Mapping metaphor research in translation and interpreting studies: a bibliometric analysis from 1964 to 2023

3.1 Data collection and screening

The main objective of this study is to dissect the characteristics of metaphor research in TIS, especially when it overlaps with CTIS, and point out future avenues of this research area. Specifically, we aim to address the following issues:

1. The evolution of research

2. Theoretical/conceptual frameworks

3. Authorship patterns

4. Publication trends

5. Characteristics of empirical research

The data in this study was retrieved from two TIS-specific databases – Bibliography of Interpreting and Translation (BITRA; Franco 2001–2024) and Translation Studies Bibliography (TSB; Gambier and van Doorslaer 2024). As of July 2024, BITRA comprises over 94,000 entries and TSB 40,000 entries in total, representing the most holistic TIS bibliographic records and allowing researchers to compile comprehensive literature representing a given area of knowledge. Both databases cover various types of publications – including monographs, book chapters, and journal articles, as well as a broad spectrum of geographical and language origins – such as English, Spanish, French, Italian, Portuguese, and Chinese.

To obtain relevant bibliographic records, we utilised the embedded functions of both databases to perform a keyword search by selecting the term metaphor. As of 18 May 2024, 756 entries from BITRA and 570 from TSB were gathered from 1964 to 2023. The records were then exported and converted into two separate spreadsheet files under the categories of publication year, title, journal, author, and abstract.^[1]

Each item in both files was separately screened by the two authors to exclude irrelevant and invalid records based upon the title, the abstract, and, at times, the body of the text. The screening process abides by the following criteria: (1) whether the entry is pertinent to metaphor and translation/interpreting; (2) whether the entry is in the form of a research paper; (3) whether the entry is complete (e.g., some items are indexed with only author information and publication year but without titles); (4) whether there are duplicate items within each database. Considering metaphor is a relatively small research strand in the broader TIS, we decided to take an inclusive approach to data treatment. That is, we kept records devoted to theoretical/conceptual and empirical investigations of metaphor in TIS, as well as comparative studies across cultures and languages. No limits concerning the publication languages were set during the screening process. The authors screened the results separately, and divergences were resolved by meticulously reviewing the documents and discussion before reaching a joint decision. Subsequently, both screened files were combined to remove duplicate records across databases. Eventually, we obtained 1,023 valid records.

3.2 Coding

With the ad-hoc database, we manually coded each valid record – based on the title, abstract, and body of the text – to unveil the characteristics of these publications and address the research issues.^[2] A mixture of deductive and inductive coding approaches was adopted (Kuckartz 2014): deductive categories were driven by our research objective and issues and complemented by inductive categories in the course of the coding process. We first determined two main categories: framework (the research or conceptual framework adopted) and research type (whether the research is empirical or not). Sub-categories were enriched and refined based on specific patterns found in the coding process (see Table 1).

For a closer look at the topics of empirical metaphor research in TIS, we further developed a coding frame using the same approach. The main categories include mediation of communication, research orientation, and research method, while the sub-categories emerged from the data (see Table 2).

The coding process was conducted on the combined spreadsheet. Each author performed the first coding attempt with nonidentical categories in the coding frame. Ambiguous entries and questionable categorisations were highlighted during the individual coding process, followed by a joint discussion for further confirmation. When the first attempt was completed, both authors cross-checked all codes before finalisation. We eventually imported the completed spreadsheet into the visualisation software Tableau for data presentation.

4.1 Historical development

Although we are aware that one of the earliest (though minimal) discussions on metaphor translation in the modern era scholarship can be found in Nida’s seminal work on translation (1964: 219–221), the above bibliometric analysis shows that the first metaphor research in TIS did not appear until 1967 (Beekman 1967), judging by records indexed in the two databases under analysis (see Figure 1). In fact, no publications earlier than 1967 were found in the two databases. The nearly sixty years of development captured by the databases, which have seen steady growth over the years, can be roughly and conveniently divided into two stages: the initial stage (from 1964 to 1993) and the rising stage (from 1994 to 2023). During the initial stage, metaphor research in TIS was relatively scarce, mostly registering fewer than ten publications per year, except in 1990 (13 records) and 1993 (19 records). As for the rising stage, the number of publications per year mostly reaches above 20, and almost half of the years even go above 30. The exponential trendline created by Tableau’s built-in function (log-transformed data fitting in a linear regression model; R ² = 0.80, p < 0.001) further suggests that the publication number per year has risen at increasing rates over the years. In other words, the growth is exponential rather than linear or logarithmic. Although the number seems to have peaked in 2014 (61 records) and dropped gradually, this does not suggest that researchers are moving out of the field after 2014. Firstly, databases’ collection of bibliographical data through online mining generally experiences certain degrees of delay, an effect reflected in a slight drop common to much bibliometric research when approaching the present period. Secondly, the relatively high volume of publications in 2014 was mainly due to the publication of an edited volume on translating figurative language (Miller and Monti 2014), which single-handedly contributed 24 records for that year to our dataset. This is to say, whether metaphor research in TIS will reach new heights depends on, for example, whether there will be more concentrated efforts or new theoretical advancements that lead to surging interests in the field.

