English as a lingua franca (ELF)

The most cursory review of the literature addressing English in its global context reveals a host of terms, such as English as an international language, World Standard English, literate English, English as a lingua franca, and limited agreement on definitions (Erling, 2005; McArthur, 2004). A mutually agreed starting point for most, however, continues to be Kachru's (1985) division of worldwide Englishes into inner, outer, and expanding circles. If we define a lingua franca as a contact language, i.e., “a vehicular language spoken by people who do not share a native language” (Mauranen, 2003, p. 513), English used in the expanding circle between L2 users “captures ELF in its purest form” (Seidlhofer, 2004, p. 211). As both Maurenen and Seidlhofer point out, however, English is manifestly unlike other contact languages. The enormously diverse intra- and international contexts of use and the continual movement of users routinely result in interactions between speakers from all three groups. While stipulating that this is an inherent feature of global English usage, I will continue to follow Seidlhofer (2004) and define ELF as talk comprising expanding circle speaker-listeners, also described as nonnative speakers (NNSs), competent L2 speakers, or, in Jenkins' (2000) terms, non-bilingual English speakers (NBESs) for whom proficiency may range from nearly bilingual English speakers (BESs) to beginner NBESs.

Intelligibility and Comprehensibility

Up to this point, I have used the term intelligibility in its broadest sense to mean both “intelligible production and felicitous interpretation of English” (Nelson, 1995, p. 274). There is no universally agreed upon definition of what constitutes this construct, nor is there an agreed upon way of measuring it (Derwing & Munro, 2005; Jenkins, 2000). Within the field of World Englishes (WE), a common conceptualization is Smith and Nelson's (1985) tripartite definition of intelligibility, the ability of the listener to recognize individual words or utterances; comprehensibility, the listener's ability to understand the meaning of the word or utterance in its given context, and interpretability, the ability of the listener to understand the speaker's intentions behind the word or utterance. This last level of divining a speaker's intentions is understandably difficult to measure and Levis reports that the term has largely “fallen by the wayside” (2005, p. 254). There remains a clear distinction in the literature, however, between “matters of form,” comprising formal recognition or decoding of words and utterances, and “matters of meaning,” variously described as ‘comprehensibility,’ ‘understanding,’ or ‘communicativity’ (Jenkins, 2000, p. 71). As Field (2003) suggests, a listener may use a higher level of contextual understanding to compensate for the fact that a message is unrecognizable, that is, unable to be precisely decoded. Similarly, an utterance may be intelligible and incomprehensible,¹

The Relative Nature of Comprehensibility

Although our immediate response as listeners is often to assign responsibility for comprehensibility to our interlocutor, leading scholars in the fields of discourse and WE have consistently recognized that comprehensibility is a relative not an absolute standard and is, in this sense, “co-constructed” (Duranti, 1986; Gumperz, 1992). Smith and Nelson state that “intelligibility is not speaker or listener-centered but is interactional between speaker and hearer,” (1985, p. 333) and Nelson notes that “being intelligible means being understood by an interlocutor at a given time in a given situation” (1982, p. 59). This conceptualization inextricably links comprehensibility to context of use, a complex setting involving factors related to the speaker, the listener, the linguistic and social context, and the environment (Field, 2003; Jenkins, 2000; Levis, forthcoming).

It is evident, however, that these elements have traditionally been narrowly construed, and comprehensibility studies in WE contexts have privileged inner circle speaker-listeners, i.e., focused exclusively on NS judgments of outer or expanding circle speakers (Jenkins, 2002). This situation is an unfortunate inheritance that both responds to and reinforces current inequalities in world Englishes.²

FIGURE 1 World Englishes speaker-listener intelligibility matrix (from Levis, 2005, p. 373) (© TESOL Quarterly. Permission to reproduce gratefully acknowledged.)

The first column, depicting the inner-circle listener, is well-represented in the intelligibility literature. In comparison, both the second and third columns, prioritizing outer and expanding circle interactions, are relatively new and growing areas of research. Significantly, however, recent work that targets expanding circle, or NBESs in ELF interactions (the EC-EC panel above) suggests that they are processing contextual factors including speaker, listener and environmental factors differently than inner circle speakers (Deterding & Kirkpatrick, forthcoming; Field, 2005; Jenkins, 2000; Meierkord, 2004).

Speaker Factors

Undoubtedly the most salient speaker factor in standard NS-NNS comprehensibility studies has been phonology and primarily the effects of accentedness. Using measures of intelligibility that ask listeners to write out what they heard, and Likert scales to judge perceived comprehensibility and accentedness, researchers have found that these measures are consistently distinguished and that comprehensibility of L2 speech by NSs is more significantly related to prosodic variables than segmental effects (Anderson-Hsieh, Johnson & Koehler, 1992; Anderson-Hsieh & Koehler, 1988; Munro & Derwing, 1995). This led Derwing and Munro to suggest that improvement in NNS comprehensibility (for NS receivers) “is more likely to occur with improvement in grammatical and prosodic proficiency than with a sole focus on correction of phonemic errors” (1997, p. 15).

