Efficiency in ELF Communication: From Pragmatic Motives to Lexico-grammatical Innovation

The considerable demographic shift in the use of English worldwide, with the effect that L2 speakers outnumber L1 speakers, particularly as typified in English as a Lingua Franca (ELF) communication, is now widely acknowledged in Applied Linguistics. To a certain extent the resulting impact this has in relation to key issues such as the ownership of English has also been recognised. Description of the linguistic consequences of this shift however is less established, and lingua franca corpora are still in their relatively early stages of development, though growing in momentum. This paper will contribute to the emerging body of work that does report on empirical studies in the field, drawing on two corpora of naturally occurring lingua franca interactions which have been gathered for PhD projects at King’s College London. To date reported ELF research projects have tended to focus separately on one or other linguistic system, on phonology (e.g. Jenkins 2000; 2005), and to a lesser extent on pragmatics (e.g. House 1999), and on lexicogrammatical features (cf Seidlhofer 2004). The focus of this paper is to report on findings in both pragmatics and lexicogrammar, and in so doing to identify the interrelationship between the two systems and highlight ways in which they are mutually constitutive. The paper aims to show how pragmatic motives can lead to changes in the lexis and grammar, and in turn how lexicogrammatical innovations impact on pragmatic norms and strategies.

Efficiency in ELF Communication: From Pragmatic Motives to Lexico-grammatical Innovation Alessia Cogo and Martin Dewey, King's College London Abstract.The considerable demographic shift in the use of English worldwide, with the effect that L2 speakers outnumber L1 speakers, particularly as typified in English as a Lingua Franca (ELF) communication, is now widely acknowledged in Applied Linguistics.To a certain extent the resulting impact this has in relation to key issues such as the ownership of English has also been recognised.Description of the linguistic consequences of this shift however is less established, and lingua franca corpora are still in their relatively early stages of development, though growing in momentum.This paper will contribute to the emerging body of work that does report on empirical studies in the field, drawing on two corpora of naturally occurring lingua franca interactions which have been gathered for PhD projects at King's College London.
To date reported ELF research projects have tended to focus separately on one or other linguistic system, on phonology (e.g.Jenkins 2000;2005), and to a lesser extent on pragmatics (e.g.House 1999), and on lexicogrammatical features (cf Seidlhofer 2004).The focus of this paper is to report on findings in both pragmatics and lexicogrammar, and in so doing to identify the interrelationship between the two systems and highlight ways in which they are mutually constitutive.The paper aims to show how pragmatic motives can lead to changes in the lexis and grammar, and in turn how lexicogrammatical innovations impact on pragmatic norms and strategies.

Introduction
The ongoing substantial shifts in the demographic trends of speakers of English have led to a situation which is without precedent in the history of human languages.As has become increasingly acknowledged in applied linguistics and the ELT profession (though in the latter case arguably more reluctantly so), the majority use of the language has shifted away from L1 settings, and has increasingly come to involve ELF communication.
This can be attested in the significant increase in the volume of discourses at conferences and in peer reviewed journals and books which address issues regarding the spread of English.Llurda (2004) for example provides an overview of an emerging field, with a discussion of the resulting implications for English language teaching.Seidlhofer (2004) provides an extensive summary of empirical research to date, stating that the gathering of empirical data in ELF settings continues to gain momentum.If we accept the now widely held assumption that English is used by more speakers outside of the inner and outer circles than in them (Graddol 1997;2006) 7 , then this momentum needs to continue if we are to better understand how this phenomenon will have impact on the nature of the language.Indeed Graddol (2006) suggests that trends in demography are among the most important factors affecting language spread and language change.Yet despite this trend, and even though there is at least some willingness to accept the argument that speakers of lingua franca English be regarded as legitimate English users in their own right, there is still a vast imbalance in the description of ENL and ELF in favour of the former, at the expense of the latter.
At a time when corpus linguistics has developed as a major field of enquiry it should be surprising if there were not a good number of long established projects aimed at collecting and analyzing samples of non-L1 English discourse.This is primarily however not the case, and with the exception of 'learner corpora' (e.g.ICLE, the International Corpus of Learner English) which aim to identify learners' errors, there is relatively limited data available to date.Despite the heightened interest in the field of World Englishes, there are numerous vast corpora of L1 varieties of English 8 , but relatively few projects in ELF, many of which are still in their early stages of development.While more established projects such as ELFA (Mauranen 2003) and VOICE (Seidlhofer 2001) have begun to reveal very interesting findings, there is still a vast imbalance in favour of ENL corpora.It seems in applied linguistics that scholars are happy to recognize the spread of English in terms of its functions but not so willing to accept the consequence that widespread diffusion will inexorably involve a change in language forms.
When a language travels from one domain to another it of course encounters new contexts, new peoples and new languages.Through language contact it is natural and inevitable that language change occurs.This is a necessary condition of human languages -without this inherent capacity for flexibility and variability they would not travel nearly so well since they would not be as adept at meeting the needs of the new speakers who use them.This is of course neither new or unique; languages have been in contact with each other since pre-history and have evolved in fundamental ways as a result.The changes that we describe in this paper are part and parcel of the same ongoing, unending processes of language evolution.After all a language which no longer changes, such as say Latin or Classical Greek for example, is classified in sociolinguistics as a dead language, remaining either only in a canon of literature or used for ritualised cultural and religious routines.In the case of English, a language which in particular has gone through periods of significant contact and change, these processes of change inevitably continue to occur.In addition it is significant that in English they are taking place in all contexts in which the language is found, including in settings that are far beyond the reach of any of its native speakers.We are therefore database comprising 5 million transcribed and coded words of spoken discourse; The BNC (British National Corpus) is a 100 million word collection of spoken and written discourse (10 million spoken, 90 million written); and ICE -GB, the British component of ICE (International Corpus of English), is the first completed corpus of some twenty planned national and regional (including outer circle varieties) corpora of English, consisting of one million words of spoken and written samples of British English; we can also mention COBUILD's Bank of English project at the University of Birmingham.
currently in an untenable position, a situation where the most predominant, and so arguably most characteristic English is overwhelmingly underrepresented with regard to empirical data.
