7/11/17
1
Race Ma*ers: Hearing Difference
Session 2 2017 LSA Ins;tute
Elaine Chun, University of South Carolina <[email protected]>
Announcements/Reminders • Submit 4 readings responses over the course of the month (See the first
two excellent examples) – You would have submi*ed yesterday on Baugh (2003) – You can submit a response to Philips (1972) by Thursday, 12 pm
• Begin to think about Assignment 2: Data you might collect and preliminarily analyze – Due July 21 (10 days from now) – What counts? Spoken or wri*en discourses about racial issues, Metalinguis;c
discourses about racialized language (e.g., racializing labels or descriptors); Spoken or wri*en discourses involving the indexing of racial posi4ons or iden44es (e.g., code-‐switching, style-‐shi\ing).
– Where can I find my data? Recordings of people around you, online corpora (create your own), surveys, interviews, etc.
– What methods can I use? Any (ethnographic observa;ons, interac;onal analysis, varia;onist analysis, and other forms of discourse analysis)
– What if I’m at a loss? Talk to or email me!
Race Ma*ers • Race may be socially constructed, but it is deeply embedded
in our material experiences – Who we befriend, date, or marry – Who receives housing and employment – Who we define as intelligent, authorita;ve, or a*rac;ve
• So what can a linguist do? – Understand how language keeps race relevant and how it keeps racial
hierarchies in place – Present solu;ons to race-‐based linguis;c discrimina;on
• What are we doing this week? – Examining two ways in which race can come to ma*er because of
(perceived) differences of language: Our (perceived) dis;nc;ve pa*erns of pronuncia;on (today) and our dis;nc;ve pa*erns of communica;on (Friday)
Hearing race • Examine how our prac;ces of racialized hearing can maintain
racial hierarchies • Answer some ques;ons:
– What is an accent? – Can we tell a person’s race based on their accent? (Baugh 2003) – Do our beliefs about the race of a speaker (based on what we see)
shape what accent we hear? (Rubin 1992) – What kinds of images and ideologies are we accessing when we hear a
speaker who sounds “non-‐na;ve”? (Lindemann 2003) – (When) is it wrong to hear an accent as “ethnic” or “racial”? – What are the benefits and drawbacks of the (modified) match-‐guise
technique used in these perceptual studies?
“Everyone has an accent” • What do linguists mean by this?
– Everyone has a pa*erned way of pronouncing words (phonological habits) despite the percep;on of only some people as having accents • Prosodic features (stress pa*ern: intona;on, rhythm, volume) • Segmental features (consonants and vowels) • Syllabic features: /ɛbʊriwən hɛbʊ ən aksentʊ/
• Why do linguists assert this? – This rela;vist statement decenters the linguis;c norma;vity of how
we tend to hear accents; language “difference” is always relaIve to some (usually norma;ve) perspec;ve
• When do we hear an accent? – Ideologies of that locate speakers as “other”(and some;mes
“inferior”) rela;ve to some norm: ”I don’t have an accent but you do” or “I have a fobby accent”
Will the real Barneys ad please speak up?
Because you love Chinese accents
From a collection of hand-painted ceramic ware from northern China, $1 to $5, ours alone.
7/11/17
2
The “Oriental” accent
• Breakfast at Tiffany’s (1961) – Starring Mickey Rooney as. . . Mr. Yunioshi
Group discussion ques;ons • Why do people perceive some accents as beau9ful, friendly, or intelligent
and some as ugly, evil, and ignorant? Which of these proposed answers do you buy? Reject? – “Some dialects and accents are characterized by inherently inferior sounds.“
Counter examples? – “Individuals carry personal prejudices against certain kinds of people;
nega9ve evalua9ons of accents reflects an personal prejudicial sen9ments.” – “Everyday discourses, including media imagery, reproduce links between
nega9ve quali9es with groups that have been historically marginalized” – “It is not all foreign accents, but only accents linked to skin that isn’t white, or
which signals a third-‐world homeland, which evokes such nega9ve reac9ons” (Lippi Green, 1997, English with an Accent: 253).
• Can you (never, rarely, some9mes, usually, always) tell a person’s race by how they speak? Can you tell a person’s race by how they look?
• We o\en hear accents as clues about a person’s social iden;ty. Is it wrong to hear an accent? To guess a person’s race based on her speech? Can you think of an example when it might be produc;ve?
Can you hear their race? What else can you guess about them?
