Recently, the expertise of a translator has become an increasingly catchy topic in translation studies. A professional translator, as generally agreed, needs to have varied expertise, for example: bilingual capacity (Amador 2019), internalization of external resources (Whyatt 2012; Redelinghuys and Kruger 2015; Whyatt 2018), intercultural competence (Eszenyi 2016), computational expertise (Rodríguez-Castro 2018; Shreve 2018), deliberate practice (Shreve 2006), or sociological aspects of expertise (Tiselius and Hild 2017). Expertise has different forms, and so a single study usually cannot comment on all aspects of expertise (Shreve 2018). Usually, in translation studies, its definition is limited to the translator’s experiential expertise (En and En 2019), that is, translation training experience and practice. In other words, an expert translator is distinguished from a novice translator by his/her translation experience and translation skills.

Expertise plays a significant role in a translator’s performance and behaviour. For instance, Rojo and Ramos (2018) have offered empirical evidence for the influence of expertise on emotion regulation and its consequences for translation performance. The result suggests that personality factors and expertise level are decisive in regulating emotion and guiding translation behaviour. Many scholars have also investigated expertise in association with translation competence, often raising the question about the relationship between the two, and the variations between translators with different levels of expertise (for example, Pym 2003; Göpferich 2009; Tiselius 2013; Goźdź-Roszkowski 2016; Shreve, Angelone, et al. 2018). Some see expertise and competence as intertwined concepts, with expertise entailing competence, or competence as a prerequisite of expertise (Tiselius 2013; Tiselius and Hild 2017). Others are concerned whether there is a need for the term competence in translation studies given that many widespread and empirically robust frameworks of expertise already exist in cognitive sciences (Shreve, Angelone, et al. 2018).

When it comes to the translation of specialized texts, it is argued that generic competence (that is, the ability to successfully exploit a range of genres to fulfil the goal of a specific professional community) and professional expertise should be included in the range of competence demanded from translators (Goźdź-Roszkowski 2016). As for the differences between professional and non-professional translators, they are often assumed to be self-evident, with the latter seen as less competent in producing a good translation; this has been confirmed bby several studies. For instance, Amador (2019) has found that amateur translators are less effective in their translation tasks. Olalla-Soler (2018)[1] has also discovered that professional translators apply internalized source-culture knowledge more efficiently; therefore, they are more capable of solving cultural translation problems. However, a different voice is heard from other scholars who have worked to develop a more nuanced perspective on translation tasks by non-professional translators. They argue that translations by non-professionals may not necessarily be deficient, sometime even superior in some aspects (Antonini and Bucaria 2016; Baker 2009, 2013; En and En 2019; Khoshsaligheh and Ameri 2017; Tesseur 2017). These seemingly contradictory voices signal a need to enlarge our understanding of expertise, which is why registerial expertise is the focus of the present study.

Registerial expertise, by and large, refers to what has been traditionally called a translator’s domain-specific knowledge (Goźdź-Roszkowski 2016; McCarthy and Goldman 2019) or subject-matter expertise (Holt-Reynolds 1999). Following the concept of register in Steiner (1998), Matthiessen (2009), and Matthiessen, Wang, et al. (2018), the present study re-terms this as registerial expertise: the translator’s domain-specific knowledge, developed as a repertoire of personalized meaning potentials through registerial progression, including the activities of a particular field, the social interactions in that field, and the mode of meaning-conveying.

Registerial expertise is an essential layer of expertise (Herling 2000; Shreve 2018), sometimes playing an even more important role in distinguishing a novice from an expert (Popovic 2004; Shreve 2018). A higher level of domain-specific knowledge will usually empower a writer to have a higher degree of epistemic authority in uttering his/her own voice in a text (Heritage and Raymond 2005; Stivers 2005; Stivers, Mondada, et al. 2011). It is also argued that the better you know a specific domain, the better you are able to know the subject at hand (Shreve 2018). Interestingly, Shreve and Lacruz (2017) found that a translator with extensive knowledge of the domain represented in a text is able to build a richer propositional network, thus producing more accurate meaning-inferencing in text representation. Important as it is, registerial expertise is mostly discussed at the theoretical level, leaving a desperate need for empirical findings to sustain its exploration within translation competence and translator training. This study offers such empirical insights on the role of the translator’s registerial expertise, specifically medical expertise, including his/her clinical experience and medical knowledge. It is a product-based investigation, using translations as end-products; these are the textual evidence of the translator’s semantic choices. Such an investigation, by nature, is an examination through textual analysis of the translator’s subjective interventions.

