'That' as syntactic pause: Retrieval difficulty effects on syntactic
production.

Victor S. Ferreira and Carla E. Firato
University of California, San Diego

In certain common sentences, a speaker can use or omit optional words, such as the "that" in a sentence complement structure like "The poet recognized (that) the writer was boring." What is the communicative value of the mention or omission of such optional words? Two independent research threads converge to suggest an intriguing possibility: First, research on disfluent production suggests that speakers use filled pauses like "uh" and "um" specifically to communicate upcoming retrieval difficulties (Clark & Fox Tree, 2002), implying that "upcoming difficulty" is communicatively useful information. Second, research on sentence-complement production shows that speakers are more likely to omit the "that" when post-"that" material is easily retrieved from memory (Ferreira & Dell, 2000; note that similar effects have been revealed with other alternations, e.g., Bock, 1986). What has not been shown is that speakers mention "thats" more
specifically when subsequent material is more difficult to retrieve; if so, then the communicative value of the "that," like a filled pause, might be to indicate upcoming retrieval difficulty.

To test this, we exploited the fact that speakers have difficulty retrieving words that are similar in meaning to other words that they have just expressed (e.g., Vigliocco et al., in press). Speakers produced sentence-complement structures in which the post-"that" material -- the embedded-subjects -- were either meaning-similar (and therefore more difficult to retrieve) or meaning-dissimilar (and therefore easier to retrieve) to three nouns in the main subjects (e.g., "The AUTHOR, the POET, and the BIOGRAPHER recognized (that) the WRITER was boring" vs. "The AUTHOR, the POET, and the BIOGRAPHER recognized (that) the GOLFER was boring."). (A separate experiment independently verified this effect of similarity on retrieval difficulty). Production was elicited with a sentence-recall procedure, where speakers read and produced sentences back after a short delay (which results in relatively free production of the "that"). The results confirmed the prediction: Speakers produced significantly more "thats" before more difficult-to-retrieve meaning-similar embedded subjects than before more easily retrieved meaning-dissimilar embedded subjects. Furthermore, meaning-similar embedded subjects were _also_ accompanied by more disfluencies, and "that"-mention and disfluency rate were significantly correlated. Thus, speakers mention "thats" more often when subsequent sentence material is more difficult to retrieve, suggesting that speakers may use "thats" (and possibly other choices of sentence form as well) to indicate such upcoming retrieval difficulties.

References

Bock, J. K. (1986). Meaning, sound, and syntax: Lexical priming in sentence production. _Journal of Experimental Psychology: Learning, Memory, and Cognition_, _12_, 575-586.

Clark, H. H., & Fox Tree, J. E. (2002). Using uh and um in spontaneous speaking. _Cognition_, _84_, 73-111.

Ferreira, V. S., & Dell, G. S. (2000). Effect of ambiguity and lexical availability on syntactic and lexical production. _Cognitive Psychology_, _40_, 296-340.

Vigliocco, G, Vinson, D. P., Damian, M. F., & Levelt, W. J. M. (in press). Semantic distance effects on object and action naming. _Cognition_.