Figure 1:

Number of metaphor research outputs in TIS by year (with a dotted trendline).

On the other hand, a closer look at how publication year intersects with the theoretical framework confirms that Schäffner may be correct in saying that CMT “entered Translation Studies only in the 1990s” (2017: 249) because the earliest research adopting the CMT framework was only found as late as 1990 (Königs 1990). However, CMT does not seem to be the major reason for the exponential growth of metaphor research in TIS around the turn of the millennium. Convenience sampling from several intermediate peaks before 2000, namely 1990, 1993, and 1998, reveals that the share portion of CMT in the frameworks adopted was not particularly significant: only 15.38 % (2 out of 13 records), 15.79 % (3 out of 19 records), and 29.17 % (7 out of 24 records) of publications adopted the CMT framework in 1990, 1993, and 1998, respectively. The proliferation of metaphor research in TIS starting from the 1990s possibly has more to do with the increasing establishment of TIS as an independent discipline and the overall expansion of higher education in different disciplines. As a result, there was simply more research produced in any field overall, including research on translating figurative language.

4.2 Theoretical frameworks

Only 44.57 % of publications (456 out of 1,023 records) adopt cognitively informed (metaphor) theories as their framework(s) of metaphor research in TIS. These theories are defined and coded here in a limited fashion as CMT (Conceptual Metaphor Theory), DMT (Deliberate Metaphor Theory), CBT (Conceptual Blending Theory), RT (Relevance Theory), and SFL (Systemic Functional Linguistics). On the other hand, 55.43 % of publications (567 out of 1,023 records) adopt miscellaneous frameworks from Literary Studies, Philosophy (e.g., Benjamin, Derrida), and various branches of Linguistics (e.g., lexicography, phraseology, terminology, contrastive linguistics), to name a few; a considerable number of publications also draw on frameworks within Translation Studies (e.g., those by Newmark and Snell-Hornby) without invoking any metaphor-related theories. These latter groups of frameworks coded as Others are excluded from the analysis of theoretical frameworks in this study not because of their demerits but because of our focus on modern-day theories with a cognitive edge.

Prior to analysis, we need to clarify how these theories could be said to be “cognitive”. CMT defines the cognitive label for (conceptual) metaphor. CMT claims that understanding metaphor always involves understanding one conceptual domain in terms of another (Kövecses 2017: 13). However, defining metaphor as cross-domain mapping does not mean that, in reality, processing metaphor always involves a specific cognitive process. Bringing in psycholinguistics’ perspectives of metaphor processing (e.g., Bowdle and Gentner 2005) to the development of a new metaphor theory, DMT restricts CMT’s claim on cross-domain mapping to a particular class of metaphor (i.e., deliberate metaphor) and can “be interpreted as an extension of CMT” (Steen 2015: 71). On the other hand, although not developed as a metaphor theory, CBT is also cognitively relevant to the study of metaphor. CBT proposes conceptual blending as a general cognitive operation underlying human cognition. The mechanism of conceptual blending involves the projection of two mental spaces onto a third space called generic space and the subsequent emergence of a fourth space called blended space (Fauconnier and Turner 2002: 39–50). Mental spaces in CBT are understood as “small conceptual packets constructed as we think and talk, for purposes of local understanding and action” (Fauconnier and Turner 2002: 40). Suffice it to say, CMT’s notion of conceptual metaphor, if contextualised in CBT’s understanding of the blending operation of human cognition, can be understood as involving a single-scope blend (Dancygier 2017: 32).

