Understanding Second Language Acquisition by Lourdes Ortega
Purcell, E., & Suter, R. (1980). Predictors of pronunciation accuracy: A reexamination. Language Learning,
30, 271-287.
This is a reanalysis of data collected in Suter (1976):
 n=61 NNSs of English (all post-critical period learners), judged for pronunciation accuracy by 14 judges
who were NSs of English
 Pronunciation scores were correlated with scores on 20 predictor variables (or independent variables that
help predict how good someone’s pronunciation will be):
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Age of beginning of residence
Age of first meaningful English conversation ability
Years of residence in English environment
Percent of home English language use with NSs
Percent of work/school English language use with NSs
Months of residence with NSs of English
Years of English formal instruction
Months of intensive English formal instruction
Weeks of English pronunciation instruction
Proportion of NS English teachers
L1 (ended up explaining 42% of variability in pronunciation scores!)
Number of languages with functional ability
Gender
Economic motivation
Social prestige motivation
Integrative motivation
Cultural allegiance
Strength of concern for pronunciation accuracy
Aptitude for oral mimicry
Extroversion or introversion
Step A  In the original analysis, L1 explained 42% of variability in pronunciation scores (Arabic & Persian were
“favorable languages” and Japanese and Thai were “unfavorable languages”)
Step B  In present analysis, the other 19 variables (excluding L1) were submitted to a Factor Analysis in order to
reduce the number of variables with true predictive value. The following findings were reported:
F1= ‘residency’
F3= ‘focused formal training’
F8= ‘formal training’
F5= ‘age of first contact’
F4= ‘aptitude for oral mimicry’
F6= ‘attitude’
F2= ?Percent of work/school English language use with NSs + Number of languages with functional ability
F7= ?gender + cultural allegiance
Step C  The researchers then recalculated scores for each factor, based only on high loading variables
Step D  They also checked correlations between low loading variables and pronunciation scores, and found
Percent of home English language use with NSs and Strength of concern for pronunciation accuracy
correlated at r=.25 or more.
Step E  All these new chosen variables were entered into a regression analysis: L1, eight composite variables
from the Factor Analysis, percent of home English language use with NSs, and strength of concern for
pronunciation accuracy
Results: Best single predictor is L1 (almost 42% of variance accounted)
 L1 + Aptitude for oral mimicry together account for 55.9%
 L1 + Aptitude for oral mimicry + Length of residency in L2 environment together account for 63.2%
 L1 + Aptitude for oral mimicry + Residency + Strength of concern for pronunciation accuracy together
accounted for 67.3%
Implications: “[…] teachers and classrooms seem to have had remarkably little to do with how well our students
pronounce English” (p. 285)… How much do you trust these findings and interpretations? What else has happened
www.routledge.com/cw/ortega
Published by Routledge © 2009 Lourdes Ortega and Mark Sawyer
in the field of L2 phonology since 1980 (cf. see Derwing & Munro’s work: Derwing, T. M. & Munro, M. J. (2005).
Second language accent and pronunciation teaching: A research-based approach. TESOL Quarterly, 39, 379-397.)
A THEORETICALLY SUPERIOR PRONOUNCER:
If we view [Purcell & Suter’s 1980] analysis in the predictive
perspective we can attempt a profile of the nonnative speakers
who are most likely to pronounce English well. They are native
speakers of the “favored” languages (here, Arabic or Persian).
They are good oral mimics. The have lived in an English
speaking country for a considerable number of years, and for the
most part or all of the time they have resided with a native
speaker of English. Finally, they are concerned about the
accuracy of their pronunciation of English. If we may be
permitted to indulge our imaginations, we might even create a
theoretically superior pronouncer. We might call her Nazila (or
him Mohammed, since sex is not a meaningful predictor). She is
a Persian. She arrived in the United States nine years ago and
has remained here ever since. Two years after her arrival, she
married Fred, an American, and continues to share a home with
him. Although she is a good mimic, Nazila continues to worry
about what remains of her “foreign accent.” This is all we need
to know in order to be fairly certain that Nazila pronounces
English very well. Whether or not she has had much or little
formal instruction in English or English pronunciation is of no
importance.
(Purcell & Suter, 1980, p. 285)
A THEORETICALLY POOR PRONOUNCER:
Conversely, the profile of nonnative speakers who are most
likely to pronounce English poorly would involve persons who
are native speakers of one of the “nonfavored languages” (here,
Japanese or Thai). They are poor mimics. They have recently
arrived in an English speaking country and do not reside with
any native speakers of English. They are not particularly
concerned about their pronunciation accuracy in English.
(Purcell & Suter, 1980, p. 285)
THE LIMITED INFLUENCE OF INSTRUCTION:
In fact, the variables which turn out to be important seem to be
those which teachers have the least influence on. Native
language, the most important predictor, results from historical
accident. Similarly, aptitude for oral mimicry seems beyond the
control of the instructor; it is doubtful that one can make a good
mimic out of a naturally poor one. This seems to match the
common view that some people simply have a better “ear” for
pronouncing languages, and that there is not a great deal
language teachers or learners can do about it. Length of
residence in a country where the target language is spoken
natively is largely beyond the instructor’s control. Prolonged
residence in a foreign country is typically the result of political,
financial, or long term educational or professional motivations,
rather than the direct result of wanting to pronounce a second
language more accurately. Finally, while the strength of concern
for pronunciation accuracy might be fortified by an effective
teacher, this concern is often the result of personal motivations
and attitudes established well before the student enters the
classroom.
(Purcell & Suter, 1980, p. 286)
How much do you trust these findings and interpretations? What else has happened in the field of L2 phonology
since 1980 (cf. see Derwing & Munro’s work: Derwing, T. M. & Munro, M. J. (2005). Second language accent and
pronunciation teaching: A research-based approach. TESOL Quarterly, 39, 379-397.)