Language Use Literature Review

Text-only Preview










Language Use Literature Review










August 5, 2003




Language Use Research Review



TABLE OF CONTENTS
1. BACKGROUND ....................................................................................................................... 3
2. THE HISTORICAL CONTEXT OF LANGUAGE USE RESEARCH .............................. 4
3. FACTORS ASSOCIATED WITH MAINTENANCE, SHIFT AND SHIFT REVERSAL6
3A. THE COMPONENTS OF IDENTITY ............................................................................................ 7
3B. GENERATIONS...................................................................................................................... 10
3C. SHIFT REVERSAL AND IMPLICATIONS FOR LANGUAGE USE AT HOME ................................. 12
3D. PROFICIENCY ....................................................................................................................... 13
3E. PROFICIENCY ASSESSMENT.................................................................................................. 15
4. METHODOLOGICAL IMPLICATIONS ........................................................................... 16
4A. ADDITIONAL OBSERVATIONS .............................................................................................. 16
4B. FROM REVIEW TO ANALYSIS ............................................................................................... 18
5. APPENDIX 1........................................................................................................................... 19
6. APPENDIX 2........................................................................................................................... 22
7. APPENDIX 3........................................................................................................................... 27


2


Language Use Research Review


1. Background

Presently, Nielsen Media Research utilizes a “Language Used in the Home” variable for
weighting in certain NSI markets and in its Hispanic samples. The variable has been
proposed as a potential weighting variable for the National Peoplemeter sample. Prior
audits of this variable, as it is collected and used in the NHTI and NSI services,
suggested that this household classification remains stable from one data collection
period to the next. However, a recent audit of the National Peoplemeter sample1
suggested that this household classification might be more dynamic than previously
expected. However, the stability or dynamics of this variable are more difficult to
interpret since historically, the procedures used to audit it have varied and the different
audit procedures are believed to, in and of themselves, lead to different conclusions
about stability.

In light of these findings, Nielsen Media Research made a commitment to the industry to
conduct a comprehensive research program during the summer of 2003. The purpose
of this research program is to evaluate the current Language Use variable and contrast
it with other variations of “Language Used in the Home” as well as other related
variables that might prove to be more stable than Language Use and yet serve
effectively as a surrogate weighting variable. Finally, since we believe that the different
audit methods that have been used in the past have led to different conclusions about
stability, this program will also evaluate the audit procedures and try to dimension the
impact of the different procedures on our conclusions of stability.

The stages of this research plan are illustrated in Chart 1:

Literature &
Variations
Evaluating
Statistical
Commercial
On Current
Alternative
Analysis
1 Search
Question
2
Q
3 uestions
4
Chart 1
Stages of the
Language Use
Research
POV on
Program
The Audit
Process
5

1 See Nielsen Media Research Quality Assurance Department. (2003). NPM Language Use Audit).

3


Language Use Research Review


This document presents the first stage of the project, the Literature Search, in which we
have evaluated the literature looking for:

• Alternate ways of measuring Language Use,
• Related variables that might serve as surrogates for Language Use,
• Scholarly work that explains why and how families chose to use one
language over another, and finally,
• An understanding of the best methodological practices with respect to
measuring Language Use, identity and a number of other related
variables.

Chart 1 shows that this literature review (1) is the first stage of a five-stage research
plan being conducted prior to making a recommendation on weighting by language or a
related variable. Aided by this search, we present a set of variables in Appendix A that
we are considering as future weighting variables. Chart 1 also illustrates that this
process will focus on two types of potential variables: those that are variations on the
five-way “Language Use in the Home” variable that we presently use in several of our
services (2), and a set of surrogate questions that would be highly correlated with
language use and television viewing (3).

Chart 1 shows that once we have narrowed the pool of high potential variables and
variations, we will be evaluating these high potential variables using a series of Mean
Squared Error analyses and other statistical techniques (4) for choosing weighting
variables. Finally, Chart 1 also illustrates that the project will provide an opinion (5) on
how this variable should be audited moving forward.

2. The Historical Context of Language Use Research

A review of the literature suggests that as Hispanic populations grew in the United
States, reaching approximately 5% of the U.S. population in 1970, scholarly work began
to emerge. Perhaps an oversimplification, but it seems as if this work was framed by
two orthogonal paradigms (See Chart 2). One paradigm was related to language
proficiency and the other was related to language use. As DeVries2 reports, most
research on usage focused on language use outside the home rather than in home.
This was in large part due to the fact that then (and now), the stimuli for much of this
research were (and are still) social and economic programs related to performance in
school and the workplace.

