Asymmetries of perception are seen in many situations. For example, if you were asked to visually search a group of letter Os on a page for one letter Q, the Q would appear to "pop-out" and you would identify it almost instantly. However, if you tried to find an O in a page of Qs, you would have to serially search through the page until you identified the O; a very slow process. This can also be seen with long lines in comparison to short lines, or with reversed letters compared to normally oriented letters.
Asymmetries are also present for auditory stimuli, and have been shown for changes in complexity (frequency modulated vs standard tones) and duration (long vs short tones). However, very little work has been done on changes in duration in language. While English does not have an obvious duration contrast for speech sounds, many languages use duration to distinguish word meaning. For example in Finnish /palo/ meaning fire vs /pal:o/ (with a long l) meaning ball.
We looked at this difference in Bengali speakers by playing real and mispronounced words through headphones, while they responded to related or unrelated words and nonwords that they read on a computer monitor. We measured brain waves related to the words appearing on the screen to determine whether the subjects would consider the mispronounced nonwords as acceptable variations of their real words.
We found that mispronunciations with increases in consonant duration were tolerated and treated like real words, but mispronunciations with decreases in consonant duration were not accepted. This could be because a longer consonant includes all of the information of the shorter consonant and any extra information can be ignored. This indicates that short duration sounds are a default category in speech.
This paper has now been published in Neuropsychologia and is available to download for free here: http://www.sciencedirect.com/science/article/pii/S0028393214001006