‘BROKEN ENGLISH’ AND STANDARD LANGUAGE IDEOLOGY

In a standard language culture, ‘the standard ideology decrees that the standard is an idea in the mind—it is a clearly delimited, perfectly uniform and perfectly stable variety—a variety that is never perfectly and consistently realized in spoken use’ (Milroy Reference Milroy2001:543, emphasis in original). In some situations, as we show here, language that is perceived as deviating from this uniform ideal may be described as ‘broken’. Thus, such a description serves to reinforce the ideal standard with which such language is seen to contrast.

Not all cultures are standard language cultures, as Milroy (Reference Milroy2001) points out. Grace (e.g. Reference Grace1992:122) notes that in his work in Melanesia, ‘in some areas the people seemed to have no conception of what a language is and no sense of belonging to a linguistic community’. In such situations, speakers were multilingual and were likely to think of communication ‘in terms of… making themselves understood… freely drawing on whatever means are available’ (Grace Reference Grace1992:124). When language is regarded in this way, describing someone's language as ‘broken’ could have little meaning, except as perhaps inability to communicate at all. The use of this descriptor to describe those who clearly can communicate can therefore be seen as functioning to maintain a standard language culture.

Lippi-Green's definition of standard language ideology also invokes power as relevant to what is considered standard, describing it as ‘a bias toward an abstracted, idealized, homogenous spoken language… drawn primarily from the spoken language of the upper middle class’ (Reference Lippi-Green2012:67). As both Milroy (Reference Milroy2001) and Lippi-Green (Reference Lippi-Green2012) note, however, complete standardization is an ideal rather than an actuality, and even the language of supposedly ‘standard’ speakers varies. Such variation may go unnoticed due to the process of erasure, described by Irvine & Gal (Reference Irvine, Gal and Kroskrity2000:38), which ‘renders some persons or activities (or sociolinguistic phenomena) invisible. Facts that are inconsistent with the ideological scheme either go unnoticed or get explained away’. Thus, we expect that while any speaker may use language that deviates from the idealized standard, the language of nonnative, non-upper-middle class (Lippi-Green Reference Lippi-Green2012), non-White (Shuck Reference Shuck2006) speakers is most likely to be noticed and described as ‘broken English’.

THE CURRENT STUDY

This study investigates how the label ‘broken English’ is used to construct speakers in natural contexts, when examples are not being elicited by a researcher. In order to find examples of how people naturally use the label, we searched two general-language corpora for examples of this term, focusing in particular on the internet as a corpus. The web is a ‘key space for enacting social practice and for reflecting and shaping social processes and problems’ (Mautner Reference Mautner2005:810). The topicality of internet data, combined with the proliferation of blogs and other open discussion forums, make it a valuable source for examining ideologies about language. While using the web as a corpus entails the use of data that represents uncontrolled and unknown variables, it allows investigation of a wide range of naturally occurring language data, not limited to participants the researchers could contact or observe personally. Corpus-based research methods have been employed by discourse analysts and critical discourse analysts for over twenty years (e.g. Hardt-Mautner Reference Hardt-Mautner1995; Baker, Gabrielatos, KhosraviNik, Krzyżanowski, McEnery, & Wodak Reference Baker, Gabrielatos, KhosraviNik, Krzyżanowski, McEnery and Wodak2008). Such research may be considered corpus-informed (Lee Reference Lee, Bhatia, Flowerdew and Jones2008), in that corpora supply the examples, but the analysis focuses on investigating the examples in greater depth.

‘BROKEN ENGLISH’ AS CONSTRUCTING NEGATIVELY VIEWED OTHERS

To see how the term ‘broken English’ is used to construct and position speakers as nonnative, we first consider an example from the web data in some depth, drawing on all of the corpus data to show the degree to which the findings based on the one text are representative. Excerpt (1) comes from a city's website, in an article entitled ‘Traveling thieves—like Cicadas—are back’. The bulleted list is the complete list given in the article; line numbers are added.

