Article body

1. Introduction

The need to categorise texts has long been acknowledged in Translation Studies. The basic assumption is that translation choices would differ in different types of text, which gives rise to research on translation-oriented text typology. One influential categorisation is proposed by Reiss (1971/2000; 1977/1989). She classifies texts into one of the three major types: informative texts (that is representing facts), expressive texts (that is expressing attitude) and operative texts (that is making appeals). By way of suggestion, Reiss further puts forward translation strategies for different text types. While the influence of text type (or its adjacent terms genre and register) on translation choices has been widely acknowledged (for example Nord 1991/2005; 1996; House 2014; Trosborg 1997; 2000; Schäffner 2000; 2002), empirical research on translation choices made across different text types is still rare (see studies focusing on a particular type of text, such as Byrne [2012] and Olohan [2015] on scientific and technical translation, and Pedersen [2011] on audio-visual translation). The current study has thus been conducted to investigate the leverage of “text type” on translation choices by attending to various types of text. Specifically, we focus on the translation of logical relations from Chinese to English and seek to explore the following research questions:

  1. What are the similarities and differences of the Chinese-to-English (C-E) translation choices of logical relations across text types?

  2. How to interpret the differences in terms of the particularities of each text type?

The next two sections will provide information on the theoretical foundations of our study as well as the data and methods. Section 4 will analyse the data and discuss the findings to answer our research questions in detail. Finally, a conclusion is drawn.

2. Theoretical foundations

The theoretical framework of this study is Systemic Functional Linguistics (SFL). As a meaning-oriented approach to language, SFL provides useful tools for text analysis in Translation Studies. In our case, SFL enables a systematic account of the logical relations in the source texts and target texts. The SFL theorisation of logical meaning will be introduced in Section 2.2, followed by a presentation of the range of choices for translating Chinese logical relations into English in Section 2.3. Below we will start with an introduction to text typology under the SFL paradigm.

2.1. “Text type” as identified in SFL

In this study, we use Matthiessen’s model for classifying texts (Matthiessen 2014; 2015; Halliday and Matthiessen 2014). This model, among others (for example Reiss 1971/2000; 1977/1989; Snell-Hornby 1988/1995), is particularly suitable for the current research as it would enable a theoretically consistent description of our data. Matthiessen’s taxonomy of text is based on the contextual variable of field, which refers to the socio-semiotic activity going on in the context of situation.

In Matthiessen’s text taxonomy, eight types of text can be identified: expounding, reporting, recreating, sharing, enabling, recommending, exploring and doing. The situations represented in the eight text types are given below:

  • Expounding: categorising or explaining knowledge of general classes of phenomena in the world (for example, a research article);

  • Reporting: chronicling events in the world, surveying divergent places or inventorying various entities (for example, a news report);

  • Recreating: dramatising or narrating aspects of human life in an imaginative way (for example, a novel);

  • Sharing: sharing private experiences or personal values (for example, a diary);

  • Recommending: promoting some commodity for the sake of the speaker or giving advice for the sake of the addressee (for example, advertisement);

  • Enabling: instructing people in how to undertake an activity or regulating the activity by controlling actions (for example, know-how [procedure]);

  • Exploring: arguing or reviewing societal values and positions in public (for example, a review);

  • Doing: directing or coordinating the performance of a social activity that involves one or more persons (for example, an invitation) (see for example Halliday and Matthiessen 2014: 35-36).

These text types, as demonstrated by several studies conducted by Matthiessen and his colleagues (for example Matthiessen and Kashyap 2014; Matthiessen and Teruya 2015; Matthiessen and Pun 2019), vary systematically in terms of their semantic features, as well as their lexicogrammatical realisations. For instance, regulating texts (a subcategory of enabling) tend to use more cohesive conjunctions (for example furthermore, however) than other types of text (Matthiessen and Teruya 2015). On the basis of these studies, our current concern is: to what extent the lexicogrammatical features, logical relations in particular, are retained in translation. One related issue under consideration is whether the semantic features and their lexicogrammatical realisations of different text types vary systematically across languages. This study on C-E translation serves as a primary step to address these issues.

