US Undergraduates’ Perceptions of Non-Native English

Introduction

Globalization and increased immigration to the US have made interactions between native and non-native English speakers more significant.
Native speakers' perceptions of different language varieties reveal their social organization, views on different groups, and group perceptions.
Evaluations of language varieties reflect evaluations of the groups who speak them.
Lippi-Green (1997) argues that negative reactions to non-native English are linked to non-white skin or third-world origins.
Criticism of English varieties has historically targeted the largest recent immigrant groups.
Non-linguistic reactions to non-native speech impact job discrimination and interactions with native speakers.
Beliefs about a speaker's ethnicity can trigger non-native assessments and lower comprehension.
Understanding the relationship between perceived origin and reactions to speech is vital for fair treatment of non-native speakers.
This study aims to understand this relationship to address linguistic prejudice in education and the wider public.
Verbal guise procedure research shows negative evaluations of native speakers of Spanish, German, Malaysians, Chinese, Japanese, Koreans, Italians, Norwegians, and Eastern Europeans by US listeners under certain conditions.
Giles and Coupland (1991) note that studies are “overrepresented by one-off studies in widely varying cultures, sociolinguistic conditions, situational and procedural domains. This has made it impossible to infer anything other than very general principles”.
Lack of systematicity in studies makes it difficult to predict linguistic bias in specific situations and address discrimination.
Rating scores in verbal guise studies don't reveal if descriptors mean the same to all respondents.
Terms like "broken" can have various interpretations, from frequent pauses to incomprehensible speech.

Perceptual Dialectology

Perceptual dialectology uncovers the folk's understanding of varieties by looking at evaluations and categorizations.
Respondents label maps with dialect locations or rate areas on language correctness and pleasantness.
Preston (1999) suggests that responding to varieties based on category names provides similar information to verbal guise studies while ensuring that the respondents are rating areas that are “cognitively real” for them.
Folk linguistics uses interviews and participant observation for detailed insights into language beliefs and evaluations.
Analysis reveals community members' reactions, salient aspects of varieties, and shared beliefs.

Current Study

The study applies perceptual dialectology to examine folk perceptions of non-native English.
A country-rating task, similar to Preston’s US state-rating task, compares evaluations of varieties systematically.
A map-labeling task reveals folk frameworks for evaluating varieties and the agreement level.
It illustrates non-native variety categorization, evaluation, salient world areas, and variety aspects for US English native speakers.

Addressing Biases

Understanding folk opinions is essential to influencing them, as Niedzielski and Preston (2003) pointed out.
Examining individual responses reveals agreement levels; conflicting reactions make challenging stereotypes easier.
Uniformly negative opinions may be shared as background knowledge, affecting discrimination cases.
Respondents' comments on non-native features indicate negatively stereotyped aspects for each variety.
This suggests focus areas for language learners (see Lindemann in press).
Linguists, ESL teachers, and teacher-educators may not notice the same features as laypeople.
Analyzing full descriptions clarifies the meaning of terms, addressing potential disagreement.
Variability in interpreting "broken English" impacts language discrimination cases.
A customer's comment about an employee’s "broken English" may not indicate the employee's English is insufficient for the job.
Identifying similarly evaluated/described world areas reveals non-native English categories and relevant groups.
This has theoretical and practical relevance, as evaluations of one group impact evaluations of similar groups.
After the World Trade Center destruction, Sikhs were attacked due to being mistaken for Arabs/Muslims.
Analyzing a belief system provides insights into language variety reactions and ways to address discrimination.
Findings are likely relevant to other groups' speech reactions; further studies will clarify general vs. specific factors.