1. (1) 1   Here are some indicators that should put you on alert and raise your

   2   suspicion:

   3  • They are looking for unlocked doors during daytime hours.

   4  • Elderly residents working in the yard with the garage door open are

   5    prime targets.

   6  • A female usually does the walking through yards with a male

   7   accomplice driving nearby.

   8  • If confronted, they will be looking for a dog, looking for directions, or

   9   ask for a glass of water.

   10  • They may pose as workers, inquiring about tree work or driveway

   11   blacktopping.

   12  • For the most part, they are not violent, unless they are trapped inside

   13   your house. In that case they use whatever force is necessary to escape.

   14  • They tend to prefer SUVs and have out of state license plates.

   15  • They may have a heavy accent and speak broken English.

   16  • They may use diversion techniques, where the resident is approached

   17    by one subject and a conversation is started, while the other subject is

   18    selecting your best jewelry.

Two major observations relevant to how the descriptor ‘broken English’ is used to construct speakers relate to (i) the kind of language described as broken, and (ii) whose language is so described. The first relates to a more literal definition of the term, while the second considers what social identities it is used to construct. We begin by looking at the language that is described as ‘broken’, noting the vagueness of the label ‘broken English’ (line 15). It is presented as a self-evident term; no definition or further information is given, suggesting that readers are expected to intuitively understand its meaning. Less directly, we can see that in this case it is treated as distinct from nonnative accent, since ‘heavy accent’ is listed in addition to ‘broken English’ in the same line. Looking at the rest of the list of indicators, we also can see that the travelling thieves being described as speakers of ‘broken’ English are apparently able to start a conversation well enough to distract a potential victim for at least a few minutes (lines 16–18) and are able to give excuses for their presence if someone questions them (lines 8–11). Thus, in this case ‘broken’ English is not used to indicate lack of ability to communicate, a clear example of standard language ideology.

Looking at the larger dataset, we see that ‘broken English’ is most often used in this way, to describe a speaker who is indeed able to communicate in English at least to some degree. However, an assumption that readers will intuitively know what ‘broken English’ means only seems plausible if it is used to describe any language sample that a listener (or reader) identifies as nonnative language, as it is used to describe a very broad range of language use. In the web data, there are cases in which the language described was directly observable, usually as a written text, allowing us to see what kind of language may be labelled as ‘broken English’. We found just two of the examples to be difficult to understand, at least without background knowledge of the topic being discussed, as in (2); the other example was from a handwritten letter that was difficult due to both handwriting and unfamiliar subject.Footnote ²

1. (2) Somewone can give me the link to download this file because you are into play at the version 1.6 whit the game and I want to change my files ram in real game like mario kart 64.Thx

However, excerpt (3) was also from a text described as ‘broken English’.

1. (3) The remains of the reactor were entombed within an enormous steel and concrete sarcophagus, so it is now relatively safe to travel to the area—as long as one do not step off of the roadway and do not stick in a wrong places…

This example includes some minor morphological differences from most high-status native varieties: ‘do’ without a third-person agreement marker (‘does’) and the indefinite article used with a plural (‘a wrong places’), which is presumably the reason it was identified as ‘broken English’. At the same time, it uses very sophisticated vocabulary and imagery, as in the terms ‘entombed’ and ‘sarcophagus’ (a fitting place for the ‘remains’ of the reactor). It is also a complex sentence, including multiple clauses, yet is highly comprehensible. In fact, while other examples were not as sophisticated, nearly all were highly comprehensible.

In other cases, the language labelled as ‘broken’ was not directly available, but the labeller included more in their description that helps provide a sense of how the term was intended. Examples (4)–(6) provide a sense of the range of descriptions.

1. (4) He spoke very, very broken English, but we were very well able to communicate.

2. (5) Dr. A is from Greece and spoke broken English which was a problem in our fully understanding what he was talking about. However we definitely got the jist of it.

3. (6) Ford Credit in Panama City, Panama. Could not understand their broken English.

