Three Essays in Culture and Cognition PDF Free Download
1 / 130/130
100%
Three Essays in Culture and CognitionThree Essays in Culture and Cognition
Michael Lee WoodMichael Lee Wood
Publication DatePublication Date
09-04-2019
LicenseLicense
This work is made available under a All Rights Reserved license and should only be used in accordance with
that license.
Citation for this work (American Psychological Association 7th edition)Citation for this work (American Psychological Association 7th edition)
Wood, M. L. (2019).
Three Essays in Culture and Cognition
(Version 1). University of Notre Dame.
https://doi.org/10.7274/t722h70618j
This work was downloaded from CurateND, the University of Notre Dame's institutional repository.
For more information about this work, to report or an issue, or to preserve and share your original work,
please contact the CurateND team for assistance at curate@nd.edu.
THREE ESSAYS IN CULTURE AND COGNITION
A Dissertation
Submitted to the Graduate School
of the University of Notre Dame
in Partial Fulfillment of the Requirements
for the Degree of
Doctor of Philosophy
by
Michael Lee Wood
Omar Lizardo, Co-Director
Terence McDonnell, Co-Director
Graduate Program in Sociology
Notre Dame, Indiana
April 2019
© Copyright 2019
Michael Lee Wood
THREE ESSAYS IN CULTURE AND COGNITION
Abstract
by
Michael Lee Wood
This dissertation consists of three essays on culture and cognition, with a special
focus on issues and opportunities relating to text and interview data. In the first essay I
develop a formal approach to analyzing patterns of cultural activation in moral reasoning,
adapting Swidler’s (1986) concept of “unsettledness” to discern the degree of diversity in
reasoning about a given topic. I then use the framework to identify variation in
unsettledness in American parents’ reasoning about the age-appropriateness of a variety of
different practices. In the second essay I take up the question of what interviews can tell us
about culture and action, responding to recent critiques claiming that the language used in
interviews are unable to tell us much about behavior because they evoke declarative
discourse removed from experience (Vaisey 2009; Martin 2010; Jerolmack and Khan
2014). Drawing on contemporary cognitive linguistics, I argue that by measuring
concreteness in language, interviews and text data more generally can access experiential
traces linked to past and future action. I test this hypothesis using a combination of
interview and survey data from the National Study of Youth and Religion, and I find that
the concreteness with which adolescents discuss their religious beliefs and practices is
Michael Lee Wood
positively correlated with the frequency of religious attendance when they are young
adults. In the third essay I develop a framework for thinking about culture in reasoning that
accounts for the fact that 1) patterns in reasoning vary across groups and individuals, and
2) reasoning is a situationally-variable process that changes depending on the substance
and framing of the topic at hand. Drawing on work from cognitive linguistics cognitive
anthropology, I argue that American parents share a cultural model for reasoning about
age-appropriateness consisting of basic schemas associated with the task of making
judgments and a set of affordances associated with the domain of age-appropriateness. This
model is deployed flexibly and variably depending on parents’ experiences with the task
in question. The three essays vary considerably in method and content, but each develops
new ways of conceptualizing and measuring personal culture found in text data, challenges
common beliefs about the uses and limitations of interviewing, and opens new avenues for
future research.
ii
For Shannon, Ansel, and Juniper.
iii
CONTENTS
Figures..................................................................................................................................v
Tables ................................................................................................................................. vi
Acknowledgments............................................................................................................. vii
Chapter 1: Introduction ........................................................................................................1
1.1 Overview ............................................................................................................1
1.2 The Three Essays ...............................................................................................2
Chapter 2: Unsettledness and Moral Reasoning ..................................................................6
2.1 Abstract ..............................................................................................................6
2.2 Introduction ........................................................................................................7
2.3 Response Categories and Reasoning Vocabulary in Moral Reasoning .............9
2.4 Measuring the Unsettledness of Response Categories and Reasoning
Vocabulary .................................................................................................11
2.5 Data ..................................................................................................................20
2.6 Unsettledness in Response Categories .............................................................22
2.7 Unsettledness in Reasoning Vocabulary ..........................................................32
2.8 Unsettledness in Response Categories and Reasoning Vocabulary ................37
2.9 Discussion and Conclusion ..............................................................................39
Chapter 3: Concreteness in Talk about Religion and Future Religious Participation .......42
3.1 Abstract ............................................................................................................42
3.2 Introduction ......................................................................................................43
3.3 Theoretical Background ...................................................................................46
3.3.1 Interviewing and Two Theories of Culture and Action ....................46
3.3.2 Challenges Facing Interview-based Measures of Experience ..........49
3.3.3 Construal-Level Theory ....................................................................52
3.4 Data and Measurement ....................................................................................54
3.5 Results ..............................................................................................................59
3.6 Discussion and Conclusion ..............................................................................63
Chapter 4: A Dynamic Theory of Culture in Reasoning: The Case Of Age-
Appropriateness ........................................................................................................66
4.1 Abstract ............................................................................................................66
4.2 Introduction ......................................................................................................67
4.3 Theoretical Background ...................................................................................71
iv
4.3.1 Public vs Personal Culture ................................................................71
4.3.2 Judgment vs Reasoning.....................................................................71
4.3.3 Schemas and Reasoning ....................................................................72
4.3.4 Age-Appropriateness ........................................................................74
4.4 Data and Analytic Approach ............................................................................75
4.5 Basic Schematic Models and Schematic Elaborations ....................................76
4.5.1 Behavior Exerts a Force on the Child ...............................................77
4.5.2 Behavior Opens New Paths ..............................................................83
4.5.3 Behavior Is Dirty...............................................................................91
4.5.4 Summary ...........................................................................................95
4.6 Variation in Reasoning ....................................................................................98
4.6.1 Variation by Behavior .......................................................................99
4.6.2 Variation by Personal Experience .....................................................99
4.6.3 Variation by Declarative Memory ..................................................100
4.6.4 Variation by Judgment ....................................................................100
4.7 Discussion and Conclusion ............................................................................101
Appendix A: Interview Sample .......................................................................................104
Bibliography ....................................................................................................................105
v
FIGURES
Figure 2.1: Effective Numbers ...........................................................................................16
Figure 2.2: Four Hypothetical Settledness Profiles ...........................................................19
Figure 2.3: Distributions of Common Age Words by Topic .............................................24
Figure 2.4: Effective Numbers of Age Categories by Topic .............................................27
Figure 2.5: Unsettledness in Response Categories and Restriction ...................................29
Figure 2.6: Unsettledness of Response Categories by Gender ..........................................31
Figure 2.7: Unsettledness in Reasoning Vocabulary by Topic ..........................................33
Figure 2.8: Vocabulary Unsettledness and Danger ............................................................36
Figure 2.9: Unsettledness in Reasoning Vocabulary and Response Categories ................38
Figure 3.1: Concreteness in Discussions about God ..........................................................58
Figure 3.2: Coefficient Plots ..............................................................................................60
Figure 3.3: Predicted Probabilities .....................................................................................62
Figure 4.1: Key Concepts in a Dynamic Theory of Culture in Reasoning ........................70
Figure 4.2: Schematic Diagram of Behavior Exerts a Force on the Child. .......................78
Figure 4.3: Schematic Diagram of Behavior Opens New Paths ........................................85
Figure 4.4: Schematic Diagram of Behavior is Dirty ........................................................93
vi
TABLES
Table 2.1: List of Practices ...............................................................................................21
Table 4.1: Summary of Basic Schematic Models and Schematic Elaborations ................97
Table A.1: Interview Sample ..........................................................................................104
vii
ACKNOWLEDGMENTS
That this dissertation is different than anything I could have imagined when I first
came to Notre Dame is a testament of the profound influence of the people I met during
my time there. As an incoming PhD student, I knew nothing of cognitive science and
computational text analysis, let alone planned on writing a dissertation about them. I am
deeply grateful to have been at Notre Dame at this particular moment, with the unique
combination of faculty, students, and opportunities that made my research program
possible.
