On general extenders in literary translation and all that stuff

1 Introduction

The present article addresses the problem of translating a special type of pragmatic markers, i.e., general extenders (GEs). I am especially interested in GEs that are employed as stylistic markers in literary texts, with a focus on translations from English into Romanian. The issue has been previously discussed by Furkó (2020) for translations from English into Hungarian, but Furkó’s study investigates strategies in the translation of discourse (and pragmatic) markers in general and only briefly looks into the translation of GEs. A more recent study (Overstreet and Yule 2021) discusses GE equivalents in a series of languages such as French, Spanish, Brazilian Portuguese, Dutch, German, Swedish, Lithuanian, Russian, and Persian but does not offer data on Romanian. To my knowledge, so far there has hardly been any literature on Romanian GEs’ behavior in translation or otherwise, with the notable exception of Hornoiu’s (2016) excellent study on spoken Romanian.

In addition to providing data on GEs’ behavior in Romanian, my analysis addresses issues such as pragmastylistic and metapragmatic awareness in translation and whether this is reflected or not in retranslation, since one of the sample texts in my corpus benefits from what Pym (2004, 90) refers to as “multiple presentation,” namely, alternative translations of the source text.

2 GEs in English and Romanian

A working definition of GEs is offered by Overstreet (1999, 3): GEs are structures such as and so on and so forth, and all, and all that stuff, or something, or anything, etc. that are said to be “general” due to their non-specific reference and “extenders” because they extend utterances that appear to be complete grammatically. Their functioning as extenders indicates that they tend to appear in clause- or at least phrase-final position. There is an ample body of literature on English GEs (Overstreet and Yule 2021, Hornoiu 2022, Unuabonah and Oyebola 2023, to mention but a few of the more recent studies) that underlines the multifunctionality and context dependency of these structures: GEs can be used either referentially, marking a set containing at least one element in the preceding discourse, or they can function non-referentially, either as pragmatic markers (indicating that there is common ground and shared knowledge, solidarity among interlocutors) or as textual markers (where they function as punctors, i.e., elements that segment the discourse, mark information chunks, help planning the discourse, etc.).

Consider the following examples:

The aforementioned examples contain excerpts from novels that were written years apart, beginning with 1915 (John Galsworthy’s Freelands) and ending with 2021 (Kazuo Ishiguro’s Klara and the Sun). The example under (1a) contains a GE (and all) used referentially, as a set marker, which is indicated by the list of items preceding it. The NPs preceding the GE make up the referent (or the “anchoring constituent” of the GE, Ward and Birner (1993, 208) quoted in Borreguero Zuloaga 2022, 156). A similar referential use can be identified for both examples under (3). All the first four examples contain adjunctive GEs (GEs that suggest that there is more to be said about the referent), while the fifth example contains a disjunctive GE (a GE that suggests that there are other possibilities that could be mentioned with respect to the referent). The GE and all that rot in the examples under (3) is an instance of downgrading GE, just like and all that crap/shit, and all that nonsense, etc. Most GEs are a mark of informality, but some of them, such as and so on and so forth, etcetera are frequently used formally (Overstreet 1999).

While the examples under (1a) and (3) are clearly referential, the rest of the examples are more difficult to pin down. The GE used twice under (1b) seems to fulfill a pragmatic function, providing clues with respect to the narrator’s emotions. So does the GE under (1c), which is meant to complete the apology uttered by one of the characters in the novel. In a similar fashion, the example under (2) contains a GE used as a hedge, mitigating a refusal. The example under (4) also contains a hedge, meant to mitigate the answer that the character provides to his interlocutor. The disjunctive GE under (5) can be interpreted both referentially, as an attempt to extend the array of choices for the character, and non-referentially, as part of a stalling strategy of the character. In fact, as stated before, it has been noted in the literature that GEs are multi-functional and their interpretation is context-dependent. They are also notoriously difficult to pin down semantically due to the fact that their referent is also difficult to identify in most cases (Overstreet 1999, Evans Wagner et al. 2015).

