United States

High-quality multiple-choice tests are required to assess learners&#39; abilities in language education. When creating such tests, it is important to create appropriate false answer choices. Because a wealth of experience is required in creating tests without easy correct answers, the number of people who can create such tests is limited.
In this study, we evaluated whether large-scale language models can create false-answer choices for language education. Specifically, we discussed the extent to which language models can predict false-answer choices in tests created by applied linguists.

CogSci 2023

Applying Large Language Models to Generate High-Quality Multiple-Choice Test Questions

**Welcome to EMNLP 2023!**

On behalf of the EMNLP 2023 Organizing Committee, I extend a warm and heartfelt welcome to all of you to the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP). It is with immense pleasure and excitement that we gather here in Singapore, a vibrant hub of innovation and technological advancement.

The conference program is packed with insightful presentations, thought-provoking workshops, and engaging networking opportunities. In addition to the technical sessions, we have also planned several social events that will provide you with the opportunity to connect with your colleagues.

The conference is held in person at the Resorts World Convention Centre in Singapore, and available on-line with the help of Underline.

For those in Singapore, I hope that you will find time to explore the exciting Sentosa Island, Gardens by the Bay, Singapore Botanic Gardens and the many other attractions unique to Singapore.

Of course, we are grateful to our sponsors and partners for their generous support of EMNLP 2023, whose contributions make it possible for us to host this world-class event.

Yuji Matsumoto (RIKEN AIP) 
EMNLP 2023 General Chair

To access the **EMNLP 2023** event page on Underline, you need to register for the Conference. 
Please follow **[this link](https://2023.emnlp.org/registration/)** for more details.

EMNLP 2023

EMNLP 2023 took place in Singapore from Dec 6th to Dec 10th, 2023.

High-quality multiple-choice tests are required to assess learners' abilities in language education. When creating such tests, it is important to create appropriate false answer choices. Because a wealth of experience is required in creating tests without easy correct answers, the number of people who can create such tests is limited.
In this study, we evaluated whether large-scale language models can create false-answer choices for language education. Specifically, we discussed the extent to which language models can predict false-answer choices in tests created by applied linguists.

poster

The 45th Annual Meeting of the Cognitive Science Society will take place in Sydney, Australia as a modified hybrid conference. Presentations from the main-session room will be live-streamed every day for all events. This will include all invited talks and award sessions. Recordings of all CogSci 2023 sessions will be made available for on-demand viewing after the conference. 
The in-person component will be held at the International Convention Centre Sydney at Darling Harbour. The Conference will take place Wednesday July 26th – Saturday July 29th, 2023 (Wednesday, July 26th is the Workshops Day).

Join us!

 [**The Program at a Glance can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2023/07/COgSci-PAAG-V2.pdf)
 [**The Detailed Program can be found here**](https://online.fliphtml5.com/ebtyf/mswz/)
 

#### **CogSci 2023 Program Book**
<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/tzgk/" seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

The 45th Annual Meeting of the Cognitive Science Society presents the latest research across cognitive science and highlights the theme of Cognition in Context.

The present study investigates the use of exam retake as an alternative to curving low test scores and as a learning opportunity for students. Students received their scores on a midterm but not which questions they got incorrect or the correct answer (i.e., no feedback). Instead of curving the exam, the students (n = 71) were offered an opportunity to retake the exam for points back and informed that retaking would not hurt their midterm. They received two days to re-study for the exam and completed a survey after their retake. Majority (87.32%) of students retook the exam and almost everyone (90.32%) scored higher on the retake. Post-retake surveys revealed that students used more effective studying strategies (e.g., testing, flash cards) for the retake and discovered better test-taking strategies for themselves (e.g., reading carefully, not second guessing, time management).

The Use of Exam Retake as a Learning Opportunity for the Content and Test-Taking Strategies

Space and time are highly interconnected in language and cognition. Here, we ask to what extent the conceptual and perceptual processes of space and time influence how word meanings are learned and extended in space and time domains. We taught 4-5-year-olds and adults a novel word with spatial (i.e., length) or temporal meaning (i.e., duration) and tested how well they learned and extended the word meaning to the other domain. We manipulated two aspects of learning multiple word meanings: the direction of extension and the modality of temporal perception. Children learned the spatial meaning better than the temporal one. Although they learned the spatial meaning, they could not extend it to the time domain. Adults learned both meanings equally successfully. However, they learned the temporal meaning better when they perceived time through visual modality. Once they learned the temporal meaning, they extended it to space as easily, regardless of modality.

Is word learning of spatial metaphors grounded in perceptual modality?

Creative people move in ways that seem aimless. Artists and mathematicians wander about, sometimes standing next to their easel or blackboard, other times standing across the room. Why do creatives expend energy on aimless movement? We propose that such movements facilitate insight by changing visually available information. We tested this mechanism in two online studies (N = 337). Participants solved an insight puzzle. We manipulated whether participants could only see a diagram representing the puzzle, as though they were standing close to it, or could also see a diagram from an earlier puzzle, as though they had stepped back. Visual access to the second diagram acted as a visual hint, increasing the rate of insight by suggesting an analogous solution. We argue that this mechanism explains the creative benefits of seemingly aimless movement. We discuss implications for understanding creativity as arising from interactions among brain, body, and environment.

