Term paper

1. Read all the instructions carefully

2. Read at least one experimental paper having the instructions in your mind

3. (Optionally) Submit topics and/or outlines for feedback

4. Write a short paper where you

• Propose a research question related to the course topic. 

• Motivate your question by referring to at least two papers.

• Propose an experiment to address the question.

Please take this as a learning process. Below I wrote down what I am expecting as clearly as possible. All the instructions are set up so that you can write a better paper by following it, so please read and follow them.

At the same time, most requirements are negotiable. If you have some good reasons not to follow them, you can let me know and I can be flexible.

Your research question must be about human language processing and related to the course topic, whether it be language production, lexical prediction, or the relationship between language production and prediction. For example:

• Do phonologically related words in the native language facilitate or interfere with the language production in the second language?

• Can people predict animacy-violation verbs (e.g., Nieuwland & van Berkum, 2006) because they use alternative world models or because they update the representations of the individual objects?

• Does engagement of language production mechanism inhibit the retrieval of unexpected words in language comprehension?

The research question can be the same as the papers you cite, but in that case you have to be clear about why the paper is not sufficient and the question is still worth asking when you refer to the paper.

Your research question needs to be specific. That is, you shouldn’t be asking "What is the relationship between lexical access and structure building?" or ”How are kids different from adults in terms of language prediction?” because those are too broad. On the other hand, it shouldn’t be too specific like ”Do people predict a word X in context Y?” A better research question would be something between those in terms of specificity, like the ones introduced in the beginning of this section.

You should reconsider questions like ”What if we take this paper and use stimuli X instead of stimuli Y?” or "What if we we test German speakers instead of English speakers for the study Z?" The idea itself may be good as a starting point, but you should go one step further and think about what you can learn by that modification in the design and turn it into a question about how the human mind works. For example, you shouldn't say "What if we use German stimuli with Momma et al. (2016)'s design?" but you might ask instead "How does partial/loose verb-finality affect the incrementality in verb production?" (compared to Momma et al., 2016), and decide to test German speakers as a way to address the question.

Refer to at least two papers as backgrounds to motivate your study, at least one from the class and at least one that we didn't read in the class. (OK to use the papers I shared but didn't discuss in the class either as the one from the class or as the "new" paper.) If you really want all your citations to be from outside of the class, please consult with me early on (at least two weeks before the deadline at the latest).

The relationship between the two papers and your experiment can be whatever, as long as they are relevant:

• your study might reconcile two apparently conflicting studies.

• your study might extend a prior paper, and the other might provide reasonable methods for that.

• etc

This minimal requirement about the prior papers doesn't mean that you can ignore other papers we read for the class. If there are closely related papers from the class, I expect you to cite and discuss them whether or not you have already cited other papers (e.g., when your argument conflicts with one of those papers). On the other hand, I don't expect you to do a thorough literature review of papers outside of the class.

Propose an experiment that can address your research question. The idea is to write up a paper just like ones we read in the class, but without actual data or details about the design.

• Design an experiment including some details:
  What kind of measures will you use? (e.g. Picture description, sentence production with given words, etc.) Which population will you look at? (e.g., Speakers of Language X, Y years old kids and adults, etc.) What kind of stimuli will you use?
  You are not expected to describe the full details of the experiments, but you should describe at least the minimal details to explain what the experiment is like and how it works. For example, you don't need to provide the full set of stimuli, but you must provide one set for each condition. You are not expected to describe which university the participants are from, the number of electrodes in the EEG cap, or how long the fixation cross is presented on the monitor.

• Discuss the results with possible outcomes:
  Explain what the results would look like, and how that helps you answer the research question.
  You can either deeply discuss one possible pattern of results that you expect, or cover multiple possibilities and respective implications.

• Of course you don't need to run the experiment. However, it is OK to write the paper as if you have actually run the experiment or even report fake data (optional). But make sure you are consistent. Don't mix up "In the experiment, participants will..." and "The target word production was slower for..."

When you are planning your final paper, you should think about both research questions and testable hypotheses but separately. First identify your research question and then turn it into a testable hypothesis.

Research questions

Do people build identical syntactic structure representations in language production and comprehension?

• A question that describes what you are trying to understand through the research. Conversely, the goal of the research is to provide an answer to the question with evidence.

• Usually researchers have their own answers (theoretical claims) and they provide support by the study.

• In this course, your research question should be about how human language processing works.

Testable hypotheses 

If syntactic structure representations are identical in language production and comprehension, structural choices in language production (active vs passive) should be equally affected by priming from language production and comprehension.

• Research questions themselves are abstract, so they must be transformed into concrete and testable hypotheses. By testing these hypotheses, you can answer the research questions.

• Often in the “If… then…” format.

