Enabling Advanced EAP Students to Integrate Academic Paraphrasing Skills Effectively and Ethically with AI Assistanceá

Jerry Spring
Nazarbayev University

This assignment asks postgraduate students in an advanced EAP course to paraphrase short texts (about 100 words of an academic journal abstract). Acknowledging that students may well use AI assistance for paraphrasing, the activities aim to both widen each student’s own range and appropriateness of paraphrasing techniques and evaluate an AI assistant’s suggestions for the same text (for potential incorporation) before finalizing their own paraphrase. This process helps raise students’ critical awareness of, and ability to use AI paraphrasing suggestions in academic writing effectively and ethically. While linguistically demanding, this assignment could also be adapted for undergraduate students.

Learning Goals

Recognize and practice using various lexical and grammatical paraphrasing techniques to widen the students’ range and appropriateness of paraphrasing techniques for academic writing.
Critically analyze an AI assistant’s use of paraphrasing techniques when prompted to paraphrase a short text in comparison with their own unassisted paraphrase to deepen the students’ ability to evaluate the paraphrasing suggestions of AI assistants.
Paraphrase the same text as effectively and appropriately as possible by synthesizing their own and the AI-generated version, and include appropriate APA-style citations and references to acknowledge the source and their use of AI, to develop the students’ ability to effectively and ethically draw on AI assistants while using their own skills in paraphrasing academic texts.

Original Assignment Context: This assignment was designed for advanced (IELTS band 7) students at an English-medium university in Kazakhstan as part of a two-semester skills- and genre-based EAP course during their first year of a two-year master’s in education program. Although these students enter with advanced language skills, the EAP course (taught in both on-line and off-line forms) is needed to prepare them for the high-level genre-specific linguistic demands of preparing a thesis in English. In addition, it has become essential to acknowledge the increasing availability of AI assistance for such students, and adapt teaching and assessment practices accordingly. The assignment itself appears about 4 weeks into the first-semester course as preparation for later modules on summarizing and essay writing.

Materials Needed

Internet access: The lesson is designed to be conducted via Google drive (or equivalent) to enable students to work on Google-doc-based activities in various modes (individual, group, whole class).
Laptop access: To work effectively and efficiently on class activities, students will need laptops (rather than mobile devices).
Teaching / learning materials: Staged access enabled by the teacher as preferred through links, QR codes, etc.
Assignment document: Provided by the teacher with detailed instructions and template for completing the assignment.

Time Frame: A minimum of three 1-hour lessons for introducing and practicing the paraphrasing techniques, along with group discussion and whole-class feedback for analyzing the AI paraphrases. Ideally, teaching should be spaced so as to maximize opportunities for practice and formative feedback before completing the assessed task. Students find the assessed task intense and time consuming, so they choose 90-110 contiguous words (forming whole sentences) from the text for their final paraphrase.

Overview: Although paraphrasing sources effectively and ethically in academic writing and speaking remains a key skill for university students, two factors have radically changed students’ use of this skill. The older factor is the widespread use of computers to access sources online, and then plan and write assignments on screen. The newer factor is the rapid advance of AI tools to paraphrase texts. Ideally, paraphrasing mostly occurs ‘organically’ as students read sources, take notes, select and integrate ideas, and produce their own argument in their own words. In practice, however, students (partly due to working moving directly on screen between sources and their emerging document) quite often paraphrase more ‘mechanically’ (or simply rely on AI) after pasting in useful text from a source.

To help students engage more critically with AI-supported paraphrasing after practicing a range of paraphrasing techniques, I ask them to prompt an AI assistant to paraphrase a short, self-contained, coherent text (hence, the use, albeit artificial, of journal article abstracts) and then closely analyze which techniques it uses, and how accurately and appropriately it paraphrases the original text. For the assessed task, students first paraphrase a text without any AI support (at no risk), then prompt the AI assistant to do it. They write a brief critical analysis comparing the two efforts before producing a final paraphrase incorporating the most effective elements from each and acknowledging that support in APA style.

I designed the assignment for an advanced EAP course for education master’s students in an EMI university in Kazakhstan (now used with six groups of students over three years). Based on the tasks produced and most students’ feedback, it appears to be an effective way to integrate teaching paraphrasing techniques themselves with raising critical awareness of how to use AI support more effectively and ethically while paraphrasing.