Seeing the connection: Manipulating access to visual information facilitates creative insight

Understanding the relationships between gesture, speech, and thought bears on how we conceptualize language and embodied cognition. We asked whether the gestures U.S. politicians produced when presenting ideas as similar (listing) or different (contrasting) varied in handshape, form, or movement. We found that location was more likely to change during contrasting, while movement pattern was more likely to change during listing. In a follow-up experiment, we showed that these gestures impacted participants’ perceptions of how similar the items discussed by the politician were, even when the auditory information was unavailable. When participants could see politicians’ gestures, they rated listed topics as more similar and contrasted items as more different, even when the videos were muted. We conclude that gesture during political speech can both reflect and affect judgements of similarity and difference.

Gesture during political speech reflects and affects judgements of topic similarity

The directionality hypothesis says that low senses [TOUCH, TASTE] usually describe high senses [SOUND, VISION] in synesthetic metaphors, so color-taste metaphors should be inaccessible; however, newer work suggests that some color-taste metaphors are acceptable because of the types of adjectives in them. In this study, we examine how adjectival classification of color terms affects color-taste synesthetic metaphors and provide a frame-based explanation for metaphors that violate the rule. English speakers (N = 116; M = 55, F= 55, NB = 4, N/A = 2) judged acceptability in an online questionnaire of color-taste metaphors with different adjective types. To confirm, we assessed the mean accessibility for the items (1 = accessible, -1 = inaccessible) wherein a mean of zero was the reference level and would suggest no significant influence. The results show that there is a significant color-taste directionality (p < .001) for the metaphors.

Modeling Red Tastes: Quality Color Adjectives in English Color-Taste Synesthetic Metaphors

Evidence suggests that the processing of graspable object nouns elicits specific motor programs related to potential hand-object interactions. Notably, adjectives specifying manipulative features of these objects are integrated into this sensorimotor representation. The present experiment investigated the effect of adjectives denoting the position of the object in space on the sensorimotor representation of graspable object nouns. We used a reach-to-grasp compatibility task, in which participants had to categorize object nouns as artifact or natural, by performing either a power or precision grip matching or not the typical grip associated with the object. On each trial the object noun was presented with a near or far adjective. While reliable grasp-compatibility effects emerged for object nouns on RTs, this was not modulated by the spatial position denoted by the adjective. Spatial adjectives appear not to be integrated into the noun sensorimotor representation, supporting the distinction between stable and variable affordances.

The space of words: on the sensorimotor processing of variable affordances in noun-adjective combinations

Psycholinguistic studies are now more frequently considering the role of embodied information in language processing. While the perceptual features of nouns lend themselves well to cataloguing and norming, verbs also contain important perceptual information that may be useful to researchers. We present preliminary analyses of an ongoing study norming the directionality and strength of motion implied by verbs in three dimensions. We sought to assess the convergent and divergent validity of our pilot database using data derived from other psycholinguistic norm databases and distributional semantic techniques. Results suggest that this norming method is supported by convergent validity while capturing unique psycholinguistic information. More specifically, these norms may contribute to stimulus selection and refinement, generating computational representations of verbs, and novel research designs in future studies.

The spatial role of verbs in embodied language processing

Mathematical concepts are paragons of abstraction. The actual practice of mathematics, though, is decidedly concrete, material, one might even say fleshy. This raises the question, What is the body doing, when the body is doing mathematics? We answer by analyzing a video corpus of mathematicians in their natural habitat: at the blackboard, chalk in hand. Mixing qualitative and quantitative analyses, we describe systematically the ways mathematicians use their bodies to write, move, and gesture. Some surprises arise, such as the observation that mathematicians point nearly constantly but seldom produce representational gestures; and that they spend most of their time away from the blackboard at which they are writing. We discuss implications for creativity, mathematical cognition, and theories of embodied and distributed cognition more broadly.

What does the body do, when it's doing mathematics?

Current approaches to empathetic response generation typically encode the entire dialogue history directly and put the output into a decoder to generate friendly feedback. These methods focus on modelling contextual information but neglect capturing the direct intention of the speaker. We argue that the last utterance in the dialogue empirically conveys the intention of the speaker. Consequently, we propose a novel model named InferEM for empathetic response generation. We separately encode the last utterance and fuse it with the entire dialogue through the multi-head attention based intention fusion module to capture the speaker's intention. Besides, we utilize previous utterances to predict the last utterance, which simulates human's psychology to guess what the interlocutor may speak in advance. To balance the optimizing rates of the utterance prediction and response generation, a multi-task learning strategy is designed for InferEM. Experimental results demonstrate the plausibility and validity of InferEM in improving empathetic expression.

InferEM: Inferring the Speaker's Intention for Empathetic Dialogue Generation

Which auditory features drive everyday listeners’ preferences for and emotional experiences of their favorite music? To test how various features contribute to listeners’ experiences of popular music, we created a novel corpus of covers of 50 popular songs from 2008 to 2019, capturing a wide variety of genres and styles. Cover versions of popular songs offer a rich space of natural musical stimuli that match original songs on some dimensions (e.g., lyrics) but not others (e.g., genre). We pseudo-randomly selected one cover per song and asked subjects to rate similarity between clips of original and cover versions. Other subjects rated emotional valence and arousal of the same clips. We found a relationship between similarity ratings and differences in arousal ratings between originals and covers (p = 0.00069) but not differences in valence ratings (p = 0.27), suggesting that similarity ratings are more driven by perceived arousal differences.

Premium content

Downloads

Next from CogSci 2023

The Use of Exam Retake as a Learning Opportunity for the Content and Test-Taking Strategies

Similar lecture

Generating and Evaluating Tests for K-12 Students with Language Model Simulations: A Case Study on Sentence Reading Efficiency | VIDEO

Stay up to date with the latest Underline news!

PRODUCT

COMPANY

RESOURCES