Abstract

A comparison of the frequencies of English phrasal verbs in two corpora of simultaneous interpreting into English, one B into A, and the other A into B, showed the A to B interpreters used phrasal verbs much less often, which confirms the theories that non-natives find it hard to acquire phrasal verbs, and that non native interpreters, because of the pressure of the SI process, find it hard to access them. In particular, phrasal verb lemmas with idiomatic and figurative meanings were very much less frequent in the language of the A into B interpreters, as were phrasal verbs with aspectual meaning. Despite this, metafunctional analysis showed that the A into B interpreters did use phrasal verbs with interpersonal function, to pursue the role of interpreter as mediator. The most striking finding on metafunctions was that phrasal verbs used by the B into A interpreters for the process of textualisation, and which are therefore crucial for meaning assembly, were almost entirely absent in the language of the A into B interpreters. In short, phrasal verbs are clearly an important resource for SI and the lack of them in the language of A into B interpreters suggests an urgent instructional need.

Keywords: simultaneous interpreting, directionality, collocations, phrasal verbs, interpreter training

1. The scope of this article

The primary purpose of this article is to make the case for the teaching of English phrasal verbs in language lessons for trainee interpreters. Phrasal verbs are a major sub-component of formulaic language, and the use of formulaic language is a major factor in the fluency that makes simultaneous interpreting a feasible activity. Despite this, to the best of my knowledge, nothing has so far been published that focuses specifically on the role played by phrasal verbs in simultaneous interpreting.

The article assumes that simultaneous interpreting is a process of assembly of meaning, as illustrated by Setton (1998/2002), and that it often necessarily involves deletion and summarising of aspects of the source text. For these reasons the article deliberately does not take a “translation” oriented perspective, and is not concerned with observing the items of language that phrasal verbs are translating, but rather with the more general questions of whether phrasal verbs are used by native and native-like interpreters, and whether there is therefore a need for non-native interpreters to use them. In seeking to find this out, the methods used are primarily those of counting and the comparison of frequencies in corpora. Given the lack of availability of transcribed interpretations, the corpora are necessarily small, but Aston (1997) argues that interesting results and applications can be derived from small corpora, and Flowerdew (2004) points out that if a corpus is specialised enough, small size is enough to provide satisfactory results – albeit when the corpora are very small, as in this research, an appropriate degree of caution will need to be exercised when generalising.

Assuming that phrasal verbs are used by interpreters, the article aims to discover which phrasal verbs are particularly characteristic of interpreting in comparison with the English language as a whole, as represented by the British National Corpus. Finally, by studying the context of use as represented in concordance lines, the article seeks to show what functions interpreters use phrasal verbs for, with a view to possible use in a language syllabus for trainee interpreters.

2. Directionality, Proficiency and Formulaic Sequences

The cognitive complexity of the process of simultaneous interpretation requires extensive knowledge of the source and target languages, while time constraints imply there must be swift access to that knowledge both in the sense of plausibly accurate comprehension and in the sense of adequately synchronic production. The initial response to this challenge was that interpreters should work only towards their mother tongue (Herbert 1952: 61), and that simultaneous interpreting, in particular, should ideally be the exclusive preserve of native speakers (Seleskovitch 1978: 100), working from their foreign active or passive language(s) (or B and C languages, respectively) to their native language (or language A according to AIIC’s classification). This favouring of C/B to A directionality is reflected in the position taken by AIIC (as reported by Bartłomiejczyk 2004: 247). Yet there are plausible arguments on the other side. Denissenko (1989: 157) argues that A to B directionality, which implies the ability to comprehend of the native speaker, but the more limited ability to produce language of the non-native speaker, is more likely to lead to reproduction of all or most of the original message. In addition, there is the pragmatic argument, as highlighted by Bartłomiejczyk (2004: 247), that circumstances lead to market demand for A to B in addition to retour, and that interpreting students might as well be properly prepared for this. One essential aspect of such preparation for working towards B, in other words for working towards the non-native language, is the effort to bring the productive language skills of the non-native speaker nearer to the proficiency level of the native speaker.

So what is it about native speakers that non-native interpreters, in seeking to be native-like, should seek to emulate? From the points of view both of comprehension and of production, acquiring the phraseological knowledge of the native-speaker would appear to be both a feasible and a useful strategy for reducing stress and processing load in SI – especially for trainees, since, according to Setton (1998/2002: 199), the attention taken by selective suppression, working in two languages, and by meaning assembly, makes access to phrasal expressions the only automatic mechanism that is likely to be available. From the point of view of comprehension, such phraseological knowledge is an important component of pragmatic processing in oral comprehension (Rost 2011: 138). From the point of view of production, the fact that native speakers know formulaic sequences (Wray 2002), in other words ready-formed phrases or strings or slot and filler patterns, brings processing advantages in reducing cognitive load and freeing up attention. This is because formulaic sequences are retrieved from memory as whole units – as it were, automatically, so that the time and effort needed to encode meanings through combining grammar and lexis is not needed (Pawley & Syder 1983: 192; Wray 2002: 9). In the words of Pawley and Syder (2000: 164), ‘It is knowledge of conventional expressions, more than anything, [...] that is the key to nativelike fluency’.

3. Phrasal verbs: description and use

3.1 The value of phrasal verbs

Multiword verbs, together with their collocations, constitute a major subclass of formulaic expressions. Following Biber et al. (1999), the subclass can be further divided, on the one hand, into free combinations of verbs and prepositions, which are not formulaic sequences, and on the other, into prepositional verbs, phrasal verbs, phrasal-prepositional verbs, and other multiword verb constructions, which are formulaic sequences (Biber et al: 1999: 403–427). Among these, the principal category is phrasal verbs, which are formed of a lexical verb in combination with an adverbial particle.

Many phrasal verbs are polysemic – they are quasi-empty vessels, with a vague tendency towards meaning which ends up being defined by the words used around them. As a consequence, many of these verbs can be used in quite a wide range of contexts – and the cotexts (in other words the adjacent text with which the phrasal verb is closely linked semantically, such as its regular collocates) may themselves extend the phrasal verb into longer strings.

As phrasal verbs are often short, with monosyllabic lexical verbs and particles, they also arguably offer value to the interpreter in reducing articulation time and effort. Additionally, they are fairly frequent in English, occurring in the British National Corpus (henceforth the BNC) at a rate of approximately one every 192 words (Gardner and Davies 2007: 347). For reasons then of processing advantages and expediency, combined with their high frequency in the English language as predicted by the BNC, one would expect to find a reasonably high occurrence of phrasal verbs in English produced by English native-speakers and native-like simultaneous interpreters.

3.2 Phrasal verbs: structural and semantic processing difficulties for non-native interpreters

The difficulty non-native speakers experience in acquiring English phrasal verbs is well documented in the literature (Dagut and Laufer 1985; Laufer and Eliasson 1993; Celce-Murcia and Larsen-Freeman 1999: 425; Liao and Fukuya 2004; Siyanova and Schmitt 2007). The difficulty is experienced even by speakers of Germanic languages, such as Dutch, which themselves have phrasal verbs (Hulstijn and Marchena1989). While one would expect experienced native and native-like interpreters working towards English to have moved beyond such difficulties, this is not necessarily the case where interpreters are non-natives working towards English as a B language, particularly if they are trainees or beginning professionals. The difficulties can be divided into two types. The first type of difficulty is structural, and applies particularly to speakers of languages with few or no phrasal verbs. The problem here is that, given the pressure experienced during SI, as detailed by Setton (1998/2002: 199; see section 2 above), interpreters working towards B may retreat to more automatised structures analogous to those of their mother tongue, avoiding the use of phrasal verbs. Given that phrasal verbs are so difficult to acquire, and that even in non-interpreters of advanced proficiency levels there is a tendency to avoid using them (Siyanova and Schmitt 2007: 129), one would expect that interpreters will to some extent avoid using them when working towards English as a B language in the booth.

