Innovations in Practice

RELC Journal
1–11
A Pedagogical Chatbot: A © The Author(s) 2022
Article reuse guidelines:
Supplemental Language sagepub.com/journals-permissions
DOI: 10.1177/00336882211067054
Learning Tool journals.sagepub.com/home/rel

Lucas Kohnke
The Education University of Hong Kong, Hong Kong

Abstract
This innovations in practice article introduces a chatbot that was developed to support and moti-
vate second language learners during the COVID-19 pandemic. The chatbot was designed to facili-
tate active, out-of-class language learning to supplement in-class input. It can adapt to learners’
abilities and pace by chatting with them, thus providing differentiated instruction. I describe the
rationale for the development and implementation of the chatbot and its pedagogical implications.
My aim is to motivate and assist chatbot developers and second language teachers in designing and
implementing chatbots that will enhance motivation and independent learning in language students.

Keywords
Chatbot, English for academic purposes, COVID-19, language learning, technology, independent learning

Introduction
In second language (L2) courses, teachers use technological tools such as mobile appli-
cations, student response systems, interactive websites and augmented/virtual reality
(Godwin-Jones, 2016) to facilitate language input and output (Kessler, 2018). While
these tools have become mainstays of language pedagogy, the use of chatbots appears
to be rare. A chatbot is a ‘text-based, turn-based, task-fulfilling’ software program
(Jain et al., 2018: 904) ‘that interacts with users on a certain topic’ (Smutny and
Schreiberova, 2020: 1). Chatbots motivate L2 learners (Kohnke (in-press,a); Fryer
et al., 2017), helping them to become more comfortable conversing in English by
engaging in human-like dialogue (Fryer and Carpenter, 2006).
Interaction ‘is an important factor for successful L2 acquisition’ (Ellis, 1999: 16). It nudges
learners to process language output (Swain, 1985) and offers opportunities to produce modified
output, which is important for language acquisition (Mackey, 2012). Carefully designed chatbot
tasks (e.g. branching, right or wrong answers with feedback) allow L2 learners to negotiate
meaning by interacting with the chatbot. Noticing, a crucial aspect of language learning

Corresponding author:
Lucas Kohnke, Department of English Language Education, Education University of Hong Kong, 10 Lo Ping
Road, Tai Po, New Territories, Hong Kong.
Email: [email protected], [email protected]
2 RELC Journal 0(0)

(Schmidt, 1990; Swain and Lapkin, 1995), can be encouraged by chatbots as well. Interacting
with chatbots can help L2 learners notice mismatches between their language production and
target forms (e.g. active vs passive voice, simple vs perfect tense). Especially when face-to-face
classes are impossible, chatbots can give L2 learners opportunities to communicate in the target
language. In addition, they can answer routine teaching questions instantly, freeing up teachers’
time and allowing learners to complete tasks without delays and to feel less isolated. However,
chatbots, despite their benefits to learners, are rarely utilised by language teachers (Fryer et al.,
2020; Lee et al., 2020). To make this technology more accessible, I explain in this article how
English teachers and non-programmers can build a pedagogical chatbot for L2 teaching and
learning. I also describe how such a chatbot was received by teachers and students at a university
in Hong Kong.

The Teaching Context: A Need for Innovation

This innovation was developed at an English-medium institution in Hong Kong with a com-
pulsory first-year course – English for University Studies – offered to approximately 2500
students each academic year. The course helps students succeed in their classes by develop-
ing English competencies needed for academic purposes, including referencing sources,
paraphrasing, summarising written and spoken texts, writing expository essays and deliver-
ing oral presentations. As a teacher and course coordinator, I have found that students: (a)
show little interest in completing out-of-class activities; (b) require a great deal of individual
guidance in developing their academic skills; and (c) ask similar questions each semester.
This remained the case after the suspension of face-to-face classes, first in November
2019 due to social unrest (Wong and Moorhouse, 2020) and then due to the ongoing
COVID-19 pandemic (Moorhouse and Kohnke, 2021). During the transition to fully
online teaching and learning, teachers resorted to assigning videos and extra reading to
make up for the lack of face-to-face input. Although this works for the proficient students,
the less skilled fell behind due to the lack of directed input, difficulties in self-regulating and
lack of learning strategies. This motivated me to develop an educational text-based chatbot
for students in this course that could answer common questions and provide active learning
direction and support during the suspension of in-person classes.

