Interpreting Universals: A study of explicitness in the intermodal EPTIC corpus

By Niccolò Morselli (University of Bologna, Italy)

Abstract

This paper presents a study on explicitness in the European Parliament Translation and Interpreting Corpus (EPTIC). EPTIC (Bernardini et al. 2013) is a bilingual, bidirectional and intermodal corpus of EP plenary session speeches in English and Italian. It contains transcripts of both source speeches and their interpreted versions, as well as their written counterparts in the form of minutes and their translations. The study set out to test the findings of the quantitative analysis on explicitness in English interpretations carried out by Kajzer-Wietrzny (2012). The indicators of explicitness for the investigation of English (linking adverbials, apposition markers and optional that) were matched by comparable indicators for the investigation of Italian and applied to the relevant sub-corpora of EPTIC. First, a quantitative analysis was carried out, both from a monolingual comparable perspective (comparing speeches and interpretations in the same language), and from an intermodal perspective (comparing interpretations and translations). Second, a parallel qualitative analysis was performed. Some interesting differences according to language direction emerged, such as the Italian interpreters’ preference to add apposition markers, or the tendency of English interpreters to leave out linking adverbials.

Keywords: explicitation, intermodal corpus, apposition markers, linking adverbials, simultaneous interpreting, interpreting universals, italian

©inTRAlinea & Niccolò Morselli (2018).
"Interpreting Universals: A study of explicitness in the intermodal EPTIC corpus"
inTRAlinea Special Issue: New Findings in Corpus-based Interpreting Studies
Edited by: Claudio Bendazzoli, Mariachiara Russo & Bart Defrancq
This article can be freely reproduced under Creative Commons License.
Stable URL: https://www.intralinea.org/specials/article/2320

1. Introduction

This study focuses on the thorny issue of translation and interpreting universals in general, and on the highly debated universal of explicitation in particular. Many translation scholars have so far questioned the existence of these concepts (Mauranen and Kujamäki 2004).

In translation studies, the so-called Explicitation Hypothesis was first put forward by Blum-Kulka (1986), who analysed explicitation from the perspective of discourse analysis, focusing on shifts in cohesion and coherence, and which led to subsequent studies. Much work has been done so far with the help of corpora to put this hypothesis to the test from many other perspectives.

After also Baker (1993 and 1996) listed it as a potential universal, explicitation has in fact been studied both as an S-Universal[1] in parallel corpora, for example, by Øverås (1998), and as a T-Universal, that is to say from a comparable monolingual perspective,for example by Olohan and Baker (2000) and Puurtinen (2004). These two perspectives have also been combined, for instance, by Pápai (2004) and Konšalová (2007). As for the different forms of explicitation, research has moved from Blum-Kulka’s text cohesive elements to a wide variety of linguistic phenomena, as in Pápai and Konšalová’s studies, or to the analysis of a single form of explicitation such as the optional that in Olohan and Baker’s research.

Of the few studies on explicitation in interpreting studies, most are still based on very small collections of texts, mainly produced by advanced students of conference interpreting. This is the case of the study carried out by Schjoldager (1995/2002), the first study also focusing on explicitation in interpreted texts in the simultaneous mode, using parallel and intermodal corpora. From this first study on explicitation in simultaneous interpreting, no cases of the interpreter making explicit something implicit in the source speech emerged, while subsequent studies have on the whole confirmed Blum-Kulka’s initial Explicitation Hypothesis. Shlesinger (1995) noticed that when the source text had an elliptical structure omitting an element mentioned before, the advanced interpreting students in her sample tended to repeat the missing element or to find a synonym, ‘thus making the connection more explicit’ (Shlesinger 1995: 201). Gumul (2006) maintains that an increased level of explicitness in the target text is often due to interpreting-specific factors, such as the interpreters’ need to rephrase the utterance to add a new piece of information or to correct themselves. Gumul also noticed that sometimes her student interpreters tended to add words without adding information, filling a pause while waiting for the next piece of information to come. For this reason, she finds it difficult to maintain that these are examples of explicitation implied by the interpreting process itself. Ishikawa’s study (1999) represents a rare example of investigations on explicitation in simultaneous interpreting carried out on texts interpreted by professional conference interpreters – in this specific case Japanese professionals working into English, that is not into their mother tongue, adding yet another variable to an already complex task. In her study, she singled out some cases of ‘pure explicitations’ thus confirming Blum-Kulka’s hypothesis, but she also argued that in many cases the interpreters preferred implicitation (1999: 252).

As for universals in general, Chesterman pointed out that ‘some have been corroborated more than others, and some tests have produced contrary evidence, so in most cases the jury is still out’ (2004:39) and explicitation makes no exception. Other authors have also put Blum-Kulka’s Explicitation Hypothesis to the test (Pym 2005; Becher 2010a, 2010b), focusing on explicitation in translation, and they reached the conclusion that explicitation as a universal should be regarded as a myth to be debunked because of the vagueness of the very concept of explicitation. Notably Pym (2005) explained this phenomenon within his model of risk aversion, and Becher (2010a and 2010b) maintained that the studies confirming the explicitation hypothesis featured serious weaknesses as for the methodology adopted, because they did not stick to the definition of explicitation as a phenomenon inherent to the translation process, but included also different types of explicitation (as in Øverås 1998; Pápai 2004). Further criticism was raised by Becher (2010a) because the corpora used were unbalanced or because one could not have access to source texts (such as in Olohan and Baker 2000). He therefore suggested (2010b) that the assumption of explicitation being a universal should be discarded for good. Also Baumgarten, Meyer and Özçetin (2008) critically investigated explicitation both in translated and interpreted texts, and concluded that the increased level of explicitation in interpreted renditions from Portuguese into German of the term Amazônia were to be ascribed to other factors, such as interpreter’s style, interpreting mode and other social and cultural variables (2008: 198).