2 De Vries, J. (1985). Some methodological aspects of self-report questions on language and ethnicity. Journal of
Multilingual and Multicultural Development
, 6, 5, 347-368.

4


Language Use Research Review



Chart 2
In studying the evolution of the
methods used by the U.S.
There are two broad but orthogonal
dimensions found in the historical
Census, Kominski3 reports that
analysis of language: maintenance/Shift
and proficiency.
the earlier Census had asked
more general language questions
such as “Mother Tongue”4 but
that in the 1980’s, the Census
Maintenance
Shift
had begun to ask questions
about language ability or, more
Maintenance/Shift Analyses
focus on age, generation,
specifically, language limitations.
years in country, self
evaluation, reference group
Up through the 2000 decennial
and identity.
census, the U.S. Census has
continued to ask about language
Proficiency analyses focus on
measuring components of
proficiency rather than usage.
language: speech, writing,
Proficiency
listening, thinking, reading, …

In the intervening years, several
federal and state and local programs have mandated that educators test English
Language proficiency. As a result, there is a very large body of literature including
Guerreo and Del Vecchio5 and Loop6 that is dedicated to providing educators with tools
for assessing both English and Spanish proficiency among LEP students (Limited
English Proficiency). These proficiency tests frequently decompose language
proficiency into its core component parts. These typically might include reading,
listening, speaking and writing components, although as we shall see, this proficiency
assessment currently taps much more complex and deep dimensions than these simple
ones.

While educators and governments may have focused on proficiency, scholarly work
focused on the factors that explained who, when and how English was adopted. The
work of Marin and Sabogal7 are illustrative of this paradigm which tried to explain and
then predict which families would shift to English and under what conditions. This

3 Kominski, R. (1989). How Good is “How Well”? An examination of the Census English-Speaking Ability
question. U.S. Bureau of the Census, Population Division. Presented at the American Statistical Association.
August 6-11, 1989. Washington, DC.
4 “Mother tongue” is asked in a variety of ways, but perhaps one of the most popular is in the form “Language
spoken at home as a child.”
5 Guerrero, M. & Del Vecchio, A. (1996). Handbook of Language Proficiency Tests. Evaluation Assistance Center –
Western Region, New Mexico Highlands University.
6 Loop, C. (2002). Which tests are commonly used to determine English and/or Spanish language proficiency?
National Clearinghouse for English Language Acquisition. AskNCELA, Frequently Asked Question #25. Available
http://www.ncela.gwu.edu/askncela/index.htm
7 Marin, G., Sabogal, F., Marin, B., Otero-Sabogal, R., & Perez-Stable, E. (1987). Development of a short
acculturation scale for Hispanics. Hispanic Journal of Behavioral Sciences, 9, 2, 183-205.

5


Language Use Research Review


Chart 3
paradigm is relevant to our
Shift reversal is illustrated in the following case diagnoses
current analyses in that we
reported in the NPM Language Use Audit, April 2003:
have sought to evaluate

whether predictors of
HH 280384, El Paso, TX. Seven-member HH. The daughter, age 15,
changed from Only English to Mostly English because she is taking a
language use (that is
Spanish class in school. The daughter, age 21, changed from Only
language-related variables or
English to Mostly English because her mother and father want her to
surrogates) might provide
speak and learn Spanish. The son, age 26, changed from Only English to
similar control but prove to be
Mostly English because his mother and father want him to learn Spanish.
a more stable variable.

HH 283189, Tucson, AZ. Four-member HH. The Owner/Renter

(female), age 35, changed from Mostly English to Mostly Spanish
More recently, authors such
because she wants her daughter to learn to speak Spanish. The auditor
as Hildago8 have extended
notes showed that the language for the three daughters were Mostly
the Maintenance/Shift model
English before and after the audit.
to consider observations that

HH 289929, Los Angeles, CA. Three-member HH. One child is under 2
have been phrased reverse-
years of age. The spouse (female), age 34, changed from Only English to
shift. This occurs where a fully
Mostly English because she is trying to teach Spanish to their new baby.
bilingual or English dominant

family takes explicit steps to
HH 460484, Chicago, IL. Four-member HH. The other relative (male),
relearn or improve family
age 5 changed from Only English to Mostly English. The reason given
was that the Grandmas want him to learn Spanish and more Spanish is
members proficiency in their
spoken to him. The son, age 12, changed from Only English to Mostly
“mother tongue.” Readers of
English. The reason given was that the responsible HH member did not
the NPM Language Use Audit,
understand the question the first time. The Owner/Renter (male), age 41,
April 2003, will recall that
changed from Only English to Mostly English. The reason given was
there were multiple
that he was trying to teach his nephew.
observations where families
cited a desire to maintain proficiency in Spanish as the reason for speaking more
Spanish then when previously measured (See Chart 3). The importance of this shift-
reversal factor is illustrated by the fact that 4 of the 22 changes in Language Use (3 of
which changed the household Language Use classification) were due to an effort to
improve Spanish proficiency among household members.