I would like to express my deepest appreciation to my committee, for all their
support and useful criticism. I am deeply indebted to Omar Lizardo, without whom this
project would not have been possible, and without whom I would be a very different kind
of scholar. His 2015 graduate seminar on cognitive science introduced me to a body of
literature and way of thinking about the social sciences which would come to define my
graduate school career. Omar believed in my ideas and gave me the resources and
encouragement I needed to develop them. I cannot thank him enough. Terry McDonnell
has been a friend and mentor since we met during a visit-day interview in February 2014,
during which I pitched a wacky project about charitable organizations organized by fans
of television and movie series. I am extremely grateful for the chance I have had to learn
from him and work alongside him. Christian Smith’s innovative work drew me to Notre
viii
Dame, and his rich data made my analysis possible. Over the years I have thoroughly
enjoyed and benefited from the conversations and debates we have shared. David Gibson’s
innovative projects inspired me to pursue new forms of text analysis, and his feedback has
been invaluable. Jessica Collett helped me develop my ideas early on and see new potential
avenues of research.
So many people beyond my committee have been equally indispensable. Over the
years, Kraig Beyerlein has given exceptional advice and counsel and answered many
questions about statistics and religion. The late Mary Ellen Konieczny encouraged my
theoretical interests and inspired me with her brilliance, openness, and generosity. I miss
her dearly. David Hachen gave me many opportunities to increase my skills in research
design and computational analysis as a research assistant for the Net Health project. Lyn
Spillman helped me develop a reading list for an exam in cultural sociology which was
foundational for my thinking in this area. Erin McDonnell, Abi Ocobock, David Sikkink,
and Barbara Walters all took a sincere interest in my work and gave me valuable feedback
when I was finishing the third chapter.
I have also benefited immensely from interactions with scholars outside Notre
Dame. In May 2015 I sat in on a special conference on “measuring culture” at Notre Dame,
which brought many of the field’s leading scholars for discussion and debate. The
stimulating conference raised questions and ideas which shaped my thinking in the next
years. In August 2017 I had the opportunity of presenting an early version of the fourth
chapter at a Session called “New Directions in Contemporary Theory,” organized by Aliza
Luft. I am grateful for her belief in my project and for providing valuable feedback. In the
summer of 2018, I had the singular opportunity of attending the Diverse Intelligences
ix
Summer Institute at the University of St. Andrews. There I had many enlightening
conversations on cognition and computation with Jacob Foster, Stephen Vaisey, Lynette
Shaw, Andrei Boutyline, Jon Atwell, Alina Arseniev-Koehler, and Joshua Doyle, which
shaped my thinking in new and exciting ways.
My classmates at Notre Dame have been more important to my education and
research than I ever expected, and have become some of my dearest friends. I am beyond
grateful to have been in a cohort with Dustin Stoltz and Marshall Taylor, who are two of
the most brilliant and prolific scholars I know, and to whom I owe so much. So many of
the ideas contained herein emerged in conversations with Dustin, to the point that it
sometimes felt like we were dissertating together. He is a dear friend and colleague, and I
feel lucky to know him. Marshall introduced me to computational text analysis, answered
my millions of questions, and has been a loyal companion and friend. Justin Van Ness has
been a constant friend, conversation partner, and inspiration. I will always be grateful to
these three for debating and fighting with me until we ended up with a co-authored
publication in Sociological Theory and a theoretical framework for our dissertations. I give
special thanks to Brandon Sepulvado, who early on pointed me to theoretical work that
would become essential to my thinking. I am also especially grateful to Peter Ryan, Bridget
Ritz, Feyza Akova, and Kelcie Vercel for the many conversations that furthered my
thinking, as well as all the participants of the Notre Dame Culture Workshop who gave me
invaluable feedback on early drafts of the second and fourth chapters.
Finally, I express my warmest gratitude to my family and long-time friends. I thank
my parents for always believing in my abilities, even when I doubted myself. I am grateful
to Keith Cooper, Jeannie Burnett, Harry Wood, and other wise mentors I had growing up
x
who instilled within me the value and importance of getting an education. I wish I could
speak with them now. Most of all, I am eternally grateful to my steadfast companion,
Shannon Wood, for her unwavering support, incisive criticism, and brilliant creativity, and
to my children, for reminding me how wonderful and special life is and making me laugh.
1
CHAPTER 1:
INTRODUCTION
1.1 Overview
Cognitive processes have long been a part of sociological investigation, appearing
early on in the works of Durkheim, (Lizardo 2013; Durkheim 2008 [1912]), Mauss
(Durkheim and Mauss 2009 [1903]), Tarde (2010), Hertz (2013 [1909]), Dewey (2012
[1925]), Mead (2015 [1934]), and Mannheim (2013 [1936]), and later in the works of
Berger and Luckmann (1967), Goffman(1974), Waner (1978), Cicourel (1974), and
Bourdieu (1977, 1990; Lizardo 2004). The turn of the century saw increasing interest in
cognition among cultural sociologists (DiMaggio 1997; Zerubavel 1997; Cerulo 2002),
which has only increased in subsequent years. The most recent form of cognition research
among sociologists stands out for its heightened interdisciplinary engagement with the
cognitive sciences (Cerulo 2016), which has opened new avenues of research and
challenged older conceptual frameworks and methodological approaches (Martin 2010;
Turner 2018, 33–34).