The focal point will be given to the translators’ logical choices in the translations of the most ancient classic of Chinese medicine, Huang Di Nei Jing. The reason for this choice lies in the unique linguistic features of logical vagueness/indeterminacy in classical Chinese, a peculiar burden in the communication between Chinese medicine and Western medicine (Maciocia 1989/2015). Moreover, logical indeterminacy allows to examine the translator’s subjective interventions: when a text is certain and determinate, the space left for the translator’s interpretation will be small, as s/he is constrained by the source text. Then, the variations between translators’ choices tend to be minimal. However, when a translator is dealing with a highly indeterminate and uncertain text (for example, TCM texts written 2000 years ago, where the contextual distance is so great that a univocal meaning is almost unreachable), the translator’s personal intervention can then be maximally revealed. Indeed, such a text provides more room for the translator to exercise his/her subjectivity. Naturally, the role of a translator’s expertise in his/her translation choices can then be maximally examined and thus the purpose of this article is better served.

Studied within semantics, indeterminacy, or vagueness (Zhang 2015), is an intrinsic feature of natural language (Russell 1923; Weatherson 2010). It occurs when meanings are “continuous” or “fused” with many hypothetical options (Halliday and Matthiessen 1999: 548). As an intricate linguistic phenomenon, its metalinguistic categories are often indeterminate as well (Halliday 2009; Zhang 2015), depending on which perspective is adopted to approach the meaning of language (for example, formal, semantic, pragmatic). Indeterminacy occurs when there is a fuzziness in semantic categorization and it can be caused by various factors, be they epistemic, metaphysical, pragmatic or grammatical (Kies 1990; Trueswell, Tanenhaus, et al. 1994; Santos 1998; Williamson 2003; Niederdeppe 2005; Spranger and Loetzsch 2011; Zhang 2015).[2] The present study approaches the meaning of language from a SFL perspective, more specifically, logical meaning, but the term indeterminacy needs to be understood from the tension of the unique logical system between classical Chinese and English in the process of translating.

Unlike English, classical Chinese is a concise language that does not provide many logical resources. Also, ancient Chinese texts originally had no punctuation nor delimiter that marks the end of the word or sentence (Huang, Sun, et al. 2010; Xu, Wang, et al. 2019). As a result, the linearity of syntactic relationships needs to be worked out by subsequent modern Chinese scholars through judou, that is, sentence segmentation (Jia 2009; Xu, Wang, et al. 2019), which often provokes controversies and indeterminacy. The lack of lexico-grammatical logical resources and the unique grammatical uncertainty in the classical Chinese system (Jiang 2010), together with contextual distance between ancient and modern China, between the East and West, present a near insurmountable challenge for the translator, causing indeterminacy in his/her decision-making. As Veith, a well-recognized translator of Huang Di Nei Jing, points out:

Classical Chinese constantly presents a great many problems. Punctuation is completely unknown and there is no indication whatsoever as to where one sentence ends and the next one begins, and the same Chinese sentence admits many grammatically different interpretations.

Veith 1966: xii-xiii[3]

Needless to say, translating such technical but indeterminate texts is highly consequential since the translator is constructing medical knowledge. The indecision or inconsistency caused by translation may eventually become an issue of life or death in clinical practice (Luo 2005; Matthiessen 2017[4]).

