The final project assignment

In plain language: explain something about how a piece of talk works, and why it matters.

More exactly: analyze the communicative effects of some aspects of one or more linguistic performances, attending to at least two different levels of linguistic analysis; and draw some interesting conclusion from your analysis.

Stage 1

For stage 1 (already done and handed in!), you've found or created a recording (or a text, but a recording is better). You've submitted a copy (or described the resource if it's inconvenient to submit a copy), along with

• an explanation of the two kinds of linguistic analysis you plan to do,
• an example of each type of analysis,
• and a brief sketch of the general point that you plan to make.

Stage 2

For stage 2 (due 11/24/2003), you need to do a serious amount of analysis, estimate the type and amount of additional analysis to be done, and explain concretely how the results of your analysis bear on your general point. You should also make any appropriate mid-course corrections in your original plan.

For example, suppose that you're analyzing interviews with two different non-native speakers of English, whose native language is the same, but whose degree of proficiency in English is different. You're looking at use of definite articles and simplification of syllable-final consonant clusters. Your point is that the more proficient speaker makes the same sorts of mistakes but at a lower rate. For stage 2, you might hand in transcripts of the two interviews, "translations" of the interviews into more idiomatic English, and a specific list of the noun phrases where you plan to analyze presence or absence of "the", and the syllables where you plan to analyze consonant cluster pronunciation. You provide a detailed analysis of definite article use in 4 (of 27) non-pronominal noun phrases in the first interview, and 4 (of 24) in the second, etc. You explain the pattern that you expect to find.

Or to take another example, suppose that you plan to test and extend Sarah Levien's "system for predicting the outcomes of Supreme Court cases based on the justices' questions and comments during oral argument." In order to do this, you've downloaded the transcripts and MP-3 audio files for ten cases from oyez.org, and given some examples of friendly, hostile and neutral questions. For your stage 2 submission, you identify all the sentences with the form of questions in the oral arguments for one of the cases. You also identify a number of examples that have the form of the statements, but seem to have the speech act function of questions, and you explain your criteria for choosing these. You then characterize each of the questions (of both kinds) according to whether it seems friendly, hostile or neutral, and you compare the results to Levien's theory, given how the case was decided. In the course of doing this, you decide that the reliability of subjective judgments of friendliness is a problem, especially if the evaluator has a stake in the outcome of the study. So you design a way for two of your friends to give you a double-blind evaluation (basically independent judgments made without knowing the outcome of the case), and you propose to complete this study for the final stage of your project, rather than analyzing the rest of the cases.

The point of this stage is to do enough of the proposed analysis to ensure that

• it's feasible to do (in a sample of specific cases) and you know how to do it
• the estimated amount of work to complete it for the whole project is not too large (or too small)
• the analysis really does bear on your general point

If you decide that your original plan needs to be scrapped, or you devise a new plan that seems much more promising, by all means change course. However, you still should aim to get to the same (stage 2) point with your new idea.