Transcript Second Language Acquisition Research Methodology

Research Methodology
Research Traditions
What is research?
Research is “the systematic approach to finding
answers to questions.”
“Questions” comes first – questions drive inquiry;
questions will inform the kind of research we do.
Quantitative and Qualitative research traditions
Both are “empirical,” involving the collection of
original data (from human subjects).
Both are drive by identifiable conceptual and
methodological assumptions.
Research Traditions
Quantitative research emphasizes:
• the systematic measurement and quantification
of variables
• the statistical analysis of data
• the use of mathematic models and causal
inferencing
Qualitative research emphasizes:
• thick description of the interactions of individuals
and interpretations of these interactions
• heuristic (discovery oriented analysis of data)
• leading to “grounded theory.”
Research Traditions
Assumptions about:
1. discovery
2. variables
3. data collection
4. data analysis
5. use of data
Research Traditions
Quantitative
Qualitative
Discovery
There is a “T”ruth that
can be discovered.
Individuals socially
construct meaning /
“t”ruth, so truth is as
relative as the number
of individuals involved.
Qualitative research is
sometimes referred to
as “interpretive,” for this
reason.
Research Traditions
Quantitative
Qualitative
Variables
Few, parsimonious –
goal is to isolate a
single variable so that
causation can be
established.
Research Traditions
Quantitative
Qualitative
Variables
Parsimonious
L. parsimonia
"sparingness, frugality,"
from pars-, stem of
parsi, perf. tense of
parcere "to spare, save"
+ -monia, suffix
signifying action or
condition
“Thick description.”
Goal is to consider all
variables that might be
pertinent.
Margaret Mead
Claude Lévi-Strauss
Hunter Thompson
Research Traditions
Quantitative
Qualitative
Data Collection
Deductive:
Researchers bring a
theory to the study
setting which is then
tested in that setting.
Inductive: Theory arises
from data – hence
“grounded theory” –
grounded in data.
Theories inform
researchers’ preparation
for study, but doesn’t lock
them into particular ways
of understanding.
Competence
data:
•error
recognition /
correction
•grammatical
judgments
•cord sorting
Research Traditions
Data Collection
Performance
Affective data:
data:
•questionnaires
•reading aloud
•matched-guise
•completion
techniques
tasks
•diary studies
•elicited
•focused
imitation
introspection
•reconstruction
•role plays
•oral interviews
•composition
Research Traditions
Quantitative
Qualitative
Data Analysis
Statistical. Searching
for statistically
significant differences
among samples around
specific variables, in
order to understand
effects of interventions.
Heuristic. Searching
data variously
(taxonomic analysis,
semantic relationships [x
is a kind of y, etc.]) in
order to understand a
specific culture.
Research Traditions
Quantitative
Qualitative
Use of Data
By random sampling,
generalize results from
a small sample to a
large population.
Not interested in
generalizability. Results
inform further study and
help build theory (our
interpretive
understanding of a
cultural setting).
Statistics
1. Descriptive Statistics
The use of numbers to describe results.
• mean = average scores (etc.) from all
participants
• SD = standard deviation ≈ average distance
from the mean of all points in a data set
Statistics
1 SD = 68 % of population (red)
2 SD = 95 % of population (red and green)
3 SD = 99 % of population (red, green, and blue)
Statistics
1 SD = 68 % of population (green)
2 SD = 95 % of population (green and blue)
3 SD = 99 % of population (green, blue, and red)
Statistics: Height of US Women 18–74 (n =
6,588)
mean =
63.5 in.
SD = 2.5 in.
61 in.
58.5 in.
66 in.
68.5 in.
Statistics: Example
Statistics: English Students
2.42
1.28
3.56
Statistics: ESL Students
2.30
1.43
3.17
Statistics: Example
Statistics
2. Inferential Statistics
Analysis of descriptive statistics. How do we know
if the difference between two data sets (means,
frequencies, correlations, and so on) is significant –
i.e., attributable to something other than chance?
A variety of tests used to determine significance:
• Anova (analysis of variance)
• chi-square
• F-test
• t-test
• Pearson product moment correlation
Statistics
Significance – means “attributable to something
other than chance.”
Example: old teaching method v. new teaching
method
Experimental method semester-end test scores are
significantly higher than control method tests
If we have set up the study correctly, we may be
able to conclude that the new method is the cause
of higher student achievement.
Statistics
Probability. Significance is usually expressed in
terms of probability, “p” – for example:
p < .01
p < .05
refer to the probability that differences occurred
through chance alone (as opposed to the influence
of the experimental method).
Statistics
p < .01 means there is less than 1 percent
probability that chance alone explained our result;
there is more than 99 percent probability that
something other than chance explained our result.
p < .05 means there is less than 5 percent
probability that chance alone explained our result;
there is more than 95 percent probability that
something other than chance explained our result.
p = 0.026 means that our inferential analysis has
determined there to have been a 2.6 percent
probability that chance alone explained our result;
and a 97.4 percent probability that something other
than chance explained our result.
