Assessing correlations between lingual movements and formants

First and second formant histories have been used in studies of both normal and disordered speech to indirectly measure the activity of the vocal tract. The purpose of the present study was to determine the extent to which formant measures are reflective of lingual movements during diphthong production. Twenty native speakers of American English from the western United States produced four diphthongs in a sentence context while tongue movement was measured with a magnetic tracking system. Correlations were computed between the vertical tongue movements and the first formant, as well as between the anteroposterior movements and the second formant during the transition phase of the diphthong. In many instances the acoustic measures were clearly reflective of the kinematic data. However, there were also exceptions, where the acoustic and kinematic records were not congruent. These instances were evaluated quantitatively and qualitatively in an effort to understand the cause of the discrepancy. Factors such as coarticulation, motor equivalence (including the influence of structures other than the tongue), and nonlinearities in the linkage between movement and acoustics could account for these findings. Recognizing potential influences on the acoustic–kinematic relationship may be valuable in the interpretation of articulatory acoustic data on the individual speaker level.

► We examined the extent to which formant tracks follow tongue movement.
► Many formant tracks correlated well with the kinematic signal.
► There were frequent exceptions to the predicted acoustic-kinematic association.
► Coarticulation and motor equivalence appear to influence the correlation between movement and acoustics.