The predominance of ENL corpora is a further indicator of the extent to which the significance of the native speaker is customarily overstated in any discussion about the current status and future development of English.This is especially the case if we take account of Graddol's (1999) projected trajectory of English, in which he argues that by 2050 speakers of nativized Englishes will far outnumber speakers of native English, that English will be used primarily as a second language in multilingual contexts.If ELF scholars are accurate in their predictions this figure will consist not simply of speakers of English in the outer circle but also, and in fact largely, of L2 users of English in the expanding circle.
Thus far however this has remained a concept that has not been accepted in the expanding circle (incidentally the only context where it is still customary to refer to 'English' solely in the singular form), in other words where it is customary to momentarily suspend, or perhaps forget, the fundamental sociolinguistic reality of the pluralism inherent in language.We talk of 'World Englishes' in contexts where an L1 dialect is involved, or where an indigenized variety is the topic of discussion, but occurrence of 'Englishes' in discourse that focuses on ELF settings is very much limited to those conducting research in this field.Even then the use of the plural is sporadic and many researchers in ELF will refer to the subject of analysis in the singular form, unwittingly complicit perhaps in the perpetuation of the myth that English can be understood as a monolithic entity.It is certainly the case at least that our discussion of the 'E' in ELF is most often inferred in this way, and interpreted to mean that one of our intended goals is to legislate about language use.Perhaps it would suit our purpose better if we discussed lingua franca Englishes or if we were even more explicit than we already are in stating that the 'E' in ELF is to be interpreted plurally.

Corpora and Empirical Data
The establishment of several large scale corpora, such as ELFA (English as a Lingua Franca in Academic Settings), and VOICE, (Vienna Oxford International Corpus of English) opens increased opportunities for systematically studying the nature of ELF interactions and greatly facilitate the initial stages of the codification of lingua franca English.In addition to these macro level projects there is also now an increasing number of more micro oriented, mainly qualitative investigations into lingua franca communication (see for example Cogo 2005;Dewey 2003;Cogo and Dewey 2006).In describing the tendencies that have emerged in two smaller scale corpora of ELF, it is our aim to contribute to addressing the dearth in empirical data.To this end we approach the issue from a micro-level perspective with the aim to give qualitative description of the most frequent features of pragmatics and lexicogrammatical forms occurring in the data.
The findings presented in this paper are drawn from two small scale corpora of spoken ELF communication, one where the data are analysed from a pragmatic perspective and the other where the focus is on lexicogrammar.In the pragmatics corpus, data consist of recordings of mainly dyadic conversations, but also small group interactions and informal meetings.The common denominator is that all conversations arise naturally, they are not constructed or planned in advance and they are collected in non-instructional settings.The interactions consist of mainly small talk conversations, some are more personal and others are work-related exchanges.The recordings amount to a total of around 50 hours, out of which roughly 13 hours are transcribed.There are 4 main participants (French, German, Italian and Japanese) and a few secondary speakers that represent European and Asian first languages.In the case of lexicogrammar, the full set of data comprises 42 different communicative events, ranging from informal entirely unplanned conversations to formal seminar presentations, with a heavy bias towards naturally occurring non-institutional interactions.38 of these communicative events have been fully transcribed, totalling approximately 8 hours in duration.The participants number 55 and between them there are 17 first languages represented.
In both corpora all of the participants involved in the research are highly competent speakers of English, regarded here not as learners of English but as accomplished L2 users in their own right.This viewpoint is very much in line with the perspective of the L2 user as described in Cook (2002), in that all participants are regarded as independent speakers of English, and their language use is treated as legitimate variation not as failed or incomplete native speaker English.
It is important here to make clear that without exception all of the features presented below have been included for their typicality.That is, all are deemed to be indicative of emerging trends in ELF use and are judged to have met the following four key criteria.Firstly, they are systematic in nature -individual features are seldom random or isolated cases, and they often reveal larger patterns of change.Secondly, they occur frequently in the data -they are all produced on numerous occasions, and by numerous speakers from a variety of L1 backgrounds.Thirdly, they are communicatively effective -in none of the attested examples do they lead to a breakdown in communication.In the case of the lexicogrammatical features the data have been subjected to statistical analysis to substantiate claims about frequency, and use of concordance software to establish patterns of language use and determine levels of systematicity.Finally, in meeting the above criteria these features may therefore be considered non-L1 variants (not errors), differing from standard L1 equivalents but not regarded as erroneous or deviant.

The Pragmatics of ELF
Research in the pragmatics of ELF can be seen to start with studies in cross-cultural communication, which can be divided in three strands (see Kasper 1998 andKasper andRose 1999): the first is the bulk of research dealing with speech act realization strategies such as requests, apologies, compliments and refusals, where data are usually elicited through roleplays, visuals and discourse completion tasks.The second strand concerns those studies that have examined discourse interactionally, focussing on the turn-taking system, openings and closings and backchannels.The third group of studies is on communication strategies, i.e. "the problem-solving behaviours adopted by language learners when they lack requisite linguistic resources" (Kasper 1998: 99).