Description of the Innovation

As a teacher and non-programmer, I chose to develop the chatbot using Dialogflow (https://
cloud.google.com/dialogflow), part of Google’s artificial intelligence platform.
Dialogflow uses a visual flow builder that makes it easy for people with little or no previous
experience to see, edit and share work. It supports omnichannel implementation, making it
possible to reach students on multiple platforms (e.g. websites, mobile apps and messenger
services), and integrates natural language understanding models capable of recognising
complex language queries – a particularly important feature as students may misspell or
be unable to clearly articulate their questions. The text-based chatbot I developed
follows a set of response rules based on a pre-established flow/pattern derived from ques-
tions posted by students. The chatbot was implemented on the language centre’s Facebook
page to accommodate L2 learning in a more social, open context. Facebook was considered
the most appropriate platform as the interface was familiar to the learners. They would not
have to download and install extra applications and they could access the chatbot 24/7.
Kohnke 3

Development and Implementation of the Chatbot

This section describes the chatbot’s development and implementation in an English for
Academic Purposes (EAP) course.

Phase 1: Needs Analysis

In the first phase, I performed a needs analysis by identifying the most common topics
students searched for on our independent learning website (see Figure 1). As seen in
Table 1, they most often searched for information about referencing, the International
English Language Testing System (IELTS), spoken English and grammar.

Table 1. Website hit rates during the 2019–2020 academic year.

No. Topic Hits Visitors

1. Referencing (general) 239,049 227,381
2. Referencing APA 60,258 40,482
3. IELTS (general) 40,261 32,586
4. Grammar 31,981 301,141
5. Spoken English 25,980 24,761
APA: American Psychological Association; IELTS: International English Language Testing System.

I also arranged interviews with 9 EAP teachers and 14 students to gain further insight
into student needs and wants. Involving EAP teachers at this stage helped me to under-
stand what the chatbot could contribute pedagogically. Both teachers and students con-
sidered referencing skills important (see Table 2). Teachers believed that the chatbot
could also be used to help students sign up for workshops, appointments with English
teachers and massive open online courses (MOOCs) and small private online courses
(SPOCs). Students indicated that they would like to use a chatbot to learn more about
grammar, speaking skills and essay types. Both teachers and students wanted the
chatbot to steer them to new resources.

Table 2. Teachers’ and students’ suggestions for chatbot content.

No. Teachers Students

1. Referencing skills (integral and non-integral) Referencing
2. Speaking/writing appointments Grammar
3. Workshops Speaking
4. MOOCs/SPOCs Course information
5. Paragraph structure Essay types
MOOCs: massive open online courses; SPOCs: small private online courses.

Phase 2: Developing and Training the Chatbot

After completing the needs analysis, I developed over 100 questions covering those
topics identified in the analysis, with multiple answers to each arranged using branching.
Figures 1 and 2 illustrate the branching for ‘referencing’ and ‘tense’, respectively.
4 RELC Journal 0(0)

Figure 1. Example of referencing branching.

Figure 2. Example of tense branching.

Kohnke 5

After the content was entered into the chatbot, I invited 3 teachers and 12 students to
pilot the user experience to ensure it provided correct and useful answers. This process
revealed that the chatbot struggled to recognise spelling errors and lengthy user input.
Therefore, multiple keywords and spelling variations, including misspellings, were pro-
grammed for each possible search term to benefit students of varying English ability
levels.