In (4), we see an example in which there was apparently no difficulty in communication in spite of ‘very, very broken English’, while in (5) communication is described as being hampered by Dr. A's English. These types of descriptions were most common, but there was a small number of examples, as in (6), in which ‘broken English’ is used to imply total or near-total lack of ability to communicate in English (although as we note in our discussion of (11) below, such claims should not necessarily be taken at face value). In other words, we see that the term was used to describe a very wide range of language, usually language that was mostly understood and that sometimes included only negligible differences from native speech. Thus, consistent with standard language ideology, the basic meaning of ‘broken English’ appears to denote language that deviates from an undefined norm, reflecting a focus on accuracy rather than communicative ability. One gloss would be ‘incorrect’, which like ‘broken’, emphasizes a lack or deficit compared to that which is ‘correct’ or ‘whole’. The emphasis on lack or deficit suggests that use of the term also indexes a negative stance towards the language being described, with the idealized ‘correct’ speech to which it is being adversely compared left implicit.

Returning to the analysis of our text in example (1), we consider our second question: whose English is described as ‘broken’? What social identity or identities does it help to construct? In this case, nearly the only information given about the speakers of ‘broken English’ is that they are traveling thieves. The reference to the speakers’ language is particularly troubling here because it is not clear how or why it is relevant. Since it is on a list of ‘indicators that should put you on alert and raise your suspicion’ (lines 1–2), we may assume that this particular point is intended to help identify thieves. The label's lack of specificity, however, suggests that the reader should treat any ‘broken English’ speaker—however they interpret that term—as a potential thief, and the data above suggest that at least anyone identified as a nonnative English speaker could be included. In addition, it is worth noting that this is not even presented as a description of a specific suspect, but as a generic or typical thief. Thus, readers are encouraged to view any nonnative English speaker with suspicion.Footnote ³

Looking at the larger dataset, including both the web and the COCA examples, we see that this association between nonnative English and criminal behavior is not an isolated example. In fact, fully 20% of all examples from the web are descriptions of speakers as criminals or suspected criminals. The COCA examples were less likely to be descriptions of criminals, although such examples comprised 8% of the COCA data. In both corpora, there is a definite tendency for ‘broken English’ to be used to describe highly negative or marginalized speakers, however. The web data provide more examples of extremely negatively perceived speakers; in addition to criminals, two other substantial groups were of (bad) customer service representatives and of noncriminal persecutors by whom the writer apparently felt harassed or victimized in some way, including collection agents and telemarketers, as seen in (7).

1. (7) Broken English. Selling vacation package. Has called twice today despite being asked not to call again and informed we are on no-call list the first time.

In the COCA data, by contrast, the term was more likely to be used for highly marginalized speakers, such as low-status workers or vendors, as in (8).

1. (8) Vendors called in broken English, imploring them to buy as if their lives depended on it. (fiction)Footnote ⁴

Many further examples in both corpora were of speakers who were described as victims of discrimination, wrongful accusation, murder, or other bodily harm, although this tendency was more prominent in the COCA data, as in (9).

1. (9) She and the others don't understand, don't know if they're being kidnapped or arrested. … Everyone is crying. This girl says in broken English that she doesn't know what to do or where to go. (spoken)

Ultimately, in both corpora, about half of all examples were of speakers who were presented as highly negative or as marginalized. This is not to suggest that the other half were presented positively; rather, many examples in both corpora were indeterminate or described a group in general. We found relatively few examples of high-status nonnative speakers being referred to in our data set—the doctor in excerpt (5) being one of these—and no instances of native speakers in high-status positions or situations described as using ‘broken English’. There were, however, two instances of low-status (young and possibly non-White) native speakers whose language was so described. While the denotation of ‘broken English’ appears to apply to any language that is perceived as differing from an ideal standard, it is highly likely that the people describing the language have been exposed to high-status native speech that also differs, but have not noticed it or remarked upon it in writing on the web. In other words, such usage has undergone erasure (Irvine & Gal Reference Irvine, Gal and Kroskrity2000). The description of speakers’ language as ‘broken’ therefore already involves a level of interpretation, as it is used not only to identify language as deviating from an imagined standard, but to construct its speakers as nonnative or low-status (and usually both), consistent with Lippi-Green's (Reference Lippi-Green2012) characterization of standard language ideology. In so doing, it establishes a clear power differential between the labeler, who claims the right to assess the other's language and find it deficient, and the speaker whose English it is considered acceptable to judge. In all of the examples, using ‘broken English’ to refer to someone's language can be seen as constructing them as Others, people whose language is worthy of note because of its deviation from assumed implicit norms. It is a linguistic analogue to the out-of-state plates mentioned in line 14 of excerpt (1).