Method

Two tasks were used: map-labeling and country-rating.
Participants considered international students at the university.
Map-labeling task: Participants labeled a world map with descriptions of English spoken by students, choosing descriptions and areas.
The map showed country borders without names to provide space for writing.
The participants were encouraged to ask if unsure where an area was, a few odd descriptions (e.g. an arrow pointing to Poland with “French speaking” in the description) were probably due to their sometimes limited knowledge of geography.
Participants were asked for clarification on unclear areas after task completion.
After the first 30 respondents, a labeled political map was provided to improve response accuracy.
Country-rating task: Participants rated the English of university students from 58 countries on correctness, friendliness, and pleasantness (1-10 scale).
They also rated their familiarity with each country's English (1-10 scale).
Countries were selected via pilot tests from a list of all countries with a population greater than two million. If most pilot participants left blanks for a particular country, it was omitted from the final country list.
213 US undergraduates (162 women, 51 men) participated; 208 completed the country-rating task, and 79 completed the map-labeling task.
All were native English speakers who had grown up in the US speaking only English in the home.
Ages ranged from 17 to 47 (mean 20, median 19, mode 18).
Participants completed the map-labeling task first to avoid influence from the country-rating task.
Both tasks were completed in groups of two to seven people.

Results: Country Ratings

Many '1' ratings were given for familiarity due to participants' unfamiliarity with some countries’ English.
Some participants guessed on correctness, pleasantness, and friendliness; a small minority gave 1s for all unfamiliar countries.
Participants giving 1s on correct, pleasant, and friendly as well as on familiar to show that they were unfamiliar with the English of speakers from that country.
Data from respondents giving '1' on all scores for unfamiliar countries (without intent clarification) were excluded.
Analysis was based on 195 respondents.
Table 1 shows the fifteen countries whose English was rated highest on familiarity, together with each country’s rank and average score for each of the four characteristics.
There's a steep drop in familiarity after the US (9.96), with no country rated above 8.5. The most familiar non-native variety was rated 7.9, and the fifteenth most familiar, Russia, was rated only 5.1.
The most familiar varieties include some but not all primarily native English-speaking countries and some primarily non-native English-speaking countries.
They include countries whose English was rated quite low, at least on some traits, as well as countries whose English was rated very positively.
Correct, pleasant, friendly, and familiar scores were closely related (factor analysis explaining 90% of variance, with loadings of 0.967, 0.973, 0.969, and 0.879, respectively).
Relatively familiar English varieties didn't necessarily translate into correct, pleasant, and friendly English.
Familiar Englishes included those spoken by people from China, Japan, India, and Russia, all rated below the median in at least one of the other scores.
Mexican English was the most familiar non-native variety but wasn't rated as particularly correct or pleasant.
Non-stigmatized groups (e.g., France, Germany) were rated positively, at least on correctness; stigmatized groups (Mexico, Japan, China, India) were rated as less correct.
Of the countries with primarily non-native English speakers, Italy was rated the most positively overall.
The countries whose non-native English was most familiar that are rated positively are in Western Europe, have had comparatively favorable relationships with the US during the respondents’ lifetimes, and do not have large populations of recent immigrants in the US.
Countries tended to have similar average scores for correct, friendly, and pleasant, but there are clear differences in a few cases.
Ratings for Russian and German English on pleasant and friendly are markedly lower than their ratings for correct.
The split between correctness and other scores is most obvious for Germany, which is rated fairly high on correct and below the median on pleasant and friendly; in contrast, all three scores are low for Russia.
The high correctness rating for German English is consistent with the generally non- stigmatized status of Western Europeans’ English, the lower ratings on pleasant and friendly are consistent with some stereotypes of Germans, especially those associated with World War II films.