Kim Pegula Co-‐owner of Buffalo Bills
Gordon Chang Professor at Stanford
John Rennie Short Professor at UMBC
John McWhorter Professor at Columbia U
John Baugh Professor at Wash U
Baugh’s (2003) opening • What does Baugh find problema;c about courts accep;ng listeners’ claims
that they can “tell” a person’s race based on speech alone? Doesn’t he believe that we can o\en “tell”? – OJ Simpson trial (California v. Simpson) and NY Superior Court (People vs. Sanchez 1985)
Cri;ques of linguis;c profiling • A listener may incorrectly ‘hear’ a person’s race and
subsequent ac;on that is taken based on this false iden;fica;on may be detrimental – Heterogeneity: Not all members of a racial group speak the same – Style-‐shi\ing: Speakers do not speak the same way all the ;me
Cri;ques of linguis;c profiling • A listener may correctly ‘hear’ a person’s race and
discriminate on this basis (e.g., housing discrimina;on) – The listener may claim she isn’t discrimina;ng based on race, but she
may very well be doing so
• LinguisIc profiling (Baugh 2003): Racial idenIficaIon (whether correct or incorrect) and discriminaIon based on linguis;c (auditory) cues
7/11/17
3
Housing discrimina;on study • “Perceptual and Phone;c Experiments on American English
Dialect Iden;fica;on” (Purnell, Idsardi, and Baugh 1999) in the Journal of Language and Social Psychology
• Racial discrimina;on is visual and auditory • Four experiments and findings
1. Dialect discrimina;on takes place 2. Ethnic group affilia;on is iden;fiable in speech 3. Very li*le speech (hello) needed to dis;nguish dialects 4. There are phone;c correlates (syllable dura;on of the first syllable
of hello) of dialects
Experiment 1 • Ques;on
– Do landlords discriminate in housing interviews over the phone?
• Method – Baugh (tridialectal speaker) conducted telephone interviews with
prospec;ve landlords
– Called each 3 ;mes (at least 30 minutes in between), once with each accent
• Results – White areas had strongest biases against non-‐standard dialect—i.e.,
Dialect discrimina;on takes place in certain neighborhoods
Lowest rate of appt confirma;on for non-‐standard speakers
Highest concentra;on of white residents
Lowest rate of appt confirma;on
Experiment 2 • Ques;on
– Is ethnic group affilia;on be iden;fiable in speech?
• Method – S;mulus tokens recorded by 20 total speakers of 3 dialects,
including Baugh – 421 undergraduate and graduate student judges – Forced choice of race or ethnicity: “African American, Hispanic
American, and European American”
• Results – Baugh’s dialect in each of the three guises was correctly iden;fied – e.g., Ethnic group affilia;on is iden;fiable in speech
Experiment 3 • Ques;on
– Can just a short hello be enough to iden;fy the race of the speaker? • Method
– 50 undergraduates (Caucasian na;ve SAE speakers) guessed dialect of tokens – hello extracted from s;mulus (10 tokens x 3 dialects x twice = 120 tokens
• Results – Very li*le speech (hello) needed to dis;nguish dialects
33% = perfect
7/11/17
4
Experiment 4 • Ques;on
– What acous;c cues dis;nguish the three dialects in the word hello? • Method
– 30 tokens (10 for each dialect) – Measurements
• Segment, syllable, word dura;ons • F1 and F2: Midpoint formant frequencies • H1:H2: Ra;o of amplitude of first two harmonics • F0: Midpoint pitch and pitch peak
• Results – 4 significant measurements
• F2 in /ɛ/: ChE has highest F2 (frontest /ɛ/) (see next slide) • Pitch peak ra;o • Dura;on of first syllable • HNR (harmonic to noise ra;o)
SAE backest vowel = lowest F2
ChE had rela;ve late pitch peak
AAVE had rela;vely long first syllable
AAVE had lowest HNR ra;o (ar;culatory implica;on unclear; also, unclear if the measurement taken
was noise-‐to-‐harmonic or vice versa)
Race discrimina;on h*p://www.eeoc.gov/laws/types/race_color.cfm
• Title VII of the Civil Rights Act of 1964 (Title VII): This law
makes it illegal to discriminate against someone on the basis of race, color, religion, national origin, or sex. . .!
• Race discrimination involves treating someone (an applicant or employee) unfavorably because he/she is of a certain race or because of personal characteristics associated with race (such as hair texture, skin color, or certain facial features). Color discrimination involves treating someone unfavorably because of skin color complexion. . .!
!
• Do you believe that linguis;c profiling, as defined by Baugh, can be legally classified as “race discrimina;on” as defined in Title VII of the Civil Rights Act?
Race/class discrimina;on • Note that SAE callers did not fare well in East Palo Alto
(Experiment 1). Is this linguis;c profiling/racial discrimina;on? • If one of the guises sounded like a “working class white”
Californian, would this speaker have been linguis;cally profiled? • Baugh’s (2015) follow up study: When black speakers use
Mainstream American Standard English (MASE), they are o\en granted appointments (see also Massey & Lundy 2001). Which is most likely explana;on? – Explana;on 1: Class is more important than race – Explana;on 2: It’s s;ll about race: MASE speakers are heard as white – Explana;on 3: Race and class are o\en intertwined: Racial language cues
also indicate class (‘sounding white’ means ‘sounding privileged’)
• How would you design a study to tease out the effects of race and class? Should they need to be teased apart?
