a peewit, very remote. ” – Notes on quotatives in literary translation

: The present article focuses on strategies of translating ﬁ ction quotatives from English into Romanian. Starting from the de ﬁ nition of quotatives as structures that in their simplest form consist of a subject and a quoting verb and accompany a quotation, I have selected two samples of literary text and their respective multiple versions so as to investigate patterns in which these structures are translated. Because, as pointed out in the literature, ﬁ ction quotatives describe narrative - advancing events and con tribute to the development of characters, the investigation of how ﬁ ction quotatives are translated ( in particular how say , the most frequently used verb in quotatives, is treated in translation ) might prove to o ﬀ er valuable insight for literary translation studies, correlating tendencies that seem to be cross - linguistic. For instance, it has been demonstrated that in Spanish there is a tendency of replacing the generic quoting verb say with other manner of speaking verbs. This may be seen as a form of “ enrichment ” as a translation strategy. The article advances the hypothesis that a similar phenomenon can be attributed to Romanian and links this phenomenon to parametric variation in English and Romance.


Introduction
The aim of the present article is to investigate strategies of translating quotatives in literary texts (narratives) from English into Romanian. I am especially interested in the translation of say quotatives, which are the mark of fiction in English, as proved in a series of studies (Rojo andValenzuela 2001, Sams 2009 inter alia). It has already been shown that in the case of Spanish (Rojo Valenzuela 2001) say quotatives are frequently replaced by other structures, such as manner of speaking verbs (henceforth MoS verbs). I intend to look at a corpus consisting of samples taken from two fiction texts in English and their respective Romanian versions in order to see whether a similar phenomenon can be observed in the case of Romanian. Given the fact that, to my knowledge, there are no studies investigating what happens to quotatives in Romanian translations, I believe that this study might shed some light into this matter and offer a basis for future research that can be extended cross-linguistically.
In discussing quotatives and their behaviour in English and Romanian, I start from Sams' (2009, 148) tentative definition for English quotatives. Thus, a quotative is a structure consisting of a quoting verb and a subject which is associated with a quotation such as in the following examples: (1) (a) "I thought I gave you my word," he said.
(H. G. Wells The examples under (1) contain a "simple quotative," made up of a subject and a predicate, while the examples under (2) contain a complex quotative, formed from a subject, a predicate and an adjunct. As pointed out in the literature (Biber et al. 1999, Sams 2009), quotatives are especially frequent in written language (newspapers, fiction books) and exhibit a great degree of variability with respect to verb selection (from general verbs of saying, such as say, to all sorts of verba dicendi), as indicated in Table 1.
(R. Kipling, Stalky & co., translated by N. Steinhardt, 1977, 31) If one looks at the examples under (1) more closely, it is clear that the subject he may or may not be lexically realized in the Romanian translation, depending on whether the larger context offers enough clues related to the reference of the missing subject. I have marked this situation by placing the subject el between brackets. It thus appears that an important difference between English and Romanian quotatives has to do with subject realization and subject position, as summed up in Table 2. While Romanian is a prodrop language and can do without a lexically realized subject, in English, the subject has to be lexically realized and normally precedes its predicate. However, quotatives allow for both the canonical word order (SV), when the subject precedes the predicate, and the non-canonical word order (VS), when the subject comes after the predicate. In Present Day English, the VS pattern is the less frequent pattern, therefore the marked variant, as opposed to older stages of English (Chichosz 2019), and is known under the name of quotative inversion (Collins and Branigan 1997).
It turns out that inversion between subject and verb is an important feature of quotatives. This is true of both English and Romanian. However, if you look at the examples under (4) and (5), it is apparent that in Romanian both positions of the subject (VS and SV) are available in non-quotatives, but only one position is available in quotatives (the VS order). Conversely, in English, only one position is available to the subject in non-quotatives, but two positions are available for the subject in quotatives: Compare: (4) (a) First the woman said something, then *said the man something.  The important point to make here is that for Romanian, inversion between subject and predicate is the only possible option in quotatives. In that, Romanian behaves like other Romance languages, such as French, Spanish, Italian, and Portuguese, none of which allow for the SV pattern in quotatives.