K-Means Cluster Analyses

K-means cluster analyses were carried out on the correct, pleasant, and friendly ratings for all 58 countries to investigate overall patterns of evaluation.
Four clusters provided the most informative results.
Table 2 shows a striking pattern of country ratings.
Countries are organized within their cluster by geographical area, showing the high degree to which English ratings are predictable based on region and familiarity.
Within each geographical area, countries are listed from highest combined correct–pleasant–friendly rating to lowest.
Poland was rated the lowest of these countries, as might be expected since it could also be classed as Eastern Europe, the other countries of which fell into the lowest-rated cluster.
Egypt is the lowest rated of the African countries in the third cluster and could logically also fall with the other Middle Eastern countries in the bottom cluster.
The highest-rated group unsurprisingly consists of the six countries of primarily native English speakers that respondents rated as most familiar.
The next highest cluster includes another country of primarily native speakers and South Africa, plus six Western European and three Latin American nations.
All other countries fell into the bottom two clusters.
The third cluster covers the remaining countries of Latin America that were rated by respondents, including Mexico, as well as the four Asian countries that respondents rated as most familiar.
The few countries of Africa that were rated also fell into this group, as did Israel and several countries in Central Europe.
The lowest-rated cluster includes the remaining Asian countries, plus Eastern Europe and the Middle East.
As with Alfaraz’s (2002) findings for Miami Cubans’ ratings of pre-Revolution and post-Revolution Cuban Spanish, political factors clearly play a role in these ratings, as many countries in this last group are much more easily classified in terms of political relationships with the US than in terms of recent immigrant groups.
Several have had poor relations with the US within the respondents’ lifetimes, including former communist bloc countries, Iran, Iraq, North Korea, Vietnam, and Afghanistan.
It is notable that, geographically and linguistically, Israel might be expected here, yet it is rated more highly than the Arab-ruled nations of the Middle East.
It is also possible that respondents were aware that the close relationship between the US and Israel has included migration between these two countries and considered this in giving it a somewhat higher rating.
Egypt may have also fared slightly better than other Middle Eastern countries that are well known to US students as fundamentalist (Iran, Afghanistan, Saudi Arabia) or long out of favor with the US (Iraq).
Other sociopolitical factors may be relevant in the ratings of countries which are ranked more highly than this bottom group.
Turning to countries of primarily Spanish speakers, we see that Spain falls in the highest cluster that includes countries of primarily non-native speakers; it is rated higher than any Latin American country, most of which fall in the next cluster down.
In this case the pattern may be better explained with reference to recent patterns of immigration to the US.
Cases other than Asia where familiarity appears to play a role include Morocco, whose English had the lowest familiarity rating (2.2) except for Bulgaria.
It falls in the lowest group, although it is not likely to be associated in respondents’ minds with fundamentalism or poor relations with the US.
The Eastern European countries show a similar pattern, where both a more familiar traditional foe of the US (Russia) and less familiar countries (Romania, Ukraine, Bulgaria, Czech Republic, Bosnia) are rated lowest, with a slightly more familiar country (Poland) rated higher.
The highly rated countries classed in Table 2 as “Western Europe” were likewise nearly all rated as having more familiar English than the lower-rated ones classed as “Central Europe”, except for Germany, which respondents may associate with World War II, especially through films.
In summary, it appears that sociopolitical factors and familiarity can largely explain respondents’ patterns of evaluation of the English of these countries, with countries that may be identified as adversaries of the US and less familiar countries being rated most negatively.
While the number of immigrants from these countries to the US also appears to be relevant in such evaluations, as has been suggested by Lippi-Green (1997), this factor seems more important for the countries that are not rated in the bottom- most group.

Map Labels

Data from the map-labeling task were investigated in three major ways.
- The number of responses about different areas suggest what respondents see as speech areas and which of these are most salient to them.
- The types of descriptors used indicate what general aspects of language are most salient to respondents.
- The details of the actual descriptors shed light on how respondents regard different varieties and how they see these varieties compared to each other.

