This research program investigates speech errors using both psycholinguistic and phonetic methods with the long-term goal of determining the roles played by articulatory gestures and abstract linguistic units in speech production planning. This study lays the foundation for the research program, focusing on the articulatory characteristics of single gestures in normal speech and speech errors that can be viewed using ultrasound. These studies will also provide data on ordinary articulatory speech errors that can be compared to other populations that produce errors due to impairment, such as children with phonological misarticulation or adults with dyspraxia or aphasia. Data will be collected in three production experiments with 10 participants each. Experiment 1 examines the normal articulation of velar stops and how velar closure location varies by vowel context (so-called 'velar fronting'). In experiment 1, velars will be produced in CV(C) word and VCV nonsense syllables for each English vowel. Experiment 2a provides baseline data on speech errors using tongue twisters that alternate velar and alveolar onsets. Experiment 2a also aims to replicate previous phonetic studies of speech errors using acoustic analysis and magnetometry, which found that an error can occur to various degrees, creating a slightly abnormal production or creating a categorical error. Experiment 3a investigates the influence of a similar adjacent consonant on the production of errors. In this experiment, tongue twisters will be used in which the coda consonant preceding the onset either shares place of articulation or contrasts place with the onset, to examine the influence of gestural coordination on speech errors. Perceptual speech error experiments will also be conducted to provide a comparison with the articulatory measures from experiments 2a and 3a. Experiment 2b will use the recordings from experiment 2a. Recordings of these productions will be played to naive subjects to score for errors. Error perception will be compared to the articulatory measures to assess the reliability of speech error perception for both categorical errors and gradient errors. Experiment 3b will replicate experiment 3a. Additional participants will produce the tongue twisters of experiment 3a, but transcription will be used as the data collection method. Findings based on error transcription will be compared to the findings from the articulatory study.
Frisch, Stefan A; Wodzinski, Sylvie M (2016) Velar-vowel coarticulation in a virtual target model of stop production. J Phon 56:52-65 |