172 Students at a Japanese National University Rated Self-Access Language Learning Services Above AI Chatbots for Speaking, Motivation, and Personalized Guidance, Seeing the Two as Complementary

Learners view SALL services and AI chatbots as complementary resources serving different purposes rather than as competitors or substitutes. Prior work has focused largely on AI's benefits for independent learning and given limited attention to SALL services themselves, and has rarely examined how the two might coexist; this study places learners' relative evaluations of both forms of support in a single survey for direct comparison. Survey data from 172 participants (99% response rate), with paired-samples t-tests on five parallel items and content analysis of two open-ended questions; participants strongly agreed the two serve distinct purposes (M = 3.76, SD = .43) and that SALL services fulfill roles AI chatbots cannot replace (M = 3.61, SD = .56).

Across all five parallel items, SALL services were rated significantly higher than AI chatbots, with the largest gaps for speaking development and motivation. Much prior research on AI-assisted language learning reports positive chatbot effects; this study offers a direct paired comparison of both forms of support from the same learners, with effect sizes. Paired-samples t-tests: speaking t(171) = 15.56, p < .001, d = 1.19; motivation t(171) = 13.27, p < .001, d = 1.01; personalized advice and feedback d = 0.55; overall contribution d = 0.49; writing d = 0.43, with effect sizes ranging from moderate to large.

Learners valued AI chatbots for convenience and accessibility while identifying limitations in accuracy, independent thinking, and authentic human interaction. The study presents learners' reservations about chatbots alongside their expectations for SALL services, complementing a literature that has emphasized AI's benefits. The composite mean for chatbot limitations was M = 3.17 (SD = 0.84), with the highest-rated items being that responses may not always be accurate (M = 3.59), that chatbots may weaken independent thinking (M = 3.50), and fewer opportunities to interact with people in English (M = 3.34); in the open-ended responses, 29% of participants considered chatbots more important, citing anytime-anywhere access (18 coded instances).

Learners believed SALL services should remain and adapt to AI developments, with their distinctive value lying in authentic communication, interpersonal support, and motivation maintenance. The study asked about SALL's future role explicitly in relation to AI chatbots, capturing learners' visions of how institutional support should adjust. Participants agreed SALL services would continue to play an important role (M = 3.29, SD = .65), were less likely to agree their necessity would decrease as AI develops (M = 2.55, SD = .76), and agreed they should be used in combination with chatbots (M = 3.39, SD = .70); 142 participants answered the open-ended question, emphasizing speaking practice (41 coded instances), face-to-face interaction (13), human interaction (9), and motivation maintenance (15).

The study concerns 172 undergraduates at one Japanese national university (mostly first- and second-year English majors) and examines their perceptions and future expectations regarding SALL services and AI chatbots rather than actual usage effects; because the institution has no dedicated self-access center, most participants had never used SALL services (97.1%), and their answers rested on a five-minute English presentation and written Japanese explanation provided before the survey, so the results should be read as learners' anticipated perceptions of SALL services. This framing can guide universities in designing self-access services when introducing AI chatbots, in addressing barriers such as not knowing what services exist or how to access them, and in providing a starting point for research with larger and more diverse populations.

Most participants had no actual experience using SALL services, so their ratings reflect expectations rather than experience, and whether evaluations would match after real use remains an open question; the study draws on a single institution, the open-ended questions were not formally piloted, and one question asked participants to compare the two forms of support directly while another framed SALL's future in relation to AI, which may have encouraged comparative responses, leaving open whether alternative wording would yield different perspectives; in addition, the loaded text is an incomplete version, and some tables (such as Table 1, Table 2, Table 3, and Table 5) are not fully reproduced in the body, so details of scale reliability and some descriptive statistics can only be drawn from the narrative.