Bottleneck influences on the processing stages of word production

Victor S. Ferreira and Harold Pashler
vferreira@ucsd.edu
University of California, San Diego

To produce a word, a speaker goes through a sequence of three processing stages: Lemma selection (where semantic constraints lead to the selection of a single word representation), lexeme selection (where a frequency-sensitive phonologically defined morpheme is selected), and phoneme selection. During which of these stages can a speaker simultaneously perform another unrelated task? In two dual-task experiments, speakers named pictures as quickly as possible and promptly discriminated low-, medium-, and high-pitched tones with button-press responses. Tones began 50, 150, or 900 ms after picture onset (though the effects below appeared equally at all these SOAs). We manipulated the duration of lemma, lexeme, and phoneme selection, and measured picture-naming latencies and tone-discrimination latencies. If a word-production stage allows simultaneous processing, then manipulating its duration should only affect picture naming; but if that word-production stage does not allow simultaneous processing, then manipulating its duration should affect both picture naming and tone discrimination equally.

In Experiment 1, speakers named pictures that completed visually presented cloze sentences (after Griffin & Bock, 1998). The duration of lemma selection was manipulated with high- and low-constraint cloze sentences, and the duration of lexeme selection with high- and low-frequency-name pictures (see examples below). The results, shown below, reveal that both manipulations affected tone discrimination as much as picture naming. In Experiment 2, speakers named pictures while ignoring visual distractor words (after Damian & Martin, 1999). The duration of lemma selection was manipulated with semantically related distractors and phoneme selection with phonologically related distractors (each compared to unrelated distractors). The results (shown below) reveal that while the lemma-selection manipulation again affected both tone discrimination and picture naming, the phoneme-selection manipulation only affected picture naming. Thus, only the phoneme selection stage allows simultaneous processing, while lemma and lexeme selection act as 'bottlenecks,' inhibiting processing in a concurrently performed unrelated task.

Experiment 1: Cloze constraint and lexical frequency affected picture naming and tone discrimination equally.

Constraint Freq Example Cloze Sentence Ex. Picture Pic RT Tone RT
High High Bob was tired, so he went to [bed] 721 1043
Low High She saw a picture of a [bed] 840 1171
High Low The man was stung by a [bee] 703 1014
Low Low There was a drawing of a [bee] 900 1245

Experiment 2: Compared to the effect of unrelated distractors, semantic distractors affected tone discrimination as much as picture naming, but phonological distractors only affected picture naming.

Relatedness Ex. Picture Ex. Distractor Pic RT Tone RT
Semantic [bed] couch 844 1102
Phonological [bed] bend 742 1004
Unrelated [bed] duke 794 1007



References

Damian, M. F., & Martin, R. C. (1999). Semantic and phonological codes interact in single word production. Journal of Experimental Psychology: Learning, Memory, & Cognition, 25(2), 345-361.

Griffin, Z. M., & Bock, K. (1998). Constraint, word frequency, and levels of processing in spoken word production. Journal of Memory and Language, 38, 313-338.



AMLaP Conference, Saarbrücken, September 2001