Logical indeterminacy is inherent to classical Chinese’s writing system, the understanding of which is context-dependent but most often unsolvable. As previously discussed, this means that investigating logical indeterminacy in that language is the best way to showcase the translator’s subjective intervention. Furthermore, in the context of TCM translation, there is a fierce debate regarding who should bear the responsibility of such a translation, with some claiming that only clinicians, that is, translators with medical expertise, should be entitled to do so (Rosenberg 2013).[5] Determining who would be the ideal translator, in my view, seems futile when considering translation as a social activity, since each translation has its unique value and individual traits. This study thus seeks to characterize such individuality by examining the role that the registerial (medical) expertise of the translator plays in his/her logical choice during translation. It will compare two groups of medical translations, one by clinicians and the other by non-clinicians, that is, translators with and without medical expertise. It looks at different translations of Huang Di Nei Jing, the earliest and most important classic that has laid the foundation for TCM theory and practice. Two research questions will be answered to fulfil this purpose:

1. In terms of logical choice, what are the differences in the translations by clinician and non-clinician translators?

2. How are the differences related to the level of medical expertise?

The analytical framework in the present study is the logical system of CLAUSE COMPLEX in SFL (Halliday and Matthiessen 1985/2014), as it allows the translators’ logical choices to be investigated in a very dedicated and systematic way (see Figure 1).

The CLAUSE COMPLEX system firstly has to do with the choices of clauses and clause complexes, which is closely related to the concept of grammatical intricacy quantified by the ratio of the number of clauses to the number of clause complexes in a text (Eggins 1994: 61), an important measurement of text complexity. The higher the ratio of grammatical intricacy, the higher the degree of text complexity is (Halliday 2009). See Example 1 for an illustration:

There are three clauses in the one clause complex, thus the grammatical intricacy is 3. The CLAUSE COMPLEX system is a resource for construing our experience in the world as sequences of “quanta of change,” with each quantum of change being a configuration of a process (Matthiessen 2002: 263). It contains two sub-systems: TAXIS and LOGICO-SEMANTIC TYPE. TAXIS describes the type of interdependency between clauses, known as parataxis and hypotaxis. Paratactic relationships are represented by numbers (1, 2, 3…), indicating that two clauses are potentially independent of one another, each constituting a proposition in its own right. Hypotactic clauses are marked by Greek letters (α, β, γ…), signalling that two clauses are dependent on each other, with the dependent clause being brought in to support the main clause. The interpretation of a clause nexus frequently depends on more delicately differentiated process types.

LOGICO-SEMANTIC TYPE between clauses is first classified into projection (where the secondary clause is projected by the primary clause) and expansion (where the second clause expands on the initial one). Projection relates our experience to a phenomenon of higher order (what people say and think). Expansion relates to a phenomenon as being of the same order of experience and is sub-divided into elaboration (=), extension (+), and enhancement (x). In elaboration, one clause elaborates on the meaning of another by further specifying or describing it. The secondary clause does not introduce a new element, but rather provides further characterization of one that is already there, simply by restating, clarifying, refining or adding a descriptive attribute or comment. In extension, one clause extends the meaning of another by adding something new. What is added may be just an addition (for example, and), a replacement (for example, but) or an alternative (for example, or). In enhancement, one clause enhances the meaning of another by qualifying it in terms of time, place, manner, condition or cause (see Halliday and Matthiessen 1985/2014: 428-548). See Example 2 for an illustration:

Example 2 explains the consequences to the human body if the winter law is violated. It also provides reasons from the ;五;行 [five elements] theory in TCM. Four clauses with 3 nexuses have been identified, two hypotactic and one paratactic. The first clause provides a condition for the second one, which is then enhanced by the next two clauses, providing reasons why the lungs will be hurt. There is no elaborating relationship in this instance.

The first finding concerns the role of expertise in the grammatical intricacy of the translations (see Table 3).

As indicated in Table 3, the average number of clauses per sentence in the translations by Wu and Ni is 2.5 and 2.3 respectively, a higher ratio of grammatical intricacy than in the translations by Unschuld and Veith, with 2.0 and 2.3. This finding suggests that the complexity of a translation tends to vary according to the translators’ expertise: the higher the level of expertise, the higher the degree of complexity of the translation. However, the two sets of data happen to show differences based on another variable, the translators’ mother tongue. This means that the translators’ mother tongue might also play a role in the discrepancy of grammatical intricacy. To further delimit the major contributing factor for this difference, Li’s translation is included as an independent variable for further comparison. As shown in Table 3, the ratio of Li’s grammatical intricacy, 1.9, is closer to the pattern of the translations by Unschuld and Veith. This means that the translator’s mother tongue (the only variable that distinguishes Li from Unschuld and Veith) does not exert any significant impact on grammatical intricacy. This comparison further consolidates the idea that it is the registerial expertise that plays the major role in the grammatical intricacy of a translation.