Salient Areas

Respondents varied widely in the areas covered in their descriptions, including anywhere from four to twenty-six speech areas outside the US.
Most respondents described a number at the lower end of this range, with ten as the average and six the most common number of areas described.
Some of the apparently more detailed maps, including the two with more than twenty speech areas, use the same labels (e.g. “accented”, “broken”, or “proper”) for many of the areas, so that they still do not differentiate these varieties of non-native English.
Table 3 shows the countries that were included in more than half of the 79 respondents’ descriptions of non-US Englishes and were described individually as speech areas by at least a quarter of the respondents.
The ten countries listed in Table 3 correspond closely to the fifteen countries with the most familiar English shown in Table 1.
Geography almost certainly played a role in the absence of Jamaica, which is small and tended to be confused or classed together with Cuba, Haiti, and the Dominican Republic.
Similarly, Ireland was sometimes described separately but often circled together with Great Britain and labeled as “England” or “UK”, making it difficult to systematically distinguish respondents’ descriptions.
On the other hand, the English of these countries was also rated among the most familiar on the country-rating task, so we might expect respondents’ descriptions of them to be more specific.
This group of most commented-on countries includes countries of chiefly native speakers as well as countries of chiefly non-native speakers; both positively and negatively evaluated countries are represented.
However, the non-native Englishes that were most often commented on were those that were also evaluated relatively negatively, namely China, Mexico, and Russia.
China appears to be the major representative for Asia or the Far East, as does Mexico for all of Latin America.
In contrast, no one country appears to stand for (Western) Europe; France, Germany, and Italy are described with approximately equal frequency.
No country in Africa appears among the most frequently commented-on countries, either as a part of an area typically included with one of these countries or on its own.

Salient Aspects of Language

Table 4 provides an overview of the types of comments most often made about the countries of primarily non-native English speakers listed in Table 3 (See Hartley and Preston 1999 for a somewhat different classification of labels used by their respondents for varieties of US English.)
Global evaluations such as “broken English” or “speaks English well” were by far the most common descriptors.
For the countries that were highly rated on correctness – Germany, France, and Italy – no other type of comment was made as frequently, with the exception of name or comparison descriptors for Italy (e.g. “Italian accent”, or “a bit similar to English w/Spanish accent”).
This descriptor type was much less common than global evaluations but was the second most frequent type of comment, followed by descriptions of specific sounds that differed and comments on clarity or comprehensibility.
In general, respondents had more to say about the countries that were rated negatively on correctness, both in terms of overall number of comments and variety of comment types.
For example, comments on clarity or comprehensibility, while one of the most common types of descriptors, were only common for the countries rated negatively for correctness.
The most negatively-rated of these ‘familiar’ countries, China, had large numbers of comments in several categories, and Mexico had the largest number of comments overall.

Descriptors of Salient Speech Areas

The analysis begins with descriptions of the English of salient negatively evaluated areas, focusing on China, India, Mexico, and Russia.
Germany is described next as an area which was highly rated on correctness but which shared some negative pleasantness-related descriptors with Russia, is described next, followed by France and Italy, the most positively described.

China

Respondents showed surprising agreement, with comments focusing on speed, choppiness, and pronunciation of /r/ and /l/.
Chinese English is described as quick, choppy, and missing unimportant words (verbs; linking verbs; is, are, and other small words).
Unclear or difficult to understand.
Some global evaluations are ambivalent, with some acknowledging that some speak English well.

India

Descriptions of Indian English showed less agreement, with two different frameworks for understanding it.
Seven respondents associate it with British English, whereas three associate it with “kwicky marts”, a reference to the US animated television show The Simpsons, in which an Indian character owns a convenience store by that name.
Those who associate it with British English either describe it positively or neutrally.
One additional respondent mentions both labels: “Although this is a stereotype, we hear a lot about Indian speakers of English having the ‘Welcome to Dairy Queen’ accent. They are taught British English in school & I guess this is why we like to make fun.”
The people I’ve met who were raised in India seem to speak “precisely”. Maybe it has something to do with the “British” English so to me it sounds more enunciated b/c that’s the way British English also sounds to me.
The English from here is very pronounced and seems clear/is spoken slowly.
Global evaluations are mixed, with some describing it as eloquent and others as very broken English.

Mexico

Respondents frequently provided a name for the English of Mexico, including “Spanish” or “Spanish-like” (11 respondents), “Mexican English” (4 respondents), and “Spanglish” (2 responses); one labeled an area around the US–Mexico border as “Tex-Mex ‘English’”.
Common descriptions include pronunciation of /r/ and stressed vowels.
Speed is described as both fast and slow, with some suggesting the speed is just wrong.
Many comments on clarity and global evaluations have a modified- positive or mixed orientation.