The current era of culture and cognition research has seen significant opportunities
and challenges for text analysis in particular. On one hand, the combination of
technological innovation, big data, and accessible analytic software has created
unsurpassed opportunities for analyzing text (Bail 2014). At the same time, the link
between text and cognition is relatively underdeveloped within sociology (Ignatow 2016),
2
and recent theories of action cast doubt on the utility of text in sociological explanation
(Vaisey 2009; Jerolmack and Khan 2014). Taken together, these developments raise two
important questions: First, to what extent can text be used to be studied culture and
cognition? Second, how can cognitive processes be measured via text? I address these two
questions in the three essays comprising this dissertation.
1.2 The Three Essays
In the first essay (chapter 2) I develop a new framework for identifying patterns of
cultural activation in moral reasoning to facilitate the analysis of variation between topics,
groups, and time. In contrast to what I identify as the common, “content-based” approach,
which realizes comparison by identifying variable cultural elements manifest in people’s
reasoning, I develop a formal approach to comparison, adapting Swidler’s (1986) concept
of “unsettledness” to discern the degree of diversity in reasoning about a given topic. I
identify two lexical features of moral reasoning by which “unsettledness” is manifest—
“response categories” and “reasoning vocabulary”—and develop methods for measuring
the unsettledness of each. I then apply the framework by measuring the unsettledness of
response categories and reasoning vocabulary in American parents’ reasoning about the
age-appropriateness of a variety of different topics.
The essay contributes to culture and cognition research in two main ways. First, it
identifies two forms of personal culture—response categories and reasoning vocabulary—
which are evoked in moral reasoning. Response categories and reasoning vocabulary are
forms of “declarative culture” which come to be associated with particular topics (Lizardo
2017; Lizardo et al. 2016). Second, the essay develops a way of thinking about personal
3
culture in collective terms. By aggregating responses to a given topic and calculating the
diversity of these aggregates, one can measure the “cognitive extent” (Milojević 2015) of
a given topic in a way that could not be observed by simply taking averages of the diversity
of individual responses.
In the second essay (chapter 3) I take up the question of what interviews can tell us
about culture and action, responding to recent critiques claiming that interviews are unable
to tell us much about behavior because they evoke declarative discourse removed from
experience (Vaisey 2009; Martin 2010; Jerolmack and Khan 2014). According to these
arguments, interviews are well-suited to study retrospective accounts or justifications of
action, but not action itself.
I argue that the argument does not apply to all analyses of interview data. Drawing
on contemporary cognitive linguistics, I show that under certain conditions, interviews and
text data more generally can give access experiential traces which are associated with past
and future action. I argue that the key to this endeavor is accessing relevant “backstage
cognition” (Turner and Fauconnier 1995; Turner 2000) manifest in verbal communication.
More specifically, the concreteness with which a person talks about a particular practice
(i.e. whether they tend to use more concrete or more abstract words) emerges from their
experiences with that practice in a way that is predictive of their dispositions. The reason
for this, according to recent discoveries in psycholinguistics, is that people tend to use more
concrete language when they discuss things psychologically proximate to them, and more
abstract language when they discuss things more distant. I test this hypothesis using a
combination of interview and survey data from the National Study of Youth and Religion.
I find that the concreteness with which adolescents discuss their religious beliefs and
4
practices is positively correlated with the frequency of religious attendance when they are
young adults.
The essay contributes to the study of culture, cognition, and religion in several
ways. First, it identifies language concreteness as a heretofore unrecognized form of
nondeclarative personal culture accessible via the way people talk about a given domain.
Second, the essay develops a method to analyze concreteness using interview data. Third,
the essay challenges common assumptions about the usefulness and limitations of
interviewing and opens up new possibilities for empirical research. In addition to being a
promising new method of measuring peoples’ experiences, concreteness in talk may also
be a promising way to get beyond social desirability bias, as psychological distance is not
something that can be easily faked, if at all. Regarding religion, the essay introduces an
innovative way to combine interview and survey data in the analysis of religious retention
and decline, and offers a way to measure the extent to which religious training is “sticking.”
The third essay (chapter 4) is a qualitative analysis of parental reasoning about age-
appropriateness. In the essay I develop a framework for thinking about culture in reasoning
that accounts for the fact that 1) patterns in reasoning vary across groups and individuals,
and 2) reasoning is a situationally-variable process that changes depending on the
substance and framing of the topic at hand. Drawing on work from cognitive linguistics
cognitive anthropology, I argue that American parents share a cultural model for reasoning
about age-appropriateness consisting of basic schemas associated with the task of making
judgments and a set of affordances associated with the domain of age-appropriateness. This
model is deployed flexibly and variably depending on parents’ experiences with the task
in question. The essay makes an intervention in schema analysis within sociology, which
5
tends to treat schemas as unique constructions of particular historical groups, rather than a
shared collection of “conceptual building blocks” which are evoked and combined flexibly
in response to immediate practical demands.
The three essays vary considerably in method and content, but there are strong
common threads uniting them. First, each essay is motivated by the same underlying
theoretical framework. Central to this framework is the assumption that culture consists of
two forms—personal and public—comprised of cognitive patterns and material objects,
respectively, that have a social causal history. I understand interview prompts as public
“frames” which activate personal culture, resulting in observed interviewee responses
(Wood et al. 2018). Related to this, each essay develops new ways of conceptualizing and
measuring personal culture found in text data. Each essay rejects the notion that text is
limited to the study of declarative accounts and seeks to open new avenues for investigating
cognition, whether it be by observing unsettledness, measuring concreteness, or
uncovering cultural models. Each essay is optimistic and forward-looking, focused on new
opportunities for sociological analysis made possible by contemporary developments in
theory and method. Going forward, I hope these essays may serve to further develop
thinking and measurement of culture and cognition in many domains of research.
6
CHAPTER 2:
UNSETTLEDNESS AND MORAL REASONING
2.1 Abstract
In this essay, I develop a new framework for identifying patterns of cultural
activation in moral reasoning to facilitate the analysis of variation between topics, groups,
and time. In contrast to what I identify as the common, “content-based” approach, which
realizes comparison by identifying variable cultural elements manifest in people’s
reasoning, I develop a formal approach to comparison, adapting Swidler’s (1986) concept
of “unsettledness” to discern the degree of diversity in reasoning about a given topic. I
identify two lexical features of moral reasoning by which “unsettledness” is manifest—
“response categories” and “reasoning vocabulary”—and develop methods for measuring
the unsettledness of each. I then analyze the unsettledness of response categories and
reasoning vocabulary in American parents’ reasoning about the age-appropriateness of a
variety of different topics.
7
2.2 Introduction
Moral reasoning is a thoroughly cultural activity. When presented with moral
questions or dilemmas, people with different experiences and group identities respond in
different ways (Schalet 2011; Homan, Valentino, and Weed 2017). Additionally, people
adjust their public moral reasoning depending on how they perceive the standing of their
own positions relative to others (Strauss 2005; Swidler 2013). Though some moral codes
appear near-universal (Schwartz 1981), the way people think about moral concepts and
issues often varies across time periods (Strand 2015; Bargheer 2018) and periods of life
(Malti and Buchmann 2010).