Complexi;es of linguis;c profiling • Baugh men;ons two other related processes:
• Preferen;al linguis;c profiling – Racial iden;fica;on can be used to connect with (rather than
discriminate against) others – Though Baugh concedes that, “one person’s linguis;c pride can easily
serve as another person’s source of linguis;c displeasure” (164)
• Linguis;c adora;on – Favorable linguis;c stereotypes of Bri;sh and French accents – Though Baugh notes the example of the black Bri;sh colleague who
was mocked for her Bri;sh accent.
Baugh’s conclusion • Baugh (2003) writes,
– Each example confirms global linguis9c profiling in human an9quity. On the basis of our keen auditory skills as a species, I believe that linguis4c profiling will exist as long as human language exists. The challenge to Americans and our fellow ci9zens elsewhere is to have wisdom, pa4ence, and sufficient tolerance of others whose linguis9c backgrounds differ substan9ally from our own. To do so would accentuate the benefits of preferen4al linguis4c profiling, while discarding the tradi9on of discriminatory linguis9c profiling that fans the embers of racial discord.
• Do you agree that linguis;c profiling will always exist? • What kind of wisdom, pa;ence, and tolerance is Baugh referring to? Does
he presuppose that those who discriminate are not wise, pa;ent, or tolerant. Do you agree that this is the case?
• What would an accentua;on of preferen;al linguis;c profiling without discriminatory profiling look like?
Sight and sound • What is the rela;onship between visual or auditory in
processes of racializa;on? Does visible race always take priority over audible race?
• Does a person’s looks like affect how you hear them?
• Is there a “racial” parallel for the McGurk Effect?
PERCEIVED LANGUAGE
PERCEIVED RACE
PERCEIVED RACE
PERCEIVED LANGUAGE
or
7/11/17
5
The McGurk Effect Imagined accent
• Philip Rubin’s (1992) study of percep;ons of NNSTAs – “Nonlanguage Factors Affec;ng Undergraduates' Judgments of Nonna;ve
English-‐Speaking Teaching Assistants” in Research in Higher Educa9on • Study 1: Does instructor ethnicity and lecture subject affect percep9ons of
her ethnicity, accent, and quality as well as student comprehension? – 62 undergraduates listened to a 4-‐minute lecture on tape. – 2 projected slide photographs: white vs. Asian – 2 lectures: science vs. humani;es – Both recordings made by same speaker (na;ve speaker of English
from central Ohio) – Test of listening comprehension + ra;ngs of accent, ethnicity, quality
of teaching. 1 2 3 4 5 6 7!
American accent ! : : : : : : : ! Foreign accent !
Caucasian/European ! : : : : : : : ! Oriental/Asian !
Poor teacher ! : : : : : : : ! Effective teacher !
Study 1: Results TOPIC: HumaniIes Science INSTRUCTOR PHOTO: Caucasian Asian Caucasian Asian
N= 16 17 16 13 MEAN SD MEAN SD MEAN SD MEAN SD
Perceived accent 3.44 2.56 4.94 1.98 2.75 2.44 3.77 1.88 Perceived ethnicity 2.75 1.98 5.53 1.97 2.06 1.39 6.23 0.93 Comprehension 11.94 4.34 9.93 5.70 12.50 5.90 7.31 4.70 Teaching qualifica;ons 8.69 3.61 8.18 1.81 7.25 2.46 9.15 2.73 A�tude homophily 5.56 5.51 6.00 4.95 2.38 4.24 3.38 4.37 Background homophily 12.75 4.12 12.18 4.03 13.25 4.48 10.31 3.54 Values homophily 15.88 4.53 14.94 2.88 14.81 3.21 14.46 3.86 Appearance homophily 11.13 5.04 10.71 5.77 10.13 4.77 7.69 3.22
Asian photo perceived as more Oriental/Asian
Caucasian photo perceived as less
Asian photo perceived as more foreign
Caucasian photo perceived as less
Asian photo resulted in worse comprehension
Caucasian photo resulted in be*er
more or less?
more or less?
be*er or worse?
Study 2 • Study 2: What factors predict undergraduates' ra9ngs of NNSTAs and their
comprehension of NNSTA speech? • Par;cipants: 148 students • S;muli variables
– Accent: high Chinese, moderate Chinese, or SAE – Subject: science or humani;es – Photo: Asian, Caucasian/European, or no photo
• Ra;ngs performed just as in Study 1 • Background ques;onnaire
– Number of classes in which they had been instructed by NNSTAs – Number of weeks they had traveled outside the United States – A�tudes toward nonna;ve speakers in general (e.g., "Our campus benefits by having
nonna;ve English speakers a*ending").