To complete this short overview of this field of research, Kajzer-Wietrzny’s corpus-based study (2012) on interpreting universals is important to mention for two main reasons. The first one is that in this study professional conference interpreters’ performances are analysed. Secondly, this study was conducted using a larger corpus, namely the Translation and Interpreting Corpus (TIC) (2012: 57), which is one of the few intermodal corpora used in interpreting research so far. More specifically, it is an English monolingual comparable corpus including ten subcorpora in total. The oral part consists of one subcorpus of speeches in English and four subcorpora of interpreted versions into English of speeches given at the European Parliament from four different languages, namely Spanish, French, German and Dutch. The written part contains transcripts of the English speeches and of the interpreted versions of the speeches pronounced in the other languages, as well as further four subcorpora of texts translated into English from the same four other languages. In particular, in her study on interpreting universals, Kajzer-Wietrzny investigated explicitness, analysing linking adverbials, apposition markers and optional that as explicitness indicators.

Besides the weaknesses of the studies on explicitation in interpreting already referred to above – analysing principally small-scale collections of texts often produced by student interpreters – existing studies on explicitation in interpreting have applied extremely varied methodologies, thus making the findings considerably difficult to compare with other investigations of the same linguistic phenomenon.

Additionally, scholars have investigated explicitation not always clarifying the concept itself. As it emerges from this review, many have looked at explicitation (sometimes implicitly) defining it as the process of rendering covert information in the source text in an explicit way in the translated text. For instance, Baker (1996: 180) rather generally defined explicitation as ‘the tendency to spell things out’. On the other hand, Kajzer-Wietrzny analysed explicitness, unequivocally defining it as a feature of target texts, rather than a strategy or a technique. In our study the term ‘explicitness’ is also adopted, since the subject of this investigation is not the process of making overt in the target text implicit information in the source text, but the tendency of target information of being more explicitly encoded, namely a textual feature not necessarily deriving from an explicitation process.

All these controversial aspects might also be some of the causes why clear and sound evidence of explicitation being an interpreting universal has not been yet been forthcoming. Nevertheless, explicitation is still considered one of the most popular ‘candidates’ in this quest. The investigation described in the following study is an initial attempt at searching for explicitness in Italian and English texts, both translated and interpreted (in the simultaneous mode). One of the aims of this study is to produce comparable results with future research projects, by adopting a quantitative approach, which is typical of corpus-based studies and has already been adopted in Kajzer-Wietrzny’s study mentioned above. For this reason, in our study we applied the methodology used by Kajzer-Wietrzny (2012) in order to obtain comparable data, though working also on different language pairs.

In addition, a qualitative analysis was also performed to combine the invaluable contribution provided by corpora with a qualitative approach aimed at corroborating quantitative data, enriching it with further insights.

2. A study of explicitness in oral and written texts in English and Italian

2.1 Materials

The texts used for this study come from the European Parliament Translation and Interpreting Corpus (Bernardini, Ferraresi and Miličević 2013), a machine-readable multifaceted resource recently developed at the Forlì campus of the University of Bologna. EPTIC evolved from EPIC, the trilingual European Parliament Interpreting Corpus (Sandrelli and Bendazzoli 2005; Bendazzoli 2010) including speeches in Italian, English and Spanish and their interpretations in the same languages. Overall, EPTIC is an extension of EPIC with a difference - it does not contain Spanish subcorpora but it includes four subcorpora of written texts, both in Italian and in English. As a result, EPTIC is to date a unique bilingual, bidirectional and intermodal corpus of European Parliament plenary session speeches in English and Italian, containing both transcripts of source speeches and their interpreted versions paired with their written counterparts in the form of source-language minutes and their translations.

Plain text transcripts and headers containing extra-linguistic information (metadata) were taken from EPIC, while their respective minutes and their independently produced translations were downloaded from the official website of the European Parliament. Transcripts were POS-tagged and lemmatized[2], and subsequently sentence-level alignment for parallel and intermodal pairs was carried out. For this study the corpus has been accessed through a UNIX server with a SSH client, and investigations were performed by using CQP syntax with a command line interface.

Thanks to its structure (Figure 1), EPTIC offers the key advantage of supporting investigations from various perspectives: its content can be examined from a comparable monolingual point of view, by comparing interpreted and translated subcorpora with oral and written source production in the same language, and from a monolingual intermodal perspective, by contrasting interpreted and translated texts, as well as from a bilingual parallel perspective that enables the investigation of target texts and their source texts.

img1

ST: source texts

IT: Italian

ST-IN: transcripts of the original speech

TT: target texts

EN: English

TT-IN: transcript of the interpreter’s rendition

IN: interpreted

 

ST-TR: minutes in the original language

TR: translated

 

TT-TR: translated minutes

FIGURE 1: Structure of EPTIC

As far as the corpus size is concerned, EPTIC subcorpora were not very much balanced in its original version (Table 1). For the purposes of this study, the first EPTIC enlargement was carried out adding transcripts, thus raising the number of tokens from 175,122 to 253,818 (Table 2).

SUBCORPUS

NUMBER OF TEXTS

TOTAL WORD COUNT

 

% OF EPTIC

ST-IN-EN

81

41,869

23.91

ST-TR-EN

81

36,685

20.95

TT-IN-IT

81

33,675

19.23

TT-TR-IT

81

36,876

21.06

ST-IN-IT

17

6387

3.65

ST-TR-IT

17

6234

3.56

TT-IN-EN

17

6577

3.76

TT-TR-EN

17

6819

3.89

TOTAL

392

175,122

100

TABLE 1: Size and composition of EPTIC (Bernardini, Ferraresi and Miličević 2013)

SUBCORPUS

NUMBER OF TEXTS

TOTAL WORD COUNT

 

% OF EPTIC

ST-IN-EN

81

41,869

16.50

ST-TR-EN

81

36,685

14.45

TT-IN-IT

81

33,675

13.27

TT-TR-IT

81

36,876

14.53

ST-IN-IT

61

26,088

10.28

ST-TR-IT

61

25,244

9.95

TT-IN-EN

61

26,113

10.29

TT-TR-EN

61

27,268

10.73

TOTAL

568

253,818

100

TABLE 2: Size and Composition of EPTIC after the first enlargement

2.2 Objectives and methods

Our study set out to test the findings of the quantitative analysis on explicitness in English interpretations and translations carried out by Kajzer-Wietrzny (2012). A second objective was to perform the first investigation into explicitness in Italian applying the same research design to the EPTIC Italian subcorpora. Furthermore, the study was also aimed at identifying tendencies in terms of higher or lower explicitness of interpreted and translated texts, in order to see whether some generalizations could be drawn.

