5.1 Explicit regulation: drawing boundaries of acceptability and correctness
5.1.1 Correcting spoken language
To consider the boundaries between unacceptable and acceptable spoken English, this section turns to instances where spoken language is corrected by interlocutors. Such intervention with a speaker’s language reflects the interlocutor’s notions of correctness, and defines what is considered ‘correct’ and ‘incorrect’. This sheds light on the boundaries of acceptable English: what is corrected and what not, in what circumstances corrections are done and who does the correcting.
Language corrections as a form of other-repair
I define language corrections as instances where an interlocutor changes a linguistic detail in the previous speaker’s turn. In CA, such corrections fall under the category of other-repairs, or instances of interaction where participants temporarily stop the course of action in progress in order to solve some communicational ‘trouble’ (Brouwer et al. 2004;
Schegloff 1992; Schegloff et al. 1977; Schegloff, Koshik, Jacoby & Olsher 2002). In the case of outright corrections, then, the correcting forms a side sequence in the interaction.
This is illustrated in the example (5.1) below.
(5.1)
S5: [yeah (there) are economics] but of course er i haven’t write i haven’t numbers or a study yeah it’s something like er (O) it’s about O-N-G yeah O-N-G er
45 Some of the findings in this section along with some of the examples used here are further discussed in Hynninen (2012).
88
S2: N-G-O
S5: N-G-O yeah O-N-G is in spanish sorry S2: [(that’s okay)]
S5: [N-G-O yeah] N-G-O and er they have a lot of pages but of course nothing about er numbers.
(SG, V07A1GW) In the example, S5 refers to a non-governmental organisation with the acronym ONG, which is then corrected by S2 to NGO. The correction is followed by S5’s acknowledgement of the correction, and his explanation that he mixed up the Spanish and English versions of the acronym (S5 is an L1 speaker of Spanish). S2’s L1 is Portuguese and she also reported speaking some Spanish, which may have helped her in deciphering what S5 meant with ONG. However, we can see that S2 still corrects S5. The correction is done towards ENL, but it is made by an L2 speaker of English. It is possible that in this case S5’s hesitation (ONG yeah ONG er) may have justified the correcting for S2.
From the example we can see that the correction creates a side sequence in the interaction: the correction is followed by S5’s repetition of the ‘correct’ item, that is, before returning to the topic, the participants negotiate which linguistic item to use. As in the example, studies in CA (e.g. Jefferson 1987: 90; Schegloff et al. 1977) have shown that corrections are typically followed by repetition of the correction, or if the correction is rejected, repetition of the original item (also known as the ‘repairable’).
In CA, especially in relation to L1–L2 interaction, corrections have been treated as a subgroup of repairs, referring only to instances where an interlocutor replaces an
‘erroneous’ linguistic item used by the speaker – ‘erroneous’ meaning incorrect when compared to ENL standards (see Brouwer et al. 2004). Repairs more generally refer to all instances where an interlocutor produces an alternative version of what a speaker has said. In addition to linguistic repairs (i.e. corrections), this includes repairs that, for instance, deal with the factual content of a speaker’s utterance. Since such repairs do not concern language regulation, this study focuses on linguistic repairs only.
In L1 interaction, it is assumed that everyone speaks the same language and knows how to speak it, which means that the norms of speaking are in a way given, and errors can be defined as deviations from these norms. With L2 speakers present, however, the norms can be problematised. In L1–L2 research, such problematisation is usually not done, and the approach adopted relies on a pre-determined definition of correctness
89
norms, often equated with a NS’s understanding of ‘normal language use’ (see Brouwer et al. 2004; Kurhila 2003: 44). These studies also seem to exclude instances where a correct form is replaced by an incorrect one, which means that an incorrect form used as a repair is not counted as correcting. If we, however, want to approach interaction from the perspective of what participants construct as correct, we cannot automatically assume that they follow some pre-determined correctness norms. We thus need to start by considering all instances where an interlocutor changes a linguistic item in a speaker’s turn and where these instances create a side sequence in the interaction – thus suggesting that the change is treated as a correction by the participants.
Since my focus is on items that are modified, instances where an interlocutor anticipates what a speaker is trying to say, for instance, by completing a speaker’s utterance, and where the intervention thus takes place without modification, are excluded from the analysis (see example 5.2). These include instances where the speaker seems to hesitate as if searching for a word, which is then followed by interlocutor intervention.
This is illustrated in example 5.2.
