Summary and discussion

73 Sometimes translators working for multinational translation companies do not even receive the source text dialogue in written form but they have to translate the dialogue straight from the screen. In these kinds of situations, other background noises on the screen might impede hearing the original dialogue. It should be noted, though, that poor working conditions do not automatically mean the subtitling quality is poor. It nevertheless makes it more difficult and time-consuming for translators to produce high-quality subtitles. As translation fees are low, speed is essential in earning a living. In other words, producing high-quality subtitles, which take time to prepare, does not necessarily pay off.

In conclusion, this set of data implies that there have been changes in the quantity and quality of translation errors since MTV outsourced its translation services to BTI. The quantity of translation errors is considerably higher in the episodes aired after the outsourcing but the quality of the errors is not necessarily poorer. In fact, it was found that MTV’s

subtitles contain nearly as many major errors than BTI’s subtitles but when evaluating the overall quality of the subtitles, it can be said that BTI’s subtitling quality is poorer as they contain a higher number of errors. The subtitles of MTV contained an average of 6.1 errors per episode, whereas the subtitles of BTI contained an average of 11 errors per episode. The changes can to some extent be explained by the deteriorated working conditions of translators but there are other factors as well, such as the translators’ work experience and the knowledge of the series, which might have an effect on how prone the translator is to make mistakes.

Nevertheless, the results give subtle indications that the working conditions of the translators have become poorer after MTV outsourced its translation services to BTI which, in turn, might have an effect on subtitling quality.

74 more dynamic approach to quality assessment, the error categorisations for subtitling should be adjusted to fit the data. These adjustments could be done according to genre, text and content type. By adjusting the error categorisation to fit the data might also reduce the subjectivity of the assessment. When the error categories are defined according to genre, text and content type, this could leave less room for interpretations and application of subjective views as the evaluator would know how errors are defined in each commission. Also, the space and time restrictions of subtitling set some limits in identifying errors. Not knowing the company-specific conventions about, for example, the number of characters or linguistic conventions poses some challenges in establishing what constitutes as a translation error and in evaluating what should not be omitted from the subtitles. This suggests that in case this quality evaluation model was applied in quality control of professional subtitling, the television channel or translation company should give the evaluator the use of company-specific guidelines and conventions. This would help in delimiting the definition of a translation error and could possibly reduce the level of subjectivity of assessment as well.

My subjectivity and my personal attachment to Emmerdale became visible in the analysis in that I may have been stricter in counting and classifying errors and more

demanding towards the language of the subtitles than ordinary viewers or evaluators who do not follow the series might have been. Language experts, as discussed in Chapter 3.4. are often more critical towards the language of translations than ordinary readers (Suokas 2012).

In other words, awkward word choices, grammatical errors and inconsistencies in stylistic features as found in this study might be more easily noticed by language experts such as myself but might not bother ordinary viewers at all. Some of the errors might seem trivial and harmless but when concentrating on the language of the subtitles alone they are violations against the target language. Also, each error category in O’Brien’s (2012) model were loosely defined which left too much room for interpretations. There is, thus, a need for less subjective criteria.

The subtitles of twenty episodes of Emmerdale contained a total of 167 errors. Of these errors, 166 were translation errors and one was a subtitling error. The quantitative analysis showed that the most common type of translation error in this set of data was a linguistic error. Linguistic errors occurred 86 times which means they accounted for 51.5 per cent of the total number of errors. Stylistic errors were the second most common type of error in this material. With 43 occurrences they accounted for 25.7 per cent of the errors. Accuracy errors, in turn, formed the third largest part of the errors. With 37 occurrences they accounted for 22.2 per cent of the errors. On the other hand, terminological errors did not appear at all in

75 this set of data. The large number of linguistic errors means that the degree of variation in the quantities between different types of translation errors was quite moderate. Another positive finding was that most of the errors found in the subtitles were evaluated as minor errors. The quantity of minor errors is directly linked to linguistic errors as all of them were evaluated as minor errors. Major errors are not very likely in subtitles, in my opinion, because the image and tone of the speaker can help the viewer understand the subtitles. In other words, even though the subtitle would be, for example, extremely condensed version of the source text dialogue, the viewer may not have problems in understanding what is happening because they can see and hear the context. The context-bound nature of subtitling could be one reason why major errors are not frequent in this material.

The qualitative analysis showed that the most typical types of linguistic errors were violations against Finnish punctuation rules, namely, comma errors. On the basis of the material it seemed quite typical, especially in the subtitles translated by BTI, to leave the comma out when the sentence continues on the next line, as in example 10 of the analysis.

Still, the subtitles were not consistent in the use of this punctuation rule as most of the

subtitles analysed followed the Finnish punctuation rules by the book. This raises the question whether or not the translators have been instructed to leave the comma out when the sentence continues on the next line in the company’s subtitling guidelines or quality manual. Even though YLE subtitling guidelines (see Appendix 1; Vertanen 2007) advise the translators to follow punctuation rules according to the Finnish grammar, these kinds of conventions may differ between television channels and translation companies. As was discussed in Chapter 3.1., some of the subtitling companies’ guidebooks provide the translators very detailed instructions on, for example, typographical techniques or the use of punctuation, while others concentrate on more general topics (Lång 2013). For some companies the edge of the screen might serve as a punctuation mark and it might well be that the translators of the episodes analysed here have been instructed that using a comma in a situation where the sentence continues in the next line is not necessary. Knowing the company-specific conventions and guidelines would help in classifying errors.

