Experimental Design and Data Collection: Socially Stratified Sampling in Laboratory-Based Phonological ExperimentationMethods For Studying Spontaneous SpeechMethods and Experimental Design For Studying Sociophonetic Variation

The article covers the factors that need to be considered in data collection and analysis related to sociophonetics. The recording methodologies are generally employed for acoustic phonetic analysis, obtaining stimuli for perception studies, or the development of automatic speech recognition (ASR) systems. The map task is used to elicit relatively spontaneous, conversational speech while maintaining considerable control over target words. Perception experiments on reduced speech require stimuli containing reduction, which are even harder to obtain than good spontaneous acoustic recordings. The most direct method for obtaining stimuli is to extract stimuli from large, relatively natural corpora that have been collected using the most spontaneous and conversational speech methods. Another method is to record spontaneous speech, extract usable stimuli, and then bring the same speaker back to read those word strings again as careful speech, out of context. The listeners' reactions to spontaneous vs. careful speech can then be compared using the same targets, with words and voice controlled. Intonation may differ unpredictably between the spontaneous and read utterances. A final method is to record a phonetician intentionally producing reduced and unreduced (careful) forms of target items. Acoustic measurements can partially confirm that stimuli match natural reductions. One of the methods employed to measure the variable and unexpected segments of reduced speech is to transcribe a corpus at phonetic and word levels and then compare the segments in the phonetic transcription to the segments given for the same words in a searchable dictionary.