3. Factors Associated with Maintenance, Shift and Shift Reversal

Having identified the major theoretical approaches to studying language use, we will
presently review some of the most important conclusions that have been reached about
Spanish language maintenance within the family and in the alternative, a shift to English
language use.

8 Hildago, M. (2001). Spanish language shift reversal on the U.S.-Mexico border and the extended third space.
Language and Intercultural Communication, 1, 1, 57-75.


6


Language Use Research Review



Our review yielded a number of empirical analyses which were directed at explaining a
family’s decision or effort to maintain Spanish language use or to shift to English.
However, many if not most studies that we reviewed report on small sample community
studies of a qualitative nature. In this section,
we will summarize both the qualitative and
Chart 4
quantitative findings. We believe that if we can

understand the reasons and conditions under
Factors Associated with Spanish
which Language Use changes, we will be better
Maintenance or Shift
equipped to develop questions that could

capture change, but only that change as would
• Identity
be appropriate for our television methodologies.
o Mother tongue

o Community
o Country of Origin
The factors associated with Spanish
o Number of Children
maintenance or shift fall broadly into two
o Parental Endogamy
categories: identity and proficiency.
o Proficiency
Chart 4 highlights the significant components of
• Proficiency
each factor that have been identified by this
o Generation
review. The literature suggests a considerable
o Age
amount of interaction of these components,
o Years in Country
which is why proficiency is shown as the last
o Parents Education
component of identity and identity is shown as
o Identity
the last component of proficiency.

3a. The Components of Identity

Figure 1
Currently, the U.S. Census asks
respondents the identity question
captioned in Figure 1.

Some authors argue that identity is better
measured as a two- or three-dimensional
process9 which differentiate between the
purely subjective component of identity
and the more demographic components
such as decent.


9 DeVries, op.cit.

7


Language Use Research Review


Chart 5
The Pew Hispanic Center10 conducted
The First Term that Hispanics Use to
research on terms that Latinos use to
Describe Themselves
identify themselves. This research
Source: Pew Research 2002
indicates that the large majority, 88%,
of Latinos have identified themselves
54
60
by country of origin, where they or their
40
24
21
parents were born. This is significant in
20
that it is slightly larger than the 81%
0
Country of Origin Latino or Hispanic
American
who have ever identified themselves as
“Latino” or “Hispanic” and much larger
than the 53% who have identified themselves by the term “American.” More significant,
the data reproduced in Chart 5 shows that over half of all Latinos use their country of
origin as the first term that they would use to describe themselves. This finding is
consistent with prior work conducted by Audits & Surveys and IBOPE11 in Latin America
where the concept of ethnic origin was almost alien to many respondents who usually
answered the question in terms of national identity. As the Pew report concludes:

Hispanics see themselves more as having separate and distinct cultures
based on country of origin rather than sharing a single culture as
Hispanics or Latinos…


This fact is supported by the discriminant function analyses of Hart-Gonzalez and
Feingold12 who found that one, if not the best, predictor of whether a family would
maintain its use of Spanish in the home was the size of the respondent’s own country
subgroup living in the state. In this analysis, age, age minus years in U.S. and the
number of Hispanics originating from the same country were the best discriminators of
whether a family would maintain Spanish, shift to English or had already moved to what
we would call English-only. The discriminant function in this analysis could correctly
classify respondents 54% of the time, in contrast to the 33% one would expect by
chance alone. These discriminant analyses will become increasingly important when we
consider whether surrogate variables or acculturation indices can provide a
discrimination in television viewing similar to Language Use at Home but perhaps
providing additional stability.