The second major factor determining the difficulties of English phrasal verbs for non-natives is semantic. The non-literal nature of the meaning that many phrasal verbs convey arguably creates processing problems during construal of the message. To be clear, I am using the term “literal” with the sense intended by Grant and Bauer (2004: 39), who quote Lakoff’s definition of literal as ‘nonmetaphorical literality: directly meaningful language – not language that is understood, even partly, in terms of something else’ (Lakoff 1986: 292). According to Celce-Murcia and Larsen-Freeman (1999), cited in Darwin and Gray (1999: 68), phrasal verbs can be divided into three categories on the basis of degree of literalness: literal, idiomatic, and aspectual. Proceeding according to Lakoff’s definition, in Celce-Murcia’s first category, a literal phrasal verb is exemplified by take and down in take down the poster. The second category, idiomatic phrasal verbs, is exemplified by make up with the meaning of ‘become reconciled’, where (to use the analytical method proposed by Grant and Bauer 2004: 44, which relies on Frege’s principle of compositionality as cited in Lyons 1995: 24) the meaning of the phrase is not recoverable from any dictionary definition of the word make combined with any dictionary definition of the word up – in other words, the meaning of make up/(=become reconciled) is non-compositional. Alongside such non-compositional items as make up, which Grant and Bauer would call ‘core idioms’, there are phrasal verbs whose meanings are recoverable by means of a shared understanding, between utterance producer and utterance recipient, of figures of speech such as metaphor (Grant and Bauer 2004: 49). One example is stand out, as in his writing stands out among that of his contemporaries, but while this example is metaphorical, Grant and Bauer (2004: 49) point out that figurative language (obviously including figurative phrasal verbs) includes all figures of speech, whose meanings are all equally recoverable through ‘taking a conversational untruth and extracting probable truth from it by an act of pragmatic interpretation’ (Grant and Bauer 2004: 50). This quotation can be taken as a definition of figurative language.

It should be stated here that most writers on phrasal verbs, and on idiomaticity in general, do not distinguish between idiomatic meaning and figurative meaning. In Grant and Bauer’s view (Grant and Bauer 2004) there are three categories of “idiomatic” meaning – first, core idioms, which are completely non-compositional, second, phrasal items with one element that is non-compositional, and third, figurative language. But it appears from Siyanova-Chanturia and Martinez (2015), who review the literature on processing of non-literal phrases, that in this field of research, the three categories are usually conflated. Hence in the context of the current research on interpreted language with its concomitant processing constraints, I will refer to “idiomatic and figurative” phrasal verbs.

Celce-Murcia’s third category of aspectual phrasal verbs in not unproblematic either, in involving overlap with the conflated idiom/figurative category. This is the definition of aspectual phrasal verbs in the words of Darwin and Gray (1999: 68):

Whereas the verb proper in aspectual phrasal verbs can be understood literally, the particle contributes meanings, not commonly understood, about the verb’s aspect. For example, up in They ate up all the chips and drank up all the soda signals that the actions are complete.

Aspectual verbs are thoroughly discussed by Side (1990), but many of his examples are figurative in addition to being aspectual, for example took off in his business really took off (Side 1990: 148) derives its meaning not only from the aspectual off meaning departure, but also by analogy with an aircraft taking off.

It is worth commenting on Darwin and Gray’s phrase ‘not commonly understood’, for if a non-native speaker is unaware of the aspectual meaning of particles, aspectual phrasal verbs become in effect non-compositional. On the other hand, in the examples quoted by Darwin and Gray (1999, above), for comprehension purposes the aspectual particle does not need to be understood as the meaning is easily recoverable in the context from the lexical verbs ate and drank. In these examples the potential problems for non-native speakers (henceforth, NNS) deriving from lack of knowledge of the aspectual meaning of up would lie in production, in the inability to produce the pragmatic effect of completion.

Whatever the semantic classification as far as non-literal meaning goes, there is general agreement that non-literal phrasal verbs cause additional difficulty to NNS. Siyanova and Schmitt (2007: 132) cite Laufer (1997), Moon (1997) and Wray (2000) as finding that ‘both teachers and learners find idiomatic multi-word units more difficult than their nonidiomatic counterparts, which is likely to lead to avoidance behaviour’ (Siyanova and Schmitt 2007: 132). A reading of Siyanova-Chanturia and Martinez (2015: 555) demonstrates that these difficulties have been observed empirically, and while the cited studies did not focus specifically on phrasal verbs, their conclusions about idiomatic phrases in general can be inferred to apply to idiomatic/figurative phrasal verbs, which are part of the same category. Hence, out of four studies comparing idiom processing by native speakers (henceforth NS) and NNS, three (Cieslika 2006; Underwood, Schmitt and Galpin 2004; Siyanova-Chanturia, Conklin and Schmitt 2011) reported NNS as processing literal uses of words used in idioms more quickly than they processed the same words used figuratively or idiomatically, while the reverse held for NS, who were shown to gain processing advantages from the use of idiomatic phrases (Siyanova-Chanturia and Martinez 2015: 555).

Since phrasal verbs form an integral part of the formulaic expressions that facilitate fluent production in English, it is worthwhile highlighting them for interpreter trainees and on inservice professional development courses. The best way to do this is to identify and describe situated use of phrasal verbs in interpreted language, produced during SI by experienced professional interpreters. This can be done by identifying the phrasal verbs in a corpus of interpretations, so that they can be presented in their context in materials used in language courses that supplement courses of training of conference interpreters.

3.3 ELF and the survival of the international phrasal verb

A further point concerning phrasal verbs is the issue of what has become known as ELF or English as a Lingua Franca (Jenkins 2001; McArthur 2003; Modiano 2003; Seidlhofer 2007), involving the simplification of English to exclude those features that non-natives find hard to use. If English in Europe is increasingly used as a medium of communication between non-native speakers, and if phrasal verbs are inherently difficult to acquire, does that mean that they are falling out of use in international forums? And that interpreters will therefore use them less? There may be a certain precedent for this in frequencies of phrasal verbs in some ex-colonial varieties of English, such as Indian English, which Schneider, working on ICE (the International Corpus of English) found to be rather low (Schneider 2004: 235). On the other hand, in Singapore, where English is a second language for nearly all of the population, yet is the national language and therefore universally taught to a reasonably high standard, the frequency of phrasal verbs is higher than in British English (Schneider 2004: 235). Additionally, research shows that phrasal verbs are relatively frequent in the international context represented by written EU documents. A study using CEUE (the Corpus of EU English), a 200,000 word corpus of EU documents intended for the general public, found a frequency of 1 every 200 words (Trebits 2009: 276), which is comparable to the frequency of 1 every 192 words found in the BNC (Gardner and Davies 2007: 347). These arguments suggest that, in international contexts where high levels of proficiency are a priority, phrasal verbs not only survive, but thrive. And this in turn suggests that the high level of proficiency expected of interpreters working towards English will guarantee a reasonably high frequency of phrasal verbs in interpreted English.