Further construals of the ‘broken English’ are added to this basic connotation of Otherness. Based on the people who are most commonly described as speaking ‘broken English’ in this dataset, the label is likely to construct this Other's identity as undesirable, as it is typically associated with negative circumstances, with the association of ‘broken English’ with criminal activity as in excerpt (1) as a relatively common example. In these cases, Otherness is interpreted as being associated with negative traits, such that one further interpretation of a ‘broken English’ speaker might be ‘criminal’. Sometimes this interpretation is made very explicit, as in excerpt (10).

1. (10) While there are some complete retards and uneducated people in the military (you're reading one right now), most Soldiers and yes, even Marines, know basic grammar. If the person you are talking to speaks in broken English and can't put simple sentences together, it's probably a scam.Footnote ⁵

A final observation about excerpt (1) is that it is unclear who the author of the ‘broken English’ label is (i.e. the one ‘who has selected the sentiments being expressed and the words in which they are encoded’, Goffman Reference Goffman1981:144). For example, it is possible that a theft victim used the term in describing a specific thief, or that victims used some other description that the police then labeled as ‘broken English’, or that it is the writer's own term. Since the use of the term in this context could be considered prejudicial (imagine if the warning read ‘they may be African-American’), the writer might have been expected to distance themselves from the term, but they do not do this. Instead, it is presented as if it were a completely unproblematic descriptor to help readers identify suspicious persons. Thus, ‘broken English’ appears to be a socially acceptable way not only of referring to the language spoken by a nonnative English speaker, but a socially acceptable way of linking such language and such speakers to criminal activity.

FUNCTIONS OF THE LABEL ‘BROKEN ENGLISH’

Although the first excerpt and many other examples from the data link ‘broken English’ to criminal activity, the overt function of the label ‘broken English’ in many of these cases might be simply construed as one of describing. The data from COCA, which comes from highly edited, polished sources, overwhelmingly uses the term in this way. Such uses may be intended as a ‘neutral’ description of any English that is identified as nonnative, used by writers who do not have (or who do not expect their readers to have) more technical vocabulary or understanding to more precisely describe the language, although as we have noted, the use mainly to describe highly negative or marginalized speakers suggests an additional, less explicit, function of constructing the speakers as Other.

In other cases, the label is used to explicitly criticize someone's language. While such usage was nearly nonexistent in COCA and rare in web data that was highly edited (like example (1)), it accounted for over half of the examples of apparently spontaneously produced web data such as reader comments on blogs and discussion forums, as in (11), highlighting an important difference between candid language directed toward a specific audience and scripted language intended for a general audience.

1. (11) And sorry LordTorok but I cant understand a damn thing you said there… I mean…. was there a question in that broken english ass ramble?

Here, the writer uses a formula used by Shuck's (Reference Shuck2004) participants, also found by Subtirelu (Reference Subtirelu2015) to be common in written comments about ostensibly nonnative-speaking instructors, ‘I/you/we couldn't understand a word he/she/they said’ (also including ‘I couldn't understand a thing’ in Shuck's data (Reference Shuck2004:203)). As with their data, it is unlikely that the poster literally could not understand ‘a damn thing’. Subtirelu argues that this formula uses ‘the social acceptability of hyperbole in complaints or narratives to erase [nonnative English-speaking] instructors’ English competency’ (Reference Subtirelu2015:52). Similarly, this poster's criticism suggests that the previous poster may be safely insulted and their message ignored because of their language use, an explicit example of Othering.