2.2. Logical meaning and clause complexing system

In SFL, three strands of meaning are postulated: ideational, interpersonal and textual meaning. Ideational meaning is composed of experiential meaning and logical meaning. Experiential meaning construes human experience within the unit of clause; logical meaning builds relations between one human experience and another. It is noteworthy that SFL recognises “clause” as the highest-ranking unit in lexicogrammar and there is no separate rank for sentence. When a sentence contains a single clause, it is a clause simplex (no logical relation in it). In cases where a sentence constitutes at least two clauses, it is called a clause complex; clauses are logically related within a clause complex.

Logical meaning can be accounted for through the system of clause complexing, as mapped out in Figure 1. It consists of two basic systems: taxis and logico-semantic type. The system of taxis is concerned with clause interdependency, with two options of parataxis and hypotaxis. In a paratactic relation, clauses are of equal status, annotated by Arabic numbers: 1, 2, 3, and so on. Clauses are of unequal status in a hypotactic relation; the Greek letter α represents the dominant clause and β, γ, etc. for the dependent clauses. The system of logico-semantic type incorporates projection where the secondary clause is projected by the primary one and expansion where the secondary clause enlarges the primary one. Projection links the activity of saying or thinking to its projected content, which is manifested in either locution (”) or idea (’). Expansion is classified into elaborating (=), extending (+), and enhancing (×). In elaboration, one clause elaborates the meaning of another by restatement, specification or exemplification. In extension, one clause extends another through adding something new to it. In enhancement, one clause enhances another by qualifying it in one of the following possible ways: time, place, manner, cause or condition.

Figure 1

Two basic systems in clause complexing (adapted from Halliday and Matthiessen 2014: 438)

Two basic systems in clause complexing (adapted from Halliday and Matthiessen 2014: 438)

-> See the list of figures

The two systems in clause complexing are applicable to both languages of English and Chinese, as described in Peng (2000), Halliday and McDonald (2004) and Li (2007). In English, one can mostly identify taxis and logico-semantic types by conjunctions.[1] However, clauses in Chinese are often linked in an implicit way only through juxtaposition (Hu, Zhu, et al. 2005; Kim, Heffernan, et al. 2016). This feature in terms of explicitness of logical relations in Chinese presents challenges to readers, requiring a great effort of interpretation and also introducing a high degree of ambiguity.

In such cases, we draw on Hao (2020) for analysing taxis in Chinese: parataxis allows for zero conjunction while hypotaxis needs to be marked through conjunction(s). In other words, whether the conjunction is obligatory or not can be seen as an indicator to distinguish parataxis from hypotaxis. Also, proposals put forward by Halliday and Matthiessen (1999: 301-305) and Hsu (2017: 151-168) are adopted as the complementary criteria for analysing our data (for details on ways to identify taxis in Chinese, see Li and Yu 2021). As to the logico-semantic type in Chinese, we analyse it mainly based on how the experiential meanings of clauses are logically related in semantics (see also Li 2019), rather than relying heavily on the presence of explicit conjunctions.

2.3. Translation choices of logical relations from Chinese into English

For the choices of translating logical relations, two studies have been found particularly pertinent to the current research. One is Li and Yu (2021), which explores C-E translation choices of taxis in expounding texts; the other is Li and Kim (2021), which investigates choices of logico-semantic type in recreating texts.[2] The translation choices found in the two studies can be consolidated into a more comprehensive system network as shown in Figure 2 (“—” representing the opposite feature of the corresponding one).