The non-referential use of GEs has been linked to a tendency towards grammaticalization (Cheshire 2007), which is supported by phenomena such as desemanticization and phonological attrition (as is the case with and stuff, and all, or something). While GEs in other languages seem to behave in a similar fashion (having developed non-referential uses of their own), as is the case with Italian (Fiorentini 2018), French (Secova 2017) and Spanish (Borreguero Zuloaga 2022), this doesn’t yet seem to be the case with Romanian. From what I could see in the examples consulted in the CoRoLa corpus (CoRoLa (racai.ro), Tufiș et al. 2019), most GEs in Romanian seem to be adjunctive and to have a referential use, as indicated by the enumerations that precede them and that constitute their referents. Let us consider a few examples:

The examples under (6) contain instances of the GE și așa mai departe, literally translated as “and so more far,” that is frequently used in standard Romanian and could be seen as an equivalent of the more formal and so on and so forth. (6a) has as a referent string of noun phrases (NPs), while (6b) extends a set of verb phrases (a set of events that are its antecedents in a temporal anaphoric chain). The examples under (7) contain instances of an informal GE, alea alea, which literally translates as “those, those” and is [-animate]. The examples are both intriguing: (7a) because in the set of NPs that the GE extends, there is also an instance of code-switching and (7b) because the GE is preceded by the same preposition that precedes all the NPs in the referent set, which is not a frequent combination in Romanian. This only goes to show that the GE is considered to be a symmetrical part of the preceding enumeration and is definitely used referentially, as are most of the GEs in the Romanian examples I provided. The sentences under (8) contain illustrations of non-prototypical, or special, GEs (of the more expressive kind but still formulaic): chestii trestii, literally translated as “things reeds,” which is a reduplicative-rhyming phrase, and și tot tacâmul, which literally means “and the whole (piece of) cutlery.” Both these GEs are expressive and informal, possibly humorous, and might be the equivalent of and all that stuff. The examples under (9) illustrate the use of the disjunctive GE sau mai știu eu ce, literally translated as “or I don’t know what,” which might be an equivalent for the disjunctive GE or something. In (9a), the GE has a set of NPs as a referent, which means that it is employed referentially. (9b) is more interesting because the referent is a whole clause (că m-am născut în comunism “that I was born during communism”), which means that the preceding referent contains only one element in the set. The meaning of the GE becomes more difficult to pin down: it may have a pragmatic nuance implying shared knowledge of other such reasons to use as excuses between interlocutors when discussing their choices in life. In my opinion, it is only the GE in the examples under (10) that is showing clear signs of grammaticalization. The example under (10a) indicates that etcetera has lexicalized in Romanian: the Latin conjunction et “and” is no longer perceived as meaningful, which is proved by the fact that it is no longer spelled separately from the noun cetera “the rest,” but also by the fact that the GE is preceded by the Romanian conjunction și “and.” This indicates that the phrase et cetera has undergone univerbation (Brinton and Traugott 2005, 68) and has petrified in Romanian, becoming semantically opaque, a phenomenon that is also observable in Spanish etcétera (Borreguero Zuloaga 2022) and in Italian eccetera (Fiorentini 2018). Moreover, the example under (9b), where the GE is repeated, indicates that in that context, the GE is no longer used just as an extender to a set (in this case, the set contains adjective phrases as a complex referent) but also as an intensifier (Borreguero Zuloaga 2022, 176), which makes the contrast between the first sentence and the second even sharper. This is probably why in the source text no adversative conjunction is used as a connector between the two sentences although I did feel the need to insert it in translation. Last but not least, under (11) there is a very interesting example that contains two instances of what Borreguero Zuloaga (2022, 162) refers to as a “reported discourse extender” (employed by a speaker when s/he has to report the speech of another speaker). In this case, it is not a set that is extended but a chunk of discourse. The first, more informal, GE is c-o fi c-o păți, literally translated as “that might be, that might happen,” while the second, more standard, GE is c-așa și pe dincolo, literally translated as “that so and on beyond,” both meaning and so on and so forth, this and that and the other, but being used exclusively to report some other speaker’s words. Both structures are interesting because they are formulaic structures that are preceded by the complementizer că “that,” just like the ones discussed by Borreguero Zuloaga (2022, 164) for Spanish.