Kajzer-Wietrzny’s three indicators of explicitness for the investigation of English were borrowed for our study; they include the use of linking adverbials, apposition markers and optional that, since they are widely considered to be linguistic signals for an increased level of explicitness (Kajzer-Wietrzny 2012: 76). The set of linking adverbials comprised: as a consequence, as a result, consequently, hence, in consequence, therefore and thus. The apposition markers adopted indicated reformulations with an explicitation function: that is, that is to say, to be (more) precise, to be (more) specific, to be exact, namely and in other words. The third indicator, the optional that connective after reporting verbs, differed from the previous two because it indicates increased syntactic explicitness rather than content reformulations. The reporting verbs chosen were: admit, believe, claim, hope, know, suggest, say and tell. All these indicators are exactly the same as in the study by Kajzer-Wietrzny.

For the investigation of Italian, the aforementioned indicators were matched by comparable Italian ones, identifying similar sets of linguistic elements characteristic of this language. So, for the purposes of this study, some adaptations have been made to the set of Italian indicators. In particular, an authoritative Italian grammar text was taken as a reference, specifically the Grammatica Italiana by Serianni, which provides lists of both linking adverbials and apposition markers (1989: 541-542). From these lists, archaic items such as onde and laonde were left out, since they are virtually never used today, neither in prestige high-register, while other elements, such as per cui, di conseguenza and in altre parole, were added in order to have the same number of elements in each set (English and Italian). The linking adverbial infatti was excluded, given its numerous pragmatic meanings and functions and its high frequency in spoken Italian language which might have distorted the results. The Italian linking adverbials selected were: dunque, quindi, perciò, pertanto, sicché, per cui and di conseguenza, while the set of apposition markers included: cioè, ossia, ovvero, se vogliamo, vale a dire, per essere precisi and in altre parole. As for the English optional that, this syntactic indicator was replaced with a similar one, as in Italian the connective after reporting verbs has always to be stated. In Italian, in fact, when the subordinate clause after reporting verbs has the same subject of the main clause, the speaker can choose between an explicit clause structure, using the connective che followed by the subject and the conjugated verb, and an implicit clause structure followed by the connective di and an infinitive verb, without having to repeat the subject. The reporting verbs chosen were: affermare, aggiungere, ammettere, annunciare, asserire, avvertire, comunicare, confessare, considerare, credere, dichiarare, dire, dubitare, esclamare, informare, negare, promettere, raccontare, ricordare, rispondere, ritenere e trovare (Serianni 1989: 552). In this subset of indicators, 22 elements have been analysed, as opposed to the only 8 items in English. In this case, we could not keep the same number of reporting verbs in English, since these reporting verbs are all currently in use in Italian.

Firstly, a quantitative macroanalysis[3] was carried out both from a monolingual comparable perspective, comparing speeches and interpretations in the same language, and from an intermodal perspective, comparing the occurrences of the selected sets of items in interpretations and translations. Secondly, a parallel qualitative microanalysis was performed to check if the trends of increased explicitness highlighted by the quantitative analyses were actually confirmed. The underlying tenet of the chosen methodology is that combining both a quantitative and qualitative approach gives researchers the opportunity to verify that the observations that emerged from the quantitative investigation are in fact the phenomenon really looked for, thus allowing sounder conclusions to be drawn.

2.3 Statistical significance test and limitations of the applied methodology

In order to verify the null hypothesis, that is that there were no differences between the raw frequencies of the investigated explicitness indicators, Fisher’s exact test was applied and carried out thanks to R, a freely accessible software for statistical computing[4]. This test is expected to produce more precise p-values for not too large counts (Baayen 2008:113), as this is the case of the chosen items’ frequencies, given the size of EPTIC and the kind of investigated linguistic phenomena[5].

The first limitation of this methodology is related to the statistical significance test. To carry out Fisher’s exact test, the number of the occurrences of the indicator (x) and the total number of tokens (N) of the investigated subcorpus must be inserted in the matrix (N – x). Since some indicators consist of more than one token, it is impossible to take this into account while performing the subtraction, otherwise each token of the same occurrence would count as separate occurrences. Therefore, every indicator was considered as if it were composed by a single token. It was thought that this does not affect the statistical relevance result.

Secondly, for an analysis to be as similar as possible in two different languages, first a preliminary contrastive analysis should be performed in order to verify that the sets of selected indicators are in fact exhaustive and equally represented in that language. Nevertheless, this would have led to a different methodology and to results not comparable with Kajzer-Wietrzny’s study (2012).

Thirdly, it must also be acknowledged that it would have been better to use a corpus with perfectly balanced subcorpora, while some EPTIC components are larger than others, in spite of the efforts made within the scope of this study, as mentioned in 2.1.

3. Results

3.1 Introduction

In the following sections the results of the investigations conducted in the eight EPTIC subcorpora are grouped according to the three different sets of selected explicitness indicators, and presented from a traditional monolingual perspective. All the results are shown together with raw and normalized frequencies[6] and with the outcomes of Fisher’s exact test for statistical relevance. First, the tendencies which emerged are discussed carrying out a comparable and an intermodal macroanalysis. For the purposes of the intermodal analysis, source subcorpora have also been statistically compared, to ensure that no significant differences due to external factors could affect the comparison between the target subcorpora, in terms of occurrences of the selected items, between original speeches and written minutes. Second, a parallel qualitative analysis is performed to confirm or disconfirm two quantitative tendencies that have emerged from the previous investigations and were selected as particularly interesting.

Even though a quantitative macroanalysis follows, the corpus size allowed us to ascertain that every single occurrence is precisely the linguistic phenomenon looked for.

3.2. Macroanalysis on English subcorpora

3.2.1 Frequency of explicitness indicators

Table 3 shows the results obtained from the different queries launched in the English oral and written subcorpora, both from a comparable and an intermodal perspective.

As regards linking adverbials, the data related to the oral subcorpora, namely original speeches and interpreted production into English, showed 57 occurrences in the interpreted subcorpus, normalized to 100,000 words, which was exactly the same number of normalized occurrences in comparable original speeches in English. As for written subcorpora, source texts (minutes) featured 65 occurrences of the seven selected items, while in the translated subcorpus the number of occurrences was almost three times as much, that is to say 183 in total. In this case, there were no statistically significant differences between the two subcorpora of oral and written source texts that could distort this comparison, counting respectively 57 and 65 occurrences.

