Where is the disadvantage for reduced pronunciation variants in spoken-word recognition? On the neglected role of the decision stage in the processing of word-form variation

ABSTRACT

This study examines the role of the decision stage in the recognition of reduced pronunciation variants, specifically variants produced with nasal flaps (e.g. “counter” produced as “couner”). We investigate whether the previously reported disadvantage for nasally flapped variants is located at an early perceptual processing stage or at a post-perceptual decision stage. Experiment 1 replicates the variant effect with a lexical decision task. Experiment 2 uses the psychological refractory period paradigm, suggesting that the variant effect is located at a stage that requires central processing resources. Experiment 3 employed the shadowing task, indicating a significantly smaller variant effect compared to the lexical decision task. In concert, our results suggest that the disadvantage for nasally flapped variants has at least in part a decisional locus, which challenges representational and perceptual accounts of the effect. A mechanism is discussed that describes how decisional factors could cause the disadvantage for reduced variants.

KEYWORDS:

• Spoken word recognition
• speech reduction
• flapping
• psychological refractory period
• shadowing
• lexical decision

Acknowledgements

We thank Micah Geer for comments on earlier versions of the manuscript and Ashley Williams, Hital Patel, and Katie Simon for assistance in testing research participants and coding the verbal responses from Experiment 3. Most of this work was done while the first author was at the University at Buffalo.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Note that one could argue that in cases in which the reduced form is more frequent than the citation form, the term “canonical” could be used to refer to the reduced form. However, we will use the term “canonical” to refer to the citation form in this article in order to be consistent with previous studies on the processing of phonetic reduction.

2 Note that for this model, the only random effect was a random intercept for participants because more complex random effect structures resulted in convergence errors.