Furthermore, RT is also not a metaphor theory but can be cognitively relevant to metaphor research in TIS. As a theoretical approach to understanding human communication and cognition, RT proposes the cognitive principle that “[h]uman cognition tends to be geared to the maximisation of relevance” (Sperber and Wilson 1995: 260) and the communicative principle that “[e]very act of ostensive communication communicates a presumption of its own optimal relevance” (Sperber and Wilson 1995: 260). Within this relevance-theoretic approach, metaphor is but one kind of loose use in communication that can achieve optimal relevance in some situations but that requires “no mechanisms or processes specific to the recognition and comprehension” (Carston 2017: 43).

Lastly, being the odd one out, SFL is a general approach to Linguistics rather than an independent theory. However, it is included in this list because publications in our dataset have exclusively used it to study the grammatical metaphor in translation and interpreting (nominalisation and denominalisation).

Figure 2 shows the distribution of cognitively informed theories adopted by metaphor research in TIS. Noticeably, CMT alone contributed to almost 90 % of the publications, even if discounting the 2.19 % of publications that combine CMT and other theories as their frameworks. This confirms the claim that CMT is “the most frequently adopted theoretical framework for research into metaphor in translation” (Shuttleworth 2014: 60). One of the reasons for this is that, proposed in 1980, CMT has had a long history of more than four decades of theoretical development, attaining high levels of maturity and divisionalisation. As a result, CMT is relatively easy to apply when conducting case studies in TIS. Besides, it is possibly the only cognitive metaphor theory that comes to mind for most people and has been the go-to theory for a cognitive approach.

Figure 2:

Percentage of cognitively informed theories adopted by metaphor research in TIS.

On the other hand, DMT, with less than 1 % of the total share, has yet to receive much exploration from TIS scholars. This is partly because, despite early theoretical proposals in the 2000s (Steen 2008), DMT only gradually established itself as a metaphor theory over the years, as late as the label was coined in 2015 (Steen 2015). As a rule of thumb, the introduction of a theory to TIS can take as long as ten years, as in the case of CMT. Besides, CBT and RT, as theories of cognition and pragmatics (respectively), have only occasionally drawn the attention of TIS scholars, even when their histories can be traced to the late 20th century. This is probably because their theorisation did not focus exclusively on metaphor.

4.3 Authorship and top productive authors

Our bibliometric analysis of publication authorship reveals that the majority of publications are single-authored (80.84 %). In contrast, publications with two, three, and four authors only amount to 15.54 %, 3.13 %, and 0.49 %, respectively. No publications have more than four authors. Contrary to our initial anticipation, the relatively low level of collaboration between scholars does not seem to sanction a network analysis. Besides, if the maturity of a field is partly determined by the degree of collaboration between scholars, it is not unfair to say that metaphor translation as a field (19.16 % co-authorship as of 2023) lags slightly behind CTIS as a whole (24.6 % co-authorship in CTIS as of 2015, according to Olalla-Soler et al. 2020).

Furthermore, Table 3 presents a count of the top 11 productive authors of metaphor research in TIS, ranked according to the number of publications in which they are the first author (also mostly the sole author). Their total number of single-authored and co-authored works is also included in the table but is for reference only. This ranking of publication volume should not be directly taken as their relative impact and influence on the field because metrics such as their total citation were not taken into account given the incompleteness of citation metrics in BITRA and their unavailability in TSB. Still, several interesting patterns could be identified. Firstly, scholars usually either work on the translation of metaphor (e.g., Popescu, Samaniego, Schäffner, Shuttleworth) or on metaphor of translation (Guldin, St. André). Secondly, except for a few who primarily work on theoretical and conceptual issues, scholars usually specialise in metaphor translation in one particular type of discourse and work almost exclusively on that: for example, political discourse (Schäffner), popular science (Shuttleworth), and business and economic discourse (Dobrota, Fernández, Popescu, Silaški). No scholars have been found to specialise in literary discourse. Thirdly, there are yet to be scholars working on metaphor and interpreting who make it onto the list. Last but not least, no scholars, if judging by the research orientation of their publications indexed in the dataset, are primarily associated with CTIS (except perhaps Martín, who is now much engaged with CTIS, but her earlier publications on metaphor in the 2000s are mainly theoretical ones).