Statistics: Example
Statistics: Example
So, while the results may be interesting …
Statistics: Example
… they are not significant – i.e., we can’t contribute
them to anything other than chance.
Research Traditions
Quantitative
Qualitative
Research Methods
Correlation studies
Survey research
Experimental research
Case studies
Ethnographic research
Research Methods
Correlation Studies (Quantitative tradition)
Purpose: to understand relationships among
characteristics.
Issue: Correlation is NOT causation. Examples?
Cancer
clusters
Research Methods
Correlation Studies (Quantitative tradition)
Analysis: Statistical, based on calculation of
correlation coefficients.
Research Methods
Correlation Studies (Quantitative tradition)
Analysis: Statistical, based on calculation of
correlation coefficients.
Research Methods
Correlation Studies (Quantitative tradition)
Example: Goldstein, L.M. (1987). Standard
English: The only target for nonnative speakers of
English? TESOL Quarterly, 21, 417-436.
The author examined the use of African American
English (BVE) among Hispanic students. She
found that those who had extensive contact with
African American students used features of BVE
more than students with less contact.
http://www.youtube.com/watch?v=kkcTpfcno-E
Research Methods
Survey Research (Quantitative tradition)
Purpose: to learn about the characteristics of an
entire group (population) by asking questions of a
small component of that group (sample).
Issue: Random sampling.
Analysis: statistical, both descriptive (percentages,
etc.) and inferential (i.e., looking at the
relationships or correlations in the results.
Research Methods
Survey Research (Quantitative tradition)
Example: Duran, R.P. (1987). Factors affecting
development of second language literacy. In S.R.
Goldman and H.T. Trueba (Eds.), Becoming literate
in English as a second language (pp. 33-55).
Norwood NJ: Ablex.
The author surveyed Hispanic freshmen
concerning their basic language characteristics,
their own rating of their academic skills, and their
SAT scores. As a result of correlation analyses,
the author found a positive relationship between
high self-ratings and high SAT scores.
Research Methods
Experimental Research (Quantitative tradition)
Purpose: to establish a cause / effect relationship
between an independent variable (IV; the cause
agent) and a dependent variable (DV; the agent
acted upon).
Requires experiment and control groups and
careful control of all but the IV so that causation (if
any) can be established.
Research Methods
Experimental Research (Quantitative tradition)
Issue: “true-ness” of experiment
Control and
Experimental Groups?
Pre-Experiment
No. Only 1 group, preand post-test structure
Random
Sampling
No. Use of
extant group
Quasi-Experiment
Yes.
No. Use of
extant groups
True Experiment
Yes.
Yes.
Research Methods
Experimental Research (Quantitative tradition)
Analysis: Inferential statistics to determine if
difference between groups are significant – that is,
caused by the IV, not by chance alone.
Example: Carrell, P.L. (1985). Facilitating ESL
reading by teaching text structure. TESOL
Quarterly, 19, 461-481.
The author found that teaching rhetorical structures
to ESL freshmen resulted in improved ability to
recall information in related texts.
Research Methods
Case Study Research (Qualitative tradition)
Purpose: to focus attention on a single entity (or a
very small number) in a naturalistic setting.
Data is gathered through interviews, observations,
diary accounts, text analysis, video / audio
recording, etc.
Analysis: heuristic, piecing together patterns in the
data.
Research Methods
Case Study Research (Qualitative tradition)
Example: Kravin, H. (1992). Erosion of a language
in bilingual development. Journal of Multilingual
and Multicultural Development, 13, 307-325.
The author analyzed the bilingual development of a
six-year-old Finnish-American child. The study is
primarily concerned with the processes by which
one of the child’s languages begins to dominate
the other. The author speculated in conclusion that
the subordination of Finnish in the child is the result
of limited linguistic input in that language.
Research Methods
Ethnographic Research (Qualitative tradition)
Purpose: to examine culture, society, schooling,
etc. for the purpose of understanding the
processes of acculturation, socialization, learning,
etc. from the perspective of individuals in the
setting.
Data is gathered through interviews, observations,
diary accounts, text analysis, video / audio
recording, and participant observation.
Analysis: heuristic, discovery oriented
Research Methods
Ethnographic Research (Qualitative tradition)
Example: Hornberger, N.H. (1987). Bilingual
education success, but policy failure. Language in
Society, 16, 205-226.
The author investigated bilingual education in a
particular community in Peru through a two-year
ethnographic study. Findings focus on the
perceptions of the value of Spanish and Quechua
among students and their families.
• Are there male/female differences in how invitations
are extended between native speakers? How does
non-native speaker behavior compare?
• Is there a sequence in which second language
pronouns are acquired? If so, what is it?
• Does practice with sentence-combining result in
learners producing longer T-units in their classroom
compositions?
• What word-attack skills do learners naturally use
when they encounter a word they don’t know?
• Is there a relationship between the age at which
second language instruction begins and the level of
SL proficiency achieved?