A less common, but closely related, function that showed up mostly in this same category of relatively spontaneously produced web data was as an apology for one's own language use, as in (12), perhaps as a defense against a possible attack like the one above.

1. (12) I'm pretty sure someone did asked this before but I cannot find it anymore…How many catapults are needed to destroy a building depending on its level? Thank you for your time guys and sorry for my broken english.

Thus, the term ‘broken English’ is most likely treated as an explicitly negative descriptor—that is, as a criticism of someone's English, whether someone else's or one's own—in the least edited data. This provides further evidence that the more basic definition of the term is negative, consistent with its deficit-oriented denotation, although it may be pressed into service as a more neutral descriptor when a writer wishes to give a sense of the speaker's language but does not have the technical vocabulary to describe it. However, the negative sense, together with its connotations of Other, appears to be more fundamental and pervasive, calling into question the neutrality of any use of the descriptor. As the descriptor is commonly understood as criticism in more candid language use (as in examples (10), (11), and (12)), it is likely that the phrase can be interpreted as indexing a negative stance even within seemingly neutral contexts. Thus, although it is not used to explicitly criticize in highly edited discourse intended for a general audience, it is noteworthy that the label is still used in such discourse, in spite of its connection with sometimes highly disparaging criticism of nonnative speech and negative constructions of its speakers.

POSITIVE USES OF ‘BROKEN ENGLISH’?

We next consider a very different excerpt, this time from the COCA data. Excerpt (13) describes ‘The New Fabio Romance Line: 900-90-FABIO’, which allows a caller to take various ‘romantic quizzes, administered and graded by the Fab himself’ for $1.98 per minute.

1. (13) Fabio, that pectorally correct romance novel cover boy, puts his charmingly broken English to work. After an appropriate introduction (‘The man. The legend. Fabio.’), the previously taped Fab-voice purrs ‘Welcome… mon amour, to The Language of Love Line, where we can share our fantasies towards a more fulfilling understanding of love and romance.’ (newspaper)

There is one major similarity between this excerpt and excerpt (1): the lack of clarity about what the label ‘broken English’ refers to. There is nothing in the quoted portion to suggest ‘brokenness’. While in other cases in the data switching to another language (in this case, French) for a few words was associated with ‘brokenness’, such an explanation does not make sense here, as there is no implication that Fabio does not know ‘my love’ in English, and he is not a native French speaker either. (In fact, the use of French here is more likely to index romance than ‘brokenness’ for US readers.) The quotation is completely nativelike, and since all of Fabio's speech is prerecorded and certainly scripted, it is unlikely that he used any nonnative grammar, lexicon, or pragmatics at all. Thus, the only likely source of identifiable nonnativeness is pronunciation, suggesting that in this case, contrary to excerpt (1), ‘broken English’ can indeed be used to refer to accent.

Perhaps more striking, especially in contrast to the previous data, is the pairing of the adverb ‘charmingly’ with ‘broken English’, a rare example of a positive descriptor put together with the term. There are only two other modernFootnote ⁶ examples of this: another from COCA that refers to Sammy Sosa's ‘delightful broken English’ and a comment on a fanfiction story from the web data about the Russian duo t.A.T.u. (‘Yuli's broken English was so cute!’). There are a number of observations to be made about these positive examples. First, each of these cases refers to a specific famous individual, although one example refers to a fictionalized description of that individual. Second, they also all refer to people who are presented as highly attractive for nonlinguistic reasons, as in the description of Fabio as ‘pectorally correct’. Third, this example refers to someone who was not encountered in the US; instead, the writer had to call—and pay—to hear a recording of the speaker on the telephone. Only the reference to Sammy Sosa appears to be based on an actual interaction with the ‘broken English’ speaker, and while the interview likely took place in the US, the interviewer sought out Sammy Sosa for the interview; he is not a person that anyone might expect to encounter by chance in the US. Thus, these speakers are already quite distant from the in-group, but have considerable social capital. In the case of these distantly superior speakers, the negative stance indexed by the term ‘broken English’ may be interpreted as condescending. This condescension of calling attention to their language ‘deficits’ actually makes them seem more accessible and therefore ‘charming’, ‘delightful’, or ‘cute’. In other words, their high status is reduced with the use of the ‘broken English’ label, making them more relatable.