Figure 2

Translation choices of logical relations from Chinese into English

Translation choices of logical relations from Chinese into English

-> See the list of figures

As can be seen from Figure 2, a logical relation in a Chinese clause complex can be translated with or without a shift. The notion translation shift was first defined by Catford (1965: 73), meaning a “departure from formal correspondence in the process of going from the SL [Source Language] to TL [Target Language].” In our study, the term shift specifically refers to changes in the domain of logical meaning. The choice of shift can be further divided according to the number of logical relations; that is, a logical relation can be either removed or added from the Chinese source text to the English translation. This category of “shift in number” is also identified as “major shift” because the original combination of clauses within a clause complex is destructed in translation. The corresponding category is “no shift in number” (also named “minor shift”), where the logical nexus of the source text is retained, only with taxis, logico-semantic type and/or explicitness shifted within the nexus. In cases showing a “shift in taxis,” a certain type of tactic relation between Chinese clauses is shifted to the other type in the translated English text. Such shifts are almost all from parataxis to hypotaxis, perhaps due to the difference that Chinese tends to have more paratactic and fewer hypotactic relations in comparison with English (Wang 1954; Lian 2010; Yip and Rimmington 2016). In cases with a “shift in logico-semantic type,” an elaborating/extending/enhancing relation in Chinese is shifted to another in English. In cases with a “shift in explicitness,” an implicit relation without conjunctive markers between clauses is shifted to explicit in translation, or vice versa. These three kinds of shift in taxis, logico-semantic type and explicitness can occur singly or in combination with another, as indicated by the curly bracket in Figure 2.

3. Data and methods

Our data set consists of Chinese passages totalling 10,982 characters and their English translations (7,263 words). These texts were extracted from a set of translation textbooks: Translation Practice between Chinese and English: Intermediate Level (Lu 2009/2017a[3]) and its associated workbook (Lu 2009/2017b[4]). This choice was made for the following two reasons. First, these textbooks contain texts from several different text types, allowing the comparison of translation choices to be carried out across a range of text types. Second, the translation quality of the books is assured by the editorial committee of the China Accreditation Test for Translators and Interpreters (CATTI), the most authoritative test of this kind in China. Yet, it needs to be noted that we are not aiming to assess the quality of the translations in this paper. Rather, assuming that the translations can serve their purposes, we focus on providing a systematic description of the translation choices as well as on interpreting these choices in terms of the particularities of different text types.

To enable meaningful comparison, we selected passages totalling 200 Chinese clauses and their English translations for each text type. In this we included expounding, reporting, sharing and recommending texts. The enabling, doing, recreating and exploring types were excluded, as the number of passages found in the selected textbooks were too limited to generate sufficient data for analysis.

To be specific, the expounding texts are exemplified by passages in the field of science, the reporting texts focus on news reports chronicling the flow of events, the sharing texts are from three pieces of lyrical prose that share personal feelings towards an object or event and the recommending texts have to do with the promotion of Chinese tourist sites and cultural phenomena. Table 1 presents the titles of major chosen passages for the four text types.

Table 1

Major selected passages in the four text types

Major selected passages in the four text types

-> See the list of tables

This study was carried out in the following steps. First, the selected Chinese source texts and the English target texts were compiled into a parallel corpus for each text type. Then, from the 800 (200x4) identified clauses in the source texts and their corresponding translations, tactic relations between clauses within a clause complex and their logico-semantic types were analysed and annotated, in accordance with the criteria introduced in Section 2.2. Explicitness of logical relations, as an aspect showing distinct typological difference between English and Chinese, was also labeled. Next, translation choices for all the logical relations in our data were identified, based on the system network summarised in Section 2.3. Finally, numbers and percentages of the translation choices were compared between the expounding, reporting, sharing and recommending texts. This provides a clear picture of similarities and differences in choices across text types as well as further exploration of the particularities of each text type that may influence the differences. In the next section we will present the findings of our comparison in detail.

4. Analysis and discussion

In broad terms, the comparison reveals three major similar tendencies among the four types of text. The first similarity is that the choices of “no shift” and “removal” are the most widely adopted, together taking up around 70% of all the choices. Second, there are certain instances involving “addition,” “shift in taxis” and “shift in explicitness” across the text types, each occupying the percentage of around 10%. One further similarity is that the shift in logico-semantic type rarely occurs. See the statistics in Table 2.

Table 2

C-E translation choices of logical relations across the four text types

C-E translation choices of logical relations across the four text types

-> See the list of tables

A closer look at Table 2 reveals some differences in certain choices between the four text types. The rest of this section will attend to the differences and discuss possible influential factors for notable differences. The focus will be on the primary choices (Section 4.1) and the delicate choice of “removal” (Section 4.2) respectively. Other delicate choices in Figure 2 are not dealt with in this article due to their limited occurrences in our data (see Table 2).

4.1. Comparison of primary choices across text types

Statistics on the two primary choices of “shift” and “no shift” are presented in Table 3.

