Almost any method of collecting language samples can be used with PCAD -- the utility, validity, and reliability of a collection technique are beyond the control of the program. However, the user must be aware that like any other computer program, the quality of the output from PCAD is determined by the quality of the input, and care should be taken to ensure that verbal transcriptions are accurate and complete, that subject selection (where appropriate) satisfies the standards of good experimental design, and that any potential bias in the elicitation instructions is clearly recorded along with the verbal sample. In particular, users should be aware that PCAD results cannot be considered reliable on input samples shorter than about 85-90 words, and that the reliability and accuracy of the system improves with the length of the sample.
For research involving spoken language, the verbal behavior can be elicited in many different ways, depending on the research purposes and design of the study. The content of psychotherapeutic interviews can be analyzed: an individual can be asked to report their feelings and attitudes towards another person or person, the speaker may be asked to relate their angry or anxious experiences or the speaker may be asked to report dreams.
"Standard procedure" for eliciting samples
Using written materials as input