The general impression is that the majority of studies in crosscultural pragmatics are constrained in at least three ways.Firstly, the setting.In the majority of research data are usually collected in formal school settings, like language classrooms, which are indeed an authentic setting of language use but not real life communication.The hypotheses drawn from these investigations can only be proved against the classroom contexts but would not necessarily translate into real life interactions.Secondly, most research is conducted in the field of SLA.Most studies concentrate on the effects of pragmatic instruction on pragmatic learning, i.e. they tend to evaluate whether the pragmatic strategies taught in class are replicated in the classroom interactions.However, these studies are accountable for their findings within the framework of SLA, they cannot be extended to other contexts, such as naturally occurring exchanges in ELF.Finally, most studies tend to see the participants as learners of English and the learners' communication as deficient and problematic.Their spoken production is viewed as 'interlanguage', it is therefore deficient, since it lacks target-like qualities.The participants are usually labelled as 'native' or non-native', since the criterion of reference is still the native speaker pragmatic and linguistic behaviour, and an L2 speaker's performance is still seen as less appropriate (House 1993) or acceptable than the native interlocutor's.
More recently, there has been small-scale research dealing explicitly with ELF interactions, which looked at the function of ELF in everyday communication and from an interactional discourse perspective (House 1999;Meierkord 2002).The overarching idea seems to be that ELF is used for restricted purposes and it is emptied of cultural reference, which is reflected in the fluidity of norms that the participants operate.According to House, ELF is a "useful tool", a language that can be employed for certain purposes, such as academic, scientific and business talk.She states: English as a lingua franca is nothing more than a useful tool: it is a "language for communication", a medium that is given substance with the different national, regional, local and individual cultural identities its speakers bring to it.English itself does not carry such identities, it is not a "language for identification".(House, 2001: 2) The functional restriction of ELF to a language used exclusively for transactional purposes is supposedly confirmed interactionally by other studies that highlight the limited amount of discourse markers employed by ELF speakers (Meierkord 2002;House 1999: 75), and the few utterance completions used by the interactants (Firth 1996).However, the findings seem to result from "constructed" collection of data: most studies that have explored ELF communication have done so in artificial settings with simulated exercises/tasks, very few record naturally-arising conversations.
Another category frequently analysed in ELF pragmatics is misunderstanding.According to House (1999) communication breakdowns are not frequent, when they occur they tend to be resolved through topic change and, less often, through negotiation.Speakers tend to adopt a 'let-it-pass principle', which gives away a general tendency to self-centred talk.Firth (1996), on the other hand, maintains that speakers 'let-it-pass' when they judge the misunderstanding to be inconsequential for the successful ongoing of the conversation, which implies a generally cooperative and consensus-oriented interactional behaviour.However, Mauranen (2006) and Pitzl (2005) get rather different results -no evidence of 'let-it-pass', and very few misunderstandings in Mauranen's research, and many instances of negotiation of meaning in Pitzl's investigation of non-understandings.
In the following extract, the data will be used to corroborate the findings of Mauranen and Pitzl, and in turn challenge the positions highlighted above, i.e. the functional limitation of ELF, the restricted use of interactional features, and tendency to let-it-pass or avoid negotiation of meaning.'The French wedding' conversation reported below takes place during a coffee break, when the speakers, all identified with pseudonyms (Karen, German; Jean, French and Anna, Italian), are talking about Bertha's leaving party.Jean tells the other participants that he will not be able to go to Bertha's leave (line 1) because he has to attend a wedding in France, and then he starts elaborating on the website that the couple have set up for their wedding and the pictures that they have posted on it.The extract above can be analysed both for its interactional work and for the negotiation of meaning.First of all, from an interactional discourse perspective, it incorporates all the features that make communication cooperative and supportive.For instance, there is a high amount of occurrences of overlaps of the co-operative type (ll.9-10, 21-22, 23-24, 27-28, 31-32, 35-36, 38-39, 41-42, 44-45), which indicate engagement and interest in the on-going conversation.There is an instance of utterance completion in line 24-25, when Jean starts the turn with it's a bit and Anna finishes it off with too much.Utterance completions, when an interlocutor tries to complete the utterance of the current speaker, are not designed to take over the speaking turn or change the topic, on the contrary they show the listener's involvement and support.
Another frequent interactional feature of the exchange is backchannelling.Backchannels are those verbal and non-verbal utterances, such as mhm, uh huh, yeah, right, head nods and smiles, whereby the listener signals that they are paying attention to what is being said and that they want the speaker to continue talking.Apart from encouraging the current speaker to continue their turn, it seems that backchannels may serve other functions, among which that of ensuring the efficiency of the communication.The whole feeling of supportive and cooperative type of exchange is confirmed by the latchings especially in the opening lines.A latching (when a turn follows another without pauses) tends to indicate two things: the informality of the atmosphere (as a matter of fact the participants are having a break from the work routine) and the synchrony and involvement in the conversation (the speakers do want to know whether Jean will be going to the pub for Bertha's leaving party).
The exchange is particularly interesting as an example of negotiation of meaning.It is an instance of an a-priori clarification of an idiomatic expression, blue flower (line 30), that could have created misunderstanding, and maybe breakdown in the communication.The French expression fleur bleue recalls the old-fashioned pictures of angels that used to hang on the wall by children's bed as a symbol of protection.The angels were usually carrying blue flowers, hence the expression.
The we say in line 30 constitutes a key clue given by Jean to signal that an expression, with the potential of being misunderstood, is being used and the interlocutors should pay attention to that.Jean seems to be aware that idiomatic expressions are culturally sensitive and could be easily misunderstood by participants of other cultural backgrounds.Therefore, by using we say Jean provides a frame, which is intended to influence the interpretation of what follows, it allows the participants to make inferences about the use of the idiom and place it into context.This is subsequently picked up by Anna in the next turn (line 31), who seems to initiate a repair sequence by asking for clarification and confirmation of the meaning, i.e. that the idiomatic expression means cheesy.