Phase 3: Implementation
The chatbot was employed for two months during the 2021 spring semester when
face-to-face classes were suspended to help students complete assigned tasks and find
additional resources (e.g. online guides, mentoring and workshops). To promote the chat-
bot’s use, teachers asked students to refer to it when completing various activities. An
icon that accompanied certain activities indicated that they should be completed with
the help of the chatbot. For example, after completing Unit 3, Using Sources in
Academic Writing, the students were asked to review the differences between referencing
systems and then could choose to use the chatbot to test themselves by completing the
activity shown in Table 3.

Task Example 1 – Activity 4 – Referencing Styles.

Table 3. Referencing styles.

Which referencing style … APA Harvard IEEE Vancouver

Uses in-text and end-text citations √ √ √ √
Uses an author-year style
Uses a numbered style
Includes years in the in-text citations
Includes page numbers for quotes
Has a reference list in alphabetical order
Has a reference list in numbered order
Puts the surname before the initial in the reference list
APA: American Psychological Association; IEEE: The Institute of Electrical and Electronics Engineers.

Another chatbot-assisted activity, shown in Table 4, followed the introduction of in-text

citations. Students were asked to consolidate their knowledge by using the chatbot to
place citations in one of four categories.

Task Example 2 – Activity 5 – Identifying Integral and Non-Integral Citations

1 = Integral citation – Paraphrase 3 = Non-integral citation – Paraphrase

2 = Integral citation – Direct quote 4 = Non-integral citation – Direct quote
6 RELC Journal 0(0)

Table 4. Categorising citations.

Citations 1 2 3 4
A Lee and Chan (2020) argue that this more diverse discussion has the potential
to create a better-informed society that understands the complexities of
living in a world of competing interest groups
B Exams can also be motivating because they provide a concrete goal for learning
and enable students to develop a study plan (Dembo and Seli, 2018).
C Breslawski (2021: 74), for example, found that ‘extrinsic motivational methods’,
such as competition between students for receiving high grades, encourage
them to spend more time on studies.
D The existing cyberculture has created various levels of ‘digital dependency’
among the public, which could lead to a decline in genuine learning (Parker,
2019: 12).

Task Example 3. To improve their spoken English, students planned a five-minute aca-
demic presentation that included academic sources. They were referred to the chatbot
for guidance in aspects of the process (pronunciation, planning, developing a checklist,
assistance with phrasing and optional tutorial sign-ups) (see Figures 3 and 4).

Figure 3. Chatbot speaking help.

Kohnke 7

Figure 4. Presentation guidance.

Paragraph Development Example. The chatbot was also employed to introduce paragraph
writing and offer supplemental information on topics such as paragraph structure, topic
sentences and supporting sentences (see Figures 5–9 for branching).

Figure 5. Paragraph.

Figure 6. Paragraph structure.

Figure 7. Options.
8 RELC Journal 0(0)

Figure 8. Topic sentence.

Figure 9. Supporting sentences.

To introduce the two types of essays required in the course (For and Against and
Problem and Solution), the students were instructed to begin by asking the chatbot
about the differences between them (see Figures 10 and 11) and then carry out a
general search to supplement their ideas before presenting to them to the class.

Figure 10. For and against essay.

Figure 11. Problem and solution essay.

Kohnke 9

Furthermore, students were asked to use the chatbot if they were unsure about the
tense to use (see Figure 2) or the difference between the active and passive voice. I
hoped that the students would be able to self-regulate their learning by employing the
chatbot to decide how much input was needed and obtain guidance on completing tasks.