Cultural analysts employ a range of concepts to describe the cultural elements at
play in moral reasoning, including frames (Ghaziani and Ventresca 2005; Feagin 2013),
narratives (Silva 2012; Ewick and Silbey 2003), schemas (Thibodeau et al. 2015; Williams,
Blair-Loy, and Berdahl 2013; Edgell and Hull 2017; Ecklund et al. 2017), beliefs
(Shepherd and Marshall 2018; Strand and Lizardo 2015), values (Myyrya, Juujärvi, and
Pesso 2010), logics (Kazyak, Burke, and Stange 2018), boundaries (Guhin 2016; Lamont
and Lamont 1992; Lamont and Molnár 2002), among others. Abend (2014) offers a useful
framework for thinking about the relation between these cultural elements, distinguishing
between “first order morality,” consisting of the moral claims, beliefs, and behaviors that
comprise the “what” of moral life, and “second order” morality, consisting in the “para-
moral elements” that “facilitate, support, or enable first-order morality” (Abend 2014, 17).
These second-order “background” elements are the things which “literally make [morality]
possible” (Abend 2011, 147), and include elements such as conceptual repertoires,
8
assumptions of what may be morally evaluated, and procedures for making and evaluating
moral arguments.
Most generally, the cultural analysis of moral reasoning involves identifying the
cultural elements that either constitute or anchor reasoning on a given topic, and identifying
how they vary across different categories of interest, such as time, topic, or structural
position. The typical approach to analyzing variation consists in identifying differences in
the substance of these cultural elements, for example, by identifying that members of
different religious groups justify their judgments about reproductive genetic technologies
in schematically different ways, and depending on the precise character of the technology
in question (Ecklund et al. 2017). This content-based approach excels at at identifying
differences in type and tracing the origins of these differences.
Though less common, variation in culture and reasoning may also be analyzed by
measuring patterns in the distribution of cultural elements associated with a given topic or
question. For example, though not an analysis of morality, Lee and Martin (2015) identify
similarities and differences between members of the Frankfurt school based on networks
created by counting their use of a set of keywords. Similarly, (Bail 2012) uses plagiarism-
detection software to investigate the process by which narratives about muslims move from
fringe to mainstream. These formal methods are promising for the analysis of moral
reasoning because they facilitate the analysis of cultural change and afford a more detailed
comparison that is based on the diversity of distributions rather than differences in type
alone.
In what follows, I develop a new formal approach for analyzing moral reasoning
based on Swidler’s (1986) concept of “unsettledness.” Though Swidler herself has applied
9
the concept in a variety of ways, unsettledness has typically been used as a macro-level
descriptor of entire periods or historical moments (Swidler 1986, 278; Ghaziani and
Ventresca 2005, 524; Bail 2012, 857). In contrast, I use the concept in a new way to refer
to the degree of diversity in reasoning in response to a given moral issue, extending the
concept of unsettledness from the temporal to the cultural domain. I argue that topics may
vary in the degree to which they are “settled” for a given population 1. I identify two types
of cultural content in moral reasoning, which I call “response categories” and “reasoning
vocabulary,” and develop a way to measure the unsettledness of each.
Measuring unsettledness in reasoning is useful to address a variety of practical and
theoretical issues. In terms of analysis, a formal measure of unsettledness facilitates the
measurement of culture across time, which is an enduring problem in formal cultural
analysis (Edelmann and Mohr 2018, 6). In terms of theory, in places where reasoning
precedes behavior (Omar Lizardo and Strand 2010), unsettledness about a given topic may
produce varied and uncertain behavioral outcomes. Practically speaking, the unsettledness
associated with a topic may be useful for measuring the success of interventions aiming to
disseminate a particular way of thinking (McDonnell 2016).
2.3 Response Categories and Reasoning Vocabulary in Moral Reasoning
To investigate patterns of settledness and unsettledness in moral reasoning, I
propose making an analytic distinction between “response categories” and “reasoning
vocabulary,” referring to distinct dimensions of moral reasoning. Response categories are
1 Though “unsettledness” has both an affective and a formal dimension, I focus only on the latter.
10
key words or phrases which are potential judgments in response to a question on a given
topic, and reasoning vocabulary is the set of all words used when responding to the question
asked. These may be differentiated as referring to judgment and justification, respectively.
For example, response categories for questions about age-appropriateness include different
age-related words, such as “sixteen,” “high school,” or “never.” A response category
manifest in a person’s reasoning is not the same as their actual judgment about the question
at hand, but indicates a potential judgment in the space of possible judgments. For instance,
a person may acknowledge a range of response categories, but choose only one or two as
the their preferred answer (e.g. “I know some people say x is good, but I say y is better”).
Response categories and reasoning vocabulary may be counted at either the
individual level or as an aggregate. The latter approach, which I take here, entails
aggregating all the responses to a given question and then counting words (as opposed to
counting words and then aggregating into averages). This approach results in distributions
of response categories and reasoning vocabulary for each topic, which may be understood
as the “space” of response categories/reasoning vocabulary for that topic. Once the space
of response words and reasoning vocabulary has been populated for a given topic, the
degree of unsettledness for that topic can be calculated.
A topic that is unsettled in terms of response categories exhibits a higher level of
diversity of response categories. Imagine, for example, that we asked a sample of South
Bend residents which local restaurant they thought was the best. If in the aggregate, female
residents mentioned 50 restaurants with relatively even frequency, and male residents
mentioned 40 restaurants with one restaurant mentioned much more frequently than the
rest, we would say that the space of response categories (here, different restaurants) is less
11
settled for female residents. Alternatively, a topic that is unsettled in terms of reasoning
vocabulary exhibits a higher degree of lexical diversity in the aggregated sample of words,
meaning that individuals tend to discuss the topic differently from one another rather than
drawing on a shared vocabulary.
Response categories and reasoning vocabulary are distinct concepts, and their
respective measures of unsettledness do not necessarily correlate. A topic with unsettled
response categories may or may not be unsettled in terms of reasoning vocabulary, and
vice versa. You and I may agree on, say, the possible contenders for best restaurant, but I
might justify my choice in the language of design and aesthetics, and you might justify
your choice in the language of taste. Alternatively, we may both justify our choices in the
language of taste, but have different ideas about which restaurants are reasonable
contenders. The theoretical implication is that different processes might be responsible for
producing settledness in response categories and reasoning vocabulary.
2.4 Measuring the Unsettledness of Response Categories and Reasoning Vocabulary
The unsettledness of response categories and reasoning vocabulary in moral
reasoning can be measured by adapting measures of lexical diversity. Lexical diversity is
a common measure in linguistics, where it is typically treated as an individual-level
variable that correlates with other aspects of language-use, such as language proficiency
and development (Malvern et al. 2004). However, in this paper, I treat lexical diversity an
aggregate property. When applied to aggregations of text, lexical diversity can be used to
measure the distributions of words (in this case, response categories or reasoning
vocabulary) for particular topics for a given sample.