Table 3

Primary translation choices across the four text types

Primary translation choices across the four text types

-> See the list of tables

As can be seen from Table 3, the percentages of “no shift” in expounding and reporting texts are higher than those in sharing and recommending texts. We propose that this may be associated with whether the logical relations have been clearly spelled out in the original text, that is the explicitness of the logical relations. The Chinese expounding and reporting texts in our data feature relatively clear logic, especially with explicit relations taking up 60% in expounding texts, while sharing and recommending texts have more implicit logical relations[5] (see Figure 3).

Figure 3

Explicitness of logical relations in the ST across text types

Explicitness of logical relations in the ST across text types

-> See the list of figures

Therefore, when translating explicit relations in expounding and reporting texts, translators do not need to spend much time analysing logical meanings in Chinese and hence are more likely to transfer the logic directly to English. For instance, in Example (1) from an expounding text, the first Chinese clause, dependent on the second one, functions to explain the purpose. This enhancing relation is marked by 为了 [wèile; (in order) to]. In such contexts where the logical meaning is clearly expressed in the source text, there is no shift in the target text.

In contrast, logical relations in recommending and recreating texts in Chinese are more often implicit. Hence, there is a greater challenge for translators of these types of texts both in interpreting how clauses are logically related to each other in the source text and in translating these relations into English. In such cases, translators are less likely to directly follow the original logic, but consider other translation factors along with the interpretation of logical meaning (as translation is often regarded as a negotiation of many variables, see for example Kim 2009). The influence of these factors may account for a greater degree of shifts in recommending and recreating texts. As exemplified in (2) from a recommending text, the three Chinese clauses are logically related in an implicit manner. In the process of interpreting the logical meaning, the experiential meanings of clauses might have been taken into account. Specifically, the omission of the third clause in translation and the subsequent removal of the extending relation may be related to two factors. One is concerned with the low lexical density of the third clause (only four Chinese characters; see further in the next section) and the other is concerned with the similarity in meaning between the second and the third clauses (see more about possible motivations for removing logical relations in translation in Li and Yu 2021).

The above comparison of primary choices across the four text types reflects that text types, varying in clearness of logic, do have an impact on whether to maintain the logical relation in translation. Despite this, it has also been found that no matter which type of text the source text belongs to, there is usually “no shift” in the logical relations expressed via correlative conjunctions. Correlative conjunctions refer to two conjunctions used together as a pair, such as 只要… 就 [zhǐyào…jiù; as long as…then] and 由于… 因此 [yóuyú…yīncǐ; because…so] in the source texts of Examples (3) and (4).

In the translations of (3) and (4), the correlatives are reduced to singly conjunctions, that is so long as and since. Although the forms of conjunctive markers are changed, nothing related to the taxis, logico-semantic type or explicitness of logical meaning has been shifted.

4.2. Comparison of delicate choices: a “removal” focus

This section attends to the delicate choice of “removal.” The distribution of this choice across the four text types is summarised in Table 4. We can see that the percentages of “removal” in recommending and sharing texts both exceed 50%, higher than those in the other two types of text. Expounding texts contain the lowest percentage of this major shift (34%).

Table 4

Translation choice of ‘removal’ across text types

Translation choice of ‘removal’ across text types

-> See the list of tables

One possible reason for this difference may be related to the varied orientations in different text types. Translations of expounding and reporting texts tend to be more ST-oriented, with a major concern to fully and accurately transfer the content from the source text to the target text. Given this communicative purpose, the choices of “no shift” and minor shifts in taxis, logico-semantic type and/or explicitness are more often found in these two text types. For instance, in (5) from a reporting text, the two enhancing relations in the source text are translated faithfully to a certain degree, with only minor shifts in explicitness (that is two explicit markers of by and and are added) and taxis (that is the paratactic relation in the inner layer becomes hypotactic).

By contrast, the translations in recommending and sharing texts are more TT-oriented, mostly aimed to “promote” Chinese cultural phenomena to English readers via the introduction of China’s tourist sites and the expression of personal attitudes towards objects or events in the Chinese context. As a result, more shifts may be needed in these text types to make the English translation more readable and understandable for readers. Such shifts are often major ones, including shifts in experiential meaning within the unit of the clause, and shifts in logical meaning. In the latter case, the shifts typically involve the removal of logical relations in C-E translation.

