Investigating trainee translators’ contrastive pragmalinguistic competence: a corpus-based analysis of interclausal linkage in learner translations

ABSTRACT

This article aims to investigate trainee translators’ contrastive pragmalinguistic competence, starting from the assumption that – although more elusively than knowledge about culture-specific references – it represents an important subcomponent of intercultural competence, which can determine the adequacy of translated texts. The study focuses in particular on the translation of interclausal linkage, as it is a form of cohesion which displays different preferences across languages. A multi-parallel corpus of English-to-Italian learner translations of the same source text is analysed to detect regularities and variation in the language behaviour of trainee translators. The frequency of connectives in target texts is compared to both the respective source texts and comparable non-translated Italian texts, in order to determine whether translations are mainly shaped by interference or normalisation. The results of the quantitative analysis confirm previous findings that interference is predominant, with students closely reproducing source text conjunctive patterns at the risk of making translations sound unnatural; more refined qualitative observations, however, reveal that there are also attempts at normalisation. A discussion of the relevance to translator training of the insights obtained is provided, together with suggestions for inclusion in training programmes.

KEYWORDS:

• Learner translation
• translation competence
• contrastive pragmalinguistics
• interference
• normalisation
• multiple translation corpus

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. The term normalisation is used in this article to indicate ST–TT shifts which go in the direction of conformity with TL standards, and not with the negative connotation of ‘conservativism’ or ‘lack of creativity’ attributed to it in part of the literature on translation universals (see, e.g. Baker’s definition of normalisation as ‘a tendency to exaggerate features of the target language’ [1996, 183, my emphasis]).

2. Toury’s law of interference, which postulated that ‘in translation, phenomena pertaining to the make-up of the source text tend to be transferred to the target text’, provided for two kinds of interference: on the one hand, ‘negative transfer’, resulting in deviations from normal, codified practices of the target system, on the other ‘positive transfer’, to be conceived as a higher frequency of features which do exist in the target language (1995, 275). As was interestingly pointed out by Mauranen (2004), negative and positive transfer can be conceived as points on a cline: The former represents a gross deviation from the TL norm, whereas the other is virtually indistinguishable from original TL production for ordinary readers, and may probably be detected only by means of large-scale frequency comparisons.

3. These were removed because of their more ‘implicit’ nature: In this way, it will be possible to consider shifts from ST non-finite to TT finite subordinate clauses (possibly revealed by a higher frequency of subordinators introducing finite clauses in TTs) as cases of conjunctive explicitation.

4. Normally, only p-values <0.05 are considered to be significant when applying the binomial test (Baroni and Evert 2009). However, in an attempt to offset the impact of language-related differences in text length, I decided to adopt looser significance levels and raise the significance threshold to p-value <0.20. Interestingly, 24 out of the 36 translations found to be significant by the binomial test were also singled out by standard deviation, which suggests that those TTs are significantly deviant from both the relevant ST and concurrent translations.

5. It is worth pointing out that the fact that percentage frequencies turn out to be lower in TTs compared to STs does not contradict the tendency towards higher conjunctive explicitness that we observed in Section 3.4.1: Although the actual number of connectives in TTs is on average higher than in STs, percentage frequencies are lower for TTs because they depend on text length, and Italian TTs tend to be longer than English STs.