Another important feature of quotatives is that they can be lexically realized, as illustrated under (1) and (2), or they may not be lexically realized when reference of the interlocutors is already established, and in this case, they are null quotatives, as in (6) Wells, Omul invizibil, 2018, 73) This fact can be correlated with an observation made by Ramalle and Maria (2009) regarding the function of quotatives in discourse. She points out that quotatives are not obligatory in text or discourse in the sense that they are not inserted in dialogues just to mark conversation turns. They must have a specific function, other than just marking direct speech for which there is already typographic marking (quoted in Caballero 2018, 49). According to Sams (2009), quotatives are genre-specific in that they advance the narrative and help with the development of the characters. Moreover, they provide a plus of expressivity to literary texts, because authors can be very creative in the type of quotatives employed (lots of MoS verbs and other special verbs of saying, combination with adjuncts, etc.). I would also add that they contribute to the cohesion of the text due to the information (person and tense) that they provide.
In summary, the following observations can be made with respect to fiction quotatives in English and Romanian: • For both English and Romanian, fiction quotatives fulfill a function in the texts where they appear: they add to the specificity of the genre and strengthen cohesion in the text. However, there are certain differences that pertain to their syntax and to lexical choices with respect to the quoting verbs employed by authors.
English written narratives exhibit a high degree of frequency with respect to say quotatives. They also employ a fair number of MoS verbs and other special verbs of saying (which contribute to the expressivity of the narrative). Two syntactic patterns -SV or VSare available, where VS is the slightly less preferred variant (Biber et al. 1999, 923); therefore, the default pattern seems to be the regular syntactic pattern SV and the marked pattern seems to be VS. Biber et al. (1999, 922) link the authors' preference for the SV pattern to information packaging: "To some extent, the conditions of inversion in reporting clauses are similar to those applying more generally to subject-verb inversion, reflecting the weight and communicative importance of the subject v. the verb: whichever is placed last becomes relatively more prominent." Romanian translated narratives seem to have placed a ban on lexical repetition, which might be at least one of the reasons why translators make moderate use of the general verbs of saying a spune/a zice (as briefly noticed by Vișan 2015, 67). This may be translators' attempt to observe the rules of Romanian fiction with respect to repetition and the variety of quoting verbs used, but I have no data to support this hypothesis yet. On a syntactic level, VS appears to be the only pattern available to quotatives (therefore, the distinction marked/unmarked does not apply, as it does in the case of English quotatives). Just like English authors, Romanian translators make use of a fair number of MoS verbs and other special verbs of saying which contribute to the expressivity of the text. There is also a tendency of contemporary translations to use the old preterite (Perfect Simplu) for quotatives while employing the new preterite (Perfect Compus) as the main narrative tense in the text, which is a pretty novel development for Romanian. This indicates that Romanian quotatives tend to become formulaic, and may be in incipient stages of grammaticalization, as mentioned by Vișan (2012Vișan ( , 2020. Based on what I know so far, I expect Romanian to pattern with its Romance sibling, Spanish. Therefore, I expect that the generic verb say will appear much less frequently in translation. I predict that the strategy used in translation is that of the replacement of say with other general verbs of saying available (such as a rosti, for example) or with MoS verbs, or with special verba dicendi, resulting in enrichment (which may be seen as a counterpart to Berman's (1984Berman's ( /2000 quantitative impoverishment).

The data and their interpretation
The predictions stated above regarding the frequency with which formulaic structures such as subject said/ said subject are rendered in Romanian are checked below with the help of a corpus consisting of two source texts, i.e. The Invisible Man by H. G. Wells, published in 1897, and Stalky & co. by R. Kipling, published in 1899. I have selected these texts because they abound in dialogue, which facilitates access to quotatives. Moreover, both source texts benefit from at least two target texts, since they have been retranslated in Romanian at least once. In Table 3, I have marked the dates of publication and the notations used in this article for each of these texts: The analysis of the data will be mainly conducted with respect to TT1 and TT2 for both source texts. As you can see, both the source texts and their first two target texts were produced in the same decade, which is relevant for an accurate analysis, I believe. I have added information from TTa3 for a very good reason: because the third version of The Invisible Man is my own translation, I have the notes made by my editor and his suggestions (which are marked between brackets) of replacing the quoting verb say with various other synonyms (all sorts of quoting verbs). This translation was published quite recently after being modified in accordance with the suggestions of the editor. I believe that it makes a good addition to the point I am trying to make with respect to the translation of say quotatives from English into Romanian.
