One of the difficulties inherent in the investigation of comprehension of nonce utterances is the necessarily anecdotal nature of the evidence. If the utterance is repeated and will consistently elicit the same response, it might properly be argued that the subject has been conditioned or trained to respond to a cue, and no higher degree of comprehension is involved. If, on the other hand, an utterance arises naturally from a life-event and is situationally motivated, arguments can be made that: (1) nonlinguistic circumstances provided the subject with cues and (2) the evidence is incapable of being reproduced and is therefore of little probative value.
Techniques have been developed in order to create scientifically sound environments for testing a subject's ability to comprehend nonce utterances. Savage-Rumbaugh (1993) has pioneered a method for testing comprehension in both young children and apes by asking them to perform unusual acts which they are unlikely to anticipate or to have seen performed by others. Likewise, situational and emotive cues are diminished by hiding the interlocutor's face or creating a barrier between speaker and hearer.
Such experiments provide useful data, but they should not be relied on exclusively. The removal of situational cues and other reinforcement creates a much higher burden of proof for comprehension than that which is generally met by most humans. Language develops within a contextual environment, and its acquisition of necessity requires nonlinguistic cues. Engagement with the subject matter is just as important for comprehension as is the ability to form abstract conceptual and linguistic categories.
Data on domestic canines provides evidence that there is a distinction between obedience to commands (rote training) and the comprehension of nonce utterances relating to the subjects' field of interest. Comprehension and engagement appear to be closely linked.
Since language and context are not separable, we might do well to examine language comprehension in context for both human and nonhuman subjects. A sociolinguistic and anthropological approach which takes cognizance of both text and context is in order.
One way to separate linguistic and extralinguistic elements of language in context is the use of a control group that does not comprehend the test language. This has been done in the past using unenculturated nonhuman primates (Savage-Rumbaugh 1993). An alternative method is suggested here. Given the same scenario, with a natural integration of contextual cues and spoken language, performance of both human and nonhuman subjects can be compared to that of a control group of normal humans who are not familiar with the language employed.
Experiments such as these would provide evidence on the composition of linguistic communication in a social setting. Extralinguistic elements of natural language use can be measured and given appropriate weight in assessing the performance of human and nonhuman subjects alike. By the same token, the contribution of linguistic input to overall communication can be pinpointed and evaluated.
© Aya Katz