Assignment

Activity 1: Developing awareness of lexical and grammatical paraphrasing techniques

Whole class: Brainstorm possible paraphrasing techniques (expecting less knowledge of grammatical techniques). Give access to the table below. Students match examples to grammatical paraphrasing techniques by completing the yellow column. (Note: the yellow column is shown completed in the table below.) Discuss pros/cons of different techniques (e.g., 2 & 3 difficult to use appropriately). Add other techniques students may also know. (Note: The list below is pragmatic rather than exhaustive or theoretically based, and emerged through repeated use of the materials and previous teaching experiences.)

	Write the letter into the yellow column to match the number.
1	Synonyms (no matching item for this technique)	----	---	The study explored Kazakhstan's trilingual policy. <-> The study examined Kazakhstan's trilingual policy.
2	Use an opposite verb in the negative form to give the same meaning (or vice versa).	J	A	Astana is in Central Asia. It is one of the world’s newest capital cities. It is the capital of Kazakhstan. Astana, which is in Central Asia and the capital of Kazakhstan, is one of the world’s newest capital cities. Astana, which is one of the world’s newest capital cities, is in Kazakhstan in Central Asia.
3	Use an opposite adjective / adverb / noun with a negative verb to give the same meaning (or vice versa).	E	B	uncritical <-> uncritically: Avoid uncritical use of AI. <-> Don’t use AI uncritically.
4	Use the opposite comparative adjective / adverb to give the same meaning by changing the sentence structure.	D	C	for cause-effect logic: We studied online because there was a pandemic. The pandemic caused us to study online. There was a pandemic. Therefore, we studied online. Studying online resulted from the pandemic.
5	Change active into passive voice (or vice versa).	H	D	fast <-> slow: She is a faster reader than me. <-> I am a slower reader than her.
6	Use nouns instead of verbs (or vice versa).	F	E	He is rich. <-> He isn’t poor.
7	Use nouns instead of adjectives (or vice versa).	G	F	grow <-> growth: The economy grew. <-> There was economic growth.
8	Use adjectives + nouns instead of adverbs + verbs (or vice versa).	B	G	happy <-> happiness: I felt happy. <-> I experienced happiness.
9	Combine sentences (particularly with which, that, who, etc.) or separate a complex sentence into several sentences. Note: Often there are different solutions that give a different emphasis to the complex sentence.	A	H	My family sold their car. <-> This car was sold by my family.
10	Choose a different linker / transition / verb for logical connections. Note: different solutions need different grammar, punctuation, joining/separating sentences.	C	I	Nazarbayev University was founded in 2010 on the outskirts of Astana in Kazakhstan. In 2010, Nazarbayev University was founded in Kazakhstan on the outskirts of Astana.
11	Change the order of phrases / clauses within the sentence.	I	J	She likes chocolate. <-> She doesn’t hate chocolate.
12	Change the person doing the action	K	K	We tested two groups of students <-> The study tested two groups of students.

Activity 2: Identifying opportunities for using lexical and grammatical paraphrasing techniques in a text (journal article abstract)

For this activity, the table should be reproduced in as many tabs as there are groups, with each group working on the relevant tab.

Group work: Students work together to identify potentially relevant techniques for each sentence by writing the techniques’ identifying numbers into the yellow column (can also highlight key words/structures in the sentences; can also add new techniques to the list if they find them).

Feedback: Students and/or teacher can compare across tabs to discuss differences in the groups’ suggestions.

Note: The aim here is not yet to do any paraphrasing; just to identify as many possibilities beyond merely synonym paraphrasing.

1	In this article, we present a tool and a method for measuring the psychological and cultural distance between societies and creating a distance scale with any population as the point of comparison.	1. Find synonyms. 2. Use an opposite verb. 3. Use an opposite adjective / adverb / noun with a negative verb. 4. Use the opposite comparative adjective / adverb. 5. Change active into passive voice (or vice versa). 6. Use nouns instead of verbs (or vice versa). 7. Use nouns instead of adjectives (or vice versa). 8. Use adjectives + nouns instead of adverbs + verbs (or vice versa). 9. Combine sentences (which, that etc.) or separate complex sentences. 10. Choose a different linker / transition / verb. 11. Change the order of phrases / clauses. 12. Change the person doing the action.
2	Because psychological data are dominated by samples drawn from Western, educated, industrialized, rich, and democratic (WEIRD) nations, and overwhelmingly, the United States, we focused on distance from the United States.
3	We also present distance from China, the country with the largest population and second largest economy, which is a common cultural comparison.
4	We applied the fixation index (FST), a meaningful statistic in evolutionary theory, to the World Values Survey of cultural beliefs and behaviors.
5	As the extreme WEIRDness of the literature begins to dissolve, our tool will become more useful for designing, planning, and justifying a wide range of comparative psychological projects.
6	Our code and accompanying online application allow for comparisons between any two countries.
7	Analyses of regional diversity reveal the relative homogeneity of the United States.
8	Cultural distance predicts various psychological outcomes.
Source: https://doi.org/10.1177/0956797620916782

Note: to save space, the full (e.g., APA) citations and references for the source text and use of AI are not reproduced in the materials here, but they should ideally form part of the class activities and final assignment. All AI paraphrases were made using various iterations of ChatGPT.