3.4 Phrasal verbs in interpreted English – hypotheses and questions

Hypothesis 1. On the basis of the arguments in sections 3.1–3.3 above, in other words on the basis of (a) the processing advantages of multi-word expressions of which category phrasal verbs form a part (sections 2 and 3.1), of (b) their expediency in the sense that they are short and quick to say (section 3.1), of (c) their high frequency in the English language as shown by the BNC, and of (d) the finding that they occur frequently in international English as shown by Trebits (2009) (section 3.3), it was hypothesised that phrasal verbs will occur frequently in the English of native speaker and native-like interpreters.

Hypothesis 2. On the basis (a) that non-native speakers find it difficult to acquire phrasal verbs, as shown in section 3.2 above, and (b) of the likelihood that non-native interpreters will respond to the pressure experienced during SI by retreating to more automatised structures analogous to those of their mother tongue (also in section 3.2), it was hypothesised that phrasal verbs will not occur reasonably frequently in the English of non-native interpreters working towards English as a B language. The meaning of “reasonably frequently” is clarified in the next paragraph.

To operationalise these hypotheses, the frequencies of phrasal verbs were measured in two corpora. Hypothesis 1 was tested by measuring the frequency of phrasal verbs in a corpus of English language interpretations produced by interpreters of native-speaker standard, which was called INT-A. Hypothesis 2 was tested by measuring the frequency of phrasal verbs in a corpus of English language interpretations produced by non-native interpreters, which was called INT-B. The notions of “reasonably frequent” and “reasonably frequently” were operationalised by comparison with frequency in the language as a whole, as represented by the BNC. On this basis, if both Hypothesis 1 and Hypothesis 2 were confirmed, then the frequency of phrasal verbs in INT-A would be fairly close to the frequency of phrasal verbs in the BNC, while the frequency of phrasal verbs in INT-B would be much lower than the frequency of phrasal verbs in both INT-A and the BNC. This would be regardless of genres and written/spoken production, on the principle that the structural and semantic difficulties outlined in section 3.2 above are intrinsic to the phrasal verb in itself, independently of the genre or mode (spoken or written) in which it is used.

With the aim of providing descriptions of situated use that could be used in language lessons for trainee interpreters, as mentioned in section 3.2 above, it was decided (in addition to investigating Hypotheses 1 and 2) to use the study in an exploratory way, to find out precisely which phrasal verbs are shown by the data to be used in interpreted English, and for which functions they were used. To help structure an exploration of exactly how the phrasal verbs were used, and to permit evaluation of the role that phrasal verbs might play in furthering the task of the interpreter, reference was made to the concept of the three metafunctions in Hallidayan systemic functional grammar – interpersonal, textual and ideational (Halliday and Matthiessen 2014: 30).

The three metafunctions in Hallidayan linguistics – ideational, interactional and textual – are ‘dimensions of language’ (Halliday and Matthiessen 2014: 30), and as such, they are complementary, and they co-occur, as is clear from the quotations from Halliday and Matthiessen that I will now cite. Halliday and Matthiessen (2014: 30) refer respectively to the ideational and interactional metafunctions as follows:

every message is both about something and addressing someone, and [...] these two motifs can be freely combined – by and large, they do not constrain each other.

They describe the role of the textual metafunction as follows:

this can be regarded as an enabling or facilitating function, since both the others – construing experience and enacting interpersonal relations – depend on being able to build up sequences of discourse, organising the discursive flow, and creating cohesion and continuity as it moves along. This, too, appears as a clearly delineated motif within the grammar (Halliday and Matthiessen 2014: 30).

Complementarity and co-occurrence mean that two or three of the metafunctions can be realised in the same clause. This multifunctionality has an implication for processes of language analysis in Applied Linguistics. Items of contextualised language – words or phrases – should not be assumed to be exclusively associated with any single metafunction, on the principle of mutual exclusivity of categories. The contribution any linguistic item makes to the meaning is dependent on the other linguistic items it is associated with and on the context, and vice versa. For example, Hyland (2005: 59) cites Hood and Forey (1999), who show that in the context of shifting from a positive to a negative judgement, but and however realise interactional function, in addition to the more predictable textual function. Conversely, the same words and phrases used in different structures and different contexts can represent different metafunctions, as Martin (1992:8) shows in a worked example. Hence any judgements made in the current research about the predominant metafunction expressed by a given phrasal verb reflect the researcher’s view of the role of that phrasal verb in that context. The researcher’s judgement of the ideational, textual or interpersonal metafunction of each phrasal verb examined is necessarily subjective, although this subjectivity was constrained by repeated analysis after a month’s interval, which resulted in an index of consistency of 98 per cent.

One reason here for distinguishing in the data between the three metafunctions is that it permits evaluation of the potential value of the effort taken to learn a particular phrasal verb, through distinguishing between interpreting process, which is predictable, because it recurs, and interpreted speaker’s translated product, which is unpredictable. I will argue that phrasal verbs with textual and interpersonal metafunction are used as part of the interpreting process, and that there is a good probability that the contexts where they are used will recur; these verbs are worth practising to make them automatically retrievable. On the other hand, with phrasal verbs with the ideational metafunction, the probability of their recurrence is a function of the extent to which the interpreter is or is not always dealing with the same subject matter – and for interpreters of both parliamentary debates and of conferences, subject matter is inherently variable. This distinction is subject to the qualification that some aspects of ideational metafunction, such as some mental processes or some material processes, are likely to recur in repeated contexts and are likely to prove productive.

The need for vocabulary to express textual metafunction is somewhat more predictable. As Setton describes it, the re-creation of textual structure is part and parcel of the process of simultaneous interpretation:

relevance (coherence) is sought as a matter of routine. Context is seen as a nested set of assumptions: a background about the world and the situation, then assumptions based on previous discourse, on the previous utterance, and finally those taking shape from the utterance-initial cues about the illocution or modality of the ongoing argument […]” (Setton 1998/2002: 193).

So it is clearly worthwhile identifying phrasal verbs used with textual metafunction, describing their exact functions, and presenting them to trainees as high priority lexical items.

Similarly, the interpreter’s function as mediator implies that any phrasal verbs used with interpersonal metafunction are likely to prove useful on repeated occasions; as with phrasal verbs used with textual metafunction, it is worth describing the exact functions to present to trainees. Hence identifying metafunctions points the way to phrasal verbs used in the interpreting process, which constitute high priority lexical items for inclusion on a language syllabus for trainee interpreters.

On the basis of the probability that they are likely to be experienced at textualisation towards English, and at the same time are likely to take advantage of the processing advantages of use of phrasal verbs as outlined in sections 2 and 3.1, it was predicted that the native or native-like interpreters of INT-A would tend to use phrasal verbs for textual metafunction. On the basis of the difficulties that non-natives have in producing phrasal verbs as outlined above, and on the basis that non-native interpreters have relatively little experience of textualisation towards English, it was predicted that non-native interpreters working towards English as B would tend not to use phrasal verbs with textual metafunction. Expressed as a hypothesis, this is

Hypothesis 3. Native or native-like interpreters will use phrasal verbs with textual metafunction more frequently than non-native interpreters.