As a sort of conclusion to this section, while Romanian does possess a number of GEs of its own, most of them seem to be used mostly referentially, as proved by the examples provided here and culled from the CoRoLa corpus. I have noted some non-referential use, such as intensification, expressed by repetition (see also examples of spoken Romanian with the adjunctive GE (și) tot, in Hornoiu 2016, 71) or by additional linguistic material (for instance, the adverb acolo in the disjunctive extender ceva acolo; Table 1). The data also indicate that Romanian evinces what Unuabonah and Oyebola (2023, 81) refer to as “low preference for disjunctive extenders.” Most of the examples in the corpus I consulted are, however, fragments of written Romanian, so a detailed quantitative analysis of a larger corpus should be conducted so as to confirm my observations. In the table above, I have compiled a list with all the Romanian GEs I have identified. I used the typology proposed in the study by Borreguero Zuloaga (2022) to classify these structures into adjunctive, disjunctive, and reported discourse extenders.

3 Fiction GEs in translation

A literary translator faced with the task of translating fiction GEs from English into Romanian might find this task tricky, considering that in fiction, English GEs tend to acquire more than one function and can be interpreted both referentially (as set-markers) and non-referentially (as pragmatic and as stylistic markers). Given that Romanian GEs seem to be more restricted with respect to their functioning as pragmatic markers, I expect that an experienced literary translator will find various other pragmatic markers as equivalents, not necessarily GEs (employing the strategy of compensation). If GEs are used by the author of the source text strategically (i.e., with marked stylistic effect), the translator might try to copy the strategy but will probably select a larger number of equivalent Romanian GEs to translate one particular English GE that has been consistently employed in the source text. This strategy is convenient because formal-functional equivalents may be difficult to find in the target language (as suggested by Crible et al. (2019, 153) for the translation of discourse markers). In this case, the translator will opt for lexical variety and dismiss the consistent repetition that stylistically marks the source text. A third possibility is, of course, for the literary translator to omit the GE without trying to compensate for the translation loss that this omission causes.

I intend to check these predictions with the help of a corpus subdivided into two sub-corpora. The first sub-corpus consists of one source text (STa) selected from the novel The Catcher in the Rye, by J.D. Salinger, published in 1951, and its multiple target texts are as follows: two Romanian translations (TTa1 and TTa2), one Italian translation (TTa3), and a Spanish translation (TTa4). I have decided to investigate strategies of translation in this particular novel because it is famous for its GEs (Overstreet 1999, Overstreet and Yule 2021). As pointed out in the literature (Costello 1959, Bloom 2007), GEs are used by J.D. Salinger strategically in order to characterize the protagonist of the novel, Holden Caulfield. Their frequent presence in Holden’s speech patterns identifies him as a teenager (for GEs in the language of teenagers, see Tagliamonte 2016). The translation of GEs in this case is doubly challenging: first, because the use of GEs (and all, or something, or anything, plus a number of non-prototypical ones such as and all that crap) is mostly non-referential, which might pose problems for the Romanian translator since in Romanian GEs seem to be mostly used referentially. This is why I have also decided to consult translations made in other languages, such as Italian and Spanish. I expect a smaller number of omissions in Italian and Spanish because these two languages seem to be advanced in developing non-referential uses for their GEs (Borreguero Zuloaga 2022, Fiorentini 2018). A second reason why finding equivalents for the GEs in Salinger’s novel might pose problems to a translator is the fact that the GEs in this novel fulfill a threefold role: (a) at the level Author-Reader, they serve as “strategies of authentication” (Furkó 2020, 208), namely as stylizing strategies meant to persuade the reader that the textual world created by the author is authentic; (b) at the level Narrator-Narratee, they reinforce the truthfulness of the Narrator (Holden), who hates all that is “phony” and who, apart from GEs, repeatedly uses the parenthetical “if you want to know the truth” when telling his story; (c) at the level Character-Character, where GEs fulfill the pragmatic, non-referential function of mitigators, solidarity enforcers, or textual signposts. An experienced translator will attempt to recover all these pragmastylistic features in translation.