For obvious reasons, I have decided not to offer back-translation for the bulkier examples, focusing on quotatives only, which I have underlined and further explained under each series of examples.
The first sample of text under analysis, under (7), is also quoted in the title of this article because it contains two say quotatives, one with a [+human] subject, another with a [−human] subject. I believe the latter is the one say quotative that a Romanian translator might have had reason to replace, due to the [−human] semantic feature that characterizes the subject (a peewit). The example is a good one to start a  Vișan (2018) debate with respect to stylistic choices in the source text and their accurate rendering in translation. As can be noticed, both say quotatives have been replaced by other quoting verbs in all their Romanian versions. Of special interest is the second version, TTa2, signed by C. Clarus, but also by Antoaneta Ralian, one of the greatest literary translators in Romanian culture. The sample is interesting because, in spite of the different semantic features of the subjects and in spite of the translator's alleged reluctance to avoid repetition, it makes use of the same verb (făcu, which is a preterite form of the verb a face "to make" and can be used as a quotative only in its preterite form). The fact that the translator uses it twice in the same paragraph may be seen as a good counterargument to the hypothesis that repetition is the only reason why such replacement is recurrent in the Romanian translation of say quotatives. I believe that the reluctance of translators to resort to repetition is only one of the causes of this strategy of replacement.
-Auzi prostie, cirip! a spus (rostit) domnul Thomas Marvel. N-am eu timp de prostii! (TTa3, p. 79) The first say quotative is replaced in both TTa1 and TTa3 by a verb of physical perception (a auzi "to hear") which is used as a quoting verb here and constitutes a safe option to translate the original quoting verb without committing the translation to employing personification. The second instance of say quotatives (said Mr. Thomas Marvel) is replaced in TTa3 by another verb of saying (a rosti, "1.to articulate, pronounce 2. to say") generally used as a synonym for the two general verbs of saying Romanian possesses (a spune, a zice). TTa1 opts for a speech act verb, a exclama ("to exclaim"), and provides pragmatic interpretation for the quotation, which in the source text is not made explicit. This is a phenomenon that is recurrent in translation and has been noticed for Spanish as well (Rojo and Valenzuela 2001).
The example under (8) contains an excerpt where there are four instances of say quotatives and two instances of other quoting verbs in the source text (marked differently here). The only version that uses a general verb of saying is TTa1: out of the four instances of say quotatives, TTa1 translates the second one (said the voice) with spuse Vocea, resorting to equivalence. The other three instances are translated by an archaic (and poetic) verb of saying (a grăi), a verb of movement (a tresări "to start, to flinch"), and the preterite of a face "to make." The use of the verb of movement a tresări is another instance of semantic enrichment and an attempt to explain the attitude of the character by pragmatic explicitation. A similar strategy is employed by TTa2, where the verb say is translated by a MoS verb such as a îngâna ("to mumble"), a speech act verb such as a contrazice ("to contradict"), and another MoS verb, i.e. a se tângui ("to lament"), all of which are instances of enrichment and pragmatic explicitation. As for the two MoS verbs employed in STa, TTa1 provides equivalents for both of them: a repeta ("to repeat") and a șopti ("to whisper"). TTa2 on the other hand chooses to replace to whisper with a verb that has extra semantic features, i.e. a bolborosi ("to gurgle, to babble"), which results in semantic enrichment. The solutions suggested by my editor for quotatives in TTa3 can be found between brackets and are all tributary to TTa2 (a contrazice, făcu, repeta, bolborosi), which makes me believe that my editor consulted TTa2 while proofreading my translation, in the good old tradition of many an editor/translator involved in the business of retranslation. This might provide good arguments for another, future, discussion on retranslation and the retranslation hypothesis.
(8) (a) "So help me," said Mr. Thomas Marvel, shuffling his coat on to his shoulders again. "It's the drink! I might ha' known." "It's not the drink," said the voice. "You keep your nerves steady." "Ow!" said Mr. Marvel, and his face grew white amidst its patches. "It's the drink," his lips repeated noiselessly. He remained staring about him, rotating slowly backwards. "I could have swore I heard a voice," he whispered. "Of course you did." "It's there again," said Mr. Marvel.
