Lexical Prediction in Human Sentence Processing

Basic course information

This is a seminar on what and how people predict about upcoming words in sentences.
When: 16:15 - 17:45, Wednesday
**No class on 16th, October!**
Where: A2 2 / Seminar room 2.03
Prerequisites: None. First couple of classes introduce background knowledge.
You MUST join the seminar Teams channel to be in this seminar. If you haven't join it here.
All course-related communications should be done via Teams, but in case of any problems in Teams, I am also available at nakamura@lst.uni-saarland.de.

Overview

In everyday life, we hear and read hundreds and thousands of utterances, and understand what they mean (including this website!). Since it is such a usual part of our life, we rarely realize what a complicated task it is, and how quickly and effortless we complete it. This might be all the more surprising given that computers with huge computational resources and with state-of-the-art models have yet to reach what humans can normally do. Why we humans are such experts of language processing, even we are equipped with only limited computational abilities?

Psycholinguists have argued that predictive processing is one key to solve this puzzle. That is, people use what they have already heard/read to facilitate processing of expected continuations. For example, when you hear "I would like to have some coffee with ...", you guess cream or sugar among other candidates are likely to follow, and you start processing them even before you hear them.

This course will explore the predictive processing in human sentence processing. We will focus on the process of identification of words, and explore how prediction is involved in it.
We will be asking following questions:

How people use contexts to make predictions about upcoming words?
Which information can be used for prediction and which cannot be?
How can we know what people are predicting and how much they are predicted?

In this course, we don’t just aim to learn about the relevant existing studies, but also to think about what is not known in the literature and how we can address it.

Course policies

Student responsibility

Each student is expected to attend, prepare, for and actively participate in every class. Please communicate in advance with me if you have to miss a class. If you cannot do so due to an emergency, please communicate it as soon as you can.
Turn in assignments on time and fully completed.
Check Teams regularly and receive notifications for posts. I assume that you see a post or a message within 24 hours during weekdays.
Ask for help!

Communication with Instructor

You can call me Masato. (If you want to address me by the last name for some reason, I would prefer "Dr. Nakamura" over "Mr. Nakamura".)
Feel free to contact me to set up a meeting.
All online communications from students or from instructors should be via Teams. You can also contact me at nakamura@lst.uni-saarland.de only if there is any issue using Teams.

Names/Pronouns and Self-Identifications

I recognize the importance of a diverse student body, and we are committed to fostering inclusive and equitable classroom environments. I invite you, if you wish, to tell us how you want to be referred to in this class, both in terms of your name and your pronouns (he/him, she/her, they/them, etc.). Keep in mind that the pronouns someone uses are not necessarily indicative of their gender identity. Additionally, it is your choice whether to disclose how you identify in terms of your gender, race, class, sexuality, religion, and dis/ability, among all aspects of your identity (e.g., should it come up in classroom conversation about our experiences and perspectives) and should be self-identified, not presumed or imposed. I will do my best to address and refer to all students accordingly, and I ask you to do the same for all of your fellows.

Excuse policy

It is your responsibility to participate in every course and to submit assignments in time. However, it does not apply if you have a decent excuse and if you communicate it in a timely manner.

You must report the issue that prevents participation or timely submission as soon as you can after it happened. For example, if you were sick, ideally you should inform me about it right after you realized that you are sick, or when you feel better enough to send me a message about it at the latest. You cannot tell me a week after you got better that you had been sick and turn in your term paper late. It is always OK to tell me that you MIGHT not be able to submit an assignment in time due to illness and then end up making it before the deadline.
If the problem is very sensitive and you prefer not to tell me about the details, that is OK. However, you have to at least tell me that a problem happened.
Unexpected and unavoidable problems (e.g., sickness) would be considered as good excuses. On the other hand, issues such as assignments in other classes or planned travels would not count. You are responsible for managing your own schedules.

Grading/Evaluation

Overall structure

4 credits

50%: Active participation

50%: Reading responses

7 credits

50%: Paper

25%: Active participation

25%: Reading responses

Active participation

Discussions are the most important component of the learning experience in this course. Therefore, everybody must take this very seriously. I highly value coming to the classroom and participating in discussions. You are expected to actively participate in the discussions, such as asking questions, answering other students' questions, sharing your own ideas, etc. Since you should be well prepared for discussions, if you silently sit in the classroom throughout the class, you are deemed to not have done your job.

In order to encourage active participation of each student, I expect everybody to be respectful to what other students say in the class.

Reading responses

Each student submit a reading response for the paper. In each reading response, you will be asked about the content of the paper (e.g., What is the research question?) and to provide questions or thoughts. The aim of this assignment is to make sure each of you is ready for the in-class discussions. Therefore reading responses will not be graded if you do not participate in the discussion.

Submit this via Teams by the end of the day before the class. (e.g., if the class is on 10th November, the deadline is 23:59 of 9th November)

Questions / Thoughts

The aim is develop ideas/questions that can stimulate discussions.

For example, you can think about:

How is the paper related to other studies?
Is there any concern about the study? Is the method valid?
Is there another way to test the hypothesis?
Are there any unanswered questions?
Is there any follow up study you can think of?

It can be a thoughts (e.g., "Although the authors interpreted the results in this way, they can be also interpreted as ...") or questions (e.g., "This paper seems to be conflicting with A & B (YYYY) in this specific results. How can we reconcile the two results?"). But if you are asking questions, you should write about your answer or at least which direction you are heading to if there is not yet a concrete answer.

When you are proposing a follow up study, it is not enough to just propose a possible extension but you should write about what you would newly learn from it and/or why it is interesting to do. "What if we do the same experiment in another language?" is not really an interesting question, but "What if we do it in Language A, which has a feature B, because it can address this unanswered aspect of the original research" is a very interesting one.

Bad examples:

"What does this study tell us about prediction in language processing":
Too vague. Which specific aspect of prediction? What do YOU think? Is your thought the same as the author?
"How can this be used to improve large language models / teaching / etc"
Too vague. What do YOU think? Why specifically this study would be useful?

Summary diagram

We will do a summary discussion in the final class or the one before. Instead of reading responses, I will ask you to create a diagram of a model of language processing to prepare for the discussion.

Term paper

If you register for 7 credits, submit a final paper where you propose a research related to the course. Details will be provided later in the semester.