Discipline-specific academic wordlists (DSAWLs) have become central tools in English for Specific Purposes (ESP) instruction, yet scientometrics -a rapidly growing interdisciplinary subfield of library and information science concerned with the quantitative study of scholarly communication- has lacked a validated, corpus-derived vocabulary resource despite its highly specialized lexicon. This dissertation reports a four-phase, mixed-methods research programme that addresses this gap by (a) systematically reviewing the methodological evolution of DSAWL research, (b) developing a discipline-specific academic wordlist for scientometrics, (c) empirically testing its pedagogical value, and (d) situating the resulting resource within the changing landscape of generative artificial intelligence (GenAI). Phase One conducts a systematic review of 44 DSAWLs published between 2000 and 2025, revealing a strong disciplinary concentration in medicine, engineering, and applied linguistics, a continuing reliance on research-article corpora, persistent dominance of the word-family counting unit, and a widespread absence of rigorous validation procedures. Building on this methodological benchmark, Phase Two compiles an 8.5-million-word corpus of 1,800 scientometrics research articles and applied frequency, range, and dispersion criteria to derive a 1,105-word-family Scientometrics Academic Word List (SAWL), which achieved 88.64% coverage of the corpus and substantially outperformed Coxhead's (2000) general Academic Word List. Phase Three tests the pedagogical efficacy of this resource in a twelve-week quasi-experimental intervention with 60 undergraduate Library and Information Science students, showing that explicit instruction using the discipline-specific list produced significantly larger gains in vocabulary knowledge, reading accuracy, and reading speed than instruction using the general Academic Word List, with large effect sizes across all measures. Phase Four extends the enquiry qualitatively, interviewing 42 ESP/EAP students and instructors about the opportunities, challenges, and future of wordlist pedagogy amid the proliferation of GenAI tools, revealing a shared vision of a hybrid model in which a curated, discipline-specific list supplies a stable lexical backbone while artificial intelligence supplies contextualized, adaptive practice. Read together, the four phases trace a coherent arc from methodological diagnosis to resource construction, empirical validation, and forward-looking contextualization, and they jointly advance the field's understanding of how discipline-specific vocabulary instruction should be designed, evaluated, and integrated with emerging technologies. The dissertation concludes with an integrated discussion of the programme's theoretical, methodological, and pedagogical contributions and offers a validated, ready-to-use lexical resource for scientometrics ESP curricula.