-Iar începe, a spus (a bolborosit) domnul Marvel (TTa3, p. 88) The reluctance of the editor to observe the stylistic features of the original (repetition) can be noticed in the example under (9), where the noun interlocutor, which the author chooses to use twice in the same paragraph, is replaced by the editor with a pro-form (celuilalt "to the other guy"). The stylistic effect of the repetition is undone and it actually hurts the translation. Only TTa1 manages to apply equivalence, rescue the repetition, and preserve the stylistic effect. A similar strategybut less damaging in this caseis used in TTa2 in the previous example: Ralian chooses to replace Mr. Marvel with vagabondul "the tramp" in an attempt to avoid repetition (the proper name has already been used twice in the same excerpt). Full repetition: "the boots of the man who was speaking to him" (c) Întoarse capul peste umărul drept, să vadă ce fel de ghete poartă celălaltvoia omul să facă o comparațiedar, vai! nu-i vedeai nici măcar picioarele, darmite ghetele! (TTa2, p. 81) No repetition: "to see what sort of boots the other guy was wearing" (d) Și-a întors capul și a aruncat o privire peste umărul drept la ghetele interlocutorului său, ca să facă o comparație, și, ce să vezi? Unde ar fi trebuit să fie ghete în picioarele interlocutorului, (celuilalt,) nici urmă de ghete și nici de picioare. (TTa3, p. 89) Partial repetition: "interlocutor" The next example, under (10), is from the second source text and contains three instances of say quotatives. It is interesting to notice that TTb1 opts for equivalence for all three instances, employing the verb a zice as the main general verb of saying. TTb2 however chooses to replace all three instances of said with three different quoting verbs (speech act verbs a declara "to declare", a se oferi "to offer (to explain)," and the verb of thinking a chibzui "to ponder (in a loud voice))." These options have as a consequence semantic enrichment and pragmatic explicitation. A similar situation can be noticed in the example under (11), where each of the target texts replaces the verb say with a pragmatically charged quoting verb (a declara "to declare"). In TTb1, the second instance of said is omitted, which is a strategy that is recurrent in all target texts discussed here.
-Asta-i cea mai mârșavă nedreptate din câte s-au pomenit vreodată, se plânse Stalky,dacă stai să te gândești numai cât m-am ostenit eu să-l amanetez. (TTb2, p. 58) The example under (12) contains three MoS verbs and one say quotative. The MoS verbs seem to be troublesome for TTb1 (for instance, gasp is mistranslated as a exclama "to exclaim"), but the translators try to provide synonyms (a gângăvi "to babble," a urla "to howl"). The complex say quotative is translated in TTb1 by a simple quotative with the speech act verb a exclama ("to exclaim"), which is an instance of undertranslation. TTb2 on the other hand chooses to overtranslate and overdramatize the complex say quotative: said Stalky with a yell is translated by țipă atunci Stalky din răsadâncul bojocilor which literally means "yelled then Stalky from the deep bottom of his lungs". The MoS verb gasp is translated by a paraphrase, resulting into a complex quotative (rosti cu greu Stalky "uttered with difficulty Stalky"). The other two MoS verbs receive equivalents: shriekedhohoti "guffawed," howledurlă "howled." (12) (a) "Moi! Je! Ich! Ego!" gasped Stalky. "I waited till I couldn't hear myself think, while you played the drum! Hid in the coal-lockerand tweaked Rabbits-Eggsand Rabbits-Eggs rocked King. Wasn't it beautiful? Did you hear the glass?" "Why, he-he-he," shrieked McTurk, one trembling finger pointed at Beetle. "Why, I-I-I was through it all," Beetle howled; "in his study, being jawed." "Oh, my soul!" said Stalky with a yell, disappearing under water. (STb, p. 52) (b) '"Moi, je, ich, ego!" exclamă Stalky, "am aşteptat până mi-a ieşit sufletul pe când băteaţi toba.