The need to find a more suitable term for the speaker than the British English expression cheesy starts off a negotiation of the various ways of rendering the 'cheesy' / 'fleur bleue' idea in a way closer to the cultural background of the speakers themselves.After Jean uses the French idiom fleur-bleue and elaborates on its meaning (line 32 and following), Karen tries to find an expression more culturally appropriate for her and codeswitches to German, using the adjective kitchig (line 38).At this point Jean accommodates to Karen by accepting the German expression and even repeating it (line 39).Karen continues expanding on the idea in lines 42-43 using another expression, self exhibition borrowed from the German selbstdarstellung.This last expression is also endorsed by both interlocutors in a simultaneous overlapping utterance (lines 44-45).The French Wedding extract shows at least two instances of accommodation: when Jean repeats and accepts the German adjective kitschig (lines 38-9) and when both he and Anna agree on the German expression selfexhibition.
From the French Wedding extract it is possible to see how ELF speakers make use of interactional features to show support and alignment, the extent of negotiation of meaning they engage in, while avoiding 'let-it-pass' attitudes and the importance of accommodation for the successful outcome of the conversation.It is especially this last strategy, accommodation, which constitutes one of the prevailing strategies in ELF pragmatics and is in need here of further attention.9

Accommodation in ELF
Accommodation theory (Giles and Coupland 1991;Giles et al 1991) explains the different ways in which speakers may manipulate language for various purposes.The three accommodation strategies are convergence, divergence and maintenance.'Convergence' occurs when a speaker alters or shifts their speech to resemble that of the interlocutor.On the other hand, 'divergence' refers to the ways in which speakers accentuate their verbal and non-verbal differences in order to distinguish themselves from others.The third strategy, 'maintenance', is in effect a type of divergence and consists in maintaining one's speech behaviour, without trying to converge or diverge from the interlocutors.
The main purposes why speakers would want to use convergent accommodation strategies are two: one is communicative efficiency, where one speaker would change their speech to converge more closely to that of the interlocutor, in order to be more intelligible.The second reason is to "maintain integrity, distance or identity" (Giles and Coupland 1991: 66).For instance, repetition is used as an accommodation strategy in order to achieve efficiency and, at the same time, to show cooperation among speakers.In this last use, repetition is signalling agreement and listenership and engagement in the conversation.Reusing the interlocutor's expression is also a way of aligning to them and showing support and approval.
Repetition is one of the strategies used to show accommodation, and frequently occurring in the ELF data.An example of accommodation where the repetition is embedded into a longer turn is in the 'Visconti films' extract.This is a conversation about cinema, more precisely the films made by the Italian director Visconti.Karen (German), Daniela (Italian) and Anna (Italian) have been to see various Visconti films this week and they are talking about two of them in particular: The Leopard and Death in Venice.Karen is comparing the two films and she is saying that she prefers one of them, The Leopard.She is trying to say why she prefers that one, and with an utterance completion (line 39) Daniela helps her out adding that the Leopard is much more complex.Karen accepts the reason (yeah in line 41).Then after a small pause Anna adds that more things were going on in the Leopard and Karen accommodates to Anna reusing more things and changing the verb going on with the synonym happen.
Another instance of accommodation occurs in Extract 3 below.Chako (Japanese) and Sila (Mandarin) are talking about the language Mandarin, and Chako is enquiring about language variation in Mandarin.This time Sila is converging to a non-standard utterance provided by Chako in a previous turn.
Chako is interested in finding out when Mandarin started to change (lines 237-9), so she explicitly asks whether Mandarin changed because of the revolution (lines 241-3) and she uses the utterance because of revolution without definite article the.In line 246 Sila is repeating Chako's utterance and accommodating from a grammatical point of view, for she repeats the zero article that Chako uses in the previous turn.That, with the repetition of the zero article, Sila is actually converging to Chako's talk and not simply omitting a grammatical item is evidenced in the remainder of the conversation.From line 246 onwards, Sila uses definite articles three times (in lines 247 and 249) where she could have used a zero article instead.From other transcriptions it is possible to confirm that the general tendency is for Sila to use the definite articles, which corroborates the evidence that line 246 is an instance of accommodation, i.e.Sila changed her speech style to resemble that of Chako.Sila's accommodation has two different purposes: efficiency and alignment.In the first sense, Sila is repeating exactly the original utterance to provide herself with more processing time while keeping up the rhythm of the encounter.In the second, she is displaying attentiveness to Chako's earlier contribution, as well as confirming it and aligning with it and the interlocutor's knowledge.On the contrary, if Sila had repeated the utterance introducing the article she would have done a repetition with variation.This change would have probably meant that Sila wanted to perform a repair and thus show non-alignment with Chako.The fact that Sila did not make any change to the original utterance is meaningful in the sense that it proves the intention to align towards Chako.