As for apposition markers, the corresponding occurrences were 46 in interpreted texts and 26 in the comparable oral subcorpus. In written subcorpora, 51 apposition markers were counted in translated texts and 26 in comparable written texts, and therefore the normalized number of occurrences in the subcorpora of source texts is the same.

The results concerning the optional that connective after reporting verbs indicated that in every subcorpus this connective was verbalized in more than 50 per cent of cases. More precisely, the structure with optional that occurred in 78 per cent of the cases in the subcorpus of interpreted texts and in 70 per cent of cases in that of original speeches, while it appeared in 84 per cent of the cases in written translations and in 80 per cent of the cases in written texts.

From an intermodal perspective, it can be noted that the normalized occurrences of linking adverbials in the subcorpus of interpretations were 57, while they accounted for 183 in the subcorpus of translations. The apposition markers in the interpreted English subcorpus were 46 compared to the 51 occurrences in the translated English one. The data on optional that show that interpreters opted for the verbalisation of this connective in 78 per cent of cases, while translators did that in 84 per cent of cases.

LINKING ADVERBIALS ENGLISH

SUBCORPUS

st-in-en

tt-in-en

st-tr-en

tt-tr-en

RAW FREQUENCY

24

15

24

50

NORMALIZED FREQUENCY (100,000 words)

57

57

65

183

STATISTICAL COMPARISON

significance

indicated by: (*)

monolingual comparable

monolingual intermodal

st-in-en vs. tt-in-en

st-tr-en vs. tt-tr-en

st-in-en vs. st-tr-en

tt-in-en vs. tt-tr-en

p-Value

significant with

p < 0,05

1

1.72 x 10-5 (*)

0.6666

2.852 x 10-5 (*)

 

APPOSITION MARKERS ENGLISH

SUBCORPUS

st-in-en

tt-in-en

st-tr-en

tt-tr-en

RAW FREQUENCY

11

12

9

1

NORMALIZED FREQUENCY (100,000 words)

26

46

26

51

STATISTICAL COMPARISON

significance

indicated by: (*)

monolingual comparable

monolingual intermodal

st-in-en vs. tt-in-en

st-tr-en vs. tt-tr-en

st-in-en vs. st-tr-en

tt-in-en vs. tt-tr-en

p-Value

significant with

p < 0,05

0.2001

0.09197

1

0.8459

 

OPTIONAL THAT

SUBCORPUS

st-in-en

tt-in-en

st-tr-en

tt-tr-en

RAW FREQUENCY

97/138

63/81

90/112

75/89

FREQUENCY IN PERCENT

70%

78%

80%

84%

STATISTICAL COMPARISON

significance

indicated by: (*)

monolingual comparable

monolingual intermodal

st-in-en vs. tt-in-en

st-tr-en vs. tt-tr-en

st-in-en vs. st-tr-en

tt-in-en vs. tt-tr-en

p-Value

significant with

p < 0,05

0.2704

0.5792

0.3282

0.0793

TABLE 3: Results - English[7]

ST: source texts

IT: Italian

ST-IN: transcripts of the original speech

TT: target texts

EN: English

TT-IN: transcript of the interpreter’s rendition

IN: interpreted

 

ST-TR: minutes in the original language

TR: translated

 

TT-TR: translated minutes

3.2.2 Major findings

The investigations carried out in the English subcorpora reveal an overall tendency towards an increased frequency of explicitness indicators in interpreted and translated subcorpora. The only exceptions are linking adverbials, whose frequency remains constant in oral subcorpora but increases in the translated subcorpus compared to the comparable written one. Only this very last tendency of higher explicitness marked by a larger number of linking adverbials in translated English texts is statistically significant according to Fisher’s exact test. Table 4 summarises the explicitness tendencies that have emerged from the comparable monolingual analysis in English.

 

SUBCORPUS

NORMALIZED FREQUENCY

EXPLICITNESS

IN TARGET SUBCORPUS

STATISTICAL SIGNIFICANCE

LINKING ADVERBIALS

ENGLISH

tt-in-en

57

CONSTANT

NO

st-in-en

57

tt-tr-en

183

HIGHER (*)

YES

st-tr-en

65

APPOSITION MARKERS

ENGLISH

tt-in-en

46

HIGHER

NO

st-in-en

26

tt-tr-en

51

HIGHER

NO

st-tr-en

26

OPTIONAL THAT

tt-in-en

78%

HIGHER

NO

st-in-en

70%

tt-tr-en

84%

HIGHER

NO

st-tr-en

80%

TABLE 4: Explicitness tendencies – comparable perspective (English)

The English intermodal analysis has brought to light a clear tendency towards a higher degree of explicitness in translated production compared to interpreted texts, even though this tendency is statistically confirmed only in the case of linking adverbials. In no comparison between written and oral source subcorpora can a statistically significant difference be observed, thus confirming that potential tendencies are to be ascribed to the translating or interpreting process and not to differences already present in the source subcorpora. Table 5 indicates the frequencies of the three selected explicitness indicators in English from an intermodal point of view.

 

SUBCORPUS

NORMALIZED FREQUENCY

EXPLICITNESS

STATISTICALLY SIGNIFICANT

LINKING ADVERBIALS

ENGLISH

tt-in-en

57

HIGHER (*)

in tt-tr-en

YES

tt-tr-en

183

st-in-en

57

NO statistically

significant difference

between source subcorpora

st-tr-en

65

APPOSITION MARKERS

ENGLISH

tt-in-en

46

HIGHER

in tt-tr-en

NO

tt-tr-en

51

st-in-en

26

NO statistically

significant difference

between source subcorpora

st-tr-en

26

OPTIONAL THAT

tt-in-en

78%

HIGHER

in tt-tr-en

NO

tt-tr-en

84%

st-in-en

70%

NO statistically

significant difference

between source subcorpora

st-tr-en

80%

TABLE 5: Explicitness tendencies – intermodal perspective (English)

3.3 Macroanalysis on Italian subcorpora

3.3.1 Frequency of explicitness indicators

As was mentioned before (section 2.2), the three Italian explicitness indicators were chosen so as to be as similar as possible to the English ones. Since the Italian language does not have an equivalent connective to the optional that, but only an alternative construction with the connective preposition followed by the implicit form, this English syntactic explicitness indicator was replaced with the most similar Italian explicit structure. Table 6 summarises the results of the queries launched in the Italian subcorpora together with the respective statistical significance tests.