It also shows that translations by clinicians tend to have a higher degree of intervention in terms of syntactic arrangement. The clause number is either the highest (394 in Wu’s translation) or the lowest (282 in Ni’s translation). On the contrary, translations by non-clinicians tend to see a closer equivalence to the source text, with 361 (Unschuld) and 359 (Veith), respectively, similar to the 357 found in the ST (see Example 3 for illustration):

Example 3 presents the number of clauses both in the source text and the translations. Two clauses are found in Veith’s translation, which is equivalent to the source text; three clauses are found in Unschuld’s translation, that is one more than in the source text. The highest clause number (4) is in Wu’s translation, and the lowest (0) in Ni’s, suggesting a higher degree of intervention compared to the ST.

Syntactic organization is an important area to consider when studying texture, which is traditionally referred to as the “form” and “content” of a text (Hatim and Mason 1990/2014: 192). The translation strategy can be one of free or literal translation (Barbe 1996). This variation, due to translators’ expertise, strongly suggests that expertise makes a difference in the choices of translation strategy: clinicians tend to favour free translation, while non-clinicians tend to favour a literal translation. In other words, non-clinicians tend to be more grammatically constrained by the source text, while clinician translators tend to be freer from the source text grammar. This is demonstrated by Example 4:

One of the key health principles in Chinese medicine is to live a lifestyle according to the atmospheric changes of the four seasons. Example 4 is about the three consequences of disobeying this health principle in the autumn. It shows that non-clinician translators tend to seek out grammatical loyalty to the ST, while clinician translators tend to be more semantically sensitive. As indicated in Example 4, the source text is presented in 4 ranking clauses, which remains the same in the translations by non-clinicians. However, in the translations by clinicians, Wu adds 4 more clauses to make explicit the covert logical meaning, providing the reasons why the lungs will be hurt and why diarrhoea will happen in the winter. Ni also exercises a higher level of intervention, not by adding more clauses, but simply by omitting the translation of the fourth clause of the source text; this suggests a higher degree of intervention.

This variation between translators with and without medical expertise clearly demonstrates that medical expertise is closely related to a translator’s subjective intervention and their strategic choices in organizing their text. Translators with a higher level of medical expertise tend to adopt free translation, adding more information or omitting clauses deemed unimportant. In other words, medical expertise seems to grant more freedom to the translator to intervene in the logical sequence of the source text, as evidenced in Ni’s paratext where he states: “I have taken much liberty in my humble attempt to convey the content of Neijing, such as eliminating … and incorporating them into the main body of translation” (Ni 1995: xv).[7] Conversely, the non-clinician translators give priority to literal translation, which seeks grammatical faithfulness. This suggests that translators without expertise tend to be more grammatically constrained, which is further confirmed by Veith’s paratext, in which he reveals that the syntactical features of the source text served as a close guide to deal with the obscure meanings of the ST.

The second finding shows that medical expertise also plays a highly impactful role in translators’ choices regarding taxis (see Figure 4).

Figure 4 presents the results of the translators’ tactic choices in relation to their transitivity choices. It shows that clinician translators tend to slightly favour hypotaxis, as indicated by the contrast between the relative percentage of hypotaxis (50.4%) and parataxis (49.6%) in the clinician translations. This tendency becomes more salient when a comparison is done between the ST and the translations, where a sharp increase of hypotaxis, from 6.8% (ST) to 50.4%, and a significant decrease of parataxis, from 93.2% (ST) to 49.6%, are found in clinician translations.

