A corpus investigation of two non-standard features, in English as a lingua franca, native speech, and learner language: the 3rd person zero, and interchange of the relative pronouns who/which.

Abstract

Using English as a lingua franca is different from using it as a foreign language for communication with native speakers. English as a lingua franca (ELF) can be defined as the use of English among communicators of different first languages (including English itself), for whom English is the communicative medium of choice, and often the only option – a definition based on that of Seidlhofer (2011). There are non-standard features of English that are viewed as errors from a foreign language point of view, but could be viewed neutrally from an ELF perspective if they do not interfere with communicative effectiveness, or if they are a manifestation of the group identity of the speaker. This thesis follows up claims that there are non-standard features of English shared by ELF events, and at frequencies that would make ELF distinct from native speaker English and English used by advanced learners. Two such features are examined, both of which have been named by Seidlhofer (2004) as not being obstacles to communicative effectiveness: the absence of the third person singular present tense -s ending (called here the 3rd person zero), and the interchange of the relative pronouns who and which. A dataset many times larger than those used in previous studies was examined: two spoken ELF corpora of over a million words each, the Vienna-Oxford International Corpus of English (VOICE) and the English as a Lingua Franca in Academic settings (ELFA) corpus. The non-standard features were also investigated in a native speaker corpus, the British National Corpus (BNC), and in a spoken corpus of higher intermediate to advanced learners of English as a foreign language, the Louvain International Database of Spoken English Interlanguage (LINDSEI). Both ELFA and LINDSEI were tagged for this study using CLAWS7, and there is some advice for researchers on using this tool within. The two non-standard features were examined in the corpora for frequency of occurrence, and by factors that might influence the frequencies of any non-standard features found: L1 (first language), event type (function or form of an event), and domain (setting where a certain kind of speech may be appropriate). The non-standard features investigated are present in the four corpora, albeit in low percentages. Standard usage is preferred by speakers in all corpora. For the two ELF corpora, this means neither of the non-standard features can be claimed as an emergent trend or a default option for ELF. The non-standard features appear in VOICE and ELFA at a lower rate than that of LINDSEI, but not by much. The BNC generally has fewer of the two non-standard features than the other corpora, but not in the case of the 3rd person zero in two event types. The non-standard features were recorded in the speech of many different L1s, and were sorted into L1 groups for analysis. In both ELF corpora and LINDSEI, for both non-standard features, Romance L1s have a consistently higher level of non-standard features than Germanic L1s. The data suggests that L1 makes a difference. The higher percentage for the 3rd person zero in LINDSEI can be mitigated by the removal of two particular groups of speakers, Chinese and Japanese L1s; without these speakers the percentage is just a little higher than that of VOICE. There are differences between event types in the ELF corpora, but they point in different directions: the higher percentages of non-standard features can appear in both formal and informal event types, so no conclusions can be drawn about them. The 3rd person zero appears considerably more often in the BNC in two event types ( interview and conversation ), which hints that what is truly standard in native speech might be contextually dependent. The event type of LINDSEI ( interview ) does not seem to be the factor that influences its frequencies. Differences between domains were shown, but no domain was prominent in any of the corpora. The all-educational domain corpora, ELFA and LINDSEI, did not form a pattern with the domains labelled Educational in VOICE and the BNC. This investigation also examined the claim that ELF speakers choose the relative pronoun which more often than who when there is a choice between them and either use would be considered standard. This was shown to be the case – marginally – in the present data, but it was also true for LINDSEI, which shows a similarity between the use of ELF and learner language in this regard, rather than something special about ELF. No investigation of the language in ELF changes anything about the ELF perspective, which is a way at looking at the language. One can imagine that English will continue to be spoken in, for example, a professional business context, in which the speakers do not consider themselves to be learners, and do not consider native-like speech to be the target for competence.

Using English as a lingua franca is different from using it as a foreign language for communication with native speakers. English as a lingua franca (ELF) can be defined as the use of English among communicators of different first languages (including English itself), for whom English is the communicative medium of choice, and often the only option – a definition based on that of Seidlhofer (2011). There are non-standard features of English that are viewed as errors from a foreign language point of view, but could be viewed neutrally from an ELF perspective if they do not interfere with communicative effectiveness, or if they are a manifestation of the group identity of the speaker. This thesis follows up claims that there are non-standard features of English shared by ELF events, and at frequencies that would make ELF distinct from native speaker English and English used by advanced learners. Two such features are examined, both of which have been named by Seidlhofer (2004) as not being obstacles to communicative effectiveness: the absence of the third person singular present tense -s ending (called here the 3rd person zero), and the interchange of the relative pronouns who and which. A dataset many times larger than those used in previous studies was examined: two spoken ELF corpora of over a million words each, the Vienna-Oxford International Corpus of English (VOICE) and the English as a Lingua Franca in Academic settings (ELFA) corpus. The non-standard features were also investigated in a native speaker corpus, the British National Corpus (BNC), and in a spoken corpus of higher intermediate to advanced learners of English as a foreign language, the Louvain International Database of Spoken English Interlanguage (LINDSEI). Both ELFA and LINDSEI were tagged for this study using CLAWS7, and there is some advice for researchers on using this tool within. The two non-standard features were examined in the corpora for frequency of occurrence, and by factors that might influence the frequencies of any non-standard features found: L1 (first language), event type (function or form of an event), and domain (setting where a certain kind of speech may be appropriate). The non-standard features investigated are present in the four corpora, albeit in low percentages. Standard usage is preferred by speakers in all corpora. For the two ELF corpora, this means neither of the non-standard features can be claimed as an emergent trend or a default option for ELF. The non-standard features appear in VOICE and ELFA at a lower rate than that of LINDSEI, but not by much. The BNC generally has fewer of the two non-standard features than the other corpora, but not in the case of the 3rd person zero in two event types. The non-standard features were recorded in the speech of many different L1s, and were sorted into L1 groups for analysis. In both ELF corpora and LINDSEI, for both non-standard features, Romance L1s have a consistently higher level of non-standard features than Germanic L1s. The data suggests that L1 makes a difference. The higher percentage for the 3rd person zero in LINDSEI can be mitigated by the removal of two particular groups of speakers, Chinese and Japanese L1s; without these speakers the percentage is just a little higher than that of VOICE. There are differences between event types in the ELF corpora, but they point in different directions: the higher percentages of non-standard features can appear in both formal and informal event types, so no conclusions can be drawn about them. The 3rd person zero appears considerably more often in the BNC in two event types ( interview and conversation ), which hints that what is truly standard in native speech might be contextually dependent. The event type of LINDSEI ( interview ) does not seem to be the factor that influences its frequencies. Differences between domains were shown, but no domain was prominent in any of the corpora. The all-educational domain corpora, ELFA and LINDSEI, did not form a pattern with the domains labelled Educational in VOICE and the BNC. This investigation also examined the claim that ELF speakers choose the relative pronoun which more often than who when there is a choice between them and either use would be considered standard. This was shown to be the case – marginally – in the present data, but it was also true for LINDSEI, which shows a similarity between the use of ELF and learner language in this regard, rather than something special about ELF. No investigation of the language in ELF changes anything about the ELF perspective, which is a way at looking at the language. One can imagine that English will continue to be spoken in, for example, a professional business context, in which the speakers do not consider themselves to be learners, and do not consider native-like speech to be the target for competence.