In the subcorpus of interpreted texts into Italian, 273 normalized occurrences of linking adverbials were counted, while there were 207 occurrences in the original Italian oral subcorpus. As for written subcorpora, in the translated one there were 106 occurrences of the seven selected items and in the comparable written subcorpus the number of occurrences is 136. Also in this case there were no statistically significant differences between the two subcorpora of oral and written source texts.

The investigations carried out for the second indicator, the set of apposition markers, showed that 51 normalized occurrences of this linguistic phenomenon could be identified in interpreted texts, but this frequency went up to 104 in the comparable oral Italian subcorpus. The same trend was observed in the two corresponding written subcorpora, where occurrences accounted for 30 in translated texts and 76 in comparable written ones. From the control comparison, in this case, no statistically significant differences were observed between source subcorpora either.

The frequencies obtained show that the third indicator chosen, namely the explicit structure after reporting verbs in clauses with the same subject, is in fact a very rare linguistic phenomenon in this corpus, being in one subcorpus even completely absent. Therefore, it could not be used to make any kind of generalisation and it was excluded from the following analysis.

The intermodal perspective highlighted that the normalized frequency of linking adverbials was 273 in interpreted texts and 106 in translated texts, and as for apposition markers, 51 occurrences could be observed in the interpreted subcorpus and 30 in the translated one.

LINKING ADVERBIALS ITALIAN

SUBCORPUS

st-in-it

tt-in-it

st-tr-it

tt-tr-it

RAW FREQUENCY

54

92

50

39

NORMALIZED FREQUENCY (100,000 words)

207

273

136

106

STATISTICAL COMPARISON

significance

indicated by: (*)

monolingual comparable

monolingual intermodal

st-in-it vs. tt-in-it

st-tr-it vs. tt-tr-it

st-in-it vs. st-tr-it

tt-in-it vs. tt-tr-it

p-Value

significant with

p < 0,05

0.1125

0.003396 (*)

0.8448

2.472 x 10-7 (*)

 

APPOSITION MARKERS ITALIAN

SUBCORPUS

st-in-it

tt-in-it

st-tr-it

tt-tr-it

RAW FREQUENCY

27

17

28

11

NORMALIZED FREQUENCY (100,000 words)

104

51

76

30

STATISTICAL COMPARISON

significance

indicated by: (*)

monolingual comparable

monolingual intermodal

st-in-it vs. tt-in-it

st-tr-it vs. tt-tr-it

st-in-it vs. st-tr-it

tt-in-it vs. tt-tr-it

p-Value

significant with

p < 0,05

0.02186 (*)

0.0001117 (*)

0.8928

0.1884

 

CHE + CONJUGATED VERB

SUBCORPUS

st-in-it

tt-in-it

st-tr-it

tt-tr-it

RAW FREQUENCY

2/3

6/9

0

7/9

FREQUENCY IN PERCENT

67%

67%

0%

78%

TABLE 6: Results – Italian

ST: source texts

IT: Italian

ST-IN: transcripts of the original speech

TT: target texts

EN: English

TT-IN: transcript of the interpreter’s rendition

IN: interpreted

 

ST-TR: minutes in the original language

TR: translated

 

TT-TR: translated minutes

3.3.2 Major findings

As was noted in the previous section, only two out of the three explicitness indicators chosen could be taken into account for the macroanalysis in Italian, since the third resulted as being too rare in the investigated corpus. Given the longer list of reporting verbs chosen for Italian (22 items in this category, as opposed to the 8 selected for English on the basis of Kajzer-Wietrzny 2012), this finding is surprising. Maybe a preliminary contrastive analysis could help ascertain whether this indicator is not representative for the Italian language or if this lack of results is due to the corpus size.

Table 7 gives an overview of the major explicitness tendencies detected. Overall, the data show an opposite trend compared to that highlighted in the English subcorpora. In both interpreted and translated subcorpora there is a statistically lower occurrence of the indicators chosen, which might be ascribed to a tendency towards a lower degree of explicitness. The only exception to this is represented by the increased frequency of linking adverbials in the interpreted subcorpora, though this was not confirmed by Fisher’s statistical test. The data concerning apposition markers are more homogenous, since both interpreted and translated subcorpora feature fewer explicitness items than their source counterparts.

 

SUBCORPUS

NORMALIZED FREQUENCY

EXPLICITNESS

IN TARGET SUBCORPUS

STATISTICAL SIGNIFICANCE

LINKING ADVERBIALS

ITALIAN

tt-in-it

273

HIGHER

NO

st-in-it

207

tt-tr-it

106

LOWER (*)

YES

st-tr-it

136

APPOSITION MARKERS

ITALIAN

tt-in-it

51

LOWER (*)

YES

st-in-it

104

tt-tr-it

30

LOWER (*)

YES

st-tr-it

76

TABLE 7: Explicitness tendencies – comparable perspective (Italian)

The intermodal investigation carried out in the Italian subcorpora shows once again an opposite trend compared to that observed in the English subcorpora, as indicated in Table 8, where explicitness tendencies are illustrated from an intermodal perspective. Here the occurrences of the selected sets of items were higher in the interpreted texts, but only for linking adverbials is this tendency statistically confirmed.