12
My approach to measuring unsettledness in moral reasoning consists of three basic
steps. First, responses to a particular question or set of questions are collected from samples
of the group or groups under investigation. Next, individual text responses are aggregated
into group-level collections of words. When measuring response categories, this
aggregation consists of words representing the categories. When measuring reasoning
vocabulary, it consists of either all the words from each document, or n words from each
document. Finally, lexical diversity on these aggregations is calculated. If the aggregation
is created using text samples, many samples are drawn and the lexical diversity score is
returned as a distribution rather than a single number.
Measures of lexical diversity are many and varied, but all begin by counting “types”
and “tokens.” When measuring lexical diversity, type refers to unique words that appear in
a document, and token refers to any word that occurs in the document. To illustrate, the
phrase “a snake is a reptile” contains four types and five tokens. The total number of types
in a document is known as lexical richness, and represents the range of words of the
document. The total number of tokens is the document length. By dividing the total number
of types by the total number of tokens, we get the Type-Token Ratio (TTR) (Johnson 1944),
the simplest measure of lexical diversity. For example, the TTR of “a snake is a reptile” is
0.8.
TTR is a simple, elegant measure of lexical diversity, and much lampooned by
linguists since its inception. The biggest shortcoming of TTR is that it is highly correlated
with document length, so much so that it is impossible to reliably compare the lexical
diversity of two documents of differing lengths using TTR alone (Baayen 2008). This
limitation has resulted in a decades-long quest to develop a measure of lexical diversity
13
that is independent of document length (Jarvis 2013). The earliest solutions to this problem
involve holding the number of tokens constant, for example, by truncating the longer
document to match the length of the first (Johnson 1944). Though many new measures
have been proposed, holding word length constant in some form remains a common
solution (Baayen 2008; Scott Jarvis 2013; Milojević 2015), and serves as the basis for some
newer measures as well (Covington and McFall 2010).
Jarvis (2013), a linguist, argues that despite their momentous efforts to craft the
perfect lexical diversity measure which is independent of document length, linguists have
ultimately failed because their technical obsession has ignored the fact that statistical
relations between types and tokens are more accurately considered measures of word
repetition rate, rather than diversity. Perhaps the biggest limitation of measures built on
the simple logic of the TTR is that they do not directly take into account the distribution of
tokens across unique types. For example, consider the following two strings:
“ant bee wasp wasp wasp wasp”
“ant ant bee bee wasp wasp”
Each of these strings has three types and six tokens, and as a result, each has a TTR
of .5. However, the latter phrase is clearly more diverse because the tokens are distributed
evenly across the three types, whereas they are concentrated unevenly in the first. In this
way, Jarvis (2017; 2013) argues, linguists relying on measures based on the logic of TTR
fail to adequately measure the way the concept of diversity is understood and perceived.
The two strings above may have the same rates of repetition, but not diversity. As a
corrective, Jarvis suggests looking to ecology, which has its own sophisticated measures
of diversity.
14
Diversity measures in ecology share the same foundation of types and tokens,
though these are referred to as species and specimen, respectively. Additionally, unlike
linguistics, the standard measures of species diversity in ecology take frequency
distributions into account. The most common indices are Shannon Entropy (1948) and the
Gini-Simpson Diversity Index (Simpson 1949). Shannon Entropy is a measure of
uncertainty in the outcome of a sampling process. “When it is calculated using logarithms
to the base two, it is the minimum number of yes/no questions required, on the average, to
determine the identity of a sampled species” (Jost, 2006, p. 363). The Gini-Simpson index
measures the probability that any two sampled specimen will belong to different species.2
Interestingly, when applied to text, these diversity measures tend to be less correlated with
word length than traditional measures of lexical diversity (Jarvis 2013; McCarthy and
Jarvis 2010).
The Shannon and Gini-Simpson indices are useful measures of diversity but are
easy to misinterpret because they are nonlinear. To facilitate interpretation, ecologists
recommend converting them to “effective numbers,” or “true diversity” (Chao, Chiu, and
Jost 2010; Jost 2007, 2006). True diversity is calculated by figuring out how many unique
equally-abundant things are needed to equal the diversity index of the observed sample.
Figure 2.1: Effective Numbers shows how effective numbers change across different
samples.3 In sample A, there are five ants and no other species, so the effective number is
one. As the distribution of species becomes more even, the effective number of species
2 The Gini-Simpson index was rediscovered by Herfindahl in 1949.
3 These numbers were produced using the Gini-Simpson Index.
15
increases. When the distribution of species is completely even, as in sample D, the effective
number of species is equal to the number of unique types.
16
Figure 2.1: Effective Numbers
17
One fascinating discovery in mathematics is that true diversity based on species
richness, Shannon entropy, and Gini-Simpson diversity can all be calculated by changing
a single parameter (q) in the following formula (Hill 1973):
R refers to richness (the number of unique types) and Pi refers to the proportional
abundance of the ith type. When q is set to 0, the effective number equals richness, when
q=1, the effective number is calculated according to Shannon diversity4, and when q=2 it
is calculated according to Gini-Simpson diversity. The differing values of q affect how
sensitive the measure is to rare vs abundant species. q=0 is biased toward rare species
(because every new species increases richness the same amount, regardless of frequency),
q=2 is biased toward abundant species, and q=1 is considered the most neutral because it
weights each type by its frequency. In the subsequent analysis, I measure unsettledness in
reasoning about age-appropriateness by calculating the effective number of words based
on Shannon Entropy.
To measure the degree of unsettledness across multiple topics or domains of moral
reasoning, an analyst may create a settledness profile. A settledness profile is a collection
of measures of unsettledness for a sample of topics for a given sample of individuals, and
may be created for either response categories or reasoning vocabulary. Figure 2.2 shows
four hypothetical profiles created by measuring the unsettledness of responses to six
4 This is technically achieved by setting q to the limit of 1, because q=1 is undefined.
18
different verbal elicitations, or “frames” (Wood et al. 2018). Plot A represents the most
settled profile because the unsettledness of responses for this group are low for each of the
six frames. By contrast, plot B represents the least settled profile because the unsettledness
is high for each frame. Plots C and D represent equally settled profiles, but they are opposite
in terms of which frames are settled and which ones are not.
19
Figure 2.2: Four Hypothetical Settledness Profiles
20
In the subsequent analysis, I create two settledness profiles—one for response
categories and one for reasoning vocabulary—from a sample of American parents who
answered a series of questions about the age-appropriateness of a variety of practices.