Options

At this point, the activities can follow (or omit) various steps depending on time available, preference, student level, etc.

Students in groups paraphrase the sentences from Activity 2 (e.g., one sentence per group to save time), followed by group peer evaluation and class feedback.
Students in groups / whole class identify the techniques used in the example human paraphrase (see table in Activity 4) and evaluate its effectiveness.

For reasons of space, Activity 3 below jumps these steps (which can still be reconstructed from the information available in the following tables).

Activity 3: Evaluating AI paraphrase of journal abstract text

For this activity, the table should be reproduced in as many tabs as there are groups, with each group working on the relevant tab.

Group work: Students work together to identify the techniques used for each sentence by writing the techniques’ identifying numbers into the final column and highlighting changed words/structures in the sentences (can also add new techniques to the list if they find them). They then highlight the first column to show the range of techniques used overall and add notes on potentially problematic paraphrasing choices.

Feedback: Students and/or teacher can compare across tabs to discuss differences in the groups’ suggestions.

Note: The table below is shown completed to provide an example of actual class work.

Technique reminder	Original text	ChatGPT paraphrase	Techns.
Find synonyms Use an opposite verb Use an opposite adjective / adverb / noun with a negative verb Use the opposite comparative adjective / adverb Change active into passive voice (or vice versa). Use nouns instead of verbs (or vice versa). Use nouns instead of adjectives (or vice versa). Use adjectives + nouns instead of adverbs + verbs (or vice versa). Combine sentences (which, that etc.) or separate complex sentences. Choose a different linker / transition / verb Change the order of phrases / clauses Clause forms	In this article, we present a tool and a method for measuring the psychological and cultural distance between societies and creating a distance scale with any population as the point of comparison.	In this article, we introduce a tool and methodology for gauging the psychological and cultural disparity among societies, thereby establishing a scale of differences applicable to any population chosen for comparison.	1,10 Methodology? Disparity? Thereby?
	Because psychological data are dominated by samples drawn from Western, educated, industrialized, rich, and democratic (WEIRD) nations, and overwhelmingly, the United States, we focused on distance from the United States.	Due to the prevalence of psychological data originating from Western, educated, industrialized, affluent, and democratic (WEIRD) countries, primarily the United States, our focus lies in quantifying the distance from the United States.	1,5,6,10 quantifying?
	We also present distance from China, the country with the largest population and second largest economy, which is a common cultural comparison.	We also provide an assessment of the divergence from China, the nation with the world's largest population and second largest economy, serving as a prevalent cultural benchmark.	1, 12 (clause form) assessment? divergence? Prevalent?
	We applied the fixation index (FST), a meaningful statistic in evolutionary theory, to the World Values Survey of cultural beliefs and behaviors.	To evaluate these differences, we employed the fixation index (FST), a significant metric in evolutionary theory, utilizing data from the World Values Survey, which examines cultural beliefs and behaviors.	1,12 (clause form) To evaluate…?! Significant metric? Utilizing data…?
	As the extreme WEIRDness of the literature begins to dissolve, our tool will become more useful for designing, planning, and justifying a wide range of comparative psychological projects.	As the excessive emphasis on WEIRD characteristics in literature begins to wane, our tool stands poised to enhance the planning, design, and rationale of a broad spectrum of comparative psychological endeavors.	1,6,7,11 Excessive..? Order of steps? Endeavours?
	Our code and accompanying online application allow for comparisons between any two countries.	Through our provided code and accompanying web-based application, comparisons between any two countries can be facilitated.	1,5,10
	Analyses of regional diversity reveal the relative homogeneity of the United States.	Investigation into regional diversity exposes the relatively uniform nature of the United States, while the measurement of cultural divergence offers insights into various psychological outcomes (Muthukrishna et al., 2020).	1,8,9,10
	Cultural distance predicts various psychological outcomes.		While? Offers insights..?