Russia

Russia's English, not highly rated on correctness, was rated particularly negatively on pleasant and friendly.
Most respondents describe a lack of clarity, with thick accents.
Global evaluations are chiefly negative, describing it as harsh and guttural.

Germany

Respondents have two key frameworks for understanding German English.
One focuses on Germany as part of Western Europe.
The other separates German-influenced varieties from those influenced by Romance languages, with descriptions more in line with those of Russian English.
German English is described as harsh, gruff, or guttural.

France

Many comments about France are very positive, making reference to poetic or more romantic language.
Other suggest that the French can also be seen as arrogant and unwilling to speak English.

Italy

Respondents show more agreement on Italy, the most highly rated of countries with primarily non-native speakers of English.
Comments focus on an Italian accent, stress, intonation, and emotions.
Global evaluations are chiefly positive, with speakers or their language described as involving lots of emotion or always smiling.

Discussion

Respondents interpreted the task as evaluating rather than describing varieties.
The focus on evaluation leads to undifferentiated categories of good and bad non-native speech.
Salient non-native speech appeared to be negatively evaluated.

Salience

Mexico and China were described as getting the most wrong, they were described the most often and in the most detail.
Areas rated highly on correctness (France, Italy, Germany) as well as less familiar negatively rated areas (India, Russia) garnered fewer and less detailed comments.
A total lack of familiarity allowed respondents to make evaluations based completely on stereotypes, as they lacked access to counterexamples that could neutralize or soften them. They were less likely to comment unless specifically asked about those countries, however.
Listeners appear to have multiple overlapping categories for evaluations of non-native English.

Native vs. Non-Native

Categories include native vs. non-native, stigmatized non-native English, and European/Western European.
The most salient sub-categories of stigmatized non-native English were East Asian or “Chinese” English and Latin American or “Mexican” English.
Chinese English quick, choppy, poorly enunciated, hard to understand.
Latin American English more casual and less prestigious than the English from Spain.
A number of explanations are possible for the more negative evaluations of Asian English in comparison to Latin American English.
May be perceived as exhibiting greater cultural, linguistic, or even physical differences.
Reactions to the less salient negatively evaluated groups are probably based on images of these groups in the popular media, since respondents are likely to have fewer first-hand experiences with them.

Verbal Guise Research

The findings suggest that there may be qualitative differences between non-native and non-standard.
A few prestige-oriented descriptors such as “eloquent” used by respondents in the current study suggest that at least some US English speakers may consider some non-native varieties to be prestigious.
Another difference between stigmatized native dialects and non-native English is the lesser degree of familiarity with non-native varieties, which appears to be related to the looser connection between salience and perceived incorrectness.

Language Discrimination

Patterns of responses suggest expectations these US undergraduates are likely to have of non-native English speakers.
The belief that someone is from a particular part of the world may trigger the perception of a “foreign” accent where none exists (Rubin 1992), or such a belief may lead to a positive evaluation because the speaker’s English is better than expected (Brown 1992).
Lack of specificity in classifications, descriptions, and evaluations of non-native English has clear implications for addressing language prejudice.
Further investigation of folk linguistic ways of describing and understanding non-native speech is clearly needed.
Finally, further study is needed to discover how the biases of these college students compare to those of other populations, and the degree to which they apply to different non-native speaker groups.

“As numerous researchers have pointed out, evaluations of language varieties can be understood as evaluations of the groups who speak them rather than of language per se. Lippi-Green (1997) has argued that in terms of non-native English in the US, it is “not all foreign accents, but only accent linked to skin that isn’t white, or which signals a third-world homeland, that evokes such negative reactions” (pp. 238–9, italics in original).”

“For example, “broken” is a descriptor commonly used to describe non-native speech, but it may be used to refer to anything from speech with frequent pauses to incomprehensible or very low-proficiency speech.”