2.5 Data
The data for the analysis come from interviews conducted from 2014-2015 as part
of the Intergenerational Religious Transmission Project (IRTP). Researchers interviewed
parents of different religious backgrounds from different regions of the United States on
topics relating to parenting and the transmission of religion. For this analysis, I used the
transcripts from 44 interviews (22 men and 22 women) that responded to a module on age-
appropriateness. Interviewers told participants they would be asked at what age it was
appropriate to begin certain practices, and then read a list of practices, one a time, allowing
the parent to respond to each one. Table 2.1 lists the practices.
21
TABLE 2.1:
LIST OF PRACTICES
Interviewer: “I’m going to list a number of things and after each one, your job is to tell
me at what age, if ever, it is appropriate for a child to do or have that thing.”
1. Getting a smartphone
2. Having a television in their bedroom with cable
3. Having a computer in their bedroom with internet
4. Getting a social media account
5. Wearing makeup
6. Watching an R-rated movie
7. Using four-letter words
8. Having a first beer or shot of liquor
9. Drinking regularly
10. Experimenting with soft drugs like marijuana
11. Attending parties with peers, unsupervised by adults
12. Start dating
13. Having sex with a romantic partner
14.
Viewing porn
22
Before analyzing the text I prepared it with a variety of pre-processing techniques.
This included sending text to lowercase, breaking apart contractions, converting numbers
to text, removing punctuation, and removing common “stopwords.” I added some
exceptions to a standard stopword list to ensure that important topic-related words were
not omitted. I also identified and removed fillers such as “uh” and “um.” This is an
important step when analyzing lexical diversity from interview transcripts because they
often contain many filler words spelled in a many different ways, which can easily inflate
diversity measures. Finally, I stemmed words, which has the effect of ensuring that words
sharing the same stem are not counted as separate types when measuring diversity.
2.6 Unsettledness in Response Categories
To measure unsettledness in terms of response categories for the different topics, I
first needed to identify what those response categories were. I did this by first identifying
the most frequently-occurring words of all the responses, and from this I created a subset
words that could be classified as responses to the prompt, which included 18 different of
age categories such as ‘sixteen’, ‘high school,’ and ‘never’. I then calculated the
frequencies of these age categories for each topic, providing a general overview of how
age-appropriateness is evaluated across topics. Figure 2.3 shows the distributions of these
age words for each topic. The topics are organized in descending order by the relative
frequency of “never.” We cannot be certain based on decontextualized words whether the
occurrence of an age word is the same as the parent’s evaluation, but the observed
distributions of age words matches intuitive expectations. For our purposes, however, we
23
are more interested in the space of possible response categories rather than people’s
personal judgments.
24
Figure 2.3: Distributions of Common Age Words by Topic
25
The topics that elicit the highest proportion of “never” are viewing pornography,
cursing, and using drugs, and the topics that elicit the lowest proportion of “never” are
getting a smartphone, creating a social media account, and watching an R-rated movie.
Interestingly, though smartphones and social media have become relatively
uncontroversial, having a tv in the bedroom is relatively contested. In general, the
distributions of appropriate ages is proportional to the proportion of “never” responses--
where there are more “nevers”, younger ages are mentioned with less frequency. There are,
however, three predictable exceptions to this rule; drinking alcohol, having sex, and
watching r-rated movies have relatively low frequencies of “never” but skew toward older
ages.
The distributions in Figure 2.3 gives a sense of the differences in the unsettledness
of response categories. Topics for which responses are highly skewed toward a single age
category, such as “viewing porn,” exhibit lower diversity, and hence, more settledness.
Similarly, topics for which responses are more evenly distributed, such as wearing makeup,
exhibit higher diversity, and hence, less settledness. However, rather than eyeballing these
differences, we can measure them directly.
To measure unsettledness in response categories, I calculated the effective numbers
of response categories for each topic using Shannon entropy. The effective numbers of age
categories for each topic is shown in Figure 2.4. The topics in Figure 2.4 are organized in
descending order from the most settled to the least settled. As we might have expected,
viewing pornography is the most settled topic in terms of response categories. Though there
are 18 possible age categories, we would only need six equally-common categories to
achieve the diversity index obtained from the responses to this topic. By contrast, wearing
26
makeup and dating are the most unsettled in terms of response categories, with effective
numbers of about 12. This means that the age categories for dating and wearing makeup
are about twice as diverse as the age categories for viewing pornography.
27
Figure 2.4: Effective Numbers of Age Categories by Topic
28
Having measured the unsettledness of these different topics in terms of response
categories and seeing the differences between topics in the unsettledness profile, we can
then then ask, what might be driving differences in response category settledless between
the different topics? In the case of age-appropriateness, almost all of the variation appears
to be explained with how “restricted” the practice is; practices with the highest proportion
of “never” responses are also the most settled (least diverse). The relation between
unsettledness in response categories and restriction is plotted in Figure 2.5. The Pearson’s
correlation between these two measures is .88, and the adjusted R-squared is .77.
29
Figure 2.5: Unsettledness in Response Categories and Restriction
30
That there is such a strong relation between the unsettledness of age categories and
the frequency of “never” responses should cause little surprise. If a practice is ever
appropriate, then a parent must choose an age or range of ages; something a parent does
not need to do if they deem the practice never appropriate. Simply put, there is more room
for diversity when a behavior is widely deemed acceptable. At the same time, there is still
some variation that needs explaining. For example, drinking regularly is still quite settled
after controlling for the frequency of never responses. This is because there is a frequent,
institutionalized response category—21— that marks its appropriateness.
For illustrative purposes, I also plot differences in the unsettledness of response
categories between men and women, which is shown in Figure 2.6. Though generally
similar, the response categories for men are more settled than women when discussing
televisions and computers in the bedroom, and less settled than women when discussing
unsupervised parties and R-rated movies. Men and women are so similarly settled in their
response categories when discussing pornography that their points overlap.
31
Figure 2.6: Unsettledness of Response Categories by Gender
32
2.7 Unsettledness in Reasoning Vocabulary
To measure unsettledness in reasoning vocabulary, I aggregated responses for each
topic and calculated the effective number of words from strings created by randomly
sampling 550 random words from each of these aggregates. I repeated this process 1000
times for each topic, creating a distribution of effective numbers for each topic. This
sampling process ensures that diversity scores are not biased in favor of topics that elicit
more words overall (Milojević 2015, 964). I chose 550 because it is less than the word
count of the sparsest topic, but not so low that the number of types for any topic exceed it.
If the number of words sampled from a document is less than the total number of types in
that document, diversity measures for these documents may reach “saturation,” causing the
resulting scores to be artificially biased (Milojević 2015, 966). Figure 2.7 shows
unsettledness of reasoning vocabulary for each topic.