On a similar basis of the likely experience of the native or native-like interpreters working B to A of mediating aspects of interpretation, combined with their knowledge of ready-made phrasal verb expressions on the one hand, and on the other hand on the basis of the difficulties non-natives have in producing phrasal verbs combined with the added processing load of the A to B interpreter, the following hypothesis was formulated.

Hypothesis 4. Native or native-like interpreters will use phrasal verbs with interpersonal metafunction more frequently than non-native interpreters.

The final Hypothesis relates to the semantic processing difficulties highlighted in section 3.2.

Hypothesis 5. Given the evidence of processing difficulties experienced by NNS with idiomatic/figurative phrases, and of processing advantages experienced by NS, as cited in section 3.2 above, it can be hypothesised that non-native interpreters working towards their B language will use idiomatic/figurative phrasal verbs less often per 1000 words than native or native-like interpreters.

4. The Corpus Data

Two corpora of interpreted English, which I called INT-A and INT-B, were compiled out of several pre-existing corpora. The objective of INT-A was to represent the interpreted English of established professional interpreters, who should be either native speakers or native-like. It was made up of texts from three sources. The first two were corpora based on proceedings of the European Parliament, and consisted of interpreting towards English (from Italian and Spanish) taken from the EPIC corpus (Monti et al. 2005; Sandrelli, Bendazzoli and Russo 2010), plus all the interpreting towards English (from unspecified languages) in 2249, a corpus of all the English spoken in the European parliament on a single day (see Aston 2017), the non-interpreted sections from which had been removed on the basis of the tags. The third source was the DIRSI corpus of conference interpretations (Bendazzoli 2010), which contributed all of its native-speaker-interpreted English, all of which took place at two conferences about cystic fibrosis (Bendazzoli 2010: 155). The interpreters in the texts contributed by EPIC and DIRSI were tagged as native speakers, but as there was no information about the interpreters in 2249, it was assumed that they were (a) established professionals and (b) either native speakers or native-like – and a reading of the interpretations in 2249 to evaluate their fluency and accuracy provided no evidence to counter this supposition. The total number of running words in INT-A was 85,891.

The INT-B corpus was intended to represent NNS interpreter language. It was made up of all those transcripts of interpretations towards English in both the DIRSI corpus and the EPIC corpus that were tagged as being done by non-native interpreters. All the texts from DIRSI were translated into English from Italian, and were taken from the same two conferences about cystic fibrosis that supplied the native speaker material mentioned above, plus a conference about social care policy (Bendazzoli 2010: 155). The total number of running words in INT-B was 24,122. It is regrettable that INT-B was so much smaller than INT-A, but this reflects the easier availability of A-direction transcripts.

The transcription of all the words in all the component texts of both INT-A and INT-B was orthographic, and word for word, meaning that features such as reformulation were maintained. In 2249, punctuation was added, as it would be in writing. On the other hand, the transcription of all the texts taken from EPIC and DIRSI was without punctuation, with the speech divided into “units of meaning” using double slash marks (for further details, see Monti et al. 2005, section 2; Bendazzoli 2010: 184ff).

5. Interrogating the Data

Information relative to the research questions was uncovered by making KWIC concordances of phrasal verbs using Wordsmith Tools Concord, first version 6, and later version 7 – see Scott (2012/16), and by comparing the frequency data obtained with the data about phrasal verbs in the BNC in Gardner and Davies (2007). There are various definitions of the term “phrasal verbs” (for a review, see Darwin and Gray 1999), but for the sake of simplicity I followed the definition of the term in Gardner and Davies (2007), which is a simplified version of the exhaustive definition in Biber et al. 1999: 405ff). To paraphrase Gardner and Davies (2007: 344), my definition of a phrasal verb is a lexical verb followed by an adverbial particle, with meaning derived from a combination of the two.

The phrasal verb forms were retrieved from the data by searches for each of the following particles, which were those used in the searches of the BNC by Gardner and Davies (2007):

out/up/down/back/off/round/along/over/around/on/through/about/in/under/by/across.

The searches were carried out first in INT-A, then in INT-B. Each concordance line was then examined, first to eliminate instances of verbs and prepositions in free combination, and secondly, to eliminate prepositional verbs (for these two categories, see Biber et al. 1999: 405ff).

To verify whether multiword verbs in concordances were phrasal (and hence eligible for inclusion in the results) or prepositional (and thus excluded from the results) I used the following operational procedures, based on syntactic descriptions from Biber et al. (1999: 405ff):

A. is the verb+particle unit used transitively or intransitively? If – as in example (1) – the answer is “intransitively” (i.e. if the following noun answers the question where or when), then the unit is a phrasal verb, the particle being clearly adverbial, since prepositions need to be followed by noun phrase objects that answer the questions who or what – as in (2).

(1) I think the law will go through. Intransitive – so adverbial.

(2) We should go through these results before the presentation. Transitive use.

B. If the verb+particle unit is used transitively, can it be used with a pronoun object placed between the lexical verb and the particle? If the answer is no, as in (3), the unit is a prepositional verb and is excluded from the results.

(3) We should go them through X

If the answer is yes, as in (4), the unit is counted as a phrasal verb, and included in the results.

(4) It was only the whiskey that saw him through.

Other points on the selection of which forms were phrasal verbs are that as only lexical verbs are included in the definition, combinations with be were excluded, for example, be out (= be in the public domain). On phrase length, which is relevant because phrasal verbs can have noun phrase objects between verb and particle, only 2–4 word phrases were considered, following Gardner and Davies (2007: 345). Gerunds, for example by setting up, were classified as verb forms not noun forms, as they are in the BNC. Phrasal-prepositional verbs were counted as phrasal verbs, because they follow the Gardner and Davies (2007: 344) definition of a verb followed by an adverbial particle. Thus come up with was counted as the phrasal verb come up.

Once the above syntactic criteria were fulfilled, a phrase was accepted as a phrasal verb independently of any notional lists of canonical phrasal verbs, on the principle that phrasal verbs are an open-ended, productive category in which the standard adverbial particles can combine with new lexical verbs to create new meanings (Side 1990: 146).

Once it had been verified that what remained in the concordances were phrasal verb forms only, the frequencies of each form were counted and entered on a spreadsheet. The frequencies were subsequently additionally grouped by lemma, to facilitate comparison with the frequencies in the BNC presented by Gardner and Davies (2007), which were taken to represent frequencies in the language as a whole. The lemmas were also ranked by frequency. For the metafunctional analysis, the concordances were merged and then classified according to the three Hallidayan metafunctions mentioned above.

When it was necessary to find the frequency in the BNC of a phrasal verb that was not listed by Gardner and Davies (2007), the online BYU-BNC (Davies 2004) was consulted, with a collocation search starting with the POS-tagged adverbial particle and looking for the lexical verb within a span of three words to the left. In this way, the phrases containing phrasal verbs found were two to four word phrases as they were in Gardner and Davies (2007).