Phase 4: Evaluation
The chatbot was implemented successfully and perceived as useful in language learn-
ing. The evaluation involved 20 students in 4 focus groups and one focus group of 4
teachers. The conversations were audio-recorded and transcribed. The comments were
generally positive. The participants focused on the innovative nature of the chatbot and
the way it enriched the learning experience by its ease of use and provision of
just-in-time learning. S6 shared that ‘the bot was fun and something new compared
to the usual apps or websites’. Many students mentioned that it provided timely teach-
ing, clarification and reassurance when they noticed gaps in their knowledge (see
Swain, 1985). For example, S12 found it helpful ‘to self-correct my quotes’ by com-
paring what she wrote with the example the chatbot supplied. Another student, S4,
commented: ‘I think it is helpful to get a quick answer when I’m stuck on an activity.
You know, this makes me more motivated to continue with the work.’ S7 elaborated:
‘When reviewing paragraph structure and trying to complete the homework, I couldn’t
remember exactly how I should write a topic sentence. So, I just typed it in, and I got
an example and could continue writing the paragraph.’ The branching arrangement of
the answers to questions ‘made it logical to find help’ (S10) and ‘saved time over
reading a long paragraph explaining a point’ (S8). This feedback highlights the import-
ance of carrying out a needs analysis to ensure that the information included in a
chatbot and the wording of the material match students’ needs and wants. Students
felt ‘learning to be ongoing’ (S1) when interacting with the chatbot and that they
had a ‘mentor’ (S16) and ‘study partner’ (S19) (see Fryer et al., 2019). Overall, the
focus group results demonstrated that the chatbot helped students fill the gaps in
their knowledge without resorting to ‘lucky guesses’ (S9) or ‘trial and error’ (S14)
and stimulated their interest in language learning (Kohnke (in-press, b); Fryer et al.,
2017).
However, the students indicated that they would like additional learning materials
developed for use with the chatbot. For example, several thought that more course mate-
rials should be included, with questions and answers geared towards the topics covered in
each class. This suggests that students viewed the chatbot as a partner in learning. Though
course materials were not purposefully developed for the chatbot, the students seemed
to have enjoyed and learned from it and were enthusiastic about continuing to use it.
My focus was on creating and implementing a chatbot rather than developing supplemen-
tal learning materials although that is a promising pedagogical area for future
development.
The teachers in the focus group indicated that students completed tasks more effect-
ively using the chatbot. They ‘appeared to appreciate the ease’ (T1) of learning with
the chatbot, although it took some time for them to become ‘better with self-managing
their learning’ (T3). All four teachers noted receiving fewer emails asking how to com-
plete homework or sign up for workshops, and they commented that their students were
generally more proactive. I noticed this about my students as well; they seemed more
10 RELC Journal 0(0)

willing to tackle unfamiliar tasks and language-related challenges. Three teachers (T2,
T3, T4) commented that they would need training and guidance to prepare materials to
use with the chatbot if it were incorporated into the class. All agreed that the chatbot
reduced the time spent providing answers in class and that both stronger and weaker stu-
dents learned from the chatbot.

Reflection and Future Direction

Chatbots are useful in L2 instruction, having the potential to motivate students to com-
plete out-of-class work, prepare for class more confidently and permit them to study
whenever and wherever they wish. Moreover, they accommodate varied learning
needs. The question-and-answer format provides differentiated, just-in-time learning
and can be used in flipped classrooms. Although this innovation was applied in an
EAP course, it could be employed in other language-learning contexts. For example,
L2 teachers could use the chatbot to introduce details of grammar (e.g. active/passive
voice) or paragraph structure (e.g. topic sentences or thesis statements).
Nevertheless, there are some drawbacks language teachers should consider before
introducing chatbots. Some students may not have a laptop or smartphone with an inter-
net connection outside the school environment; teachers should be mindful of providing
equal learning opportunities. Furthermore, using a chatbot in the classroom may distract
students from the task at hand, increasing the teacher’s classroom management responsi-
bilities. While these are significant caveats, I believe that the pedagogical affordances of
this purpose-built chatbot for L2 learners outweigh these limitations.
I offer the following suggestions for teachers interested in developing and implement-
ing a text-based chatbot to optimise language teaching. First, consider open-source bots
such as Dialogflow or Botsify, which require little, if any, previous programming experi-
ence. Second, be aware of the time commitment required to develop a chatbot. Prepare in
advance for the next term by building up a question-and-answer bank, noting common
student errors and questions. Ensure that the questions and answers are of good quality
to avoid reliability issues. Third, orient the chatbot around task-based teaching and
begin with short, simple, controlled tasks. Slowly enhance the chatbot until it can
handle larger, less structured tasks. That said, the chatbot cannot replace human teachers.
Teachers still need to provide a constant, responsive presence to handle unexpected ques-
tions and make necessary adjustments to assigned tasks.
This chatbot was well received by teachers and students alike. I encourage language
teachers to explore the pedagogical possibilities of educational chatbots that cater to their
learners’ needs and reiterate that teachers with little or no programming experience can
develop chatbots to enrich English language learning.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship and/or
publication of this article.

