Locus equations as proxies for co-articulation lend support to the Degree of Articulatory Constraints model

The degree of articulatory constraints (DAC) model (Recasens, Pallares, & Fontdevila, 1997) proposes that consonants involving the movement of the tongue dorsum are more resistant to coarticulation than those consonants that have a more fronted articulation. The present study aims to assess this claim using locus equation (LE) slopes as indicators of coarticulation. Participants were asked to produce V 1(t) .C 1 V 2 sequences as part of two-word phrases in a scripted dialogue, where C 1 is one of /p, t, s, ?/. LE were derived by measuring F2 at V 2 onset and midpoint. Since LE slopes approaching 1 indicate high levels of coarticulation, it was hypothesized that those segments with the lowest DAC would have the steepest slopes, and vice versa. /p/ was predicted to have the lowest DAC and steepest slope, followed by /t/, /s/, and /?/. Results were highly consistent with these hypotheses, lending support to the DAC model. A secondary hypothesis assessed the effect of emphatically stressing C 1 on the LE. Participants partook in a dialogue involving a “mishearing”, which prompted them to repeat the original V 1(t) .C 1 V 2 sequence. We expected participants to emphasize the misheard segment, which was either the target C 1 (Prominent condition) or the preceding V 1(t) (Control condition). It was predicted that prominence would reduce coarticulation, resulting in a downward shift in LE slopes relative to the Control condition. Our findings indicate that only the LE slopes of sibilants /s/ and /?/ were reduced under prominence as hypothesized, and that these reductions were statistically comparable. Results are thus consistent with the DAC model, since /s/ and /?/’s being less likely to co-articulate than /p/ and /t/ in the Prominent condition may due to their relatively large DAC values.