Options: Given that the assignment asks students to comment on both their own unassisted and the AI’s paraphrase, at this point, students could write group commentaries about the AI paraphrase in activity or read an example (see below).

Example Commentary for WEIRD Psychology Article Abstract

The analysis shows that ChatGPT (OpenAI, 2023) used a wide range of techniques, including perhaps an extra category. However, it used none of the reversal techniques (2-4). Overall, its paraphrasing was largely effective and maintained the coherence and cohesion of the original (Muthukrishna et al., 2020), although its use of synonyms and additional elements sometimes changed meanings and it tended to use fewer techniques than the human paraphraser within individual sentences. The paraphrasing can be analyzed in two main sections: use of synonyms and use of other techniques.

Use of Synonyms

The ChatGPT paraphrase used a lot of synonyms. Many of these work well, but some are more questionable for four main reasons. First, the word used is sometimes not an effective synonym. For example, “method” and “methodology” (sentence 1) apply to different conceptual levels (de Aguiar, 2024); “excessive emphasis on” has a different meaning to “extreme” (sentence 5); “offers insights into” has a different meaning to “predicts” (sentence 7) and perhaps also a qualitative rather than quantitative methodological connotation.

Second, the synonyms used are sometimes more marked or with a different connotation than the originals, which may distort meaning. For example, “disparity,” refers to inequality or difference, but with a sense of unfairness (Cambridge Dictionary, n.d.-a), whereas the original text used the more neutral word “difference.” In sentence 3, “assessment” has an evaluative connotation beyond the more neutral “present” while “divergence” suggests change (i.e., increasing or decreasing difference; Cambridge Dictionary, n.d.-b) rather than just “difference” at one point. Related to this is the need to retain technical terms. ChatGPT generally did this (although some of its changes to “difference” were inadvisable). Finally, it changed “rich” to “affluent” in sentence 2, which means the word no longer matches the acronym WEIRD.

Third, ChatGPT sometimes inferred extra ideas, which the original authors may not have intended, for example the cause-effect linker “thereby” in sentence 1; “quantifying” in sentence 2; “To evaluate these differences” in sentence 4; and “while” in sentence 7 (although this arguably improves the original’s cohesion).

Fourth, readability may be reduced if the chosen synonym is a noticeably less frequent word, as in “prevalent” vs. “common” in sentence 3 or “wane” vs. “dissolve” in sentence 5 (see Google Books Ngram Viewer: https://books.google.com/ngrams/).

Use of Other Techniques

Turning to other techniques, ChatGPT most frequently used technique 10 (changing linkers). This was particularly effective in sentence 2 (“Due to”) and arguably improved the original’s cohesion in sentences 1 (“thereby”) and 7 (“while”). It made several effective part-of-speech changes (techniques 6, 7, and 8): “are dominated by” to “the prevalence of” (sentence 2); “design” and “justifying” to “design” and “rationale,” respectively (sentence 5); “become more useful” to “poised to enhance” (sentence 5); “relative homogeneity” to “relatively uniform nature” (sentence 7). It made two effective and quite sophisticated active-passive transformations: “[are] drawn from” to “originating” (sentence 2); “allow for” to “can be facilitated” (sentence 6). It used technique 9 to combine two sentences (sentence 7) and changed the order of ideas in sentence 5 (technique 11), although this perhaps changed meaning: does design logically come after planning (as in the paraphrase) or before it (as in the original)? Finally, ChatGPT modified clausal structures in two sentences, thereby demonstrating an additional technique: “which is” to “serving as” (sentence 3); insertion of “which examines” in sentence 4.

Activity 4: Comparing human and AI paraphrasing techniques in a text (journal article abstract)

Students should have editing access to annotate the shared document - if done as a whole class activity.

Whole class: Students critically compare the techniques used by the human and AI for each sentence by writing the techniques’ identifying numbers into each cell and highlighting key words/structures in the sentences. Students should consider efficiency, accuracy, register, etc.