For Hypothesis 5, phrasal verbs were classified as idiomatic/figurative if the meaning of the whole phrase was figurative, or if the meaning of the whole phrase was non-compositional in the sense that the meaning could not be deduced from summing the literal meanings (as defined in section 3.2 above) of the verb and the particle. If the meaning was entirely literal, they were classified as literal. With aspectual phrasal verbs, if figurative/idiomatic they were classified as figurative/idiomatic, and were classified as aspectual only if the meaning of the verb was literal, with aspectual use of the particle. The final test for literal meaning was any of the meanings of the individual word (verb or particle) listed in a contemporary dictionary (Cobuild 2001), according to Grant and Bauer’s (2004) method as outlined in section 3.2 above. To help ensure plausibility of the classification, the analysis was repeated after a month’s interval, with an index of agreement of 0.97.

6. Results and discussion

The INT-A corpus was used to make observations about the comparative frequency of phrasal verbs in interpreted language, while both corpora were used for the functional analysis and the analysis of literality. But before proceeding to report the results, I would like to clarify the statistical approach I have adopted to these comparisons.

When using mathematics in research, it is as well to consult a well-informed source for the most appropriate method. In corpus linguistics research, Kilgarriff, author of a review of statistical tests for corpus comparison (Kilgarriff 2001) published in The International Journal of Corpus Linguistics, constitutes such a source. As Kilgarriff has not only pointed out but also demonstrated empirically (in Kilgarriff 2005), in corpus frequency comparison the fact that language is a matter of choice, and is therefore not random, presents problems for statistical hypothesis testing (Kilgarriff 2005: 264). Hypothesis testing depends on disproof of the null hypothesis that there is only a random relationship between two sets of data (Kilgarriff 2005: 263). Clearly, if the language occurring in two different corpora is by definition not random, then to demonstrate that it is not random is meaningless – albeit one uses impressively sophisticated mathematics for this demonstration (Kilgarriff 2009). For this reason, in the research reported in this paper, the mathematics used to compare frequencies will be that of normalised frequencies, as Kilgarriff (2010: 3) recommends, without the use of chi-square, log likelihood, or any other probability test.

6.1 Phrasal verbs in interpreted English - frequency

The first finding, on the overall frequency of phrasal verbs in interpreted English, was that phrasal verb tokens were less frequent in INT-A than in the BNC. In the 100 million word BNC – as mentioned above – there is one phrasal verb every 192 words (Gardner and Davies 2007: 347), whereas the 364 tokens in INT-A work out as one phrasal verb approximately every 236 words. This raises the question of why, given the potential usefulness of phrasal verbs to interpreters, the frequency should be lower in interpreted English rather than higher. The logical explanation is the size factor. The hugely greater size of the BNC relative to INT-A means that it represents a much wider range of contexts. For example, from the point of view of mode, the BNC includes dialogue, whereas INT-A is limited to monologue. From the point of view of field, INT-A contains mainly parliamentary subject matter, whereas the BNC samples many different fields. From the genre angle, the BNC also contains many examples of texts exemplifying different literary genres, as well as journalism, both of which are absent from INT-A. In particular, it seems likely that the literature and journalism together with the dialogue present in the BNC will present narrative and interactional contexts of use of a large number of phrasal verb lemmas, whereas those same contexts of use are absent from the discourse sampled in INT-A.

A rate of one phrasal verb every 236 words is quite frequent, nevertheless. Assuming that interpreters towards English speak at 132–136 words per minute (Bendazzoli 2010: 149, 152), and taking the average of 134, it means interpreters were using on average one phrasal verb token approximately every 106 seconds; looked at another way, this means 136 phrasal verb tokens over four hours of interpreting. This is certainly a frequency high enough to confirm Hypothesis 1 (in section 3.4 above), that there would be a reasonably high occurrence of phrasal verbs in English produced by English native-speaker and nativelike simultaneous interpreters.

In terms of types, INT-A contained 131 phrasal verb lemmas. Table 1 shows the most frequent twenty-two, together with raw and normalised frequency (per million words), and, for the BNC, the frequency per million and the rank order of frequency. Additionally, the Table shows the ratio of keyness, obtained by dividing the normalised frequency of INT-A by the normalised frequency of the BNC. The keyness procedure is recommended by Kilgarriff (2010) as the most effective mathematical way to uncover differences between corpora.

Table 1: The top 22 phrasal verbs in INT-A

As the rightmost column of Table 1 shows, fifteen of the most frequent phrasal verbs in INT-A also occurred in the Top 100 in the BNC as reported by Gardner and Davies (2007: 358). But Gardner and Davies’ “top 100” is anomalous – it includes only phrasal verbs that contain one of the top twenty lexical verbs making up phrasal verbs, as contained in Gardner and Davies’ Table 5 (Gardner and Davies 2007: 350). A search of the BYU-BNC shows that some phrasal verbs with lexical components outside the “lexical top twenty” are more frequent than the hundredth ranking phrasal verb in Gardner and Davies (2007), which has a frequency per million of 4.23 (see Gardner and Davies 2007: 359). Such verbs should therefore be considered as particularly frequent alongside the “top hundred”. The f/million/BNC column of Table 1 shows that there are five such phrasal verbs in INT-A – draw up, open up, send out, sort out and end up. This means that a total of twenty out of twenty-two of the most frequent phrasal verbs in INT-A were among the most frequent in the BNC too. This in turn suggests that the phrasal verbs most often used in interpreted language are mostly used frequently in the language as a whole, although the differences are perhaps more interesting than the similarities. These differences are shown mathematically in Table 1 in the ratio of keyness column. The higher the ratio, the more characteristic the phrasal verb of interpreted English as represented in INT-A. According to this measure, the ten most characteristic verbs, in descending order, are bring on, move on, send in, send out, open up, draw up, move in, bring in, and come up (these verbs will be further discussed in sections 6.5 and 6.6). Examination of the contexts of some of the verbs that are ranked the highest in keyness in INT-A can help to define what marks out interpreted English, and to provide clues about formulae that are proven to be of use to interpreters. I will return to this topic in section 6.5.

6.2 Phrasal verbs in non-native interpreters: frequency

In INT_B, there were 59 phrasal verb tokens, representing one phrasal verb token every 409 words,

which means that the non-native interpreters used phrasal verbs as a category much less often than the interpreters of INT-A, and much less often than in the BNC, which confirms Hypothesis 2, and this in turn confirms the theory that processing difficulties make the choice of phrasal verbs by non-native interpreters less likely. It seems that, in the pressure of the SI situation the non-native interpreters had no time to search for phrasal verb items – items that, although probably known in theory, were hard to access in memory, probably due to their having being encountered only occasionally.

As far as individual phrasal verbs go, only 12 lemmas were used more than once. There was a considerable difference in the phrasal verb lemmas that were most frequent in INT-A and INT-B, as can be seen from Table 2, which shows the top 25 verbs in INT-A together with the rank orders, raw and normalised frequencies (per million) of the same verbs in INT-B, and the rank and ratio of keyness of the verbs in INT-B relative to the BNC. Only verbs with a ratio of keyness over 1.0 are ranked, as only these are statistically characteristic of INT-B relative to the language as a whole. All phrasal verbs in INT-B with a frequency of >1 are included in the Table.