33
Figure 2.7: Unsettledness in Reasoning Vocabulary by Topic
34
In the case of age-appropriateness there are some notable differences between the
settledness of response categories and the settledness of reasoning vocabulary. Most
notably, porn and drugs—two topics that exhibited high settledness of response
categories—are the most unsettled topics in terms of reasoning vocabulary. This suggests
that although parents mention relatively few age categories when discussing these topics,
their discussions about these topics are quite diverse.
Why might some topics exhibit more unsettledness in reasoning vocabulary than
others? One possible explanation is that unsettledness in reasoning vocabulary is related to
perceptions of controversy and danger. When people voice an opinion, the way they do so
is moderated by how they perceive their position relative to others (Strauss 2005: 232).
Opinions perceived as controversial are presented and justified differently from opinions
believed to be common, such as by adding more justifications and qualifications. This may
especially be the case when reasoning about a topic that is perceived as dangerous, thereby
raising the stakes. For example, consider the way June, mother of three sons, justifies her
position on viewing pornography:
Um, I think they can really handle it. I mean I guess 15 16… When they
first got their computer I would say “are you looking at boobs? Are you
looking at boobs online?” I would drive them crazy. It's like, they know that
I know that happens. They don't think “oh mom doesn't know.” If they're
curious they can look at whatever they want, but they have to kinda--their
brain has to be a little more developed.
June says that it is okay to view pornography, but she does so in a way that clearly
acknowledges the danger of the activity and justifies her position in response to this danger.
First, she claims that they can “handle it,” implying that viewing porn exerts a force that
affects the viewer (Talmy 1988). Second, she qualifies her response by saying that “their
35
brain has to be a little more developed,” implying that viewing porngraphy could be
harmful if this condition were not met.
Perceptions of controversy and danger do not necessarily produce more vocabulary
unsettledness, but it seems plausible that we would observe higher lexical diversity in
situations where a higher proportion of respondents are in a “more alert” justificatory mode
that pushes them to add more qualifications and justifications than less controversial, safer
topics. Following this logic, reasoning about viewing pornography may have an especially
high degree of vocabulary unsettledness because its defenders and opponents both feel it
is dangerous, and both believe their positions are relatively marginalized (defenders,
because porn is stigmatized, and opponents, because porn is omnipresent).
Absent any dedicated measures of controversy or danger, I performed a preliminary
test of the controversy-danger hypothesis by correlating the frequency of “dangerous”
words with vocabulary settledness. The dangerous words I counted were “harm, danger,
trauma, hurt, distort, destruct, abus, ruin, kill, addict, serious, conseq, injur, pain.”5 The
relationship between the vocabulary unsettledness and the frequency of dangerous words
is shown in Figure 2.8. The pearson’s correlation between these two variables is .83 and
the adjusted r-squared is .66.
5 Technically, I counted strings of stemmed words, such as ‘injur.’
36
Figure 2.8: Vocabulary Unsettledness and Danger
37
2.8 Unsettledness in Response Categories and Reasoning Vocabulary
As mentioned above, unsettledness in response categories and reasoning
vocabulary are analytically distinct and not necessarily correlated. We can now directly
observe the relation between these measures by plotting them against each other, as in
Figure 2.9. The dots represent the median values for each topic for each measure. The
horizontal and vertical lines are set to the mean values for each measure. The plot suggests
a two-dimensional typology. In the bottom-left quadrant are topics which are more settled
both in terms of response categories and reasoning vocabulary. Following my hypotheses,
these should be practices that are more restricted but perceived as less controversial or
dangerous. Drinking regularly, and swearing and are highly restricted practices, but
relatively uncontroversial in adulthood.
38
Figure 2.9: Unsettledness in Reasoning Vocabulary and Response
Categories
39
In the upper-left corner are topics which are more settled in terms of response
categories but less settled in terms of reasoning vocabulary. Following the proposed
explanations, these should be topics that are more restricted and more
controversial/dangerous. The most extreme points, viewing porn and trying drugs, seem to
fit. In the bottom-right quadrant are topics which are less settled in terms of response
categories but more settled in terms of reasoning vocabulary. Following the proposed
explanation, these should be practices that are neither heavily restricted nor dangerous or
controversial. Wearing makeup and dating again seem reasonable exemplars. The upper-
right quadrant contains topics which are unsettled in both dimensions, though there is less
of a clear case. The best exemplar in the sample is having sex with a romantic partner,
which remains a relatively controversial topic for American parents, even in the context of
loosening norms (Schalet 2011).
2.9 Discussion and Conclusion
In this paper I developed a formal approach for analyzing culture in moral
reasoning. I adapted the concept of “unsettledness” for the analysis of reasoning and
identified two forms of culture in reasoning—response categories and reasoning
vocabulary—to which it may be applied. Taking inspiration from computational linguistics
and ecology, I developed methods for measuring the unsettledness of these concepts using
Shannon entropy. I also introduced the concept of a settledness profile to facilitate the
comparison of settledness across topics and domains. I then used this framework to
measure the unsettledness of parents’ response categories and reasoning vocabulary in
reasoning about the age-appropriateness of 14 different practices. I identified significant
40
differences in the degree of settledness across practices, and argued that this variation could
be explained by the degree of restriction, in the case of response categories, and perceptions
of controversy and danger, in the case of reasoning vocabulary.
Though my analysis focused primarily on comparing the unsettledness associated
with different topics, the framework also allows comparisons by group identity, as
suggested in the Figure 2.6 showing differences by gender. Additionally, the framework
will be useful for measuring the degree to which culture in reasoning changes or is
consistent across time.
Arranging topics by lexical diversity is a useful formal approach that delays the
interpretive moment and brings the author and audience on the same even plane (Lee and
Martin 2015). At the same time, though the approach I have developed here is a formal
one, analyses of settledness can benefit from traditional qualitative analysis as well
(Spillman 2014; Chakrabarti and Frye 2017). Quantitative analysis of transcripts excels at
identifying patterns of settledness and unsettledness, though close reading of texts,
especially interview texts, may reveal possible mechanisms that explain the degree of
settledness for a particular group. However, these proposed mechanisms may then be
formalized with additional data.
Future research on settledness in reasoning may benefit by combining multiple
forms of data collection. Measures created from the text itself can only go so far.
Explaining variation in settledness across topics would be greatly enhanced with the
inclusion of survey variables questions capturing respondents’ perceptions about the topics
at hand. For example, immediately after respondents give a verbal response, they can
indicate where they think their position stands in relation to others.
41
The current framework should be seen as a flexible approach and general direction
rather than a rigid proscription. I introduced the concepts of response categories and
reasoning vocabulary for the analysis of moral reasoning in particular, but unsettledness
might be measurable in other ways. For instance, I measured settledness using individual
words, but the same principle can be applied to clusters of related words. Such an approach
may make use of more sophisticated measures of diversity, such as recent “similarity-
sensitive” measures developed by ecologists (Leinster and Cobbold 2012). Finally, though
I measured settledness exclusively as a property of groups, there are currently-unspecified
forms of settledness which may be measured at the individual level as well, such as
cognitive ease, or intra-individual diversity over time.