Original text (160 words)	Human paraphrase (186 words)	ChatGPT paraphrase (192 words)
In this article, we present a tool and a method for measuring the psychological and cultural distance between societies and creating a distance scale with any population as the point of comparison.	In this article, a tool and method are presented to measure how distant societies are, culturally and psychologically.	In this article, we introduce a tool and methodology for gauging the psychological and cultural disparity among societies, thereby establishing a scale of differences applicable to any population chosen for comparison.
	The study also develops a distance scale that can make any group the focus of the contrast.
Because psychological data are dominated by samples drawn from … WEIRD nations, and overwhelmingly, the United States, we focused on distance from the United States.	Due to the dominance of Western, educated, industrialized, rich, and democratic (WEIRD) data samples in psychology, especially from the USA, the focus is on distance from this country.	Due to the prevalence of psychological data originating from Western, educated, industrialized, affluent, and democratic (WEIRD) countries, primarily the United States, our focus lies in quantifying the distance from the United States.
We also present distance from China, the country with the largest population and second largest economy, which is a common cultural comparison.	In addition, because China is the world’s largest country by population, second biggest economically, and often used to compare populations culturally, this study calculates distance from this country as well.	We also provide an assessment of the divergence from China, the nation with the world's largest population and second largest economy, serving as a prevalent cultural benchmark.
We applied the fixation index (FST), a meaningful statistic in evolutionary theory, to the World Values Survey of cultural beliefs and behaviors.	Evolutionary theory makes valuable use of the fixation index (FST).	To evaluate these differences, we employed the fixation index (FST), a significant metric in evolutionary theory, utilizing data from the World Values Survey, which examines cultural beliefs and behaviors.
	This is applied in our study to the World Values Survey, which measures cultural behaviors and beliefs.
As the extreme WEIRDness of the literature begins to dissolve, our tool will become more useful for designing, planning, and justifying a wide range of comparative psychological projects.	The use of this tool will be increasingly applicable to design, plan, and justify various studies in comparative psychology once the current WEIRD literature no longer remains so extreme.	As the excessive emphasis on WEIRD characteristics in literature begins to wane, our tool stands poised to enhance the planning, design, and rationale of a broad spectrum of comparative psychological endeavors.
Our code and accompanying online application allow for comparisons between any two countries.	Any two countries can be compared using the code and associated online software presented here.	Through our provided code and accompanying web-based application, comparisons between any two countries can be facilitated.
Analyses of regional diversity reveal the relative homogeneity of the United States.	Analyzing regional diversity in the USA shows there is not much heterogeneity.	Investigation into regional diversity exposes the relatively uniform nature of the United States, while the measurement of cultural divergence offers insights into various psychological outcomes (Muthukrishna et al., 2020).
Cultural distance predicts various psychological outcomes.	A variety of psychological effects are predictable from cultural distance (Muthukrishna et al., 2020).

Activity 5: Comparing different AI paraphrases of a text (journal article abstract)

This table compares different iterations of ChatGPT’s paraphrases and compares the effects of prompt design.

Whole class: Students critically compare the techniques used by the AI for each sentence by writing the techniques’ identifying numbers into each cell and highlighting key words/structures in the sentences (note: the table is shown completed). Students can add comments about particular differences they note regarding efficiency, accuracy, register, etc. They can also compare the latest AI version back to the human version in Table 4.

Here is a further comparison between ChatGPT’s 2023 paraphrase and two prompts in 2024 (ChatGPT, 2024): #1. “Can you paraphrase the following academic journal article abstract?” #2. “Can you paraphrase the same abstract using as wide a range of paraphrasing techniques as possible?” Highlight changes in the two new versions. Bear in mind synonyms and the other techniques. In brackets, note the techniques used for each sentence. Do the latest changes improve or worsen the paraphrase?