In contrast, reporting research addressing phonological factors in interlanguage talk (ILT), Jenkins (2000, 2002) argues convincingly that the processes through which NBESs achieve understanding in ELF interaction are qualitatively different from those used by NSs in NS-NS or NS-NNS interaction. Jenkins (2000) analyzes conversational and information gap task data she collected from L2 mixed-language dyads. Following an examination of all instances of communication breakdowns, she reports that pronunciation issues comprise the biggest source of loss of comprehensibility or intelligibility, and these most commonly occurred at the segmental level. In a typical example given below, (B) a Japanese speaker is describing a small set of pictures to (A) a Swiss-German speaker. As Jenkins points out, despite only one picture containing cars, which were red, and no indication that they were for hire, (A) persists in “adjusting the context and/or co-text to bring them into line with the acoustic information rather than vice versa” (2000, p. 90).

She reasons that this predominant focus on bottom-up processing reflects a higher dependency on phonological form by ELF speakers who are less able to integrate inferences from some kind of higher contextual knowledge or from a shared background with their interlocutors.

This greater reliance on deciphering phonological form was also found by Deterding (2005) who investigated outer circle speaker-listeners listening to a nonstandard British English variety (Estuary English). Based on transcripts and commentary by undergraduate Singaporean listeners, he identified several segmental issues, ‘th fronting,’ glottalization of medial /t/, and fronting of the high, back, rounded vowel, as barriers to intelligibility. Interestingly, some of these listeners also expressed strong listener irritation with this unfamiliar inner circle variety: “he almost made my blood boil because I could hardly understand his words” (2005, p. 435).

Field (2005) tested both NS and NNS listeners in a psycholinguistic study in which lexical stress and vowel quality were manipulated on sets of disyllabic words. Words were recorded with normal acoustic cues and in conditions where stress was shifted leftward or rightward and in some cases, vowel quality altered. When tested with an intelligibility measure, although both groups of listeners were significantly handicapped by modified stress patterns particularly when the lexical stress was shifted to the right, nonnative listeners demonstrated a lower success rate in identifying the words in the standard, unmanipulated group correctly. Citing additional work (2004), Field is in agreement with Jenkins that “evidence is emerging that NNLs place greater reliance on interpretations at word level even in the face of contradictory evidence” (2004, p. 418).

This initial research in ELF supports a long held contention that certain errors have a higher cost associated with them than others in terms of comprehensibility and that this error hierarchy will vary depending on speaker and listener factors. Hence we would not necessarily want to promote strategies among NBESs that have been shown to increase comprehensibility for native speaker listeners if the former will be primarily or exclusively engaged in ELF interaction (Jenkins, 2002). For NBESs engaged in ILT, Jenkins suggests a lingua franca core (LFC), a set of core phonological features that will result in maximum intelligibility in ELF interaction. The core chiefly comprises segmental features but also includes tonic placement, or sentence stress (see Jenkins, 2002 for a complete description.) In her study of ELF interaction, she further observed that NBESs accommodated to each other, thereby increasing intelligibility, by converging on more target-like forms and reducing L1 transfer features when dealing with these high risk core areas.

Although pronunciation has been described as “possibly the greatest single barrier to successful communication” (Jenkins, 2000, p. 83), phonology is only one aspect of a speaker's language output, and other linguistic variables are perceptually salient for ELF interlocutors. Meierkord (2004) investigated the syntactic forms produced in 22 hours of conversation between outer and expanding circle speakers. Despite finding some “extreme divergences” from the rules of inner circle varieties, she found that these did not impede comprehensibility. This concurs with Jenkins (2000) who found that grammatical miscues played a minor role in comprehensibility in ELF interaction. It is noteworthy, however, that Meierkord describes ELF as a “syntactically heterogeneous form of English” (2004, p. 128) in which she found unstable and unsystematic features, and she surmises that we are some way from the development of a standard international form of English. Lexical variation is likely to impede comprehension in the form of variety specific idioms (Jenkins, 2000; Seidlhofer, 2001) or in the use of localized vocabulary terms (Nelson, 1995). But perhaps more common with the cline of proficiency that is inevitably present in ELF interaction is vocabulary that is unknown to one or other of the participants (Meierkord, 2004).

As with other aspects of language structure, pragmatics in cross-varietal contexts of English has typically dealt with NS/NNS interaction with inner circle varieties as the fulcrum (Gumperz, 1992; Pickering, 2001). Meierkord focuses on conversational data in ELF interactions and observes that there is a high degree of negotiation and collaborative achievement as revealed by participants' choice of behaviors such as pausing to allow transitions between conversational topics, choosing safe topics and using politeness strategies such as back-channels and routine formulaic expressions. She suggests that participants employ these strategies in order “to create a variety which assures a maximum of intelligibility” (2000, p. 10). A rather different interpretation of these kinds of behaviors is given by House (1999) “Interactions in English as a Lingua Franca and the Myth of Mutual Intelligibility”, cited in Seidlhofer (2001). House argues that these behaviors are rather a reflection of a “palpable lack of mutual orientation” (p. 82) which may be designed to disguise a lack of solidarity between participants and pass over trouble spots (House, 1999, cited in Seidlhofer 2001, p. 143; see also House, 2002, for further discussion).

