Prosody in linguistic journals: a bibliometric analysis

Prosody—encompassing tone, intonation, stress, and rhythm—plays central roles in speech processing across L1 and L2, including lexical activation and segmentation, syntactic parsing, information structure marking, and pragmatic signaling. Despite seminal narrative reviews (e.g., Shattuck-Hufnagel & Turk, 1996; Cutler et al., 1997) and later state-of-the-art overviews (Cole, 2015; Wagner & Watson, 2010), no bibliometric overview had charted the intellectual structure and trends of prosody research in linguistics. To address this gap, the study applies bibliometric methods to SSCI-indexed linguistic journals (2001–2021) to map influential sources, authors, and themes, and to detect emerging trends. The research questions are: (1) What is the research productivity of linguistic journals on prosody? (2) What is the intellectual structure in the field of prosody in terms of influential authors, references, and venues? (3) What are the research trends of prosody research in linguistics?

The paper situates prosody as a multifaceted phenomenon studied across linguistics, psychology, cognitive science, and computer science. Foundational narrative reviews (Shattuck-Hufnagel & Turk, 1996; Cutler et al., 1997) and subsequent syntheses (Cole, 2015; Wagner & Watson, 2010) summarize theoretical and experimental advances. Bibliometric approaches have been successfully used in related domains—second language acquisition (Zhang, 2019), applied linguistics (Lei & Liu, 2019a), and visual word recognition (Fu et al., 2021)—but not yet applied to prosody in linguistics. This motivates a systematic, data-driven mapping of prosody’s intellectual landscape and evolution.

Data source and retrieval: Bibliographic records were retrieved from Web of Science (WoS) on 14 June 2022. WoS was chosen for its wide academic use, provision of academic-only citations, and availability of co-citation data. The search used the query terms: prosod*, autosegmental-metrical, metrical structure, accent, intonation*, stress, suprasegment*, F0 (fundamental frequency), rhythm, pitch, combined with OR, and excluded “semantic prosody” via NOT. Timeframe: January 2001–December 2021. Inclusion criteria: English-language peer-reviewed research articles in SSCI-indexed international linguistics journals. To focus on substantial outlets, journals needed ≥30 prosody-related articles over 2001–2021, which covered >70% of total publications and ensured robust statistics. Non-article types (e.g., book reviews, editorials) were excluded. Data cleaning: Author names with variant formats were recoded to a single canonical form. Semantically identical keywords (including singular/plural variants) were consolidated (e.g., “Event-related Potential,” “ERP” unified to “Event-related Potential”; “boundary tone(s)” unified to “boundary tone”). Near but non-identical terms (e.g., “bilinguals” vs. “bilingualism”) were kept distinct. Analytic strategy: The 21-year span was divided into three periods: 2001–2007, 2008–2014, and 2015–2021.

• Co-citation analysis: Based on cited references within the retrieved articles, co-citation networks were constructed in VOSviewer. A threshold was set to include the top 50 most-cited items per map for interpretability, following prior practice. Clustering used the smart local moving algorithm to identify thematic groupings and influential sources, references, and authors.
• Keyword analysis: Author-supplied keywords and keywords extracted from abstracts (noun/noun-phrase n-grams up to four tokens, POS-tagged and manually curated) were combined. Raw frequencies were calculated and normalized per period using: normalized frequency = (raw frequency in period / total publications in period) × 10,000, to account for differing publication volumes. One-way chi-square tests across periods were run on keywords with ≥10 occurrences to detect significant diachronic changes (p < 0.05).