-Sfinte Sisoe! țipă atunci Stalky din răsadâncul bojocilor și făcând o tumbă în apă. (TTb2, p. 63) I have decided to discuss just four examples from my corpus, choosing the most representative ones from the samples of text I selected. The main strategies I have identified are replacement, equivalence, and omission. Omission was not very frequent, but it was consistently employed in all target texts under discussion. There were two types of omission: total omission (when the whole quotative was omitted in the target text) and partial omission, such as the one discussed in the last example I offered here, where the adjunct expressed by a prepositional phrase was deleted and the complex quotative was reduced to a simple quotative. Some instances of complex say quotatives were rendered by either partial omission or by paraphrase of the adjunct (especially when the adjunct was a manner adverb, which is in accordance with the fact that, unlike English which is an adverbial language, Romanian is only a partly adverbial language (Protopopescu 2011) and chooses to use either equivalence or omission or paraphrase in translation; this situation is also checked in (Vișan and Protopopescu 2022, to appear), where it is shown that reinforcement of a special quoting verb with a redundant manner adverb is one of the strategies employed in the translation of complex say quotatives).
In Table 4, I offer a summary of the data analysed for this article.  Table 4 shows that earlier translations favoured equivalence as the main strategy, opting for preserving the repetition of the say quotative. The situation dramatically changes in the later translations, which favour the replacement of say quotatives with MoS et al. quotatives in almost all cases: only 18 say quotatives survive out of the 132 instances counted in the source texts, i.e. 13.63% of the original number. These data seem to confirm that what happens in Spanish seems to be happening in Romanian as well and the result is a target text characterized by enrichment and explicitation. The data also point to the fact that there is a clear discrepancy between the strategies employed when translating quotatives with verba dicendi and those applied in the translation of goal of motion verbs, which, as Slobin (2005) observed, follow the lexicalization patterns proposed by Talmy (1985), leading to a typology which separates Germanic languages from Romance ones into Satellite-framed languages vs. Verb-framed languages (and confirmed by Drăgan (2021 andin preparation 2022) for Romanian). This typology does not seem to hold true for either say or MoS et al. quotatives. However, the results I discussed above mirror those obtained for the translation of verbs of saying in Spanish (Rojo and Valenzuela 2001) and for the translation of MoS verbs in Italian (Mastrofini 2014): while English narratives are rife with say quotatives but less so with other quoting verbs, it seems that their Romanian translations abound in various quoting verbs with which they have replaced the general verb of saying employed in the source text.

Parametric variation?
As shown above, both Romanian and Spanish data converge towards the same choice in the translation of say fiction quotatives from English. So far, two proposals (not mutually exclusive) have been made with respect to differences between English and Spanish, regarding differences in style and in the way in which the introduction of speech events to the reader are conceptualized. To these two proposals, I would like to add a third one that has to do with the parametric variation existing between English and Romanian at a syntactic level.
(1) BAN ON REPETITION -DIFFERENT STYLISTIC PREFERENCES The data from Spanish translations of English fiction texts indicate that while repetition of the general verb of saying is the norm in English fiction, it definitely is frowned upon and infrequent in Spanish. Translators tended to supply additional [pragmatic] information which was taken from their evaluation of the global interactional frame, regardless of the verb used in the original version. … the differences between the two languages seem to indicate differences in the narrative style of English and Spanish. English also has a high number of specific verbs of saying; however, it seems to be a conventional feature of English dialogues to use profusely the general verb to say, leaving the reader the inference of the specific role that the utterance plays in the interaction frame. On the contrary, in Spanish dialogues, the continued repetition of the general verb to say results in a lack of stylistic naturalness, and translators prefer to look for variants, by incorporating to the verb information surrounding the communicative event. (Rojo and Valenzuela (2001, 477) (2) CONCEPTUAL AND CULTURAL DIFFERENCES Another study on the use of speech framing expressions (SFEs) conducted by Caballero (2018) analyses the use of SFEs in English and in Spanish. Caballero defines SFEs as "narrators' cues to how their readers should understand and assess what the characters say" (Caballero 2018, 45). While Caballero's study does not consider translated samples, the results point to conceptual differences in which English authors present speech events: In contrast to the mental focus in Spanish, there is a tendency in English to focus on the physical and dynamic aspects of speech events. The English texts make more frequent use of manner verbs concerned with the physical-auditory component of speechmanner-human, manner-animal and manner-inanimate verb meanings (e.g. stutter, cackle, and thunder, respectively). … Spanish SFEs are INTROVERT in the sense that they provide explicit information about sensory responses and ways of thinking and understanding in relation to what the speakers communicate, while English description of speech events are EXTROVERT in the sense that thoughts and feelings have to be extracted and inferred from the behaviour of the speakers. Such findings point to the existence of cultural patterns. (Caballero 2018, 69-70) The results of this contrastive study might explain why Spanish translators make different options such as attempting to replace say quotatives with verbs that are more explicit about the speakers' feelings and manner of thinking.