The second sub-corpus consists of two source texts where GEs are not necessarily used as stylistic markers but mainly as pragmatic markers at the Character–Character level. The first source text (STb) is John Kennedy Toole’s novel A Confederacy of Dunces, where I identified 14 disjunctive GEs (or something), and its respective Romanian translation (TTb). The second source text of this sub-corpus is Paul Beatty’s novel Tuff, where I identified as many as 26 adjunctive GEs (and shit), and its respective Romanian translation (TTc). In these two novels, GEs are neither as frequently used nor as diverse as they are in Salinger’s novel: in only the first four chapters of Salinger’s novel, I identified 23 adjunctive GEs (and all), 17 disjunctive GEs of the type or something, and another 17 disjunctive negative polarity GEs (or anything). The reason I have chosen the second sub-corpus has to do with the restricted role that GEs play in these source texts: since here GEs are not as frequent and do not function as strong stylistic markers, I expect that the Romanian translators will choose to omit them in most cases, without necessarily compensating for this loss. I intend to compare the results for the two sub-corpora and see whether translators were stylistically sensitive and/or metapragmatically aware when dealing with GEs in these source texts. I believe that the information I will gather from this analysis will be accurate enough with respect to strategic choices in literary translation, since I made sure to select target texts with experienced translators who have translated numerous works of fiction. I have summed up the information about my corpus in Table 2.

Consider the following example, which is the first sentence of Salinger’s novel and contains two adjunctive GEs:

The example under (12) contains a prototypical GE (all that) that is used consistently throughout the entire novel and has a non-referential function, as well as a non-prototypical GE (and all that David Copperfield crap) used non-referentially as an expressive device so as to “downgrade routines against which the speaker or writer is rebelling” (Overstreet 1999, 136). The author places two GEs in the same complex sentence, both of them meant to establish a rapport based on truthfulness between narrator and narratee, but all translators, with the exception of the Italian one, choose to translate only the non-prototypical, expressive GE and omit the shorter, prototypical GE. More than that, both TTa3 and TTa4 also choose to omit the parenthetical clause if you want to know the truth. TTa1 does translate the parenthetical but omits the noun truth, leaving it implicit in the target text. Since both the GEs and the parenthetical are used as strategies of authentication and as pragmatic markers meant to reinforce the truth of the propositional content in this sentence, omission actually hurts the translations.

Consider also the example under (13), which is an excerpt from a dialogue where Holden is trying to convince his girlfriend to run away with him:

The example under (13) contains two instances of the adjunctive GE and all and one instance of the disjunctive GE or something. All these GEs are used non-referentially, as a strategy of negative politeness, to mitigate the imposition created by Holden by asking his girlfriend to join him in a risky endeavor. They are also part of a strategy of persuasion, together with the truth-enforcing pragmatic marker Honest to God! There is, however, a difference between the adjunctive and the disjunctive use: the disjunctive GE is used here only when the marriage proposal is tentatively laid on the table and is meant to offer an alternative to the interlocutor, something maybe even more advantageous than a possible marriage. It is interesting to see whether the translators perceive this difference: while TTa1 and TTa2 make use of adjunctive and disjunctive GEs to render the source GEs, neither TT3 nor TT4 perceives the difference and chooses to make use of adjunctive GEs. The fourth translation omits 2 out of 3 GEs, a strategy that is consistently applied by the Spanish translator in her target text. The adjunctive GE and all has two equivalents in the first target text, i.e., și așa mai departe ‘and so on and so forth’ and și tot ‘and all’. The second Romanian equivalent phrase (și tot) does not appear in the CoRoLa corpus at all, and I had a hard time finding a context in which it could be used as a GE (possibly: Mi-a furat și bani, și bijuterii, și haine și tot. ‘He took both my money, and jewellery and clothes and all.’). The second target text omits one of the instances of and all and only translates the second instance by resorting to the same GE selected by the first translator: și așa mai departe ‘and so on and so forth’. The Italian translator, who is the most consistent in avoiding omission as a translational strategy, uses two equivalents for the instances of and all: e tutto quanto ‘and everything’ and eccetera ‘and so on and so forth’. These are distinct from the adjunctive GE she used in the translation of and all in the previous example: e compagna bella ‘and what not’. This in fact confirms my intuition that the translators will select a more diverse set of GEs in their target texts and, in this way, avoid copying pragmastylistic repetition in translation. As for the disjunctive GE employed by the two Romanian translators, who seem to be the only ones who actually preserve the difference between adjunctive and disjunctive GE use in their target texts, the first Romanian translator uses sau așa ceva (literally translated as ‘or so something’, meaning or something of the kind), which is probably a shortened variant for sau cam așa ceva (literally translated as ‘or approximately so something’, meaning or something of the kind). The second translator uses a more recently developed GE (sau ceva ‘or something’), which is either an even shorter form of sau așa ceva or a calque from English. I have tried to identify contexts with it and only found a few in my corpus of fiction texts, none in the CoRoLa corpus, and an abundance in a particular novel (Ciobanu 2015) written from the perspective of a Romanian teenager (and that also uses a lot of taboo language and Romanian-to-English code-switching, which seems to support my suspicion that this use is more of a loan translation than an instance of truncation).