(5.2)
S2: but but should we set up (xx) (possibly) do something else because i just read um that there has been P-V-A put down in 2006 that predict er wh- er what is it um, population viability, an-
M2: analysis
S2: analysis yeah and that predicted that the kind of er positive um fut- future (xx) that it is the population’s going to become more stable ((…))
(GG, B09C2GGW) In the example, S2 hesitates with the term population viability analysis, which can be seen in the use of the hesitation marker um and the two pauses. This triggers M2 to intervene in S2’s turn by offering the missing element of the term, which is then acknowledged by S2 repeating the offered item. These instances were not counted as corrections, since they do not correct an item already expressed by a speaker; but rather anticipate what the speaker is trying to say and apparently help him or her out.
Also, other-initiations of repair (e.g. Egbert 1997, Schegloff 2000), such as non-understanding signals and confirmation questions as requests to repeat or reformulate an utterance have been excluded from the analysis of corrections (cf. section 5.2.2 on reformulations). Such initiations may be instances of monitoring the speaker’s language,
90
but they may also indicate trouble in hearing or understanding the contents. It thus appears that although they sometimes function as a language-regulatory practice, they do that only in part.
The focus, thus, is on linguistic other-repair, or language corrections, that is, linguistic repairs that are done by the interlocutor either without initiation from the speaker as outright corrections (other-repair) or with initiation from the speaker as self-initiated corrections (self-self-initiated other-repair). Let us now turn to the data analysis.
Types of corrections
Instances of outright language corrections of spoken English as well as self-initiations were scarce in the data. Table 5.1 shows the number of language corrections found (outright corrections and self-initiated corrections) and their distribution into three types of linguistic correction: pronunciation, grammar and lexis. Only corrections concerning spoken English are included.
Table 5.1 Number of language corrections (ca. 20 hours of data)
Outright corrections Self-initiated corrections In total
Pronunciation 7* 1 8
Grammar 4 1 4
Lexis 9 11 20
In total 20 13 33
* 5 of these corrections were made in relation to the same pronunciation.
The table shows that language corrections were not common in the data (33 instances in 20 hours of data means one correction in approximately every 36 minutes).
However, as we can see, corrections were done even if the speaker did not initiate it, and outright corrections were somewhat more common than those initiated by the speaker.
As to the types of language corrections, we notice that most corrections concerned lexis. This is in line with Smit’s (2010) study on student-teacher interaction in English-medium lectures. In Smit (2010), outright and self-initiated corrections were more common than in this study (the 49 instances given in table 5.2 are from 6 hours and 33 minutes of data, which means one correction in approximately every 8 minutes), but the distribution of the different types of linguistic correction is similar. As illustrated in table 5.2, in Smit (2010: 202–209), language corrections concerned lexis in 19 out of 27
91
instances and only 8 concerned pronunciation or grammar, and a similar uneven distribution could be seen in relation to self-initiated corrections, with all but one instance concerning lexis.
Table 5.2 Number of comparable language corrections in Smit (2010: 202–209)
Outright corrections Self-initiated corrections In total
Pronunciation 3 1 4
Grammar 5 0 5
Lexis 19 21 40
In total 27 22 49
There thus appears to be a tendency to focus more on lexis than other linguistic aspects. Then again, lexis can be difficult to classify as a purely linguistic category: as Smit (2010: 182–183) points out, it can be impossible to distinguish between repairs focusing on lexical, instead of factual elements, which means that double categorisation may be in place. In this study, all borderline cases have been included as lexical corrections (see example 5.3). Considering that most linguistic corrections in my data and in Smit’s (2010) data concerned lexis – a category which may overlap with factual repairs – it appears that even less attention may have been put on language as opposed to meaning than what the numbers suggest. Thus, the emphasis on lexis instead of pronunciation and grammar corrections, as well as the sheer scarcity of corrections implies that, in the study events, it was generally not acceptable to correct someone’s speech and it was generally not necessary to do linguistic corrections in order to achieve mutual understanding.
The scarcity of correcting is in line with earlier studies on ELF interaction. For instance, Smit’s (2010: 202) findings show that repairs were common, but that there was a proportionately higher amount of interactional repair due to, for instance, mishearing, and factual repair dealing with content, rather than language (in total 72.4% out of all the repairs), compared to linguistic repair (27.6%). Also, if we focus on only the two categories explored in this study (i.e. outright corrections and self-initiated corrections) in Smit’s (2010) data, and look at the distribution of repairs given in table 5.3, we see that slightly over half of the language corrections were done by the interlocutors (49 instances out of 94 were other-repairs), which further means that such corrections formed only ca.
14% of all the repairs in Smit’s data. Linguistic other-repairs thus appear to be relatively uncommon in ELF (see also Mauranen 2006b).