The most typical types of accuracy errors, in turn, were unnecessary omissions which most often arose from omitting the object or the adverbial from the subtitle. Even though omission and condensation are the main strategies used in subtitling, it was noticed that there was a slight tendency to condense the content of the source text even if the space and time limitations would not require it. This might imply that translators make a habit of

condensation and omission and they tend to do it when it is not necessary. This finding

76 appears to contradict Tuominen’s (2013: 32) claim in Chapter 2.2. that the Finnish translators tend to use complete sentences even when condensation could be necessary. Somewhat surprisingly, the material included none of the “amusing” mistranslations that subtitles are quite famous for (see for example the aforementioned Paakkinen 2003; 2005). These kinds of errors were subtly expected to form an even larger part of the errors. These expectations can partly be explained by the critical and sometimes negative discussions on translators’

shortcomings and errors in subtitles (see for example Tuominen 2013: 302–309, Paakkinen:

http://www.jounipaakkinen.fi/kaannos.html), which can easily mislead to think that

mistranslations occur frequently in subtitles. However, their share of the errors was modest and smaller than that of stylistic errors. The frequency of unnecessary omissions could be linked back to the working conditions of translators. It may be that the omitted elements have slipped out because the subtitles have been produced in a hurry. Also, there might not have been enough time to review or proofread the text. The method of quality control called

“preview” (James 2001) discussed in Chapter 3.2. could provide a tool to avoid the issues arising from unnecessary omissions.

Another surprising finding about accuracy errors was that meaning errors did not always arise from pure mistranslation but they could be caused by ambiguous translations or by the translator’s personal choices on what to include in the subtitle and what to leave out.

Due to this, categorising and evaluating the severity levels of the translation errors can be somewhat problematic as there might be dissent between the evaluator and the translator. The subtitle that appears on screen always reflects the translator’s personal vision of what they think is the most important and relevant information for the viewer (Tuominen 2013: 27), while the evaluation of an error as a translation error depends on the evaluator’s subjective views on what they consider to be an acceptable translation. In some cases categorising and evaluating the errors can lead to questioning the translator’s personal vision but it is difficult to avoid these kinds of situations in quality assessment which is largely based on subjective views and interpretations.

The qualitative analysis showed that the most typical types of stylistic errors in this set of data were register errors which mainly arose from inconsistent use of register. There were clear inconsistencies in the use of language varieties in the subtitles similar to

Hietamaa’s (2012) findings which might stem from the lack of communication and

collaboration between the translators. Unlike the translators working for YLE, the translators subcontracted by big companies do not usually have the same kind of possibilities to interact with one another and offer personal support. The translators of YLE interact with one another

77 and the more experienced translators familiarise the new ones with the working methods and subtitling conventions of the company (Lång 2013: 58). This kind of mentoring is only possible when the translators are working under the same roof (Lång 2013: 58) which is not often the case with subcontracted freelance translators whose work is not bound to a specific place. As Pedersen (2011: 215) notes, daily television series are often subtitled by a team of translators and communicating with all the translators about different translation solutions might be time-consuming. An increase in the number of stylistic errors can be noticed after MTV outsourced its translation services to BTI which could mean that the reasons for the inconsistencies might lie in the translators’ working conditions. When the amount of work load per translator increases and the deadlines become shorter, the translators might not have time to discuss previous translation solutions.

Furthermore, this set of data implies that there have been changes in the quantity and quality of translation errors since MTV outsourced its translation services to BTI. The

subtitles translated by MTV contained 61 errors, whereas the subtitles translated by BTI contained 106 errors. Based on the quantity of errors alone, the subtitling quality can be evaluated to have become poorer after the outsourcing. However, there were no significant differences in the severity of the errors. The subtitles of MTV contained six major errors, while BTI’s subtitles contained seven. Linguistic errors were the most common type of translation errors in both groups but the errors in MTV’s subtitles were more evenly

distributed than in BTI’s subtitles. Linguistic errors accounted for 57.5 per cent of the errors in the subtitles translated by BTI, while in the subtitles translated by MTV they accounted for 41 per cent of the errors. What was interesting, and somewhat surprising in my opinion, was the difference in accuracy errors. The subtitles of MTV contained more accuracy errors than BTI’s subtitles. Accuracy errors appeared 21 times in MTV’s subtitles which was slightly over 34 per cent of the total number of errors, whereas the 16 accuracy errors in BTI’s subtitles formed 15.1 per cent of the errors. In other words, MTV’s subtitles contained more accuracy errors with relation to the total number of errors. Stylistic errors, on the other hand, were more frequent in BTI’s subtitles than in MTV’s. Stylistic errors formed 23 per cent of the errors in MTV’s subtitles but in BTI’s subtitles the equivalent number was 27.4 per cent.

The changes can to some extent be explained by the deteriorated working conditions of translators but there are other factors as well, such as the translators’ work experience and the knowledge of the series, which might have an effect on how prone the translator is to make mistakes. Nevertheless, the results give subtle indications that the working conditions of the

78 translators may have deteriorated after MTV outsourced its translation services to BTI which, in turn, might have an effect on subtitling quality.

79