For example, in (6), the Chinese clauses are logically related in a hypotactic enhancing relation in a clause complex. The second clause contains several Chinese cultural phenomena, such as 二泉映月 [èrquán yìnɡyuè; “moon reflection in the fountain of Er Quan,” the title of a piece of classic Chinese music played on a string instrument] and 隶书 [lìshū; “Li style,” an ancient Chinese calligraphic style]. In a sharing text, these cultural elements need to be explained in translation for the sake of comprehensibility and acceptability by English readers. Perhaps due to the difficulty a single English clause would have in accommodating a detailed explanation of the cultural phenomena, the experiential meaning of the second clause is divided into two parts in translation. One part is integrated with the meaning of the first clause into a clause simplex while the other part becomes independent as another clause simplex. Accordingly, the logical relation is removed in the target text.

Another possible reason for different frequencies of “removal” in the four text types is the concept of text complexity (Eggins 1994; Matthiessen 2002; Halliday and Martin 1993; Castello 2008). In SFL, text complexity can be measured in terms of two parameters: grammatical intricacy (that is the number of clauses per clause simplex/complex) and lexical density (that is the number of lexical items per clause). Lexical items are also known as content words, which include nouns, verbs, adjectives, and most adverbs (in contrast with functional items, which are also known as function words and include prepositions, conjunctions, auxiliary verbs, pronouns and determiners).

The complexity of a text in source texts may affect a translator’s choice as to whether to change the number of logical relations in translation (Li and Yu 2021). Specifically, when the clause complex is grammatically intricate, it is more likely to be segmented into two or more clause simplexes/complexes in translation, which leads to the removal of logical relation(s). On the contrary, lower grammatical intricacy means a lower possibility of segmentation, but nevertheless sometimes combining clause simplexes/complexes into one clause complex. When the lexical density is relatively low, the “removal” choice often occurs through downgrading—a clause that is logically related to another clause in the Chinese source text is downgraded into a group or phrase in the English translation. The higher the lexical density, the less likely it is that logical relation(s) would be removed. Instead, the translator may upgrade, in some cases, a constituent in a clause to a ranking clause.

Figures 4 and 5 present statistics on grammatical intricacy and lexical density in clause complexes of the Chinese source texts across text types. We can see from Figure 4 that clause complexes consisting of less than four clauses (that is relatively low grammatical intricacy) take up 92% of all the clause complexes in expounding texts, the highest percentage among all types of text. Therefore, the likelihood of removing logical relations in expounding texts is lower than in the other three text types. As can be seen in Figure 5, sharing texts have the lowest lexical density, containing mostly clauses with less than ten lexical items (83%), while those with more than ten items only account for 17%; expounding texts have the highest proportion of content words compared to the other three types (50%). Given such features of the source texts, sharing texts tend to contain more shifts of “removal” by downgrading; conversely, logical relations are less likely to be removed in expounding texts.

Figure 4

ST grammatical intricacy in clause complexes across text types

ST grammatical intricacy in clause complexes across text types

-> See the list of figures

Figure 5

ST lexical density in clause complexes across text types

ST lexical density in clause complexes across text types

-> See the list of figures

Example (7) from a sharing text is a distinctive instance of high grammatical intricacy and low lexical density in the source text. The Chinese clause complex includes six clauses, five of which have less than five content words, with only one clause, the fourth one, having six lexical items. We present our interpretation of taxis and logico-semantic types in the source text while acknowledging that there may be other ways of analysing the logical relations, as most of them are implicit without any linking through conjunctive markers. It might be difficult for the translator to tease out the logic in Chinese and accommodate it in one English clause complex. Consequently, the clauses within a clause complex are segmented into two clause simplexes in translation; the second Chinese clause as well as the third, forth and fifth clauses are respectively downgraded into embedded clauses[6] of the first clause and the sixth one. In other words, the target text removes all five logical relations. Only the extending relation indicated by 并且 (bìnɡqiě; and) is, it can be argued, retained in an embedded clause complex annotated in “[[…]].”