Table 2: Comparison of phrasal verbs in INT-A and INT-B

Of course, it is easy to object to the finding that there was a lower frequency of phrasal verbs in INT-B than in INT-A on the grounds that the frequency differences are explicable in terms of the subject matter translated in INT-A being inherently more suitable for translation by phrasal verbs than the subject matter translated in INT-B. This can be tested through examination of one case, the material in DIRSI that is collected from two conferences on cystic fibrosis. Part of this material was translated into English by a NS interpreter, and part by NNS interpreters. Over 8436 words of text, the NS used 21 phrasal verb tokens, making one phrasal verb every 401 words on average, and covering 19 lemmas. The NNS interpreters used 22 phrasal verb tokens in 14.435 words, which makes one phrasal verb every 656 words, and covered 9 lemmas. At one every 401 words, the NS interpreter uses phrasal verbs less frequently than they are used in INT-A as a whole (see the first paragraph of section 6.1), so this would seem to confirm that discussions about cystic fibrosis do seem to require phrasal verb translations relatively seldom. On the other hand, this makes no difference to the relative frequency of phrasal verb use in the two corpora – frequency remains much higher in INT-A than it is in INT-B. So while it remains the case that the conference topic could have an effect, there is nevertheless a difference between the corpora, in terms of lower frequencies of phrasal verbs in INT-B as compared to INT-A . It is risky to restrict explanatory factors to one. So, given the evidence presented in section 3.2, the greater processing difficulties phrasal verbs present to non-native interpreters would seem to be a good candidate for an additional explanation for the lower frequency of phrasal verbs in INT-B.

6.3 Metafunctional distribution

To uncover information relating to Hypotheses 3 and 4, about phrasal verbs used for textual and interpersonal metafunctions, each phrasal verb concordance line was classified according to the metafunction that it predominantly expressed. The distribution of phrasal verb occurrences across the metafunctions in the two corpora is shown in Table 3.

Table 3: distribution of phrasal verbs across metafunctions

Table 3 shows that there was a substantial representation of all three metafunctions in the INT-A corpus. In the INT-B corpus, all three metafunctions were represented, but the proportion of phrasal verb tokens used for textual metafunction was lower than in INT-A. The data therefore confirmed Hypothesis 3, thus in turn confirming that native and nativelike interpreters use phrasal verbs as a resource to textualise their interpretations. Conversely, the data in terms of the much lower frequency of phrasal verbs used with textual metafunction in INT-B suggests that in tending not to use phrasal verbs when constructing the ongoing sense of their interpretations, non-native interpreters are neglecting a valuable resource.

For phrasal verbs used with interpersonal metafunction, the situation was reversed. The non-native interpreters used phrasal verbs interpersonally more often than the native speaker interpreters, so that Hypothesis 4 was not confirmed. It cannot entirely be discounted that the differing proportions of different types of message being translated (for example, allocation of seating, transitions of speaker, management of latecomers – which are more likely to involve mediation) may have influenced the disparity in frequency of interpersonal phrasal verbs between the two corpora. However this cannot reliably be checked, since there was no information of this type in 2249, which formed a substantial component of INT-A. An alternative explanation is that the non-native interpreters have substantial experience of phrasal verbs used interpersonally, as a result of having passed through courses of English language conducted using communicative language teaching methods. Therefore the assumption that phrasal verbs would be hard to access for non-native interpreters during SI does not apply in the case of interpersonal contexts.

In sections 6.4 to 6.7, there is a detailed examination of examples of phrasal verbs used for the three metafunctions, which reveals the prominent role carried out by phrasal verbs in the work of the interpreter. The examples also demonstrate the functions that the phrasal verbs are being used for.

6.4 Mediation using interpersonal phrasal verbs

When used with interpersonal metafunction, phrasal verbs were used to expedite the interpreter’s role of mediator, in helping to make a third party do something or feel something. The following examples from INT-A show how phrasal verbs express the process of mediation:

(5) I will give the floor back to you.

(6) I could hand out some of the tables with the relevant information

(7) I’d like to thank you for … the fact that you've stuck it out this evening.

(8) could you please come in and close the door.

As Table 3 shows, phrasal verbs with interactional metafunction were relatively well represented in INT-B, with examples like the following:

(9) please hand back the headset and receivers for the interpretation to the desk at the entrance

(10) just five minutes while filling in the questionnaire and pick up the devices the headsets

All of these examples show how phrasal verbs are used by interpreters in their integral role of mediator, to expedite event proceedings. It is interesting to note that phrasal verbs are used in this role in both types of event, parliamentary debates (5, 6, 7) and conferences (8), represented in INT-A, as well as in the conferences represented in INT-B (9 and 10).

6.5 Textualising phrasal verbs in interpreted language

Phrasal verbs used with textual metafunction are also indexical to the interpreter’s role, being involved in the interpreting process in the sense of Setton’s account of online meaning assembly, which sees simultaneous interpreters as involved in a continuous process of simulating and recreating context:

relevance (coherence) is sought as a matter of routine. Context is seen as a nested set of assumptions: a background about the world and the situation, then assumptions based on previous discourse, on the previous utterance, and finally those taking shape from the utterance-initial cues about the illocution or modality of the ongoing argument […] (Setton 1998/2002: 193).

The way that phrasal verbs convey the process of creation and articulation of textual meaning in SI is shown in examples (11) to (19) from INT-A. Here is an account of the rhetorical function of each example.

In examples (11) and (12) , the interpreter is indicating to herself and to the hearers that between propositions a relation of evidence should be perceived; in (13), (14) and (15) there is a direction to perceive a relation of presentation; in (16) there is the perception and relaying of a relation of recap; in (17) with both the initial choice focus down and the reformulation narrow down there is the evocation of a relation of elaboration involving a perception of movement from the general to the specific; and (18) and (19) are relations of resumption [1].

(11) The + data can all be documented, to back up what I said.

(12) this analysis has in fact been borne out by a number of different contributions

(13) That brings us on to the report, by Mrs Lucas

(14) we're kicking off on the basis of an assessment report

(15) We move on to the report by Mr Sakalas.

(16) And that brings me back to what I was saying at the very beginning

(17) this allowed us to focus down to narrow down the field to

(18) Now we'll go on with the er debate

(19) Well, I would like to er pick up on the point on the credit rating agencies.

Bring on is the highest ranking phrasal verb by keyness in INT-A, while move on ranks second in keyness in INT-A (see Table 1 above), and first in INT-B (see Table 2). Six of the seven occurrences of bring on in INT-A express the textual metafunction, as do fifteen of the twenty-one occurrences of move on – so these would seem to be two instances of relative prominence of a phrasal verb in INT-A which is explicable in terms of aspects of the process of interpreting, in this case textualisation. 

Phrasal verbs used for the textual metafunction were much less common in INT-B. However, the two most key phrasal verbs in INT-B (see Table 2) were used with textual metafunction. These were move on, signalling the presentation relation, and sum up, used to indicate a relation of summary. It is thus clear that textualisation of output was taking place explicitly in INT-B, as one would expect. Given this, it could be that different types of conference situation provide the explanation for the lower frequency of textualising phrasal verbs in INT-B; in INT-A, there is a preponderance of parliamentary debates, whereas in INT-B there is a preponderance of conferences with the function of updating members of a specific professional community. But given that message construction with attendant textualisation should be taking place in both contexts, it is equally plausible to explain the low frequency of phrasal verbs with textual metafunction in INT-B through a lack of experience on the part of the non-native interpreters of textualising in English, or simply through the pressure of the SI situation, both of which plausibly result in the signalling of textual structure through structures transferable from the mother tongue, rather than through phrasal verbs.