Corpus overview and productivity: 4598 prosody-related research articles were identified in SSCI-indexed linguistics journals (2001–2021). Annual output rose overall, surpassing 300 publications/year since 2019, with a dip in 2020 likely due to COVID-19. Authorship, regions, institutions: 58 authors published ≥10 articles; 5279 authors published one article (0.81% vs. 99.19%). Top authors included P. Prieto (41), K. Saito (35), M. Swerts (22), P. Trofimovich (21), B. Braun (19), T. Cho (19), C. Gussenhoven (18), M. Grice (17), S. Kim (17), Y. Xu (17). Top regions: USA (2002 publications), England (530), Germany (473), Canada (411), Netherlands (299), Australia (235), France (218), China (198), Spain (174), Japan (124). Top institutions: University of London (158), CNRS (145), Max Planck Society (134), University College London (117), Radboud University Nijmegen (100). Co-citation—top sources: Across all periods, the Journal of the Acoustical Society of America and Journal of Phonetics are the two most-cited sources, with combined citations increasing markedly (approx. 2329 in 2001–2007; 6249 in 2008–2014; 9978 in 2015–2021). Other consistently top-10 venues include Journal of Memory and Language, Journal of Speech Language and Hearing Research, Language and Speech, Cognition, and Phonetica. Doctoral theses are also highly cited sources. Co-citation—field structure over time: In 2001–2007, five clusters emerged: linguistic/phonetic investigation; L2 learning; psycholinguistics; language development/disorders; neurolinguistics. In 2008–2014, psycholinguistic and neurolinguistic sources merged into the dominant cluster. In 2015–2021, experimental vs. formal/theoretical work separated again (e.g., Laboratory Phonology vs. Linguistic Inquiry). The L2 prosody cluster expanded slightly over time. Most-cited references: Fifteen references appeared in the top-50 lists across all periods. Perennial anchors include Ladd (1996/2008), Nespor & Vogel (1986), Hayes (1995), Chomsky & Halle (1968), Pierrehumbert (1980/1990), Selkirk (1984), and Beckman & Pierrehumbert (1986). In 2015–2021, a new dominant cluster centers on statistical methods and tools: Praat (Boersma & Weenink, 2018; 693 citations), LME4 mixed-effects modeling (Bates et al., 2015; 287), maximal random-effects structure (Barr et al., 2013; 177), mixed-effects modeling (Baayen et al., 2008; 118), and R (R Core Team, 2017). Information structure/focus-related works (e.g., Rooth, 1992; Breen et al., 2010) appeared in the top-50 in the latest period. Most-cited authors: Influential across periods include A. Cutler, J.E. Flege, C. Gussenhoven, D.R. Ladd, M.J. Munro, J. Pierrehumbert, E. Selkirk, L.D. Shriberg, Y. Xu. In 2015–2021, methodological contributors (Bates, Barr, Baayen, R Core Team) also entered top ranks, reflecting the methods shift. The L2 prosody author cluster (e.g., Flege, Munro, Derwing, Kuhl, Saito) expanded over time. Keyword analysis: Of 207 author-supplied and 37 abstract-derived keywords (≥10 occurrences), 61 changed significantly across periods. The top-10 author keywords (by normalized frequency) were: prosody (decline), intonation (stable), phonology (decline), speech perception (stable), stress (stable), focus (stable), bilingualism (increase), English (stable), speech production (stable), accent(s) (stable). Notable increasing topics: bilingualism; second language/second language acquisition; foreign accent(s) and cross-linguistic influence; language attitudes; voice onset time; sound change; tone sandhi; syntax–phonology interface; electroencephalography; rhythm. Topics rising then declining: gesture; aphasia; corpus. Decreasing topics: prosody, phonology, syllable, syllable structure, lexical access—likely supplanted by more specific terms (e.g., F0, pitch, stress). Languages consistently frequent include English, Mandarin Chinese, Japanese, French, and Dutch, with English most prevalent.

The findings address the research questions by: (1) quantifying productivity (4598 articles; accelerating output since 2019) and mapping contributions by authors, regions, and institutions; (2) delineating the intellectual structure via co-citation, showing persistent core venues and references, dynamic clustering (merges/splits between psycholinguistics and neurolinguistics, and the separation of experimental vs. formal/theoretical work), and the emergence of a methodological cluster (mixed-effects modeling, R, Praat) in recent years; and (3) identifying diachronic topic trends, notably rising interest in bilingualism/L2 prosody and neurocognitive methods (EEG), and stable centrality of intonation, stress, and accent. These patterns underscore prosody’s interdisciplinarity and its methodological modernization. The expansion of L2-related clusters and keywords reflects broader sociolinguistic realities (global bilingualism/multilingualism) and theoretical developments (e.g., L2 intonation learning theory), while the methods cluster signals a field-wide shift to mixed-effects modeling and reproducible tools, reshaping empirical standards. Together, the analyses guide newcomers to influential sources and evolving themes and help situate prosody within adjacent domains (psycholinguistics, neurolinguistics, acquisition).

This bibliometric review maps two decades of prosody research in SSCI-indexed linguistics journals, revealing increased publication volume; persistent core sources, authors, and references; evolving thematic clusters; and a recent methodological turn toward mixed-effects modeling and open tools. Intonation, stress, and accent remain central, with sustained attention to speech perception and production. Bilingualism/L2 topics have grown notably, while some general terms (prosody, phonology, syllable) declined as more specific constructs rose. The study highlights the role of special issues (e.g., 2008 JML on Emerging Data Analysis) in catalyzing methodological shifts. Future research could expand search terms to encompass additional prosodic constructs (e.g., duration, emphasis), broaden journal coverage beyond SSCI-only filters, integrate qualitative syntheses to complement quantitative mapping, and continue tracking cross-disciplinary methods (e.g., EEG) and multilingual phenomena to refine understanding of prosody’s interfaces.

Limitations include: (1) Search-term scope may have omitted relevant prosody-related concepts (e.g., duration, emphasis), risking undercoverage despite their appearance in extracted keywords. (2) Journal selection (SSCI-indexed, ≥30 articles) excluded some influential venues from publication/keyword analyses (though they appear in co-citation), potentially biasing outlet-level trends. (3) The quantitative design and map interpretability constraints (top-50 thresholds) limit granularity; qualitative analysis was beyond scope and would add depth. (4) Some subjectivity is inherent in interpreting clusters and trends; expert triangulation is desirable. (5) Space and timeframe constraints preclude exhaustive treatment of all observed trends.