The interactional features and the accommodation instances investigated in this section have in common that they show engagement in a certain type of ELF discourse and that they constitute pragmatic practices for ensuring efficiency of communication.In fact, the need for efficiency appears to be a central motivation for many changes in the lexicogrammar.This will be clarified further in the following sections, where we present some of the more characteristic features of lexical and grammatical innovation in the corpora.Seidlhofer (2004) pointedly remarks that although the causes and consequences of the global spread of English have been critically discussed at some length, there has been little consideration of what effect this is having on the forms of the language.In addition she points out that existing empirical studies have focused on phonology (Jenkins 2000;2002) and pragmatics (Meierkord 2002;House 1999;2001), with very little at the level of lexicogrammar, where least description has taken place.This dearth is attributed to the likelihood that a significantly larger corpus would be required for findings to be regarded as reliable.The data gathered in VOICE can provide the kind of large scale corpus necessary for the description of language on any level.And while primarily intended to form the basis of any kind of language area ELF researchers may be interested in, Seidlhofer also indicates that a particularly useful research aim would be to build on the findings of pragmatics and phonology by focusing on the lexicogrammar.She adds that while there have been no quantitative studies of characteristic lexicogrammatical features, regular tendencies continue to emerge in the data.These are summarized as:

The lexicogrammar of ELF
-Dropping 3rd person present simple -s -Confusing the relative pronouns who and which -Omitting definite and indefinite articles where they are obligatory in ENL, and inserting them where they do not occur in ENL -Failing to use correct forms in tag questions -Inserting redundant prepositions, as in we have to study about -Overusing certain words of high semantic generality, such as do, have, make, take -Replacing infinitive constructions with that clauses, as in I want that -Overdoing explicitness, as in black color rather than just black (Adapted from Seidlhofer 2004: 240) These features are presented not as conclusive results but as hypotheses about what might constitute characteristics of ELF lexicogrammar.In making the hypotheses Seidlhofer importantly points out that these would all likely be regarded by language teachers as typical learner errors, and therefore afforded considerable classroom time and attention.There is however some dissonance in the way in which the features are described: in presenting such a strong case for regarding ELF interactions as "sui generis" (2004: 211, here quoting from House 1999), and ELF speakers as "agents of language change" (2004: 212, and here quoting from Brutt-Griffler 2002), it is a great pity that the language used to report these findings is reminiscent of language used in error analysis.The items listed are defined according to negative criteria, such as 'confusing', 'failing to use', 'overusing' 'overdoing' even though the hypotheses are, although preliminary in nature, intended as examples of ELF variants in their own right.They are therefore of particular interest to any attempt to provide description of innovations in the lexis and grammar of lingua franca communication, and have served as a starting point for our analysis of the corpus data we have gathered at King's College London.
The naming of these items also illustrates the extent of the problem involved in overcoming the stigma attached to L2 English use.Even scholars most at the forefront of ELF research can fall into the trap of describing lingua franca English in terms that reflect the old paradigm, a paradigm which they have often challenged and redefined.That a researcher who has been so central to the establishment of ELF as a valid field of enquiry has herself expressed these hypotheses in a negative light suggests the extent of the task involved in coming to terms with the formal implications of ELF.We will turn now to our own initial findings in relation to lexis and grammar, many of which corroborate the recurrent themes identified in the above list.In addition to these items the data also indicate other quite different -though in terms of the processes leading to their emergence similar -regular lexicogrammatical features in ELF.The labeling of the features below has been chosen to reflect the notion that they each represent active choices in the way a set of linguistic resources is being used.The following features confirm all but one of the hypotheses given in Seidlhofer (2004), and all occur in the data with a degree of frequency and wide distribution.
-Use of 3rd person singular zero -Extension of relative which to include functions previously served only by who -Shift in the use of articles, (among other patterns this involves preference for zero article where L1 article use is largely idiomatic, and preference for definite article to attach extra importance to a referent in a stretch of discourse) -Invariant question tags (and use of other similar universal forms, such as this for this and these) -Shift in patterns of preposition use, e.g.we have to study about -Extension to the collocational field of words with high semantic generality, e.g.take an operation -Increased explicitness, e.g.how long time in place of how long Each of these features is widely attested in the two corpora being referred to.For instance in extract 1 above, the utterance are you going there? in line 3 can be regarded as an example of increased explicitness.We have also seen in extract 3 how convergence has led to the use of zero article for specific reference.There are many other cases where these features occur in the data.However, due to restrictions of space we have chosen to report at length on only one of the items.We will focus primarily on 3rd person singular zero as this is the feature that so far has been most closely scrutinized in our analysis.In addition to these features, others that have emerged during our analysis of the data are as follows: -Preference for bare and/or full infinitive over the use of gerunds, as in interested to do rather than interested in doing, or as in to study is… and to read is…, where the infinitive is used as the subject of a clause -Exploited redundancy, such as ellipsis of objects/complements of transitive verbs, as in I wanted to go with, You can borrow, etc.
These will be the subject of further investigation, and will be reported in subsequent papers.

The pragmatics of 3 rd person singular zero
The use of present simple verbs in 3 rd person singular with omission of the s morpheme occurs particularly frequently in the data (Breiteneder 2005 reports similar findings regarding the use of 3 rd person singular verbs in a small scale ELF corpus).It is a feature common to many of the interactions recorded over a three year period of data collection, as can be seen from the total number of incidents of 3 rd person zero summarized below in Table 1.As a reflection of the frequency and regularity of this feature we see its occurrence in the data as indicative of the use of a linguistic option, and not in a more negative light as an omission or 'dropping' of an item.It is for this reason that, in line with contemporary analysis of spoken data in L1 English varieties, we have chosen to describe this feature as 3 rd person singular zero (see for example Trudgill's 2002 treatment of this item in his discussion of African American Vernacular English and East Anglian dialects).
The contrast in the labelling of this phenomenon is very striking.In L1 English varieties this is customarily regarded by sociolinguists as a stable and legitimate feature, and labelled accordingly, whereas in analyses of L2 English in the expanding circle this is mostly regarded as a 'non-feature'.It is treated simply as the omission of an item that is absent not by design but as the result of a lack of control over the target language system or ignorance of that target.The treatment of this feature in sociolinguistics is often parallel to the perception of L1 dialects in non-specialist contexts, where any variation is generally regarded as substandard deviation.Here, and throughout the data analysis, we have therefore attempted to avoid labelling which connotes a negative construct.Following this it is thus not the case that 3rd person s is being 'dropped', rather that s and zero are competing variants in ELF communication.To borrow a term used by Roach (2000) to talk about elision in phonology, we can perhaps better describe the occurrence of this feature as a zero realization as opposed to an omission of s.It also appears from the data that the 3rd person zero is the variant that is winning this competition, and is in other words the feature emerging as the default option in informal naturally occurring communications.The data for this feature have been categorised according to whether an item is a main or auxiliary verb.This was initially not part of the design of the data collection, but something which emerged as important during the analysis phase.It soon became apparent that recording items in this way mattered since there seemed to be such a significant difference between 3rd person marking depending on whether a verb functions propositionally as a main verb or merely grammatically as an auxiliary.