Study 2: Results
Perceived more similar a�tude correlated with higher teacher ra;ng
Perceived greater foreign accent correlated with lower teacher ra;ng
A*ended more NNSTA classes resulted in beZer listening comprehension
Study 3: Pilot interven;on effort in which undergraduates served as teaching coaches for NNSTAs. No detectable effect on undergraduates' a�tudes.
Ques;ons • What are the prac;cal implica;ons of this study? Do changes
need to be made in universi;es? What kind? • Rubin uses a “matched-‐guise” approach for part of his study
(the two Chinese accents). What are the benefits and disadvantages of this approach?
7/11/17
6
What does a non-‐na;ve accent really mean for hearers?
• We know that consequences may arise from hearing an accent, but what are hearers really hearing?
• Stephanie Lindemann. 2003. Koreans, Chinese or Indians? A�tudes and ideologies about non-‐na;ve English speakers in the United States. Journal of Sociolinguis9cs.
• Local ideologies vary, so a “na;ve speaker” accents can evoke different kinds of images – Not all non-‐na;ve accents are s;gma;zed: Cargile & Giles (1998)
showed that a “moderate” Japanese accent was highly rated on status (but not a*rac;veness) traits compared to a ‘Stand. American’ guise
– Hearers may be a*uned to other social dimensions (e.g., class)
• Hearers some;mes misiden;fy the race of the speaker
Lindemann’s study • Mo;va;ons: Researchers haven’t inves;gated misiden;fied
ethnicity or how local ideologies may shape percep;ons • Ques;on: How do Michiganders view (unfamiliar) Koreans?
– Background facts: Asians as a “model minority” and Koreans o\en mistaken for Chinese
• Par;cipants: 39 na;ve speakers of English at a Midwestern university
• S;mulus: 4 na;ve Midwestern English and 4 na;ve Korean speakers reading a 101-‐word answering machine outgoing message; 2 female and 2 male for each na;ve language
• Task: Listen to first sentence of 10 passages; then listen to each message twice; Iden;fy non-‐na;ve speaker ethnicity; Rate speakers on 6 status + 6 solidarity scales (7-‐point scale)
Result 1 • RESULT 1: Koreans iden;fied as (in this order) Asian (23%), Chinese,
Indian, Japanese, La;no, Korean (8%), Other. • BUT doesn’t mean judgments were of Indians/La;nos; judgments at
second (higher, subconscious) order; l/r alterna;on may be “Korean” or “foreign/strange”
Results 2a • Na;ve Korean speakers judged lower on status scales
– Ra;ngs of posi;ve quali;es (intelligent, successful, ambi;ous); nega;ve quali;es (lazy, uneducated, incompetent)
Results 2a • Na;ve Korean speakers judged lower on status scales
– Ra;ngs of posi;ve quali;es (intelligent, successful, ambi;ous); nega;ve quali;es (lazy, uneducated, incompetent)
Results 2b • Na;ve Korean speakers judged equal on solidarity scales
– Ra;ngs of posi;ve quali;es (friendly, likeable, helpful); nega;ve quali;es (unkind, insincere, aloof)
– English NSs as ‘stuck up’
7/11/17
7
Results 2c • Na;ve Korean speakers judged lower on language-‐focused
traits – Ra;ngs of posi;ve (appropriate to hire to do message, nice to listen to);
nega;ve (speaks poorly) – Raters felt it was okay to rate language poorly even if not the speaker
Lindemann’s discussion • Results may be shaped by specific site where listeners are not
threatened by Koreans (What about LA or NYC?) • Unsurprising result: Koreans judged as an outgroup and as equally
friendly (equal solidarity), less intelligent (low status) – Aligned with previous studies: – Subordinate/outgroup received lower status ra;ngs – High solidarity can be assigned to ingroup, dominant, or subordinate group
• Surprising results: Judged as less intelligent despite local stereotype of Asians as hardworking, intelligent, reserved – Suggests that listeners primarily heard them as “non na;ve” as opposed to
“Asian” or “Korean”
– Possible explana;on: Koreans are not a salient category (as seen by misiden;fica;ons)
• Korean accent interpreted as ‘foreign faul;ness’ rather than ‘Koreanness’ per se
Conclusions • Is it wrong to hear an accent?
– Problems arise not because we can (or think we can) iden;fy race but because we assign moral qualiIes based on these racialized accents (‘Friendly’ La9no; ‘Annoying’ Chinese; ‘Sexy’ blonde) as well as take acIons that reproduce racial hierarchies.
• If racial iden;fica;on is an inevitable part of our social reality, where do we go from here?