In contrast, in (8) from an expounding text, the original Chinese text is characterised by low grammatical intricacy, including two clauses within the clause complex. At the same time, it is lexically dense with more than 15 lexical items per ranking clause on average. Such characteristics of text complexity in the source text may decrease the likelihood that the logical relations are removed and, in (8), the choice of “no shift” is made in the translation.

So far we have explained how text orientation and features of text complexity in different text types may affect a translator’s choice as to whether to remove a logical relation. This, to some extent, proves the influence of text type on the number of logical relations in translation. However, it can be seen from Table 2 that “removal” is almost always the most frequently adopted translation choice. Even in expounding texts where “removal” is the least frequently used choice among the text types, the percentage is still over 30%, higher than most of the other choices. This unexpected phenomenon is probably related to the more frequent use of nouns and nominal groups in English scientific texts than in other types, as stated in Halliday and Martin (1993). The nominalised rather than verbalised language in the English translation of expounding texts would lead to a decrease in the number of clauses and thus a decrease in the number of logical relation(s), such as Example (9).

In (9), the verb 通过 [tōnɡɡuò; traverse/pass] is omitted in the English translation and the first Chinese clause is simplified and downgraded to a nominal group the effect of irregularly-shaped freight trains at current speeds. In this way, the temporal relation in the source text is removed in the target text.

5. Conclusion

This study makes a preliminary step towards the empirical investigation of the influence of text type on translation choices by attending to the case of translating logical relations from Chinese to English. Adopting the SFL framework, this article has compared the C-E translation choices of logical relations across expounding, reporting, sharing and recommending texts. While the same range of choices are present, the favoured choices are different among the four text types. Expounding and reporting texts tend to favour the choices of “no shift” and minor shifts in terms of taxis, logico-semantic type and/or explicitness. This could possibly be interpreted in terms of the relatively clear logic of the source text and the ST-oriented style of translation with limited freedom to make shifts. The features of low grammatical intricacy and high lexical density in expounding texts may also contribute to the dispreference for shifting the number of logical relations. In contrast, major shifts, especially the “removal” choice, are more frequently adopted in the translation of sharing and recommending texts, which is possibly owing to the frequent absence of conjunctive markers in the source text as well as the TT-oriented style of translation. These preferences of translation choices and their possible explanations with reference to textual characteristics of the source texts and text orientations in certain text types are briefly depicted in Figure 6.

Figure 6

Favoured translation choices across the four text types

Favoured translation choices across the four text types

GI for grammatical intricacy; LD for lexical density. * Reporting texts are excluded from the feature of low grammatical intricacy and high lexical density in our data.

-> See the list of figures

This article is one of a number of attempts to employ SFL in Translation Studies (for example Choi 2013; Yu 2019; Kim and Munday 2021; Jing 2021). Together, we have reinforced the idea that analytical tools and concepts from the linguistic approach of SFL can offer a systematic and comprehensive way in accounting for translation phenomena. Additionally, SFL can provide a more disciplined approach to addressing the question “why certain translation choices are made.” It thus complements the studies focusing on “what translation choices are to be made” by providing linguistic evidence. In this, the possible factors conditioning translation choices could be made explicit. In other words, the key to the question “why certain translation choices are made” would no longer only rely on translators’ intuitions, but could be explained with reference to linguistic features. In the current case, the SFL approach has enabled us to identify the textual characteristics at stake in translation and the extent to which translation choices are conditioned by these characteristics. This study would thus help translators make informed choices when translating texts in different types and enable them to justify their translation choices with reference to the linguistic particularities of each text type. Such justification has significant pedagogical implications for translator training: it could help translator trainees explain their intentional or unintentional choices as well as provide linguistic grounds for translator trainers to interpret and evaluate translation products in different text types. Specifically, the analysis of particularities of each text type and the favoured translation choices can concretise the linguistic evaluation of translation quality with respect to the field of register (see more about translation quality assessment in House 2014).

Future research could explore the leverage of text type by attending to other aspects of meaning, such as the experiential. It would also be interesting to see the translation choices made in other languages and whether semantic features and their lexicogrammatical realisations of each text type vary systematically across languages. In the end, we would like to call for more applications of SFL as an empowering tool for exploring translation issues based on concrete linguistic evidence.