 

SUBCORPUS

NORMALIZED FREQUENCY

EXPLICITNESS

STATISTICAL SIGNIFICANCE

LINKING ADVERBIALS

ITALIAN

tt-in-en

273

HIGHER (*)

in tt-in-it

YES

tt-tr-en

106

st-in-en

207

NO statistically

significant difference

between source subcorpora

st-tr-en

136

APPOSITION MARKERS

ITALIAN

tt-in-en

51

HIGHER

in tt-in-it

NO

tt-tr-en

30

st-in-en

104

NO statistically

significant difference

between source subcorpora

st-tr-en

76

TABLE 8: Explicitness tendencies – intermodal perspective (Italian)

3.4 Parallel microanalysis

3.4.1 Introduction

After having discussed the major explicitness tendencies observed thanks to a twofold monolingual analysis following a comparable and an intermodal approach, in this section some of the findings so far obtained are tested, performing a parallel qualitative analysis in English and Italian – a way of further exploiting the huge potential of EPTIC. In particular, the source frequencies of the selected sets of items are compared and investigated through bilingual concordances. Needless to say that only linking adverbials and apposition markers are the subject of this analysis, since these are the only indicators that can be contrasted from an inter-linguistic viewpoint.

The assumption underlying this kind of analysis, combined with the previous one, is that this is an effective way to verify if the emergent tendencies of higher or lower explicitness are in fact the result of explicitation or implicitation processes. For the purposes of this paper, only two interesting cases are discussed, namely English linking adverbials and Italian apposition markers.

3.4.2 Parallel microanalysis of English linking adverbials

In the monolingual quantitative analysis of linking adverbials, the tendency of translated texts being more explicit than comparable English written texts emerged as statistically significant. From a parallel point of view, the data related to linking adverbials (Table 9) show a drop in the number of the occurrences of this indicator from 54 in Italian original speeches to 15 in their English interpreted counterparts.

 

SUBCORPUS

RAW FREQUENCY

LINKING ADVERBIALS

ENGLISH

st-in-it

54

tt-in-en

15

st-tr-it

50

tt-tr-en

50

TABLE 9: linking adverbials - parallel perspective (English)

By looking at parallel concordances, it can be observed that English interpreters tend to leave out linking adverbials more frequently than Italian speakers, as shown in the following examples (1) and (2).

(1)

st-in-it <text_id 18>: Un precedente ripeto molto grave di censura che sembra più tipico della democrazia alla cu- turca che non di un paese sicuramente democratico e fondatore dell'Unione europea contro la forza politica che vuole autonomia e federalismo e quindi contro lo statalismo //

tt-in-en: It 's a kind of thing that one would expect more in Turkey than in a country which is supposed to be democratic and which is certainly a founding country of the European Union // Ehm we are in favour of federalism and against the state centralism [...]

st-tr-it: E' un precedente , ripeto , molto grave di censura - che sembra più tipico della democrazia alla turca che di un paese sicuramente democratico e fondatore dell' Unione europea - nei confronti di una forza politica che vuole autonomia e federalismo, quindi contro lo statalismo.

tt-tr-en: This seems more typical of Turkish style democracy than that of a country that is undoubtedly democratic and was a founding member of the European Union. I would reiterate that this is an extremely serious precedent concerning censorship of a political party that wants independence and federalism and is, therefore, against statism.

 (2)

st-in-it <text_id 46>: Credo che la scelta dell'Unio- dell'Unione europea sia corretta quella di avere lunghi tempi non di fermarsi immediatamente con una politica dove i nodi vengono tagliati come nell'antico attraverso il taglio del nodo gordiano // Abbiamo bisogno quindi di un tempo abbiamo bisogno di riflettere e di costruire relazioni //

tt-in-en: I think that the European Union has made the right choice that is looking at things on the long-term rather than simply cutting the gordian knots immediately in the form of certain policies // What we need is time we need to reflect we need to build up our activities //

st-tr-it: Credo che sia corretta la scelta dell' Unione europea di optare per il lungo periodo e di non pretendere risultati immediati attraverso il taglio del nodo gordiano di antica memoria. Abbiamo quindi bisogno di tempo per riflettere e costruire relazioni.

tt-tr-en: I think the European Union 's decision to opt for the long term and not to aim for immediate results by cutting the Gordian knot of old is the right one. We therefore need time to reflect and to build relationships.

In example (1), being against state centralism is, in the speaker’s version, a result of being federalist, while the interpreter simply juxtaposes the two concepts. Probably, it cannot be said that the interpreter’s rendition lacks coherence, but certainly English recipients need to process the concept heard to a deeper level than the Italian audience. Example (2) shows a case in which a common Italian SVO pattern is turned by the interpreter into a pseudo-cleft sentence, placing more emphasis on the concept of time. Since this sort of structure presumably requires a more considerable cognitive effort for the interpreter, it might be that the interpreter opted for this more complex structure to compensate the linking adverbial’s omission.

The observed tendency of English interpreters to omit linking adverbials that are, on the contrary, present in the respective translations, seems in line with the results of the intermodal analysis, thus confirming the higher explicitness of translated texts for this explicitness indicator.

3.4.3 Parallel microanalysis of Italian apposition markers

The quantitative comparable macroanalysis of Italian apposition markers highlighted a statistically significant tendency towards a lower degree of explicitness of the interpreted subcorpus compared to the original oral one. Table 10 shows the results of the investigations carried out on this indicator from a parallel perspective, displaying a rise from 11 to 17 occurrences in interpreted texts.

 

SUBCORPUS

RAW FREQUENCY

APPOSITION MARKERS
ITALIAN

st-in-en

11

tt-in-it

17

st-tr-en

9

tt-tr-it

11

TABLE 10: apposition markers - parallel perspective (Italian)

By examining the relevant parallel concordances, the Italian interpreters' preference to add apposition markers can be clearly observed in examples (3) and (4).

(3)

st-in-en <text_id 23>: There is sometimes a danger that we engage in the in the politics of the Book of Genesis //

tt-in-it: a volte c'è un pericolo ossia ehm fare un po'come la Genesi che luce sia e luce fu //

st-tr-en: Sometimes there is a danger that we engage in the politics of the Book of Genesis : let there be light - and there is light.

tt-tr-it: A volte si rischia di fare la politica del Libro della Genesi : " Sia la luce. E la luce fu ".