(3) COMPENSATION FOR WIDER SYNTACTIC VARIATION (ENGLISH) WITH WIDER LEXICAL VARIATION (ROMANIAN) As stated at the beginning of this article, English has two syntactic patterns for its quotatives (with or without inversion: SV/VS), while Romanian evinces only one possible syntactic pattern (VS). The pattern with inversion appears to be the marked variant in English narratives and is genre specific, as demonstrated by the data provided in Sams 2009: while quotative inversion is better represented in newspapers (23.94%), it turns out that only 7.38% of the instances analysed in fiction books were instances of quotative inversion. These data were actually not supported by my analysis, as indicated by Table 5, but one has to bear in mind that my study is qualitative rather than quantitative.
These findings seem to be in accordance to Sams' observation that "Within fiction writing, it is evident from the data that the use of inversion is also dependent on the preference of the individual author; of all the cases of inversion within the fiction data, nearly 20% of these instances are attributed to a single author" (Sams 2009, 162).
Consider also Table 6, which sums up what happens with say quotatives in both translations with respect to the syntactic patterns available to English for quotatives. The target texts from the 1930s retain more than half of both structures, while the target texts from the 1970s retain roughly a third of the structures with canonical order (SV) and only a mere 13% of the VS instances.
Both my samples of text had a very large percentage of inversion with say, which might be the mark of the writer's choice or a marked preference of the period when both books were written (late 19c), or both. The wide use of this particular syntactic pattern (which is the marked variant in fact) can be viewed as a stylistic mark of the texts, which the Romanian translators might have been inclined to compensate for because Romanian does not have the same array of syntactic choices. More than that, if one relates that to the observation made by Biber et al. (1999, 922) about information packaging, the fact that Romanian is limited to only one syntactic pattern means that the subject of the quotative is always under focus, while the verb of the quotative is never so. In translation, it might be reason enough to compensate for lack of communicative weight for the verb by replacing the general verb of saying preferred in English with quoting verbs that are semantically richer. The data from Spanish and from Italian seem to support this proposal because both these languages benefit from only one syntactic pattern in quotatives themselves and do not have the wider array of choices that English has. Mastrofini (2014) notices that the 83 MoS verbs identified in English fiction texts were matched with 148 MoS verbs in Italian translations, to the detriment of the English general verbs of saying. Although both English and the three Romance languages mentioned here have a large number of quoting verbs that can be employed in narratives, the numbers differ dramatically in translation. In fact, it has been demonstrated (Stoica 2021) that, while both English and Romanian have an impressive inventory of MoS verbs, the syntactic behaviour of MoS verbs is different with respect to extraction, double object constructions, and complementizer omission. So even in the case of this particular class of verba dicendi, there is parametric variation. To what extent it can be linked with distinctions in translation is yet remains to be seen.  So far, there are data from Spanish, Italian, and Romanian which indicate that the translation of say quotatives tends towards replacement with other quoting verbs rather than with general verbs of saying, hence towards a strategy of semantic enrichment and pragmatic explicitation. The research should be extended to analysing a larger number of English-fiction texts from different periods for the translation of say quotatives into Romanian so as to verify whether the fact that English has two syntactic patterns available triggers compensation in translation for a language that does not have the same array of patterns. My analysis has been more of a qualitative than a quantitative kind, using samples that favoured the VS syntactic pattern and therefore a larger corpus is probably needed.
Other Germanic and Romance languages need to be investigated in order to see if there is a pattern that might indicate that options of translation in these languages are dictated not only by different stylistic preferences and cultural distinctions but also by parametric variation at a linguistic level.
A closer look at the evolution of quotatives in Germanic and Romance languages might offer insight into this matter. For instance, Chichosz (2019) shows that quotative inversion is tied to the limited choices of quoting verbs in OE and in ME and that the preference lately shown for the SV pattern is due to the fact that a lot of other quoting verbs started to be available for quotatives. It would be interesting to see what happened in Romanian and to also look more closely at tense choice for verbs of saying in quotatives.