One may argue that the tactic shift from source text to translations can be taken for granted from a typological point of view, given that Chinese is widely acknowledged to be a paratactic language and English, a hypotactic one (Yu 1993; Tse 2010; Bisiada 2013; Liu and Lin 2015). However, tactic choices are more often motivated by pragmatic reasons (Khalil 2011), as indicated by the sharp contrast in tactic choices between clinician and non-clinicians. Indeed, Figure 4 reveals a patterned variation of tactic choices between clinicians and non-clinicians. Unlike with the clinician translations, where hypotactic choices are favoured, the non-clinician translators are found to highly favour parataxis, as suggested by the dominant percentage of 74% (Veith) and 68.9% (Unschuld), respectively. This variation pattern strongly shows that a translator’s medical expertise plays a significant role in their tactic choices.

The interpretation of a clause nexus often depends on more delicately differentiated process types (Matthiessen 2002: 264), which is why tactic choices are better understood when looking at the interaction with translators’ transitivity choices. Figure 4 shows that the logical sequence of events flows differently in the clinician and non-clinician translations, as indicated by the primary choices of processes made by each group of translators. Compared with the translations by non-clinicians, the translations by clinicians see a higher percentage of material process, with 67.9% and 70.4% (clinicians) vs. 59.8% and 59.4% (non-clinicians), and a lower percentage of relational process, with 12.8% and 17.7% (clinicians) vs. 22.9% and 23.8% (non-clinicians). This further suggests that domain expertise makes a difference in terms of how the translators construe experience of Chinese medicine in the translations.

The interaction between translators’ tactic and transitivity choices can be summarized as follows: clinicians tend to build their clinical experience materially and hypotactically. According to Butt and Webster (2017), our world is modelled through the combination of experiential and logical resources. It is often thought that the mainline flow of events tends to be construed paratactically (Thompson 1987). However, this finding shows a counter-expectant pattern, namely that the hypotactic construal of mainline events is possible. This is perhaps partly due to “expectancies,” that is, interactions between the experiential and logical systems (Butt and Webster 2017: 104), and partly to the medical expertise of the translator. Clinician translators are found to have a stronger sense of the relationships between medical activities, as illustrated by their higher tendency to arrange logical relationships hypotactically. See Example 5 for an illustration:

Example 5 describes a number of actions that should be taken in order to maintain a healthy lifestyle. The original text is arranged in three paratactic clauses, with three material processes in each translation. The non-clinician translations have strictly followed the ST pattern. However, clinician translations show great differences with the ST, with an increase in both material processes and hypotaxis. Both clinicians have added hypotactic clauses and give the reasons and purpose for the suggested actions. For instance, Wu not only suggests doing exercise, but he also explains from a clinician’s point of view that the purpose for doing exercise is to maintain harmony between human physiology and the atmospheric energy of spring. The same is found in Ni’s translation: he not only asks the reader to take certain actions, but also adds the reasons why these actions need to be taken. The logical connections between these actions are also made explicit.

This clearly illustrates the significant role of medical expertise in tactic choices. However, to be more accurate, there is a need to differentiate the impact of expertise from that of the translators’ mother tongue, as expertise may not necessarily be the only contextual variable that motivates a translator’s tactic choices. In fact, some researchers have argued for the role of a translator’s mother tongue in their stylistic choices, which has led to the establishment of the golden rule or into-mother-tongue principle in the directionality of translation (Venuti 1995/2017; Pavlović 2007; Wang 2011; Xu and Xu 2019). This is directly relevant to this study, as the first language of the clinician translators is Chinese, whereas the non-clinicians, Veith (English) and Unschuld (German), are non-native Chinese speakers. This extra layer of heterogeneity in the translators’ background poses an important question: how important a role does expertise play? In other words, can the observed differences in tactic choices also be attributed to a translator’s mother tongue?

To answer this question, another Li’s translation of Neijing by brought into comparison to test how important a role expertise actually plays (see Table 4):

-> See the list of tables

In Table 4, Li’s translation is compared with that of non-clinicians, ignoring the variable of expertise (as they are all non-clinicians) and allowing for a comparison to be made under only one variable, that of mother tongue. It shows that mother tongue does not play a significant role in the overall preference pattern of taxis. Indeed, regardless of their mother tongue, Li, Veith, and Unschuld all tend to favour parataxis, with a relatively higher percentage (Li: 53.4%; Veith: 74%; Unschuld: 68.9%) in each translation. This means that mother tongue does not affect the overall preference pattern of tactic choice so much; rather, it is registerial expertise (clinical experience in this case) that plays the central role in the overall preference pattern of tactic choices.

