Web Activity 10.5 (Un)natural methods for studying language production

Researchers who study language production are faced with a perennial methodological dilemma. Often, they are interested in testing very specific hypotheses about language production, hypotheses that might hinge on rigorous comparisons of multiple examples of specific words or structures during production. However, most conversational settings are too unconstrained to allow for easy exploration of these hypotheses. If the target forms are unusual ones, subjects may never get around to producing them. And a typical conversational setting involves a number of variables that can be tricky to control in such a way as to provide direct comparisons. For example, it might be hard to compare the time it takes to produce certain target forms, or the number of errors that occur in their production. So, there is an inherent trade-off between naturalness and specificity in methods for studying language production.

Below, you will see brief summaries of three different language production tasks that have been used to probe specific questions about the language production system. For each task, provide a critical assessment by addressing the following questions:

1. Does the task allow for an effective examination of the targeted question? Does it make efficient use of research resources (e.g., availability of participants, participants’ time, researchers’ time)? Can you foresee any difficulties in inducing the subjects to produce the desired forms?
2. Do you think the task relies on mechanisms of language production that are likely to reflect spontaneous and naturalistic language production? Why or why not? What evidence would you need in order to be satisfied that the task realistically reflects the aspects of language production that are under study?
3. If you are dissatisfied with the task on either of the above criteria, can you propose an alternative way to explore the question?

A. Retelling a story

The following example describes a method used by Brown and Dell (1987).

Question Certain actions are accomplished through the use of very typical instruments. For example, stabbing normally takes place with a knife, pounding a nail usually involves a hammer, and so on. Are speakers more likely to explicit mention the instrument if the instrument is an unusual one (stabbing with an ice pick) than if it is a highly typical one (stabbing with a knife)? If so, this would suggest that speakers are less likely to explicitly mention information that is predictable than information that is less predictable.

Method Subjects were presented with a stack of twenty stories printed on index cards. They were told to read each story carefully, and then immediately after reading it, to retell it in their own words in such a way that the experimenter/listener would understand exactly what had happened. Half of the stories involved typical instruments and half involved atypical instruments, as shown in the examples below. The researchers made audio recordings of the responses and scored each re-telling to indicate whether the subject made explicit mention of the instrument and whether that mention occurred in the same clause as the verb (e.g., The robber used a knife to stab the man.) or in the clause after the verb (e.g., The robber stabbed the man. He used a knife.)

Example of a typical instrument story:

The robber hid behind the door and when the man entered the kitchen, he stabbed him in the back. He wiped the blood off the knife and rummaged through the drawers. Later police investigators found his fingerprints all over the knife and had no trouble catching him.

Example of an atypical instrument story:

The robber hid behind the door and when the man entered the kitchen, he stabbed him in the back. He wiped the blood off the ice pick and rummaged through the drawers. Later police investigators found his fingerprints all over the ice pick and had no trouble catching him.

B. Targeted paraphrase

The method described below was used by Arnold et al. (2004).

Question In making choices about syntactic structure, speakers are faced with the competing demands of making sentences easy to understand, while also choosing structures that are not too difficult to plan or produce. In this study (whose results are partially discussed in Section 9.3 of the textbook, and then more fully in Chapter 11), the researchers examined whether speakers would be more concerned with avoiding ambiguity for their listeners or with minimizing their own memory demands by quickly producing short, easy-to-utter phrases in order to clear them from memory.

Method The goal was to see whether speakers would be able to avoid ambiguities for their listeners by producing an unambiguous sentence like sentence 1 rather than one like sentence 2, which contains a temporary PP attachment ambiguity (the listener might be tempted to think that Lincoln received the letters from the foundation):

1. The foundation gave a museum in Philadelphia Grant’s letters to Lincoln.
2. The foundation gave Grant’s letters to Lincoln to a museum in Philadelphia.

Moreover, the researchers also wanted to know whether the difference in the length of one of the two post-verbal phrases would affect the likelihood of producing an ambiguous structure. If speakers are acting so as to minimize their own memory demands, they should be more likely to utter the ambiguous sentence 3 than the one in sentence 2 because sentence 3 allows them to quickly utter and “dump” from memory a short, easy-to-produce phrase:

3. The foundation gave a museum Grant’s letters to Lincoln.

The researchers needed to elicit and compare the frequencies with which the above sentences were produced, while avoiding biasing the subjects toward either the ambiguous or unambiguous structures. The subjects, who always played the role of the speaker, sat at a computer terminal, reading sentences such as: A museum received Grant’s letters to Lincoln from the foundation. The subjects were told to remember the meaning of the sentence as accurately as possible. After reading the sentence, another person, in the role of the listener, read a question from a sheet of paper. The question was designed to elicit one of the above structures. For example: What did the foundation do? The subjects were advised not simply to repeat the sentence they had read, but to directly address the question by answering. This was intended to increase the likelihood that they would begin the sentence as The foundation gave…. The subjects were told to be sure to answer in a way that would make the meaning clear to the listener.

C. Uttering memorized sentences

This example describes a method used by Snedeker and Trueswell (2003).

Question Do speakers manage to successfully clarify syntactically ambiguous sentences through prosody? For example, a sentence like Tap the frog with the flower can either convey an instruction to use the flower to tap the frog or to tap the frog that has a flower.

Method Speakers and listeners played a game in which each of them sat on opposite sides of a barrier, so that neither could see what was happening on the other side. On each trial, the speaker and listener had the same array of objects in front of them and were aware of this fact. The speaker watched an experimenter perform an action out of sight of the listener. Next, the speaker was given a card with an ambiguous sentence like the one above printed on it, and was told to memorize the sentence. When the speaker had memorized the sentence, the experimenter demonstrated the action again. The speaker then uttered the memorized sentence in such a way as to communicate to the listener the exact action that the experimenter had demonstrated. The listener performed the action, while having his or her eye movements monitored as a measure of comprehension success.

References

Arnold, J. E., Wasow, T., Ash, A., & Alrenga, P. (2004) Avoiding attachment ambiguities: The role of constituent ordering. Journal of Memory and Language, 51, 55–70.

Brown, P. M., & Dell, G. S. (1987) Adapting production to comprehension: The explicit mention of instruments. Cognitive Psychology, 19, 441–472.

Snedeker, J., & Trueswell, J. (2003) Using prosody to avoid ambiguity: Effects of speaker awareness and referential context. Journal of Memory and Language, 48, 103–130.