There are several other examples of ostensibly positive uses, as in (14), from the web data.

1. (14) Our week in Suwon ended with the closing ceremonies and farewell dinner at the 5 Star Ramada Plaza. The captains collected the hardware as all the teams enjoyed a Korean hip-hop display and speeches by the Korean lacrosse community. Everyone had a great time trading gear and getting to know the other guys in the tournament through broken english and big smiles.

Such examples do not include a positive descriptor paired with ‘broken English’, but describe the interaction itself in positive terms. They also refer to interactions that took place outside of the US (or other primarily English-speaking country), in which the ‘Other’ status of the interlocutors was more likely to be treated as a given. In these examples, the people who were spoken to in ‘broken English’ did not know the language of the country they were in, so the ‘broken English’ is presented as what allowed communication to take place at all, as opposed to primarily as a potential criticism of that communication. Interestingly, these more positive usages still function to maintain standard language ideology, as the language that allows communication to take place is still described in terms of deficit. Further, they also serve to support monolingualist ideology, because ‘broken English’ only becomes an asset outside of countries of primarily English speakers, in comparison to the expected monolingualism of speakers in that country.

MEANINGS OF ‘BROKEN ENGLISH’ AND STANDARD LANGUAGE IDEOLOGY

Since ‘broken English’ was very frequently described as allowing for successful communication, it appears to be a folk assessment of a speaker's linguistic accuracy, or match to an assumed norm, rather than of communicative ability. In fact, the descriptor may be used to refer to nonnative features that may be common even at very advanced levels of proficiency, such as differences in article use. This focus on accuracy allows the descriptor to be used to indicate any speech identified as nonnative (or in limited cases, stigmatized native). As a result, its use simply emphasizes the nonnativeness or nonnative-like-ness of the speaker whose language is being described, constructing them as Other.

Whose speech is actually identified as ‘broken English’ does not seem to be determined by objective measures of accuracy or proficiency, but rather by issues of status and place, as Blommaert, Collins, & Slembrouck (Reference Blommaert, Collins and Slembrouck2005) argue is the case for the construction of language proficiency more generally. Specifically, the term is usually used to refer to speakers who are viewed negatively or marginalized when they are portrayed as using that language inside the US. In the US, it indexes a negative stance toward a speaker's language, and may additionally construct a negatively viewed identity (e.g. as criminal). The much rarer uses for more high status speakers, such as in excerpt (5), may initially appear to be inconsistent with such an index. However, these occasional examples can be used to imply a clash between someone's language and their position in society, suggesting that their language renders their high status problematic in some way (similar to negative ratings of university instructors whose language is criticized on ratemyprofessor.com, as found by Subtirelu Reference Subtirelu2015).

Outside of the US (or other primarily English-speaking country), speakers with the same degree of bilingualism appear to be viewed more positively, as their ‘broken English’ allows English speakers to function without needing to be bilingual themselves. Both characterizations of ‘broken English’ speakers appear to stem from a monolingualist ideology (Wiley & Lukes Reference Wiley and Lukes1996; Shuck Reference Shuck2006), in which everyone is expected to speak the country's single main language and not any other language. Thus, nonnative varieties of English are perceived as a deficit within an English-speaking country, but as an asset when outside such a country.