6.6 Ideational phrasal verbs in interpreted language

The ideational metafunction, by virtue of its application to content, can be stated to cover interpreted product. Phrasal verbs used for the ideational function in INT-A cover a wide range of types of meaning. As indicated above, some phrasal verbs used with ideational meaning are likely to recur whatever the subject matter being translated, such as certain mental process verbs, while others will be linked with a more restricted, specialised context. The discussion is organised according to verb types – mental processes, material processes, and relational processes. The degree of productive value for interpreters of certain phrasal verbs, and of the multiword expressions of which they form part, as revealed by the corpora, is discussed within these sections. Of course verbs cannot intrinsically be placed in a single one of the three categories of mental, material and relational processes. I classified them as belonging to one of the three categories on the basis of the type of meaning that they predominantly expressed in each single instance of use in context actually observed in the KWIC concordance.

Starting with mental processes, some of the more frequent phrasal verbs in INT-A form components of collocations or slot and filler patterns that extend the phrasal verb into longer multiword expressions which are likely to be formulaic sequences particularly worth learning for non-native interpreters. There is for example the phrasal-prepositional verb come up with, which occurred as many as fifteen times, thus contributing to this verb’s position as number ten for keyness in INT-A (see Table 1). Come up with collocated relatively frequently (seven times) with proposal (example 20).

(20) we asked the Commission to come up with a proposal to harmonise the guarantee.

At a more general level of semantic prosody, a slot and filler pattern can be observed, which is come up with + noun phrase describing something viewed positively in the context. Thirteen of the occurrences of come up with fall into this category. A further specification for this particular formulaic sequence is that the positive semantic prosody is reflected in ten cases out of fifteen by collocates that are positive adjectives (excellent, good, speedy, flexible) – see (21).

(21) we’ve been able to come up with a very pragmatic and flexible proposal.

Although the mental nature of the process represented by come up with would suggest it would occur in a wide range of situations, in INT-A it only occurred in parliamentary debates. This is almost certainly a reflection of the small proportion of INT-A (about 10 per cent) represented by conference interpreting. It is true that come up with, through its association with solutions to problems, is closely associated with parliamentary process; but it is also associated with idiom, so it should occur fairly frequently in conferences, given that they are often held for the purpose of exchanging information about innovations. This unexpected lack of instances reflects the need for bigger samples of conference interpreting, to produce more representative results.

Moving on to material processes, carry out was also observed to combine in INT-A with collocates in formulaic sequences, notably with analysis (three times), checks (three times), a study/studies (three times), actions (twice) and noun phrase modifier + activities (twice) – for example, carrying out the reception activities). This was one of the few verbs to occur really frequently in INT-B (12 occurrences), where it was used with study/studies (twice), demonstrations and research. This verb was used in both the parliamentary debates and the conference speeches, so it would seem to be particularly productive for interpreters, in the sense of having a wide application.

The material process verb set up was also fairly frequent in both INT-A and INT-B. In INT-A there are few indications of extended multiword expressions, with mechanism(s) being the only collocate of set up that recurs (twice). There were no repeated collocations in INT-B, though the contexts of use were standard (access, laboratory, model), so that one can infer that in a larger corpus of non-native interpreting, some recurring collocations would have emerged.

The material process verbs draw up and open up, as shown by their contexts viewed in the KWIC concordances, are productive in the context of interpreting in the sense that they refer to content that interpreters in international institutions are likely to repeatedly experience and then produce. The noun collocates of draw up in INT-A are law (twice), regulation, list, plan and agreement. All of the occurrences were in parliamentary debates, and closely reflect the role of parliament. It is therefore unsurprising that draw up did not occur in INT-B, which is almost entirely composed of medical conference interpretations. The association of draw up with parliamentary procedure explains why it ranked as high as sixth in terms of keyness in INT-A as compared with the BNC (see Table 1 above). The noun collocates of open up are mostly (nine times) the word market or synonyms of market, reflecting the current preoccupation of the European parliament with market-led economics, which in turn explains the number five ranking of this item in the INT-A keyness scale (see Table 1). Open up was absent from INT-B, which is unsurprising given that almost all of the subject matter is concerned with social care and health matters rather than economics. But it is surely uncontroversial to claim that, for as long as the hegemony of market economics persists, open up + market(s) will be a productive multiword pattern for non-native interpreters to learn to automatically produce.

Relational processes are quite frequently conveyed in INT-A through phrasal verbs, for example end up (four times), bring about (four times), make up – meaning constitute – three times, and bring in (nine times). One formulaic sequence was discernible in the case of bring in, which is its collocation with new (three times), and, more generally, there was an association with the semantic field of innovation (eight times), involving adjectives like extraordinary and nouns like improvement and advances. Make up (=constitute) also occurred three times in INT-B, with standard contexts of use (e.g. an audience made up of friends), but none of the collocations was repeated.

So far the ideational phrasal verbs reported on here are shown by their presence in the BNC top 100 to be reasonably frequent in the language as a whole. But (as Table 1 shows, and as mentioned in section 6.1) there is also the question of the keyness of some items, in other words of their conspicuously high normalised frequencies in INT-A as compared with the BNC. Among these are bring on, move on, come up, draw up and open up, whose keyness has already been discussed. For the remaining members of the top ten of keyness in INT-A (see Table 1), send in, send out, put in, move in and bring in, it would be tempting to claim that their higher frequency in INT-A makes them typical of interpreted English. However, there is nothing in the contexts revealed by the concordances to support this, and all of the occurrences are from parliamentary debates. So it is more realistic to attribute the higher frequencies of these phrasal verbs to the fact that the contexts where they were the appropriate translations happened to have occurred in parliamentary business in the debates concerned.

It is interesting to note that there is a reasonably wide variety of ideational phrasal verbs in INT-B, with 16 different types represented. This seems to suggest that the non-native interpreters are not inherently shy about using phrasal verbs for translated subject matter, albeit the figures show that they use them less frequently than native speaker and native-like interpreters. It is also in contrast with the lack of variety (only two types, sum up and move on) of phrasal verbs used in INT-B for the textual metafunction. Together with the low number of tokens, this suggests a precise reason why so few phrasal verb types and tokens with textual metafunction are used by non-native interpreters. This reason is that textualisation is the aspect of interpreting over which interpreters have most control. By contrast, for ideational function (i.e. content) the context set by the original speaker is probably steering the interpreter into the phrasal verbs associated with it. In this respect, with ideational function, non-native interpreters appear to be behaving like native and native-like interpreters, in using phrasal verbs as a processing resource, rather than experiencing them as a processing difficulty (though, the frequency figures suggest, with non-native interpreters this happens much less often). But in the textualisation process, with the creation of the textual framework the responsibility of the interpreter, the non-native interpreters are probably automatically falling back on textualising structures analogous to L1, and are thus avoiding phrasal verbs in the process. This is speculation, of course, but it would be interesting if some research could be designed to test this idea.