Good research proposals don’t try running experiments without expectations and see what happens, but they have clear and specific theories/hypotheses to evaluate. Therefore, when you provide your research question, you should show your hypothetical answer to the research question (= your claim), or competing theories/hypotheses that you are trying to evaluate.

For example, if your research question is “What roles do language production processes play in lexical prediction in language comprehension?”, you should show your answer such as “I claim that people use the language production processes as an emulator” and your experiment should be able to provide support for it if you get the expected results.  (This example research question is a bit vague and yours should be more specific.)

Of course your answer doesn’t need to be “correct” because you are trying to see it through the experiment. But, on the other hand, your hypothetical answers or the competing theories/hypotheses shouldn’t be random guesses. Ideally they should be supported by prior studies. For this assignment, they can be good guesses with some rationale.

The main focus of this assignment is on the processes of (i) proposing an unanswered and motivated question based a prior paper(s) and (ii) designing an experiment that can address it. In this regard, a good experiment is something that nicely addresses the question. So you don’t need to try to come up with a fancy design and there is no problem for the experiment to be simple if it addresses the question very well.

• Follow the template specified below. 

• Up to 3000 words excluding figures, captions, and the reference list. No minimal word counts.

• Double-spaced.

• Put the title and your name at the top of the first page and then begin the main text on the same page.

• Put a page number on each page.

• Other formats for better readability (e.g., reasonable font size and margins, headings, indentation for each paragraph's beginning, etc.)

You should expect average masters student in LST who haven’t taken this or a similar course (or you yourself before taking this course) as readers of your paper. Therefore you should explain jargons that those people might not know (e.g., content retrieval, SOA, etc.).

It is a good idea to include tables and figures from the prior studies and/or fake ones for your experiment. All tables and figures should be accompanied by a figure/table number and should be referred to with the number.

e.g., Figure 1 illustrates the EEG data of ...

It is your responsibility as the author to give credits to findings or ideas of other researchers. All prior works should be cited using the APA citation format, which is frequently used in the psycholinguistics literature. The website below provides useful examples for both in-text citation and the reference list: https://www.mendeley.com/guides/apa-citation-guide/

Purpose: Introduce your study and convince the reader that they should read the rest of the paper.

1. Introduce the topic (broader than your specific research question)

2. What exactly the research question is.

3. Why it is important to answer. 

• How does it contribute to understanding language processing?

• Why is it particularly important compared to other questions?

4. Expect your conclusion. (e.g., This study will show that ...)

Additional notes:

• Make sure readers know where you are going after they read the introduction.

• 2 and 3 can be swapped if that improves the flow.

Purpose: Situate your research question and your experiment in the background literature.

1. Introduce the prior studies.

• What exactly they did (methods)

• The relevant results

• The author's and/or your interpretation of the results

2. Show what is not yet known (= the gap) in the literature.

3. Briefly introduce your experiment regarding the gap.

• How your experiment fills in the gap in the literature.

• A brief description of the crux of your experimental design. You might use some examples here. If you haven't done it yet, you should introduce the linking hypotheses (i.e. theories of how your experimental paradigm reflects the cognitive processes you are interested in) of your methods.

• Your prediction about the results (If X, then participants should Y.)

Additional notes:

• Make sure readers know why you are citing the work before you provide its details.

• The above structure assumes that two (or more) studies you are citing are direct theoretical/empirical precedences of your work. If a prior study provides some useful methods, you may introduce it differently (e.g., you might insert it when you introduce your own study).

• Clearly separate the methods, results, and interpretations!

• Readers must know the high-level idea of what you are doing in your experiment and what you are expecting to find by the end of this section. You further describe your experimental design in the Design section and the expected results in the Discussion section.

1. Subjects: What kind of speaker group(s) will you test?

2. Materials: What are the conditions?
   Use examples to describe your materials if you didn't do that in the previous section. Don't just throw examples after you verbally explain everything.

3. Tasks: What kind of tasks do the participants perform?

4. Other important details specific to each study such as:

• • proficiency in English if you are looking into non-native English speakers.
    (You don't need to specify tests etc to assess the proficiency.)

  • the SOAs if you are using an interference paradigm just like Schriefers.

  • etc

Additional notes:

• Since you don't need to provide as much details as real experimental papers, you may not create subsections for participants, materials, etc as the papers we read.

• (Briefly) provide motivations for your choices of the design if they don't naturally follow from what you have written about your study in the previous section. 

• If you use non-English languages, make sure to include English translations of your example stimuli. Recommended: follow sections 6d,e and 8 from Linguistic Society of America style sheet.

1. Expected data patterns

• e.g., The production latency would be shorter in Condition A than in Condition B.

• It is useful to show a fake figure (optional)

2. Implications of the expected results

• Make sure to tie the expected results back to the research question.

3. Other discussion

• Broader implications, future studies, limitations, relationship with other studies, etc.

• List the reference following the APA format.