92
Table 5.3 Distribution of repairs in Smit (2010: 202–209)
N % out of all repairs
Linguistic repairs
• Other-repairs (self-initiated and outright
corrections) 49 14.4%
• Other types 45 13.2%
Non-linguistic repairs 247 72.4%
In total 341 100%
Interestingly, the scarcity of correcting is in line with studies conducted on L1–L2 interaction. For instance, in Kurhila (2003, 2006), NSs did linguistic corrections, but a lot of ‘erroneous’ use was left uncorrected, which places correcting in a minor position in the interaction. What is more, most corrections were found to be embedded ones, that is, they did not form a side sequence in the interaction (Kurhila 2003: 137; see Jefferson 1987; section 5.2.1). Chun, Day, Chenoweth and Luppes (1982) further show that NSs correct L2 speakers’ language, but only marginally in relation to correcting their factual errors: factual errors were corrected in ca. 90% of the cases, whereas language (word choice, syntax and omissions) were corrected only in ca. 8% of the cases. Although the correcting in Chun et al. (1982) is restricted to NSs, the interactional relevance of correcting appears to remain small, which lends support for studies that suggest the
“normality” of L2 talk (Wagner and Gardner 2004). The scarcity of linguistic corrections in this and Smit’s (2010) study suggest the same for ELF interaction. This is in contrast to language classroom interaction, where linguistic corrections have been found to be prominent (e.g. McHoul 1990). ELF thus differs from interaction in language classrooms in that linguistic form is given (much) less attention.
What was corrected and by whom?
Sometimes linguistic form, however, matters in ELF interaction as well. What, then, was corrected and by whom in my data? In the following, I look at lexical, pronunciation and grammatical corrections of spoken language, and end by focusing on examples of unconventional language that was left uncorrected.
93
Lexical corrections formed the biggest group. In example 5.3 (reproduced from chapter 4, example 4.2), the correction is done by a teacher (T1) during a student’s (S4) presentation.
(5.3)
S4: ((…)) they have different climatic conditions er ranging from (sahara) or semi-arid zone to the tropical zone where annual rainfall is er is 100 or 15000 millimetre annually but er [beekeeping]
T1: [not 15000] 1000- er 1500
S4: yeah 1500 yeah 1500 1000 and 500 millimetre annually but er it’s still in separate areas scattered area(s) ((…))
(TLC, V08D3Sp) We can see that the lexical correction is a correction of facts: the correction concerns the amount of annual rainfall in a specific area. This means that while this is counted as a lexical correction, T1 did not intervene in order to correct lexis as such, but in order to correct the contents.
The correction in example 5.1 (reproduced below as 5.4 for convenience) more clearly focuses on lexis. The example is from the student group, where no teachers were present.
(5.4)
S5: [yeah (there) are economics] but of course er i haven’t write i haven’t numbers or a study yeah it’s something like er (O) it’s about O-N-G yeah O-N-G er S2: N-G-O
S5: N-G-O yeah O-N-G is in spanish sorry S2: [(that’s okay)]
S5: [N-G-O yeah] N-G-O and er they have a lot of pages but of course nothing about er numbers.
(SG, V07A1GW) In this example, the acronym ONG is corrected by S2 to NGO. S2 is an L2 speaker of English, but as we can see in the example, she takes on the role of language expert. She
94
introduces the acronym in English, apparently to ensure mutual understanding in the group.
Also in the following example (5.5), an L2 speaker of English acts as language expert.
(5.5)
S5: we have (an excurse) on on on on saturday yeah S1: yeah
S4: mhm
S1: the field trip
S5: yeah and where we are going to g- to go S1: i don’t know
S4: to porvoo and askola
(SG, V07A3aGW) In this example, the correction is delayed as it only occurs after minimal responses by both S1 and S4. Since S5 does not take the floor after the minimal responses, S1’s turn could also be seen to function as a verification of understanding, a paraphrase of an excurse, rather than a correction. The marked lexical item is changed to a standard one, but the change is done by an L2 speaker of English.
Many other-corrections initiated by the speaker dealt with field-specific terms, as illustrated in example 5.6.
(5.6)
S7: ((…)) or two or three centimetres per day it’s very very quickly. (xx) (the growth). okay , so erm is i i don’t know is in (english is) ((monomic)) (plant) to have the both sex in the same plant ((monomic)) in english
T2: jaa it’s monoecious S7: monoecious T2: mhm
S7: okay (so it’s a) monoecious plant
95
T2: but we could stop here the very strange thing with maize is that male and female ((…))
(TLC, V08D5Sp) In the example, the student explicitly expresses that he does not know what to call a specific type of plant, and the teacher then provides the requested term. These kinds of instances placed the teachers in the position of not only experts in the field but also of language experts. Similarly, the mentors were asked about the English names of Finnish associations in the guided group (example 5.7).