If the term were used based on speech accuracy alone, we would expect to also find it used to describe ‘imperfect’ language use by more native speakers, including high-status native speakers. Such speakers may use prescriptively ungrammatical language in some circumstances; in fact, our data included many examples of prescriptively ungrammatical language used to describe others’ ‘broken English’. However, we see evidence of the erasure (Irvine & Gal Reference Irvine, Gal and Kroskrity2000) of such instances of ungrammatical language. In cases where speakers with substantial social capital are known to be nonnative speakers, and thus whose ‘ungrammatical’ language may be resistant to erasure, their English may still be described as ‘broken’, where ‘broken’ is reinterpreted as a more positive feature. Given the negative stance indexed by the term, as found in the present study, it may be that the use of ‘broken English’ is viewed as a kind of blunder, which a classic study in social psychology found rendered an otherwise distantly superior person as more human, and thus more attractive, although the same blunder decreased the attractiveness of an average person (Aronson, Willerman, & Floyd Reference Aronson, Willerman and Floyd1966). Thus, Sammy Sosa's language is described as ‘delightful broken English’, while unknown speakers of ‘broken English’ encountered in the US may be identified as potential scammers or thieves.

Thus, the descriptor ‘broken English’ plays an important role in the maintenance of standard language culture (Milroy Reference Milroy2001), as without an ideology that imagines a uniform and stable variety, describing speech as ‘broken’ when one has comprehended it perfectly would be uninterpretable. More specifically, it plays a role in the social construction of language proficiency within a monolingualist ideology, since in English-speaking countries, only deviations in the speech of those perceived as nonnative and usually lower-status speakers is likely to be identified as ‘broken’.

SLIPPAGE AMONG VARIED CONNOTATIONS OF ‘BROKEN ENGLISH’

Since the degree of explicit negativity associated with the ‘broken’ label varies, more edited examples that include quotations or references to someone else's (possibly spontaneous) speech may demonstrate slippage between supposedly neutral and obviously negative uses. Specifically, because quotations and references to others’ speech make it difficult to determine whose label ‘broken English’ is, as seen in our discussion of the first excerpt, overtly highly negative uses may appear to be sanctioned by sources that might otherwise be expected to be neutral. The first excerpt was from a police warning; another example in (15) is from a news article. Line numbers have been added; H1 and H2 are the headline.

1. (15) H1  ‘If you do not speak English, please hang up’

   H2  Telemarketers told to believe in America or don't call

   1  Veterans of Foreign Wars Post 5642 in Montrose, Pa., has a message for the

   2  many telemarketers that call its answering machine: Speak English.

   3  Those who call the post when no one's around to answer the telephone will

   4  be greeted by a pleasant, female voice on the answering machine saying, “Hi,

   5  you've reached the Montrose VFW, Post 5642. We are an English-speaking,

   6  American establishment. If you do not speak English or believe in America,

   7  please hang up.”

   8  The one-story, white VFW post is a popular gathering point for veterans

   9  along Pennsylvania's Route 706, according to a WNEP-TV report. According

   10  to its post commander, the VFW's phone number is also a popular target of

   11  telemarketers, many of whom speak in broken English.

   12  “We are living in America,” Post Commander John Miner told WNEP. “We

   13  should be able to speak English.”

In line 11 of this example, ‘broken English’ might be understood as simply referring to speakers who are identified as nonnative, based on the most common uses of the term, and its common use as a descriptor in edited text such as this type of news item. The author(s) and editor(s) demonstrate no need to distance themselves from the term ‘broken English’ itself, as it is not in quotation marks or in any way treated as a problematic description. The quote that follows in lines 12–13, however, defines the term as referring to being unable to speak English, and in the context of the full story, refers to speakers who are not worthy to speak or be acknowledged at all, as they are instructed to simply “hang up” (H1 and line 7). Although the post commander is quoted directly for this explicit definition, the term's ambiguity allows it to reinforce the notion that any English identified as nonnative may be seen as ‘broken English’, which may in turn be identified as ‘not English’—and as a problem.