 (4)

st-in-en <text_id 3>: Commissioner Byrne I welcome very much your statement here here this morning // But I understand Sir you went to Thailand and were told that it wasn't avian flu but it was chicken cholera cholera //

tt-in-it: Commissario Byrne io sono molto lieto della dichiarazione che lei ha fatto questa mattina però mi pare che lei sia andato in Tailandia e le è stato detto che non c'era l'influenza aviaria ma che era un'altra malattia dei polli e cioè il colera dei polli //

st-tr-en: Mr President, Commissioner Byrne, I welcome your statement here this morning, but I understand that you went to Thailand and were told that it was not avian flu but chicken cholera.

tt-tr-it: Signor Presidente, Commissario Byrne, accolgo con favore la sua dichiarazione di stamani, ma ho sentito che lei è stato in Tailandia e che le è stato detto che non si trattava di influenza aviaria, ma di colera dei gallinacei.

While example (3) seems to be a simple addition of the apposition marker ossia, without further relevant interventions by the interpreter on his/her production, example (4) features a more segmented rendition than the original speech. The structure “it wasn’t x, but y” is rendered into Italian more gradually “it wasn’t x, but another illness, that is y”, and is a typical case of the interpreter approaching the correct translation by stating a general term first and then correcting and/or complementing it with a more precise one immediately afterwards. The choice of a longer rendition might have been required by the interpreter’s need to recall the information from the short-term memory, or to think of a more accurate term. In both examples, apposition markers appear as good tools to add a new piece of information to an utterance that could also have been considered as concluded, without starting a new sentence.

Finally, example (5) shows a quite common phenomenon in the corpus. Here the apposition marker is added in the interpreted production as a consequence of the interpreter verbalising a logical link that s/he presumably perceives from the speaker’s prosody. In the corresponding written versions, this link is not lexicalised but it is expressed through punctuation marks.

(5)

st-in-en <text_id 23>: Mention has been made here about the slow progress in relation to the two major ehm Amsterdam imperatives the ehm the ehm the directives in relation to asylum //

tt-in-it: Si è parlato della lentezza dell'avanzamento dei due imperativi più importanti di Amsterdam ossia le di- ehm le direttive legate all'asilo //

st-tr-en: Mention has been made of the slow progress in relation to the two major Amsterdam imperatives: the directives in relation to asylum.

tt-tr-it: Si è parlato dei lenti progressi riguardanti i due principali imperativi di Amsterdam: le direttive in materia di asilo.

In the light of the cases of explicitation examined so far, it seems reasonable to conclude that the tendency of interpreted texts being less explicit cannot be observed in the parallel qualitative analysis performed.

4. Conclusions

This study found no clear evidence of more or less explicitness in interpreted/translated versus untranslated speeches, and therefore no evidence for a universal tendency in its strictest sense. In other words, no homogenous and conclusive tendency could be observed which could summarise and include all the results obtained with the methodology applied to analyse each explicitness indicator selected. On the other hand, from the monolingual macroanalysis different tendencies according to language direction have emerged.

Firstly, results in the English subcorpora in this study suggest an overall tendency of increased explicitness in translated and interpreted subcorpora, but only the tendencies of the translated subcorpus being more explicit compared to the written comparable one and to the interpreted subcorpus were confirmed by Fisher’s exact test of statistical relevance. The results highlighted by the queries related to both linking adverbials, apposition markers and optional that do not indicate a statistically significant tendency towards a higher explicitness of interpretations, while it seems that their frequency is, in a few cases, statistically significantly more pronounced in translated texts and not statistically significant in the remaining ones. The findings related to all the three explicitness indicators selected are therefore in line with Kajzer-Wietrzny’s (2012: 141).

Secondly, after having excluded the third explicitness indicator because of its overall lower frequency in the corpus, an overall tendency of lower explicitness of both interpreted and translated texts has emerged from the investigations in the Italian subcorpora. As for linking adverbials, the tendency of increased explicitness of interpreted texts compared to source oral texts in Italian was not statistically significant. On the other hand, the tendency of translated texts being less explicit than their comparable counterparts was confirmed by Fisher’s exact test. Also the lower frequency of apposition markers in both translated and interpreted texts compared to the respective comparable subcorpora was statistically proved. From an intermodal point of view, the tendency of interpreted texts being more explicit than translated ones was statistically relevant only for linking adverbials. It must be borne in mind that Kajzer-Wietrzny’s study did not investigate Italian subcorpora, and therefore these results not only corroborate hers, but also complement them.

Finally, thanks to the versatility of the EPTIC corpus and the twofold methodology (quantitative and qualitative) applied, some interesting differences according to language direction have emerged from the parallel microanalysis. In spite of the limitations that are present in this study, such as the sample size and the unbalanced subcorpora, in this paper the English interpreters’ preference to leave out linking adverbials more frequently than Italian speakers, and the tendency of Italian interpreters to add apposition markers, which they use more frequently than English speakers were discussed. In particular, the first case confirms the outcome of the monolingual analysis, while in the second case the findings of the quantitative macroanalysis were disproved by the parallel qualitative microanalysis.

The overriding aim of this study was to conduct an investigation into explicitness, trying to add new elements to the discussion on translation and interpreting universals. Hopefully, the methodology applied will aid in comparing and contrasting the findings of future studies, thus enriching the debate on translation and interpreting universals. Furthermore, parallel concordances allowed for the highlighting of some shortcomings of traditional monolingual comparable analyses, whose results could not sometimes be disconfirmed by parallel investigation. A deeper analysis on larger Italian corpora of the explicit structure after reporting verbs in clauses with the same subject is hoped to verify its appropriateness. We also noted (3.3.2) that a preliminary contrastive analysis can more effectively help choose perfectly equivalent explicitness indicators, especially if this kind of investigation is to be carried out on non cognate languages. These and other aspects may represent a fertile ground for future studies aimed at dispelling some of the confusion about this potential universal.

Nevertheless, these results highlight the huge untapped potential of bilingual, bidirectional and intermodal corpora like EPTIC, and the need to enlarge the corpus in order to gain further research insights into the nature of translation and interpreting universals.

References

Baayen, R. H. (2008) Analyzing Linguistic Data, Cambridge, Cambridge University Press.

Baker, M. (1993) “Corpus Linguistics and Translation Studies: Implications and Applications” in Text and Technology: In Honour of John Sinclair, M. Baker, G. Francis and E. Tognini-Bonelli (eds) Amsterdam, John Benjamins, 233–50.

---- (1996) “Corpus-based Translation Studies: The Challenges that Lie Ahead” in Terminology, LSP and Translation. Studies in Language Engineering in Honour of Juan C. Sager, H. Somers (ed) Amsterdam, John Benjamins, 175–86.