4.4 Journals that publish the most metaphor research

Metaphor studies are usually published in journals such as Cognitive Linguistics, Journal of Pragmatics, Metaphor and Symbol, and, since the last decade, Metaphor and the Social World. As for metaphor research in TIS, its publication patterns will probably be less straightforward. We have identified 239 journals from our database that have published metaphor research, accumulating 555 journal articles. To determine the journals that publish the most metaphor research, we counted the number of articles published in each journal and their proportions. Table 4 presents the results with some basic statistics.

Babel (36; 6.49 %) and Meta (31; 5.59 %) take the lead, contributing more than 30 articles throughout the investigated timespan. This looks reasonable, as both journals dedicate many, if not all, of their publications to literary translation, which was a central focus in the early stages of TIS. This may also explain why time-honoured journals, including Notes on Translation (11; 1.98 %; though no longer active) and The Bible Translator (9; 1.62 %) – both focusing on Bible translation – are also included in this list, contributing a combined total of 20 articles. As Translation Studies started off by looking into translating literary and religious texts, the importance or difficulty of translating metaphorical expressions within such genres may have attracted much scholarly attention at the initial stage.

Next come the leading TIS journals Perspectives (14; 2.52 %) and Target (14; 2.52 %), contributing around 30 articles combined. These two journals are well-established in publishing both theoretical and empirical research in different kinds of linguistic and cultural mediation. Our database also shows a variety of metaphor-related topics in the two journals, ranging from theoretical modelling of metaphor translation (e.g., Dickins 2005), the metaphorical conceptualisation of translation (e.g., Cheung 2005), to practical techniques for undertaking metaphor translation (e.g., Al-Zoubi 2009), and empirical investigations in specific texts, genres, and contexts (e.g., Grave 2016).

In addition, journals commenced after the millennium, such as JoSTrans (10; 1.80 %) and Translation and Interpreting Studies (8; 1.44 %), also published quite a few articles on metaphor research. While Translation and Interpreting Studies embraces all areas of language mediation like the journals mentioned above, JoSTrans emphasises non-literary translation and interpreting in subject-specific fields. Most of the publications in these two journals appear after the 2010s, which seems to coincide with the then-changing landscape of TIS driven by “the emergence and proliferation of multimodal texts and digital technology” (Munday et al. 2022: 232). For example, we can find innovative endeavours to explore metaphors in legal translation using large corpora (Božović 2022), audiovisual translation (Pedersen 2015), and machine translation and post-editing (Koglin and Cunha 2019).

4.5 Mediation of communication

Zooming in on empirical studies, we have identified 453 valid records from our database. In the following, we will explore how metaphor research unfolds in the contexts of various types of mediated communication.

Mediated communication takes various forms, especially for those involving multiple languages (or “multilectal mediated communication”; see Halverson and Muñoz 2021). For this study, we streamlined the classification of mediation and grouped the research into two main categories based on their research subjects: translation and interpreting. Translation and interpreting here are operationalised as a translational activity that involves cross-linguistic and non-linguistic transfer; the difference is that interpreting deals with a “one-time presentation” (Pöchhacker 2022: 11) of the (non-)linguistic elements while translation does not.

To gain a deeper understanding, we further sub-categorised the two main categories by assigning codes to crystallise the mediation. Translation research papers involving only written texts were coded as written; when they additionally involved non-verbal elements, the code was multimodal; when they addressed machine translation, they were labelled machine. For interpreting research, oral and sign were distinguished. The distributions of translation- and interpreting-dedicated research and their sub-categories are visualised in Figure 3.

Figure 3:

Distribution of empirical metaphor translation and interpreting research.

Several prominent trends are depicted in Figure 3. From a bird’s eye view, metaphor is investigated more dominantly in translation research than in interpreting research, with the former accounting for more than 97 % of the total empirical studies. Within translation research, written translation occupies the lion’s share (94.33 %), followed by multimodal translation (4.76 %); only a few studies look at metaphor in the context of machine translation. As for interpreting research, spoken language interpreting (83.33 %) is the main area where metaphor is probed, though sign language interpreting also captures some attention.

The small proportion of metaphor interpreting research could be accounted for from several aspects. Firstly, the community of researchers investigating metaphors in interpreting is arguably smaller than that engaged in translation research, leading to a comparatively small number of publications. Secondly, interpreting studies is often plagued by methodological constraints: an intricate challenge for interpreting scholars is to obtain (transcribed) in-the-field texts, and there are only a few options of publicly available recordings (Pöchhacker 2022: 225). Since metaphor is primarily text- or discourse-embedded, interpreting scholars, without acquiring such textual data, may have fewer choices but to rely on simulated texts in a somewhat restricted fashion.