There is considerable literature which considers the construction of such acculturation
indices. Stoessel13 suggests that while the literature reflects:

10 Pew Hispanic Center/Kaiser Foundations. (2003). The 2002 National Survey of Latinos. Retrieved June 27, 2003
from the World Wide Web: http://www.pewhispanic.org/site/docs/pdf/LatinoSurveyReportFinal.pdf
11 Soong, R., Donato, P. & Verdin, P. (1999). Television in Latin America. Audits and Surveys.
12 Hart-Gonzalez, L. & Feingold, M. (1990). Retention of Spanish in the home. International Journal of the
Sociology of Language
, 84, 5-34.
13 Stoessel, S. (2002). Investigating the role of social networks in language maintenance and shift. International
Journal of the Sociology of Language
, 153, 93-131.

8


Language Use Research Review



A large inventory of factors that influence individual’s decisions about first-
language maintenance, they have largely not addressed the issue of the
social mechanisms by which individuals negotiate and sustain their
choices.

Stoessel conducted detailed social network research among a small sample of
immigrants. She found that the frequency of English usage within the non-family
networks was a principle determinant of whether subjects would maintain their use of
their first language.

However, micro social networks will form even
within the family. Tuominen14 concluded that
Marin, et al. Acculturation Scale
children within the family often determine what
What Language do you read or speak?
language the family will speak at home. That
What language did you use as a child?
reflects the fact that children in schools are in
What language do you usually speak at
non-family networks that often represents, to
home?
most, intense connection between the family
In which language do you think?
and formal training in English proficiency.
What language do you usually speak with
Furthermore, Tuominen found that in families
your friends?
with multiple children, the “children invariably
In what language are the tv programs that
spoke English among themselves, creating a
you usually watch?
micro-network within the family which further
In what language are the radio programs
increased the likelihood that the first language
that you usually listen to?
would not be maintained.” In the case of our
In what language are the movies, tv and
own methodological objectives, this suggests:
radio programs that you prefer to watch

and listen to?
One child makes it more likely that the
Are your close friends: all
family will shift towards English but two
Latinos/Hispanics, More Latinos than
or more school-age children, further
Americans, About half and half, More
Latinos than Americans, all Americans
empowers the children to influence the
language of the family.
Do you prefer going to social
gatherings/parties with: all Latinos …. all

Americans.
Marin et al.15 present an interesting battery of
The persons you visit or who visit you are:
language, media use and social network
all Latinos ….. all Americans.
questions. These scale items are reproduced
If you could chose your children's friends,
and discussed in Appendix 1 along with many
would you want them to be: all Latinos…..
All Americans.


14 Tuominen, A. (1999). Who decides the home language? A look at multilingual families. International Journal
of the Sociology of Language
, 140, 59-76.

15 Marin et al., op. cit.

9


Language Use Research Review


other questions identified as part of this review.

The authors conducted a factor analysis against these items and, without surprise, they
derived three factors from within the data: language, media and social relations (See
Table 1).


Table 1
Marin, et al.
Language
Media
Social Relations
Validation Variables
Factor
Factor
Factor
Generation 0.69
0.43
0.53
Length of Residence
0.76
0.46
0.50
Self-evaluation 0.74
0.52 0.66
Acculturation Index
0.86
0.60
0.66
Age on Arrival
-0.72
-0.58
-0.46

The authors correlated the three factors with several variables that are frequently
considered throughout the literature as being associated with maintenance/shift in an
effort to predict language usage.

Table 1 suggests that the validation variables are most closely correlated with the factor
reflecting the language-related questions shown in the previous page (read or speak,
spoke as a child, at home, in which you think, to your friends). These potential
surrogates were next most closely correlated with the questions on social relations
(close friends, parties, whom you visit, preference for children’s friends). Finally, the
correlation with media-related questions (TV usually watched, radio usually listened,
programs preferred) was the lowest.

It is impossible to determine from Table 1 whether any of the validation variables could
serve as surrogates for language use as a control for television measurement. The
correlations are reasonably strong as far as social science relationships go, but they are
far from deterministic. Table 1 does give us some guidance though, on what we might
find with respect to the relative strength of the surrogates in predicting media behaviors
in the Nielsen Peoplemeter panel.

3b. Generations

It is worth a special note about generations. As illustrated in Table 1, generation turns
out to be less correlated than many of the other variables. It is one of the more heavily
studied predictor variables. Stevens16 points out that a majority of those who use a

16 Stevens, G. (1992). The social and demographic context of language use in the United States. American
Sociological Review,
57, 171-185 and,

10

Document Outline

  • ÿ
  • ÿ
      • ÿ
    • ÿ
    • ÿ
    • ÿ
  • ÿ
  • ÿ