Baumgarten, N., B. Meyer and D. Özçetin (2008) “Explicitness in Translation and Interpreting: A Critical Review and Some Empirical Evidence (of an Elusive Concept)”, Across Languages and Cultures 9, no.2: 177–203.

Becher, V. (2010a) “Towards a More Rigorous Treatment of the Explicitation Hypothesis in Translation Studies”, Trans-kom 3, no. 1: 1–25.

----(2010b) “Abandoning the Notion of "Translation-inherent" Explicitation: Against a Dogma of Translation Studies”, Across Languages and Cultures 11, no. 1: 1–28.

Bendazzoli, C. (2010) “The European Parliament as a Source of Material for Research into Simultaneous Interpreting: Advantages and Limitations” in Translationswissenschaft - Stand und Perspektiven, L. N. Zybatow (ed) Frankfurt, Peter Lang: 51–68.

Bernardini, S., A. Ferraresi and M. Miličević (2013) From EPIC to EPTIC - building and using an intermodal corpus of translated and interpreted texts. Paper presented at the 46th Annual Meeting of the Societas Linguistica Europea (SLE 2013) Split, Croatia. 18-21 September.

Blum-Kulka, S. (1986/2001) “Shifts in Cohesion and Coherence in Translation” in The Translation Studies Reader, L. Venuti (ed) (2001) London, Routledge: 298–313.

Chesterman, A. (2004) “Beyond the Particular”, in Translation Universals. Do They Exist?, A. Mauranen and P. Kujamäki (eds), Amsterdam, John Benjamins: 33–49.

Gumul, E. (2006) “Explicitation in Simultaneous Interpreting: A Strategy or a By-product of Language Mediation?”, Across Languages and Cultures 7, no. 2: 171-90.

Ishikawa, L. (1999) “Cognitive Explicitation in Simultaneous Interpreting”, in Anovar/anosar estudios de traducción e interpretación, A. Àlvarez Lugrís and A. Fernándey Ocampo (eds), Vigo, Universidade de Vigo: 231–57.

Kajzer-Wietrzny, M. (2012) Interpreting Universals and Interpreting Style, PhD diss., Adam Mickiewicz University, Poznań.

Konšalová, P. (2007) “Explicitation as a Universal in Syntactic Ce/condensation”, Across Languages and Cultures 8 no. 1: 17-32.

Mauranen, A., and P. Kujamäki (eds) (2004) Translation Universals. Do They Exist?, Amsterdam, John Benjamins.

Olohan, M., and M. Baker (2000) “Reporting That in Translated English. Evidence for Subconscious Processes of Explicitation?”, Across Languages and Cultures 1, no. 2: 141–58.

Øverås, L. (1998) “In Search of the Third Code: An Investigation of Norms in Literary Translation”, Meta 43, no. 4: 571–88.

Pápai, V. (2004) “Explicitation. A Universal of Translated Text?” in Translation Universals. Do They Exist?, A. Mauranen and P. Kujamäki (eds), Amsterdam, John Benjamins: 143–64.

Puurtinen, T. (2004) “Clause Connectives in Finnish Children's Literature” in Translation Universals. Do They Exist?, A. Mauranen and P. Kujamäki (eds), Amsterdam, John Benjamins: 165–75.

Pym, A. (2005) “Explaining Explicitation”, in New trends in Translation Studies. In Honour of Kinga Klaudy, K. Károly and A. Fóris (eds), Budapest, Akadémiai Kiadó: 29–34.

Sandrelli, A. and C. Bendazzoli (2005) “Lexical Patterns in Simultaneous Interpreting: A Preliminary Investigation of EPIC (European Parliament Interpreting Corpus)”, Proceedings from the Corpus Linguistics Conference Series 1, no. 1.

Schjoldager, A. (1995/2002) “An Explanatory Study of Translational Norms in Simultaneous Interpreting” in The Interpreting Studies Reader, F. Pöchhacker and M. Shlesinger (eds) (2002), London & New York, Routledge: 301-11.

Serianni, L. (1989). Grammatica Italiana, Torino, Utet.

Shlesinger, M. (1995) “Shifts in Cohesion and Simultaneous Interpreting”, The Translator 1, no. 2: 193–214.

Notes

[1] S-Universals are ‘universal differences between translations and their source texts’, while T-Universals are ‘universal differences between translations and comparable non-translated texts’ (Chesterman 2004).

[2] Part-of-speech tagging and lemmatization was performed independently of EPIC using Tree-Tagger, while Corpus Work Bench (CWB) was used for the indexing process (Bernardini et al. 2013).

[3] In this study, by ‘quantitative analysis’ we mean the analysis of the number of occurrences of the indicators chosen, on which we applied Fisher’s exact test to find out their statistical significance.

[4] For further information about the R project see: http://www.r-project.org/ (accessed: 16 June 2016).

[5] This is also the reason why chi-squared test was not applied. In this respect, Baayen maintains that ‘For tables with not too large counts, a test of independence of rows (or columns) that produces more precise p-values is Fisher’s exact test’ (Baayen 2008: 113).

[6] For the results of the different indicators and corpora to be comparable, raw frequencies have all been normalized per 100,000 words. Differently, the occurrences of the third indicator of each set of indicators are already expressed in percentages, and hence comparable across subcorpora.

[7] P-Values with more than five decimal numbers are conveniently written in scientific notation.

About the author(s)

Niccolò Morselli graduated in Interpreting (MA) from the Advanced School of Modern Languages for Interpreters and Translators (SSLMIT) at Forlì, University of Bologna
in 2014 with a dissertation on "Interpreting Universals: a study on explicitness in the intermodal corpus EPTIC".

Email: [please login or register to view author's email address]

©inTRAlinea & Niccolò Morselli (2018).
"Interpreting Universals: A study of explicitness in the intermodal EPTIC corpus"
inTRAlinea Special Issue: New Findings in Corpus-based Interpreting Studies
Edited by: Claudio Bendazzoli, Mariachiara Russo & Bart Defrancq
This article can be freely reproduced under Creative Commons License.
Stable URL: https://www.intralinea.org/specials/article/2320

Go to top of page