Original text (160 words)	Last year (2023) (192 words)	2024 Version 1 (162 words)	2024 Version 2 (182 words)
In this article, we present a tool and a method for measuring the psychological and cultural distance between societies and creating a distance scale with any population as the point of comparison.	In this article, we introduce a tool and methodology for gauging the psychological and cultural disparity among societies, thereby establishing a scale of differences applicable to any population chosen for comparison. (1,6,10)	This article introduces a tool and method for assessing psychological and cultural distances between societies, creating a scale with any population as a reference point. (1, 6)	This study introduces a novel tool and methodology for quantifying the psychological and cultural gaps between different societies, allowing for the creation of a distance scale where any population can serve as the benchmark. (1,6 )
Because psychological data are dominated by samples drawn from Western, educated, industrialized, rich, and democratic (WEIRD) nations, and overwhelmingly, the United States, we focused on distance from the United States.	Due to the prevalence of psychological data originating from Western, educated, industrialized, affluent, and democratic (WEIRD) countries, primarily the United States, our focus lies in quantifying the distance from the United States. (1,5,6,10)	Given the predominance of psychological research based on samples from Western, educated, industrialized, rich, and democratic (WEIRD) countries, particularly the United States, the study emphasizes measuring distance from the U.S. (1,5,6,10)	Due to the overrepresentation of data from Western, educated, industrialized, rich, and democratic (WEIRD) societies, particularly the United States, in psychological research, this study concentrates on measuring the distance from the U.S. (1,6,10 )
We also present distance from China, the country with the largest population and second largest economy, which is a common cultural comparison.	We also provide an assessment of the divergence from China, the nation with the world's largest population and second largest economy, serving as a prevalent cultural benchmark. (1, 12)	It also includes comparisons with China, a country frequently used in cultural studies due to its large population and economic significance. (1,6,10,11)	Additionally, it examines the distance from China, a nation frequently compared in cultural research due to its massive population and economic stature. (1,6,10,11)
We applied the fixation index (FST), a meaningful statistic in evolutionary theory, to the World Values Survey of cultural beliefs and behaviors.	To evaluate these differences, we employed the fixation index (FST), a significant metric in evolutionary theory, utilizing data from the World Values Survey, which examines cultural beliefs and behaviors. (1, 12)	The authors applied the fixation index (FST), a key statistic in evolutionary theory, to data from the World Values Survey on cultural beliefs and behaviors. (1,13?14?)	The authors utilized the fixation index (FST), a significant metric in evolutionary biology, to analyze the World Values Survey, which captures cultural beliefs and behaviors. (1,12,13?14?)
As the extreme WEIRDness of the literature begins to dissolve, our tool will become more useful for designing, planning, and justifying a wide range of comparative psychological projects.	As the excessive emphasis on WEIRD characteristics in literature begins to wane, our tool stands poised to enhance the planning, design, and rationale of a broad spectrum of comparative psychological endeavors. (1,6,7,11)	As the dominance of WEIRD-focused research diminishes, this tool is expected to be increasingly valuable for designing and validating various comparative psychological studies. (1, )	As the dominance of WEIRD-based research starts to wane, this tool is anticipated to become increasingly beneficial for the conceptualization, execution, and validation of diverse comparative psychological studies. (1,6 )
Our code and accompanying online application allow for comparisons between any two countries.	Through our provided code and accompanying web-based application, comparisons between any two countries can be facilitated. (1,5,10)	The provided code and online application enable comparisons between any two countries. (1, )	The study also offers code and an online platform that facilitate comparisons between any two nations. (1,13)
Analyses of regional diversity reveal the relative homogeneity of the United States.	Investigation into regional diversity exposes the relatively uniform nature of the United States, …. (1,8,9,10)	Analyses of regional diversity highlight the relative homogeneity within the United States, ... (1,5,14?)	Regional diversity analyses underscore the relative uniformity within the United States. (1,5,6,14? )
Cultural distance predicts various psychological outcomes (Muthukrishna et al., 2020).	… while the measurement of cultural divergence offers insights into various psychological outcomes (Muthukrishna et al., 2020).	...and the concept of cultural distance is shown to predict different psychological outcomes (Muthukrishna et al., 2020).	The study further demonstrates that cultural distance is a predictor of various psychological phenomena (Muthukrishna et al., 2020).

Assignment task instructions/template:

For the assessed task, students should paraphrase a certain number of contiguous words/sentences from the selected text (e.g., journal article abstract), first without any AI help (unassessed step). Second, they use AI to paraphrase the same text. Third, they write a commentary comparing the two paraphrases (assessed step). Fourth, they finalize their paraphrase (assessed step) drawing on the AI if they want (to a limited extent defined by the teacher). Finally, their work should acknowledge the source and use of AI appropriately according to the relevant citation/reference system.

At least in the way I have implemented it, the assessed task needs to be clearly set out for the students because it has multiple steps that students need to follow correctly to enable effective marking and feedback. Please contact me (jeremy.spring@nu.edu.kz) if you would like to see the instructions / template document that I used.

Acknowledgements

I've received very helpful feedback from my colleagues and students on various iterations of the materials over the last three years.

References

American Psychological Association. (2020). Publication manual of the American Psychological Association 2020: The official guide to APA style (7th ed.). American Psychological Association.

Muthukrishna, M., Bell, A., Henrich, J., Curtin, C., Gedranovich, A., McInerney, J., & Thue, B. (2020). Beyond western, educated, industrial, rich, and democratic (WEIRD) psychology: Measuring and mapping scales of cultural and psychological distance. Psychological Science, 31(6), 678-701. https://doi.org/10.1177/ 0956797620916782

OpenAI. (2023). ChatGPT (Aug 20 version) [Large language model]. http://chat.openai.com/chat

OpenAI. (2024). ChatGPT (Sep 6 version) [Large language model]. http://chat.openai.com/cha