4.6 Research orientation and method

The categorisation of research orientation is based on how the empirical data is acquired and analysed. We have identified three orientations from the 453 empirical studies: product-oriented, process-oriented, and participant-oriented research. Product-oriented research focuses on translation and interpreting final outputs; process-oriented research highlights the translators’ or interpreters’ online processing when the task is performed in situ; participant-oriented research emphasises the translators’ and interpreters’ own general feelings of the process and the reception of translation and interpreting products.^[3]

Figure 4 encapsulates the distribution of the three research orientations in empirical studies. Noticeably, product-oriented research takes up the most significant proportion (93.38 %) of all empirical studies, overshadowing process-oriented research (5.96 %) and participant-oriented research (0.66 %). The dominance of product-oriented research may be explained by the deductive tradition of the cognitive-linguistic approach to metaphor (our database shows that 46.68 % of product-oriented studies adopt the CMT framework).

Figure 4:

Distribution of research orientations in empirical studies.

Departing from the research orientations mentioned above, we further examined the research methods used in these empirical studies. Figure 5 depicts the methodological landscape of all empirical studies.

Figure 5:

Distribution of specific research methods used in empirical studies.

As shown in Figure 5, text-based research constitutes the majority of all empirical studies, accounting for around 71 %. This prevalence likely reflects the long-standing tradition of metaphor research using textual data. Research that leverages corpora, comprising approximately 22 %, ranks second. This seems to suggest a growing interest in utilising a large amount of (digital) textual data in complementary to the traditional textual analysis approach.

Research that adopts behavioural measures comes third, representing 5.08 % of the total. Research using only introspective measures (0.44 %) is noticeably scarce. The small proportions hint that these two measures are less frequently used in metaphor research in TIS.

Furthermore, a few studies integrate behavioural, text-based, and psycho-physiological measures with introspective measures. Though they hold a minimal share (1.76 %), their use of combined measures in metaphor research in TIS suggests an emerging trend toward methodological triangulation.

To better understand how different methods are applied and distributed, we aligned them with the three research orientations. Figure 6 shows the methodological landscape across three research orientations.

Figure 6:

Distribution of methods used in the three research orientations.

Among the 423 product-oriented studies, text-based analysis (75.65 %) is the dominant method, followed by corpus analysis (23.64 %). Even though product-oriented research enjoys the largest number of studies, the exploration of metaphor still focuses on linguistic features, whether the analysis is based on a single source or a large amount of textual data. There is also a tiny proportion of studies combining text and introspective measures (0.71 %). One example is exploring translation trainees’ strategies for translating linguistic metaphors with/without the existence of comparable conceptual metaphors through mixed methods, i.e., quantitative output rating and qualitative interview (Chen 2017).

Within a small pool of 27 studies, it is evident that process-oriented research is dominated by behavioural measures, including eye-tracking, keylogging, think-aloud protocol, screen recording, and reaction time analysis, to link the translator’s behavioural patterns with underlying cognitive processing. While behavioural measures are mainly employed without incorporating other measure types (85.19 %), this does not indicate that process-oriented research always uses one method at a time. There are plenty of studies adopting a combination of behavioural measures. A typical example would be an integration of eye-tracking and keylogging to investigate translators’ cognitive effort in translating metaphors (e.g., Koglin 2015; Sjørup 2013). Moreover, behavioural measures are sometimes used in conjunction with introspective measures (14.81 %). For example, a common choice of methods would be to combine eye-tracking, keylogging, and retrospective interviews (e.g., Schmaltz 2018; Wang 2021). These endeavours can be considered attempts to enhance methodological rigour through multi or mixed methods research.

When it comes to participant-oriented research, only three studies have been identified. These studies utilised introspective measures, such as questionnaires and interviews, to understand the participants’ subjective feelings towards metaphor while translating, interpreting, and consuming the translated/interpreted product. Additionally, psycho-physiological measures, exemplified by heart rate and facial expression variation, may provide a more objective angle for this purpose (see Ramos Caro 2016).

In summary, metaphor has been examined in various stages of translation and interpreting. However, while product-oriented research still takes the lead in the quantity of research output, process-oriented and participant-oriented research is catching up with methodological rigour, adding a more cognitive profile to metaphor research in TIS.