Table 3 shows the summary of my findings for the first sub-corpus.

The data presented in Table 3 show that TTa1 translated 25 out of 57 GEs (43.85%), TTa1 translated 25 out of 57 GEs (43.85%), TTa3 translated 53 out of 57 GEs (92.98%), and TTa4 translated 17 out of 57 GEs (29.82%). The results are interesting from a cross-linguistic perspective, since I expected that the Romanian translators would strategically employ omission and compensation to a larger extent than the Italian and Spanish translators. I also expected the Italian and Spanish translators’ figures to match, but this did not actually happen. Another interesting fact was that both Romanian translators managed to translate a similar number of GEs, although it seemed that the GE they approached differently was disjunctive or something, translated by the second Romanian translator by sau ceva, which I have identified as an instance of loan translation. As predicted, most translators (with the notable exception of the Spanish one) selected a number of synonyms for the three English GEs that are consistently repeated in the source text. From a stylistic point of view, this is not as surprising, since there are numerous studies about literary translators’ propensity to avoid repetition (Vișan 2022, Ben-Ari 1998, Rojo and Valenzuela 2001, Mastrofini 2014). This phenomenon can be either explained linguistically (by looking at linguistic variation, as I have tried) or from the perspective of retranslation studies (in the broader, multilingual sense, as proposed by Amaral 2019). From a linguistic perspective, the data presented here indicate that the non-referential, pragmatic use was indeed a problem for Romanian translators (who resorted to omission to a fairly large extent: 43.85% for TTa1 and 31.57% for TTb1), but more so for the Spanish translator (63.15% instances of omission). On the other hand, the fact that the Italian translator only omitted one GE out of 57 would indicate that she has perfect metapragmatic awareness. It is either that or she is more style-sensitive than the other three translators (which is probably the case, as confirmed by Valentino 2017, 37). The differences traced are interesting from the perspective of retranslation theory. While the Romanian translators seem to be fairly evenly matched in the strategies they apply, the Italian and the Spanish translators have almost opposite approaches to Salinger’s text: the Italian one preserves the style of the source text and foreignizes, while the Spanish one departs from it and domesticates.

My data about the first sub-corpus indicate that non-referentially used GEs are a problem in translation for Romanian and Spanish. However, they have been so far inconclusive with respect to parametric variation. From another perspective, that of multilingual retranslation, integrated into a multilingual polysystem (Even-Zohar 1979, Vîlceanu and Păunescu 2022), the data are definitely relevant as they point to various degrees of translational sensitivity to the style of the source text.