Funding
The author received no financial support for the research, authorship and/or publication of this
article.
Kohnke 11

ORCID iD
Lucas Kohnke https://orcid.org/0000-0001-6717-5719

References
Kohnke, L. (in-press, a). L2 Learners’ Perception of a Chatbot as a Potential Independent Language
Learning Tool. International Journal of Mobile Learning and Organisation.
Kohnke, L. (in-press, b). A Qualitative Exploration of Student Perspectives of Chatbot Use During
Emergency Remote Teaching. International Journal of Mobile Learning and Organisation.
Ellis R (1999) The Study of Second Language Acquisition. Shanghai: Shanghai Foreign Languages
Education Press.
Fryer LK, Ainley M, Thompson A, et al. (2017) Stimulating and sustaining interest in a language
course: An experimental comparison of chatbot and human task partners. Computers in
Human Behaviour 75: 461–468.
Fryer LK and Carpenter R (2006) Emerging technologies – bots as language learning tools.
Language Learning & Technology 10(3): 8–14.
Fryer LK, Coniam D and Carpenter R (2020) Bots for language learning now: Current and future
directions. Language Learning & Technology 24(2):8–22.
Fryer LK, Nakao K and Thompson A (2019) Chatbot learning partners: Connecting learning
experience, interest and competence. Computers in Human Behaviour 93: 279–289.
Godwin-Jones R (2016) Looking back and ahead: 20 years of technologies for language learning.
Language Learning & Technology 24(2): 8–22.
Jain M, Kumar P, Kota R, et al. (2018) Evaluating and informing the design of chatbots. In:
Proceedings of the 2018 on designing interactive systems conference 2018, Hong Kong,
China, 9–13 June 2018, pp.895–906.
Kessler G (2018) Technology and future of language teaching. Foreign Language Annals 51(1):
205–218.
Lee JH, Yang H, Shin D, et al. (2020) Chatbots – technology for the language teacher. ELT Journal
74(3): 338–344.
Mackey A (2012) Interaction, Feedback and Task Research in Second Language Learning:
Methods and Design. Cambridge: Cambridge University Press.
Moorhouse BL and Kohnke L (2021) Responses of the English-language teaching community to
the COVID-19 Pandemic. RELC Journal 52(3): 359–378.
Schmidt R (1990) The role of consciousness in second language learning. Applied Linguistics
11(2): 129–158.
Smutny P and Schreiberova P (2020) Chatbot for learning: A review of educational chatbots for the
facebook messenger. Computers & Education 151: 1–11.
Swain M (1985) Communicative competence: Some roles of comprehensible input and comprehen-
sible output in its development. In: Gass S and Madden C (eds) Input in Second Language
Acquisition. Rowley: Newbury House, pp.235–253.
Swain M and Lapkin S (1995) Problems in output and the cognitive processes they generate: A step
towards second language learning. Applied Linguistics 16: 371–391.
Wong KM and Moorhouse BL (2020) The impact of social uncertainty, protests, and COVID-19 on
Hong Kong teachers. Journal of Loss and Trauma 25(8): 649–655.