6.7 Idiomatic/figurative and aspectual phrasal verbs in INT-A and INT-B

The data on literal, idiomatic/figurative and aspectual phrasal verbs in INT-A and INT-B is summarised in Table 4. The verbs listed by name in the Table are the most frequent in each category. An asterisk indicates that the verb is neither present in Gardner and Davies’ BNC top 100 nor has a BYU-BNC frequency high enough to be included in it, in other words a frequency of 4.23 per million, that of Gardner and Davies’ hundredth ranked item. The data confirmed Hypothesis 5, which predicted that non-native interpreters working towards English as their B language would use idiomatic and figurative phrasal verbs less than do native and native-like interpreters. In INT-A, there were 2200 figurative/idiomatic phrasal verb tokens per million words, while in INT-B there were only 1200 tokens per 1000 words. These data reinforce the theory that processing difficulties make idiomatic and figurative phrasal verbs difficult to access for non-native interpreters working towards B.

Table 4 shows that, as might be expected in the circumstances, the number of different lemmas used with figurative/idiomatic meaning was much higher in INT-A at 75, while in INT-B the total was 8, which is surprisingly low even considering that the number of words in INT-B was less than a third of the number of words in INT-A. Indeed, in INT-B 12 of the 29 figurative/idiomatic tokens are accounted for by a single lemma, carry out, and all but one figurative/idiomatic lemmas are inherently frequent, as shown by their having a frequency per million in the BNC above the threshold of 4.23. This suggests that access to individual figurative/idiomatic phrasal verbs during B-direction interpreting is likely only if frequent experience has brought them relatively near to the surface of memory. Conversely, the finding that figurative meanings of phrasal verbs less frequent than the 4.23 per million threshold are hardly used at all by the B-direction interpreters (only a single occurrence of spring up – see Table 4) has a plausible explanation. It is likely that, given the time lag shown in the literature for NNS access to non-literal items, less frequent phrasal verbs (even though they may be known) take too long for the non-native to access in the SI situation.

Table 4: Idiomatic/figurative, aspectual and literal phrasal verbs

In contrast, inherently less frequent non-literal phrasal verbs did occur in INT-A, with the presence (mostly as one-off occurrences) both of non-compositional lemmas like beef up, crop up, kick in and stump up, and of lemmas used figuratively, such as chime in, float around, iron out, thrash out and whittle down, to name some examples of verbs found on the BYU-BNC to have frequencies lower than the 4.23 per million threshold. This occurrence of not particularly frequent idiomatic and figurative items in the INT-A corpus is another way in which the corpus shows that use of phrasal verbs in interpreted language resembles the use of phrasal verbs in the language as a whole, and is consonant with the theory that NS and native-like interpreters, like English speakers in general, derive processing advantages from the use of such idiomatic/figurative phrases.

The figures for aspectual phrasal verbs follow a similar pattern, with frequencies of aspectual tokens per million words much higher in INT A (600) than in INT B (100). Once again, the variety of lemmas was disproportionately lower in INT-B, with only two lemmas used with aspectual meaning in INT-B as against twenty-five in INT-A. The suggestion here once again is that phrasal verbs with aspectual meaning do not come readily to the mind of the B-direction interpreter engaged in SI; and that a phrasal resource that is used to advantage by native and native-like interpreters, seems plausibly to constitute a difficulty, to be circumvented, for the non-native interpreter.

7. Conclusions

In this article I have set out to make the case for the teaching of phrasal verbs to trainee interpreters. The case rests on six findings.

1.  The observation of reasonably frequent use of phrasal verbs by native and native-like interpreters in INT-A.
2. The observed use of phrasal verbs by the native and native-like interpreters in INT-A for the textualisation that is a crucial aspect of the SI process.
3. The observed use of phrasal verbs for interpreter-mediated interpersonal functions.
4. The observed use of phrasal verbs for mental processes which are likely to recur in situations where interpreting takes place.

Findings 1–4 are intended to show that phrasal verbs are part of the linguistic fabric of professional SI. Findings 5 and 6 concern non-native interpreters.

5. There was confirmation of the hypothesis that processing difficulties in SI would lead to a lower frequency of phrasal verb use by the non-native interpreters of INT-B.
6. There were very much lower frequencies in INT-B of phrasal verbs with textual metafunction, which demonstrate a linguistic lack, and a consequent need for learning and teaching (though there appears to be less need for instruction on phrasal verbs with interpersonal metafunction).

It is conceded that the small size of the INT-B corpus, and the restriction of the corpus to only part of three conferences with just a small section of parliamentary interpreting, must make finding number 5, that there is a relatively low frequency of phrasal verbs by non-native speakers, a tentative one. The inclusion of different subject matter, forming a larger sample, might have revealed more occasions when opportunities were taken up by interpreters for the choice of phrasal verbs in translations. Finding number 1, that there was a reasonably high frequency of phrasal verbs in INT-A, should also be received with a degree of caution, due to the low proportion of conference interpreting and the preponderance in the data of parliamentary debates. Overall, however, the results found in both INT-A and in INT-B are in accord with the theories that propose that native speakers find processing advantages in phrasal verbs while non-natives find processing difficulties, and for that reason it seems likely that replication of the research with a larger and more varied non-native interpreter corpus would result in similar findings.

The lower frequencies of phrasal verbs in the non-native interpreters corpus suggests that there is room for improvement in phrasal verb knowledge if non-natives are to more nearly approach native or native-like standards in terms of readiness of automatised formulaic sequences involving phrasal verbs during meaning assembly when working from A to B. Similarly, a need for improved knowledge is indicated by the low frequency of non-literal phrasal verbs among the non-native interpreters, which if translated from production to reception, would mean the risk of non-comprehension when working from B to A. The literature suggests that, partly because of the large number of non-literal phrasal verbs, acquisition cannot be left to simple experience, even when residence in an English speaking country is involved (Siyanova and Schmitt 2007), and this in turn suggests that for improvement in knowledge of phrasal verbs to take place, there must be some form of designed instruction, preferably before trainees begin interpreting work, and preferably involving attentional processes (Schmidt 1990). This form of instruction cannot take place in interpreting practice classes, precisely because attention there is perforce directed away from linguistic form; so it should take place in adjunct language support courses. Material for such courses has been uncovered in the current study, where analysis of the KWIC concordances revealed a number of longer multiword formulae (phrasal verbs +…) useful for interpreted language. Some of the concordances have already been converted into learning materials and used with trainees at the Department of Interpreting and Translation (University of Bologna at Forlì). The hope is that such work will continue, and that our understanding of the role of formulaic language in the process and product of interpreted English can be extended by the development of corpus research methods. This should include more representation in corpora of conference interpreting, to take its place alongside existing corpora of interpretation of parliamentary debates. Work also needs to be done on developing larger corpora of non-native interpreters. The small size of the non-native interpreted language corpus used in this research has meant that the findings presented here must be qualified as merely tentative. But this is no reason to disregard them, for progress in corpus research on interpreted language has to start somewhere, and in any case, it is unrealistic to expect this progress, particularly in a relatively unexplored field, to be anything other than gradual and incremental.

About the author(s)

Andy Cresswell has taught English Language and Linguistics in further and higher education in the UK and Italy. He has studied Sociology, English Literature, Education, and Applied Linguistics, and holds a Ph.D from Reading University, UK. His research interests are academic writing, discourse analysis, corpus linguistics, phraseology, spoken fluency and advanced learner pedagogy, with specific reference to pre-service interpreters and translators.

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

•  Previous article
• Next article