However, if we look at the discrepancy between parataxis and hypotaxis, Li’s translation is clearly much closer to the clinician translations, with 8.8% vs. 0.8%, and is significantly different from the non-clinician translations, with 8.8% vs. 48% and 36.8%. This difference demands further analysis of the differences between Li and the non-clinician translators. Apart from mother tongue, Li is significantly different from Veith and Unschuld in terms of domain-specific knowledge: Li’s education is primarily in the field of linguistics, whereas both Veith and Unschuld are principally educated in the field of medical history. This further suggests that registerial expertise (domain-specific knowledge in this case) plays a significant role in a translator’s tactic choices.

Tactic choices constitute a complicated rhetorical issue. They may also be motivated by stylistic considerations. In hypotaxis, units of information are organized according to their importance, with the most important information receiving primary emphasis in the main clause (Khalil 2011). In parataxis, the information in both clauses receives equal emphasis, creating a certain scope of contextual distance in which various dynamics are allowed (Liu 1983). Paratactic structure in classical Chinese provides enough room for readers’ imagination. For instance, Unschuld indicates that his reproduction of the source text’s format and style is there to help readers “reconstruct the ideas, theories, and practices” of Chinese medicine themselves (Unschuld 2011: 13).[8] This seems to echo Hemingway’s philosophy about his writing style (strong preference for using short and paratactic sentences), known as “the Iceberg Theory”: the dignity of movement of an iceberg is due to only one-eighth of it being above water, leaving the remaining seven-eighth as room for readers to interpret the crux of the story themselves (Darzikola 2013).

Expertise has also been found to play an important role in a translator’s choices of logic-semantic types (See Figure 5).

Figure 5 shows that the type of elaboration is statistically insignificant in all the translations, as indicated by the low percentage of 2.1%, 0%, 0.5%, and 6%, respectively. The discussion will thus only focus on extension and enhancement, which make up the dominant percentage in all the translations.

In Figure 5, a salient variation pattern of logic-semantic types has been found between clinician and non-clinician translators: clinician translators tend to favour enhancement, whereas non-clinicians, extension. This is indicated by the sharp increase in the relative percentage of enhancement, from 25.9% (ST) to 55.5% (Wu) and to 60% (Ni), a much higher percentage compared with 36.7% (Veith) and 35.7% (Unschuld). With regards to extension, the pattern is reversed: compared with clinician translations, non-clinician translations display a higher tendency towards extension, with a percentage of 62.8% and 58.3% respectively. See Example 6 for a further illustration:

Example 6 is about ancient Chinese people’s attitude towards health; it suggests that preventing disease was more important than treating it. This example shows clear differences between clinicians and non-clinicians. Clinician translators tend to favour enhancement, while non-clinician translators tend to favour extension. Both clinician translators choose to enhance the meaning by qualifying it with circumstantial elements of time and manner: Wu’s translation is organized in two ranking clauses logically linked by “when,” suggesting time. Ni’s translation is organized in three ranking clauses, with the first one enhanced by the second, suggesting manner, which is in turn enhanced by the third. In Veith’s translation, there are two ranking clauses, the internal logic relation being extension. Unshuld’s translation contains three ranking clauses, the first of which projects the second and third, where the logical relation is extension. This shows that expertise surely affects the choices of specific logical relations.

With regard to the impact of mother tongue, the same method as the one described iin Section 5.2 is applied (see Table 5).

Table 5 reveals a different preference pattern between translations differentiated by the translator’s mother tongue (the only contextual variable in this case) rather than field of expertise. It shows that Li’s translation tends to favour enhancement (51.7%), whereas Veith and Unschuld tend to favour extension, as indicated by the higher relative percentages of 62.8% and 59.5%, respectively. This means that, apart from the translators’ medical expertise, their mother tongue also plays an impactful role in their choices of logico-semantic relations.