Main Verbs
As can be seen from the above table there is a fairly even distribution of both items in terms of the total number of occurrences of 3 rd person singular -s and 3 rd person singular zero in main verbs.Respectively they represent very approximately 48% and 52% of all verbs with prepositional meaning that occur in 3 rd person singular present forms.There is however a marked difference in the nature of this distribution.Most importantly, the 3 rd person zero recurs in numerous settings and domains, constituting one of the more salient and widespread features of the current data.Out of a total 40 communicative events, 34 contain examples where 3 rd person singular forms occur in main verbs; for the significant majority of these, 24 (or 70% of the total number of conversations containing 3 rd person present forms) include examples of 3 rd person zero.The use of 3 rd person singular zero is thus not restricted by the nature of the ELF setting, the L1 of the speakers involved or the linguistic context; its use is shared by a considerable number of speakers irrespective of their first language.
The same cannot be said of the 3 rd person s however.There are for example particular restrictions governing its use.These operate on two levels: firstly, the situational context of an interaction has an important bearing on the probability of the form to occur in place of the zero marker; and secondly, the linguistic context is an important influencing factor.For the most part instances of 3 rd person s are widely dispersed, with many interactions having no or only single occurrences of this form (of the 32 communicative events where s does appear, in 11 of these it occurs only once).The number of interactions containing high frequency scores for s is relatively low.Significantly these appear in clusters of events, as is the case for example with two of the transcribed conversations, which have high frequency scores for 3 rd person s, 11 and 6 respectively, far higher than the mean score.Even more significant are the relatively low frequencies of 3 rd person zero that occur in these interactions.The settings for these two anomalous communicative events are untypical of others in the data, and it should be said untypical of ELF interactions more generally.In both of these interactions the setting is uncustomarily formal in nature: the recordings were taken in classrooms during an interval in an English language exam preparation class, and uncharacteristically in the presence of a NS teacher, who in one was a silent observer, and who in the other actively took part in the conversation.(Although these conversations took place in a classroom, they are naturally occurring exchanges, are not task driven, and thus still regarded as examples of lingua franca communication).
Out of the total of 40 ELF interactions in only 5 was there an L1 English speaker present, which was usually not by design but an unplanned and unavoidable occurrence. 10If we exclude these cases from the data analysis the overall frequencies and their relative importance to each other are altered significantly.This can be seen in Table 2 below, where the ELF settings have been categorised according to whether the interaction is L2 only or L2-L1.The table shows how the distribution of 3 rd person s and zero is affected by the presence and absence of L1 speakers.When the interactions involving L1 speakers are excluded the ratio of the zero to s form is significantly increased in favour of 3 rd person zero.In the case of 3 rd person s the frequency score has been reduced by 31 to a total of 72, which put another way means that a significant number of s forms, about 43% of all main verbs in 3 rd person singular present tense, occur in interactions where there is one or more L1 speaker present.In order to add support to these preliminary observations and to investigate further the effect of context on the use of the two 3 rd person singular options, the data were subjected to statistical analysis.A simple chi-square test was carried out on the 3 rd person singular present tense verbs, where each communicative event was categorised according to whether L1 speakers were either present or absent.The total frequencies of 3 rd person s and zero forms in the two types of event were then recorded in order to determine the significance of L1 English speakers on the relative distributions of each form.The value of χ² obtained in the test proved significant, and the null hypothesis was rejected. 11It can be concluded that the presence of L1 speakers in an interaction has a significant effect on the relative frequencies of the two forms, with a greater likelihood for 3 rd person s forms to occur during interactions in which L1 speakers (most of whom in the corpus are not teachers) are present and a far greater likelihood for the 3 rd person zero option to occur in interactions involving exclusively L2 English speakers.It then follows that the 3 rd person zero is emerging as the more characteristic, unmarked feature for present simple verb forms in ELF communication.This finding is further corroborated if we take into account the nature of the linguistic contexts in which the zero and s variants occur.In the case of 3 rd person zero there seem to be no limits to the linguistic contexts in which the variant occurs.The range of verbs is far greater for the zero form than the s form, covering a wide variety of main verbs representing a range of processes.Table 3 below shows examples of main verbs in the 3 rd person zero variant.The examples presented here represent only a small sample of those found in the data, but nonetheless they illustrate the occurrence of a wide range of verb types, including the phrasal verb grow up.The total range of different verbs occurring in the data with the zero marker is very broad indeed, almost to the point of being infinite.
In contrast, this kind of variety does not seem to be reflected in the examples of 3 rd person s, where a far narrower range can be observed.In analysing the data, a record was made of each individual verb in which the s form has been sighted.The overall range seems to be quite low, and certain verbs continue to appear time and time again.Detailed analysis carried out by hand reveals approximately half of all cases of 3 rd person s that occur in main verbs can be accounted for by only 4 verbs: has, means, looks, and depends.The frequency scores for these are as follows: Has 15 Means 15 Depends 10 Looks 9 Total = 49, or >47% of all 3rd person -s forms It is interesting to note that with the exception of has these are not necessarily high frequency items 12 .It seems to be the case that these forms recur with high frequency in the data because they form part of a prefabricated chunk of language; in many cases for example the verb cooccurs with greater than chance frequency with a preposition or adverb as part of a 'strong' collocation, as evidenced with depends on and looks like, which both appear numerous times in the data.