Let us now discuss the data in the second sub-corpus. The following examples are selected from fiction texts where GEs were mainly used as pragmatic markers and as register markers rather than as stylistic markers at a Narrator–Narratee level, as was the case in Salinger’s novel. The first text, STb, contains repeated use of the disjunctive GE or something (14 tokens), while the second text, STc, contains repeated use of the adjunctive GE and shit (26 tokens):

The examples under (14) and (15) show that both target texts tend to translate the non-referentially used GEs. For a Romanian reader though, the Romanian equivalents of the English GEs sound unnatural, infinitely more so than in the case of the Salinger translations. The translators do select some synonyms as equivalents for the original GEs, but lexical variety is lower than in the case of the Salinger translations. Table 4 shows the summary of my findings for the second sub-corpus.

The data in Table 4 show that GEs have been translated to a much larger extent than they have been in the stylistically marked Salinger text. Out of 40 GEs, the translators found formal equivalents for as many as 29 (72.5%), but these equivalents are not, in fact, functional equivalents. They do not function as believable pragmatic markers in the target texts and sound artificial and contrived, which in fact means that they have been translated literally. This is quite a paradox, since, as I explained before, both translators are experienced (for instance, Lidia Ionescu has translated Gerald Durrell’s My Family and Other Animals and has managed to flawlessly recreate the book’s humor in Romanian). If compared to the results of the Salinger translations into Romanian (43.85%), these figures indicate that in the first case, the translators have either not managed to reflect Salinger’s style or had trouble creating pragmatic equivalence in translation (probably due to the fact that non-referential use in GEs is difficult to match in Romanian, as I suggested in the first section of this article). On the other hand, the translators of the second sub-corpus went out of their way to translate all the English GEs, even if they were not crucial for the style of the novels in question. This approach resulted in literal translation, which can be motivated by linguistic variation (Protopopescu 2022, 399). Of course, it would have been interesting to see what happens in Italian and Spanish for these source texts. I believe that this is a shortcoming of my research, for I would have probably obtained a clearer picture if I had been able to compare the Romanian target texts in the second sub-corpus to their respective Italian and Spanish translations.

4 Conclusions

The data presented in this article indicate that the translation of fiction GEs from English into Romanian is an issue, probably because Romanian GEs have not yet developed as many non-referential functions, and so Romanian translators have to correctly identify the non-referential functions of the source text GEs and find appropriate equivalents for them in a larger class of Romanian pragmatic markers. The translators’ discomfort when dealing with GEs was reflected in the fact that, in certain cases, they could not distinguish adjunctive from disjunctive uses.

In the first section of this article, I made three predictions with respect to GEs in translation.

The first prediction was that translators would employ the strategy of compensation by resorting to various other pragmatic markers so as to render the pragmatic charge carried by English GEs. This prediction was confirmed, but not to the extent I envisaged, as shown in Tables 3 and 4.

The second prediction was that translators would use a larger number of lexical choices in rendering English GEs into a target language by matching one English GE in the source text with two or three (or more) available GEs in the target text. This proved to be true, especially in the case of the Romanian translators (for both sub-corpora, but more visibly in the first sub-corpus) and of the Italian translator (for the first sub-corpus). The Spanish translator is the only one who used very few synonyms and also the one who omitted more GEs than all the other translators. Since Spanish has a rich set of GEs of its own and has well-developed non-referential uses for its GEs, I can only draw the conclusion that the Spanish translator showed a lower degree of metapragmatic awareness and sensitivity to style than her colleagues.

The third prediction I made regards omission in the translation of GEs. I predicted that Romanian translators would employ omission strategically when faced with the task of translating non-referential uses of English GEs. This prediction was confirmed mainly in the case of the translators of the first sub-corpus, which was an interesting thing, since I expected them to omit less than the translators of the second sub-corpus. However, it turned out that the number of omissions was greater in the first sub-corpus and smaller in the second. This happened because the translators of the second sub-corpus chose to translate most of the original GEs literally, producing dialogues that sound quite odd in Romanian. I think this situation occurred because Romanian uses GEs more infrequently and in a more restricted manner than English does, and translators had trouble interpreting English GEs correctly. Further quantitative studies on a corpus of spoken Romanian need to be conducted to find out if this hypothesis is true.