlled natural speech stimuli were created. Discrimination
testing has shown a 100% correct identification of
difference between the two stimuli. Results to date using the MMN
paradigm have confirmed an enhanced MMN response for real word
stimuli compared to pseudoword stimuli. This real word-related
enhancement provides evidence to support the ‘‘word advantage
effect’’, thought to reflect the activation of language-specific memory
traces in the brain for words.The method outlined above will facilitate a range of accurate
natural speech stimuli for use in MMN studies. These may be more
ecologically valid than stimuli developed using previous methods.
Building on these results, investigation of this theory through MMN
studies of disordered language (aphasia) may lead to a greater
understanding of real brain mechanisms.
We would like to acknowledge Science Foundation Ireland for
funding this work.