(5.7)
S3: i think er <FINNISH> suomen luonnonsuojeluliitto </FINNISH> what’s that in english
M2: i think it’s the finnish association for nature conservation
S3: okay they they complained to EU couple of years ago i don’t know was that any help or has EU decided anything
(GG, B09C3bGGW) In this example, the student explicitly asks for a translation of a Finnish association before continuing on the topic. The findings thus to an extent correspond to Smit (2010:
362–365), who found out that teachers were relied on for subject-related terminology.
In the guided group, also the presence of the English instructor (E1) was made use of. In the second meeting, E1 stayed a while to observe the group, and during this time we can see a slight increase in questions concerning unfamiliar words in English. Typically these kinds of questions occurred only once during a meeting, whereas three questions were directed at E1, twice exploiting E1’s knowledge of Finnish, as in example 5.8.
(5.8)
S1: ((…)) protection of the species and er then um to focus er research on the seal so that’s like the four main points and under the f- first one with the the (anthropogenic) facts there’s lots of stuff about er the fishing and using these what could a <FINNISH> katiska </FINNISH> be in finnish er i mean english (xx)
E1: traps i would say [(maybe) fish traps]
96
S1: [trap] a fish trap okay instead of the net so that’s like the most important thing under this (anthropogenic) thing but er just like um to sort of (xx) the focus on the other bits as well like getting people involved and (xx)
(GG, B09C2GGW) The example shows how E1 was used as a dictionary, and thus a language expert.
E1 also attended the beginning of the last session of the group work in order to comment on the students’ mock presentation. He mainly gave general advice on presenting, and commented on some of the students’ language (e.g. pronunciation as in example 5.9).
(5.9)
E1: mhm <NAME S3> just before you sit down you did really well with the terminology and stuff (there) but just a couple of words to to keep (an an ear at o-) one was survival (xx) (incorrect) survival (but) the other one was strict strict protection i think you pronounced it /straikt/ or something like that (but they are the only two that)might confuse someone if they are (a bit) mispronounced (the others were fine I think) just strict protection
(GG, B09C6GGW) In this example, the correction is given in the form of advice, which was typical of E1. In general, E1’s presence seems to have increased the participants’ focus on language.
As we saw in example 5.9, E1 paid attention to the students’ pronunciation, but otherwise pronunciation corrections were rare in the data. The rest of the pronunciation corrections occurred in the teacher-led course, focusing on one particular lexical item, as illustrated in example 5.10 where T2 intervenes in S7’s pronunciation of maize.
(5.10)
S7: ((…)) (there are two principal) classes of of /maıs/ is the er both are the most in- cultivated in around the world er this yellow /maıs/ and white the white /maıs/ is principally use- used for the human (xx) and the yellow /maıs/ is used for the [(a- anim-)]
T2: [kindly] pronounce it /meız/ so because it would be repeated we are talking about
S7: [/meız/]
97
T2: [/meız/] not /maıs/
S7: excuse me please /meız/ (xx) /meız/ so yellow one is used for the (er nutrition of the a-) animals (not the) people, ((…))
(TLC, V08D5Sp) We can see that the student’s pronunciation of maize evokes the word mice, which is treated as improper in a presentation dealing with maize (kindly pronounce it maize, would be repeated). Considering that the teachers did not intervene with S7’s – or any other student’s – pronunciation otherwise46, there seems to have been high tolerance for variation in pronunciation. The lack of pronunciation corrections in the group-work events, except those done by E1 in the guided group, further support this conclusion.
Similarly to pronunciation corrections, correcting grammatical form was rare.
Example 5.11 comes from the first course session of the teacher-led course where the students chose presentation dates and topics.
(5.11)
S9: i’d like also to present on the 22 T1: 22nd yes
S9: 22ndand <FIRST NAME S9>
(TLC, V08D1S) In the example, T1 corrects S9 by replacing the cardinal number used by S9 with the corresponding ordinal number, which is then repeated (and thus accepted) by S9. Again, the correction appears to make use of ENL norms, but it is done by an L2 speaker of English.
What was left uncorrected?
In all, the examples above show that corrections were made towards ENL norms.
However, a lot of usage deviating from ENL norms was left uncorrected. This is
46 S7 kept going back to the corrected pronunciation and the teachers ended up correcting him also later in the interaction, and even suggested that S7 talk about corn instead of maize. However, corrections were always related to the pronunciation of maize, and no other lexical item. Also, on one occasion, another student specifically asked how to pronounce a word.
98
illustrated in the examples below. In example 5.12, S4 talks about the boundaries of reserve areas.
(5.12)
S4: do you know the (xx) effect (xx) maybe this is a reserve or you can’t fish inside but you you you can fish er outside the line <SU-M2> [mhm] </SU-M2>
S4: do you know the (xx) effect (xx) maybe this is a reserve or you can’t fish inside but you you you can fish er outside the line <SU-M2> [mhm] </SU-M2>