Finally, at the level of discourse structure, two initial studies suggest that while on one hand, ELF interlocutors may use more transparent information-structuring devices such as topicalization structures to ease listener understanding (Meierkord, 2004), at the same time, speakers may transfer variety-specific features into international settings such as the use of discourse particles rather than acoustic cues to distinguish between new and old information in Singaporean English (Ee Ling Low, forthcoming). Clearly, these studies represent the beginning of a fruitful area of research.

Listener Factors

The bulk of current research in listener factors can collectively be described as investigating the effect of listener familiarity in relation to a number of different variables. A particularly significant area is listener experience with phonological representations of the target language. Field (2003, p. 36) argues that the “non-invariance” of phonemes, i.e., the lack of a single,“ideal” representation suggests that listeners may store multiple representations of the phonemes and that the more we are exposed to a certain production, the more intelligible it will be. Major et al. (2002) suggest something similar for prosodic variables. Finding that Chinese and Japanese listeners understood Spanish accented English as well as standard American English, the authors speculate that this result is related to a similarity of prosodic rhythms in Spanish, Chinese and Japanese, specifically, the lack of vowel reduction. In relation to ELF, the effect of familiarity with phonological forms is termed “the interlanguage speech intelligibility benefit” by Bent and Bradlow (2003). The authors measured the intelligibility of recordings of English sentences produced by Chinese and Korean speakers and NSs of North American English for listeners from Korea, China and with mixed L1 backgrounds. They found that although all listeners judged lower proficiency L2 speakers to be less intelligible, nonnative listeners found high proficiency NBESs with whom they shared a native language as intelligible as the native speakers suggesting a “matched interlanguage intelligibility benefit” (2003, p. 1606). The researchers also found a “mismatched interlanguage intelligibility benefit”; that is, even when there was no shared native language, nonnative listeners found some high proficiency NBESs from different language backgrounds to their own equally as or more intelligible than the NSs. Regarding the matched benefit, Bent and Bradlow suggest that “a nonnative listener is well equipped to interpret certain acoustic-phonetic features of the speech of a native-language matched nonnative talker as the talker intended them to be interpreted, even though they may deviate markedly from the target language norm” (2003, p. 1607). This accords with Jenkins (2000), who observed that in interlanguage talk between same L1 dyads, interlocutors tended not to converge on a target-like form but retained their transfer-based replacement form.

It is less clear how the mismatched benefit works. Bent and Bradlow suggest that it may be the result of shared NBESs' knowledge of developmental characteristics in ILT phonology; however, similar investigations do not support the mismatched benefit result. In another psycholinguistic study measuring speech intelligibility for Dutch NBESs, Van Wijngaarden (2002a) found that Dutch listeners did not benefit from hearing their “own” nonnative accent in a second language (2002a, p. 1909) and that the native English speakers were more intelligible. In addition, although Deterding and Kirkpatrick (in press) report that specific pronunciation characteristics used by ASEAN³

Another factor associated with interactional success in NS-NNS research is listener attitude (Lippi-Green, 1997; Rubin, 1992). Smith and Nelson (1985) suggest that a listener who expects to understand a speaker will be more likely to find that speaker comprehensible than one who does not. Although the effect of listener attitude in ELF interactions has rarely been studied (although see Dalton-Puffer, Kaltenboeck & Smit, 1997), an innovative study by Lindemann suggests a possible model of how this research might proceed. North American English speakers who had relatively positive or negative attitudes toward Koreans were matched with Korean speakers and asked to complete a map task. Qualitative and quantitative differences between the two groups revealed that North American partners with negative attitudes often failed to provide feedback to their Korean partners and “problematized their contributions” (2002, p. 431). Only two of the pairs in this group did not successfully complete the task, yet all six pairs rated their interactions as unsuccessful. As Lindemann (forthcoming) states, “[This] suggests that listeners may react negatively to certain accents (and thus claim to find them unintelligible) even when we would expect that the features of those accents themselves do not directly impede intelligibility.”

Additional variables that may affect intelligibility or perceived comprehensibility to a greater or lesser degree for listeners include familiarity with a particular speech event, topic, or specific interlocutor, listener specific factors such as level of tiredness (Field, 2003) or situation specific factors such as environmental noise (Rogers, Dalby & Nishi, 2004; van Wijngaarden et al., 2002a; van Wijngaarden et al., 2002b). There are currently very few studies addressing these variables in ELF contexts.