Victor S. Ferreira1 & Zenzi M. Griffin2
1 University of California, San Diego, 2 Georgia Institute of Technology
On what basis do speakers select the content of their linguistic expressions? One possibility is that since speakers produce expressions to convey meaning, language-production processes select content based only on semantic properties. Such a mechanism would maximize the likelihood that production processes select and produce the most semantically precise expression. However, in most production situations, speakers and listeners can collaborate to convey meaning, and so semantic precision may not be the only factor that affects the selection of linguistic content. If so, pressures to speak in a timely and fluent fashion may cause production mechanisms to be sensitive also to how accessible the sounds of a linguistic expression are. This predicts that other things equal, speakers should be more likely to produce a linguistic expression when its phonological content is especially accessible.
We explored this prediction by presenting speakers with sentences, the final word of which was replaced by an incongruent picture. Speakers were instructed to read the sentences and accurately name the pictures. Critical trials came from the following conditions:
Example materials. Items in brackets denote pictured objects.
| Direct-competitor conditions | |
| Related | The woman went to the convent to become a [priest] |
| She put a contact lens in her [nose] | |
| Control | The woman went to the convent to become a [nose] |
| She put a contact lens in her [priest] | |
| Homophone-competitor conditions | |
| Related | I thought there were some cookies left, but there were [priest] |
| A selfish person cares only about me, myself and [nose] | |
| Control | I thought there were some cookies left, but there were [nose] |
| A selfish person cares only about me, myself, and [priest] | |
In the related direct-competitor condition, sentence completions were similar in meaning to the pictures. Since speakers select linguistic expressions at least based on semantic properties, we expected that meaning-similar sentence-completions would be erroneously produced instead of picture names ("nun" instead of "priest") more often than unrelated sentence completions ("nun" instead of "nose"). In the related homophone-competitor conditions, the sentence completions were homophones of the direct-competitor completions and thus only sounded like they were similar in meaning to the pictures. If the accessibility of phonological content also affects the selection of a linguistic expression, then sentence-completions that sound like they are similar in meaning to a picture may also be produced more often instead of the picture name ("none" instead of "priest") than unrelated sentence completions ("none" instead of "nose").
In three experiments, we found that both kinds of related sentence completions were produced instead of picture names significantly more often than their matched unrelated controls. "Nun" was produced instead of "priest" an average of 15% more often than it was produced instead of "nose," while "none" was produced instead of "priest" an average of 8% more often than it was produced instead of "nose." These results thus suggest that production mechanisms are sensitive to the accessibility of an expression's phonological content as well as its semantic properties.