Auxiliary Verbs
The situation with 3 rd person singular auxiliary verbs, which include does, doesn't, has, (as well as the contracted 's, for example in he's been/she's had) and hasn't is very different from the present tense singular marking with main verbs 13 .In these cases the zero variant appears in only 3 out of the 65 occurrences of auxiliary 3 rd person verbs or just over 4% of the total.The overall occurrence of 3 rd person present singular verbs in the data, with a total of only 211, seems relatively low for a corpus of 60,000 words plus, which may be due to something in the nature of these particular interactions or may prove to be indicative of a more general trend in ELF communication.This will require much further data collection and systematic analysis of a far larger ELF corpus, followed by comparison of similar communicative events across different types of corpora.These would ideally involve both L1 and L2 Englishes to enable broader trends, differences, and similarities in 12 Using Oxford Wordsmith, a frequency list was produced for the demographic component of BNCB (a 900,000+ word corpus of spoken English).The verbs depend, mean, look, are ranked significantly lower than the verb have, each with tokens in the hundreds, while have is ranked 19th in the overall list. 133rd person present singular forms for 'be' copula and auxiliary verbs have not been included in this section.The 'be' verb is something of an anomaly in the data, and it does not seem to be following the patterns of any of the other verbs.There seems to be a very different situation emerging with regard to the use of 3rd person forms with 'be', and for this reason the verb has been treated independently and will be discussed in a forthcoming paper.
patterns of language use to be identified.To return for now though to the high number of 3 rd person s auxiliaries occurring in the data, there are several points that need to be highlighted.Firstly, the number of cases where s features in an auxiliary verb is proportionally significant: of all uses of s with any verbs (N=165), 62 (approximately 38%) are auxiliaries.However, if we remove the cases that occur in interactions with L1 English speakers present, the ratio increases to 68% (49/72) of s forms found in auxiliary verbs.The proliferation of 3 rd person s primarily in verbs that perform functions of tense and aspect or serve as morphological marking in questions and negatives reinforces the notion that s performs no real communicative function.It is thus absent (but not missing) from the majority of verbs that carry meaning.
In addition to this, the use of the 3 rd person zero in auxiliaries seems quite idiosyncratic.Not only is the overall number very low but also, and perhaps most significantly, the distribution is very specific and limited in setting.In one the interaction takes place in London, in a multilingual setting with 5 participants present: the speakers are L1 Arabic, L1 Urdu, and three L1 Mandarin.It is significant that the zero auxiliary occurs in an adjacency pair involving two of the L1 Mandarin speakers.In another the interaction takes place in Shanghai, China during an informal meeting between colleagues where all but one of the participants were L1 Mandarin.The use of the 3 rd person singular zero in auxiliary verbs therefore may prove to be an emerging feature (but even here its occurrence is so far very limited) in Mandarin English only, or a limited number of lingua franca Englishes but perhaps not as a characteristic of ELF communication more generally.
To return to the issue of situational context, the impact of the presence of an L1 English speaker on the lexicogrammar of ELF interactions can be observed in a number of places in the data.The following, extract 4, perhaps best illustrates how accommodation operates and -at least in this semi-formal setting where deference to the authority of a NS teacher might be expected -how there is convergence towards the speech patterns of the L1 speaker.The following extracts (4 -7) are all taken from a conversation that took place at the end of an advanced level language class, in which the teacher and students had used material from unit 1 of Cutting Edge Advanced (a very popular recent ELT textbook series) which presents reading texts, listening texts and discussion tasks on the topic of English as an International Language.The participants: Vicky, L1 English, Naoko, L1 Japanese, and Lucy, L1 Russian, are here discussing their views about the topic of the material, and in the first extract Vicky, the teacher of the lesson asks the students for their opinions on the notion of ownership and the spread of English.In line 378 Naoko uses the 3 rd person -s for the present simple of the verb speak, and then in line 379 Lucy uses the -s form in belongs.It is noteworthy that both uses of the form occur in close proximity to 3 rd present singular verbs spoken by Vicky, the L1 English speaker.Lucy repeats belongs in her response to the question after Vicky has already used the form three times when posing the question, which strongly suggests the -s occurs as the result of Lucy converging towards the L1 pattern.For Naoko it seems even clearer that accommodation is at play.Her utterance of speaks occurs entirely in isolation and is given with no follow up, serving thus as purely an echo of Vicky's own utterance -its function is not to communicate a proposition; rather it appears to be a discourse strategy, there to show interest and agreement with her interlocutor and to act as a means of convergence (see also discussion about accommodation above).
The occurrences of 3 rd person -s in the above extract can be shown to be in direct contrast to the use of the 3 rd person zero in the following exchange that occurs earlier in the same conversation.Here Lucy uses in line 180 the zero form to express 3 rd person singular present tense.It is notable that this occurs towards the end of a fairly long stretch of turns in which Lucy and Naoko are more active participants than Vicky, whose presence in the discourse is far less prominent than it was in extract 4.This is further evidence to suggest that the more natural option for 3 rd person singular verbs is the zero variant, since in this extract the verb occurs not as a repetition or echo of an L1 speaker but as the expression of a proposition between two L2 speakers.Accommodation has not altered the default pattern, and the 3 rd person -s has not occurred.This use of make with zero marking is far more representative of the behaviour of 3 rd person singular verbs in the data, and thus possibly more typical of informal ELF communication in general.That the form occurs here despite the semi-formality of the situation and the presence of a NS teacher indicates the strength of this naturalness.It is also worth noting that perhaps ironically the form occurs here despite the 3 rd person -s being explicitly discussed previously in the conversation, with Lucy insisting that it had always been important for her to have teachers who corrected her if she did not use the form.There is another way in which the presence of an L1 English speaker in ELF settings can alter the patterns and trends that seem to be emerging as characteristic features of ELF communication.This can be illustrated very clearly in extract 6 below.Extract 6.