Thus, the ostensibly ‘neutral’ descriptor is placed in a more obviously problematic context in which ‘speaking English’ is equated with ‘believing in America’, such that speaking ‘broken’ (identifiably nonnative?) English suggests that one does not ‘believe in America’. This context parallels the ‘we're in America’ trope that Shuck (Reference Shuck2004) found used to equate America and English, and specifically to complain about the use of a language other than English. In this case, the answering-machine message and the post commander both make this direct connection between the US and the use of English, but the construction of telemarketers as ‘speak[ing] in broken English’ rather than speaking a different language suggests that being in America is properly associated not only with English in comparison to some other language, but with fully ‘standard’ English that is not recognizable as nonnative.

The conflation between ‘identified-as-nonnative English’ and ‘not English’ is especially apparent given that someone who literally does not speak English would not understand the instruction to hang up, since it is only delivered in English. Of course, it is possible that the intended force of the message is not to communicate with the putative non-English speakers at all, but to perform English-speaking nationalism for English speakers. However, another interpretation is that the message is indeed intended for the those who ‘do not speak English’ but implies that the issue is not that they are incapable of speaking English, but that they are simply unwilling to speak ‘properly’ because they do not ‘believe in America’.

IMPLICATIONS AND CONCLUSIONS

We have seen that the label ‘broken English’ is used to emphasize the nonnativeness of primarily marginalized or very negatively viewed people rather than as a consistent measure of communicative ability or even accuracy. Choices about what speakers it is applied to and in what contexts suggest that it is very relevant to power relations, with it used to position speakers as less powerful people whose language may be safely judged, sometimes to the extent of open mockery and abdication of any responsibility to listen or consider the speaker at all.

The negative stance indexed by the term is especially problematic in the absence of a clear or consistent definition. While it tended to be used to construct stigmatized speakers as Others, it was not used to refer to high-status native speakers who might make some of the same ‘errors’. In fact, close scrutiny of any natural language sample that is not highly edited is likely to find ‘errors’, but nonnative speech, especially of marginalized speakers, is more likely to be scrutinized for such ‘errors’ (Hu & Lindemann Reference Hu and Lindemann2009). Thus, ‘broken English’ is applied disproportionately to already negatively viewed English, but the bias may not be obvious because something can virtually always be found to point to as ‘imperfect’ and therefore ‘broken’.

Using the term ‘broken English’ is thus not only problematic in terms of its deficit orientation, but in its focus on the language of less powerful groups in society, with the language of more powerful groups assumed as a vague norm that is ‘defined’ by its not needing to be mentioned or critiqued at all, similar to Hill's (Reference Hill1999:684) observation of ‘“whiteness” as an invisible and unmarked norm’. The usage of the label thus parallels the identification of only some speakers as ‘having an accent’, while those who have an accent associated with a dominant group are considered not to have an accent, such that their accent (or even their ‘lack of accent’) is unlikely to be mentioned. In both cases, the unmarked speaker or even unmarked human being is one who speaks English natively.

The variation and ambiguity we have seen in how the term ‘broken English’ is used has other implications. This loaded term, used pejoratively in unscripted texts not written for a wide audience, is still used in highly edited, published pieces by writers who avoid or at least distance themselves from other pejorative terms. While journalists and others often use the label in a way that is apparently intended to be neutral, the frequently derogatory use of this descriptor in less edited contexts suggests that it still may be seen as indexing a negative stance even within apparently neutral contexts. This allows any identifiably nonnative English to be characterized as ‘broken English’, slipping into ‘not English’, with such characterizations ultimately seen as an acceptable way to identify a typical nonnative speaker as a public menace or even as a criminal. Thus, one danger in the usage of the label is that it provides a socially acceptable means of marginalizing, Othering, and further disenfranchising groups of people based solely on perceptions of their language use.

More generally, based on our analysis we can identify four key mechanisms by which everyday talk about language promotes and ingrains ideologies that reinforce the subordinate position of nondominant social groups: (i) scrutiny of the language of less powerful groups, (ii) accessible, decontextualizable language to refer to the phenomena identified through this increased scrutiny, (iii) lack of precision of that language, and (iv) failure to observe and/or challenge this lack of precision. We also note parallels with findings in other studies to demonstrate that each mechanism is a more general phenomenon in the naturalization of language ideologies.