In Lab 1B, You Will Use the Speech Error Corpus Collected by the Class to Try to Derive

LING440 - Lab 1b

In Lab 1b, you will use the speech error corpus collected by the class to try to derive some insights into the speech production system. 25% of your grade for Lab 1 will come from Part a and 75% from Part b. Feel free to refer to the Fromkin (1971) reading for thoughts and some history about this methodology (conversely, that reading should not be needed to complete this lab).

To ensure you receive full credit, for any answer where you refer to a particular error in the dataset, be sure to provide BOTH the content of the error AND the ID number in the Excel file for that error. Use complete sentences, as if you were writing up a brief technical report on this dataset for broader dissemination within a research group.

I would also suggest quickly reading through all of the questions first, so that you can identify the most appropriate place to mention particular points that you want to make.

Finally, I would recommend using formatting or color in your spreadsheet application to mark different error classes—this will make it easier to draw broader generalizations!

A. Exploring the dataset and methodological considerations

1. Take a few minutes to look through the dataset. (a) What kinds of variability do you observe in the data from the different collectors? (b) Why might this be expected?

2. What benefits does this technique have compared to running production experiments in the lab where participants are encouraged to produce particular kinds of utterances?

3. Conversely, what are the drawbacks?

4. What kind of challenges did you encounter in data collection? Could you suggest some best practices that investigators should follow when using this method?

5. When speech errors are made by non-native speakers, the hypothesis space is more complex (and sometimes more interesting). Some errors could reflect the same causes as errors in native speakers, but some errors could reflect incomplete grammatical or lexical knowledge of the language, and some errors could reflect non-native processing routines (e.g. I am a native Spanish speaker whose unconscious production routine is used to retrieving the lexical item for the noun before the adjective, even though I know that the English grammar puts them in the other order). Find one error in the lab that seems likely to be due to non-nativeness. What are a couple hypotheses about the factors that drove this error?

****Before going on, go ahead and mark ALL items in the corpus that seem likely to be due to non-nativeness of the speaker, and EXCLUDE them from consideration for the rest of this lab.

6. Even in native speakers, the term ‘speech error’ can be used to refer to a couple very different situations. One is cases in which the speaker unintentionally produces speech that is not consistent with any grammar/dialect that they have knowledge of. Another is cases in which the speaker (intentionally or unintentionally) uses a dialect or a term that is not recognized as ‘standard’ by the dominant culture. In this case, there is no true ‘error’, since the production is wholly consistent with the grammar of that dialect; although the speaker may feel that they made an error in switching to a non-dominant dialect. Do you see any cases in the corpus that you think might reflect the accurate use of a non-standard dialect or terminology?

****Before going on, go ahead and mark ALL items in the corpus that you think reflect accurate use of a non-standard dialect, or errors that seem likely to be due to non-nativeness, and EXCLUDE them from consideration for the rest of this lab.

B. Phonological errors

1. Provide at least one example of a phonological exchange (the position of two phonemes in the original utterance are exchanged with each other).

2. Provide at least one example of a phonological perseveration (a phoneme from earlier in the utterance is erroneously re-used later on).

3. Provide at least one example of a phonological anticipation (a phoneme from later in the utterance is erroneously copied to a position earlier in the utterance).

4. What could one conclude about how far phonological information is planned ahead from the data in this corpus?

5. How would you analyze what happened in error #72?

6. From this dataset, was it your impression that segments with shared phonological features are more likely to be exchanged or substituted? Explain using examples.

C. Lexical errors

1. Provide at least two examples of a lexical exchange (position of two words are exchanged).

2. Provide an example of a lexical perseveration or anticipation.

3. Provide at least two examples of lexical substitutions that seem to involve semantic similarity (the word is semantically but not phonologically related to intended word).

4. Provide at least one example of lexical substitution that seems to involve phonological similarity (a word was substituted that was similar in phonological form to the intended word).

5. What could one conclude about how far lemma information is planned ahead from the data in this corpus?

6. Were there any other lexical errors that struck you as particularly interesting?

7. Describe one new research question about the human speech production process that comes to you in looking at these errors (doesn’t have to be directly about a particular error, just inspired by).

D. Morphological errors

1. Provide at least 2 examples of what might be considered morphological errors from the corpus (e.g. errors of tense, agreement, etc.)

2. Take one of these errors and try to describe in words the sequence of production steps that might have led to the error.

E. Restatements

1. Provide at least 2 examples from the corpus of cases in which the speaker stopped and reframed part of the utterance.

2. What could these cases tell us about the mechanisms involved in planning and producing at the sentence-level? Did you notice any cases where the speaker appeared to ‘talk themselves into a hole’, in other words starting the sentence in a way that would be hard to finish grammatically?

F. Final/additional thoughts

1. Try to imagine an error that, if you heard it, would tell us something new and important about the production system. Explain why.

2. Imagine that you encounter someone unfamiliar with this area who hears about this methodology and says dismissively ‘What a waste of time…you are taking notes on people who just don’t know how to speak right! How are you going to get anything out of that? This is science??’ Try to explain in a couple clear sentences why/how studying speech errors can give insights into human production.