427
LUCY: and (.) for people who wants to know culture 428 and who wants to know er know deep -er 429 deeply British nation and er or American for 430 example (.) they need to know these things The 3 rd person -s occurs here as a marker of a plural verb, with the subject people.This is extremely uncommon in the data, and occurs in only one other place in the 40 communicative events analysed.It is probably the case that during the discussion of grammatical rules, teaching methods, error correction and so on, Lucy has become so aware of the -s form and its frequent use by Vicky in the conversation that she has shifted far away from her more usual, more natural zero form that she is here consciously attaching the morpheme to both singular and plural 3 rd person present simple verb forms.She has become so aware of the differences between her and Vicky's patterns of speech that she is consciously accommodating towards Vicky and 'hyper-converging' ('hypercorrection' in ELT) towards her use of this form-so much so that her language becomes unnatural, untypical and very marked for its difference from how she customarily uses English in her more characteristic ELF communications.
With regard to the use of 3 rd person singular zero we will give the last word to Vicky, an experienced teacher and teacher trainer with over 15 years as an ELT practitioner.This final extract occurs quite early on in the same conversation, and as it raises important issues with regard to current practice in ELT, it is given here in anticipation of the pedagogical considerations that need to be addressed in ELF research.

Conclusion
Each of the features described above can be investigated for the underlying processes that lead to their emergence.These processes act as motivating forces that can in many cases be observed to be in operation with some degree of transparency.Increasingly language corpora are sufficiently vast and the software available for analysing them sufficiently sophisticated and powerful that we can now far better than ever before trace shifts in language patterns and trends as they are in progress.As these emerging trends are traced it is possible to make better sense of them by considering their likely causes, and by explaining the underlying motives that lead to their appearance.In reporting on findings in both pragmatics and lexicogrammar, it has been our aim to identify the interrelationship between the two systems and highlight their mutually constitutive nature.In analysing data from our corpora it came to our attention that the two are fundamentally interconnected, and that in many instances these underlying causes of shift in the lexis and grammar are primarily pragmatic in nature.Pragmatic motives often lead to changes in the lexis and grammar, and in turn lexicogrammatical innovations have significant impact on pragmatic norms and strategies.In terms of the motivations that can be inferred from the data, these include: efficiency of communication, added prominence, reinforcement of proposition, increased explicitness, exploiting of redundancy.In the case of 3 rd person singular present verb forms, the data strongly suggest that use of the zero option occurs as the result of efficiency of communication and exploited redundancy.Trudgill (2002) in addressing why East Anglian dialects use 3 rd person zero, points out how in contemporary Standard English the -s is something of a typological anomaly.The feature is firstly unique; in present tense verb forms only 3 rd person singular displays any morphological marking.Secondly, its occurrence is according to typologists all the more unusual because it is precisely the least likely form to receive any such marking.Among the world's languages English then is something of an oddity for its inflection of only one of the present tense verb forms, and especially so for its attachment to the third person singular.The more pertinent and certainly more justifiable question then with regard to 3 rd person singular -s is not why L2 speakers and some speakers of L1 dialects use the zero form, but precisely the oppositewhy in standard varieties of English does the 3 rd person singular verb carry morphological marking?Zero marking for first, second, and third person plural makes the use of the singular -s an unexpected irregularity, and surely the phenomenon that most requires explanation.
In fact Trudgill goes on to comment that there is a considerable number of English varieties which make use of the 3 rd person singular zero, including as well as East Anglian dialects in the UK, African American Vernacular English, English-based creoles of the Caribbean and West Africa, and indigenized L2 varieties such as Singapore English.Many of these varieties have evolved as the result of language contact 14 , in situations where the irregularity and markedness of the unusual -s would likely appear quite cumbersome, largely unnecessary.Perhaps 14 Trudgill (2002:97) presents evidence to account for the existence of third person -s zero in East Anglian dialects.He suggests that as with the case of creoles and L2 Englishes the phenomenon occurs as the result of significant language contact.He observes how in the sixteenth century Norwich, at the time by far the largest urban centre in East Anglia and therefore the most significant influence linguistically, was home to some 6000 recent immigrants (at that time 37% of the city's population) from the low countries and was a setting for quite considerable language contact.Most interesting of all is Trudgill's observation that English was not only used by the immigrant population to communicate with the indigenous locals, but also to communicate with each other, that is as a lingua franca between the French speaking and Dutch speaking immigrant groups.Remarkably Trudgill explains that it is this lingua franca use which reinforces the shift from the competing native forms of s and th in favour of the non-native zero form.
predictably then, increased language contact has often in the past resulted in the -s losing out to competition from the more regular, expected and natural zero option.The shift away from the -s produces better consistency, resulting in a more systematic pattern with universal zero morphological marking for all present verb forms.This means it is entirely to be expected in ELF settings, where language contact is not only considerably extensive, but also a constitutive factor in any occurrence of ELF interaction.Any linguistic system that contains an element so marked in nature is bound to be prone to change, especially in contact situations (Trudgill 1986 remarks how the 'more natural' option in dialect contact invariably wins where a number of features are in competition), and it is thus entirely probable, not to mention logical, that the 3 rd person zero should be a characteristic feature of ELF.This increased regularity that the zero -s allows is also a likely explanation for a good number of other features found in the data.There are several reasons for this, all of which point to the underlying, largely pragmatic processes that give rise to ongoing changes.
It is therefore also important to relate these processes of change currently operating in ELF to other previous and possible subsequent linguistic changes This gives rise to fundamental questions with far reaching implications in the study of language change, and it is for this reason our recommendation that researchers in ELF engage further with historical linguistics.
the Death in Venice was … I liked it it was nice 0.5 more things going on in the Leopard= =I think yeah … more things happen because Death in Venice at the end I think nice beautiful boy yeah (laughing) and all ugly men

Table 2 .
ELF settings and the impact on 3 rd person singular verbs of L1 speakers

Table 3 .
Cases of 3 rd Person Singular Zero