Although the above analysis focused exclusively on the settledness of reasoning,
the proposed reconceptualization of settledness is applicable to cultural analysis more
generally. The framework is flexible enough to be adaptable to diverse forms of data
collection. For example, settledness could be defined more generally as the degree of
diversity of action vectors in response to particular frames (Martin 2011, 2015) With such
an expansion, there is no reason, for example, why ethnographers cannot observe the
settledness of actions during participant observation. In a given social setting, people
respond in different ways, sometimes by physically moving toward or away from
something the environment. By taking note of the different vectors in any given situation,
an ethnographer can work toward measuring the settledness of actors within that situation.
Alternatively, survey questions, understood as productive tasks, generate responses which
may be aggregated by group and used to measure consensus.
42
CHAPTER 3:
CONCRETENESS IN TALK ABOUT RELIGION AND FUTURE RELIGIOUS
PARTICIPATION
3.1 Abstract
What information about culture can sociologists glean from interview data, and
how can it be leveraged to study behavior? This question has provoked insightful debates
in recent years and has led to productive theoretical and methodological developments
(Pugh 2013; Vaisey 2014; Lamont and Swidler 2014; Martin 2010; McDonnell 2014). In
this paper I develop a novel approach to the cultural analysis of interview data that draws
on recent developments in psycholinguistics and automated text analysis. Using linked
survey and interview data from two waves of the National Survey of Youth and Religion,
I find that the concreteness with which adolescents discuss religious topics is associated
with their religious attendance when they are young adults. More generally, my findings
suggest that the concreteness with which a person discusses a topic like religion may
contain traces to their experience with that topic which are associated with future behavior.
One implication is that interview methods, analyzed quantitatively, may provide a more
direct access to experience than previously believed.
43
3.2 Introduction
What information about culture can sociologists glean from interview data, and
how can it be leveraged to study behavior? Though the value of interviews in cultural
analysis has been scrutinized (Vaisey 2009; Martin 2010), there is a growing recognition
that the kind of information accessible via interviews is vast, valuable, and compatible with
recent theorizations of culture drawing on dual-process and multiple memory system
models of cognition. (Pugh 2013; Lamont and Swidler 2014; McDonnell 2014; Wuthnow
2011; Lizardo 2017; Vaisey 2014; Lizardo et al. 2016). Consistent with these theoretical
developments, cultural analysts have used interviews to study things like schemas (Wood
et al. 2018; Homan, Valentino, and Weed 2017), cultural models (Ignatow 2006; Holland
1987), metaphors (Ignatow 2009; Robinson 2008), emotions (Pugh 2013), and so on.
Meanwhile, quantitative researchers have used forced-choice survey questions (Vaisey
2009; Hoffmann 2014; Miles 2015; Vaisey and Lizardo 2010) and implicit association tests
(Shepherd and Marshall 2018; Miles, Charron-Chénier, and Schleifer 2019) to identify
implicit values as direct measures of dispositions toward particular actions. This de facto
division of labor, where interviewers study reasoning and processes of meaning-making on
one hand, and quantitative researchers study internal dispositions associated with behavior
on the other, has been largely unquestioned, and has received some explicit support
(Wuthnow 2011; Lamont and Swidler 2014)6.
6 “For our part, we are less interested in the intra-individual than in the dynamic production and
consolidation of meaning… In this context, focusing on the simple question of whether attitudes predict
behavior is quite different than focusing on whether, how, and how much action is empowered by
vocabularies, symbolic boundaries, cultural scripts and repertoires” (Lamont and Swidler 2014, 156).
44
In this paper I argue that interview data can offer more to the study of culture and
behavior than cultural analysts have recognized. Under certain conditions, and when
analyzed in a particular way, interview data can provide traces of experience that may be
associated with behavior. In making this claim I develop a new approach to the analysis of
interview data that draws on recent developments in psycholinguistics and automated text
analysis. Using linked survey and interview data from two waves of the National Survey
of Youth and Religion, I find that the concreteness with which adolescents discuss religious
topics is associated with their religious attendance when they are young adults. Drawing
on the construal-level theory of psychological distance (Trope and Liberman 2010), I
propose that language concreteness in talk about religion is associated with future religious
behavior because adolescents for whom religion is a psychologically proximate entity in
their lives tend to discuss it with more concrete details, and this psychological proximity is
connected to experience in such a way that it is associated with past and future religious
practice. More generally, the findings suggest that in certain cases, interview methods may
provide access to more forms of personal culture than previously believed, including forms
associated with behavior.
I contribute to ongoing theoretical conversations about what culture consists of and
how it can be measured (Lizardo 2017; Mohr and Ghaziani 2014) by introducing a new
form of nondeclarative culture and a way to measure it. Drawing on recent work from the
cognitive sciences, I argue that construal-level, or level of concreteness with which a
person represents a particular object, can be considered a form of nondeclarative culture,
insofar as persons come to construe certain objects in similarly abstract or concrete ways,
and without conscious awareness. Drawing on the insight that textual analysis can afford
45
access to inner subjective states (Ignatow 2016; Quinn 2005; Geeraerts 2006), and
following recent applications by psycholinguists (Snefjella and Kuperman 2015), I
measure construal-level in discussions about religious topics using a dictionary of words
scored according to their concreteness by thousands of participants (Brysbaert, Warriner,
and Kuperman 2014).
I also develop a novel approach to integrating interview and survey responses.
Regarding mixed interview and survey approaches, Wuthnow (2011:7) observes, “One of
the common reasons for conducting qualitative interviews about religion is to supplement
results from surveys.” Interview data make results more interesting for the reader and “can
reinforce readers’ confidence that survey questions have been interpreted properly by
interviewees themselves and by the investigator.” Interviews, Wuthnow argues, provide a
“deeper understanding” to “superficial” survey responses. Most importantly, this approach
takes advantage of the qualitative strength of interviews without submitting them to
quantification (Small 2008).7 I agree with Wuthnow (2011) and Small (2008) about the
worth of interviews on their own merits and in conjunction with surveys, but I argue that
interviews can do more when combined with surveys than heretofore recognized. In
addition to providing a valuable context for interpretation and validation of survey
responses, interview data can be quantified in a way that contributes to the quantitative
analysis of survey data. Formal linguistic analysis can access internal subjective states
(Ignatow 2016) that can be used as valuable predictors in standard regression models. In
making this claim I am not saying that other uses of interview data in conjunction with
7 Wuthnow (2011) also argues talk is worth analyzing on its own merits. I share this view, though
the current paper focuses on the use of interviews with survey responses.
46
surveys are inadequate; rather, I am introducing an additional way of combining interview
and survey responses.
Measuring and explaining changes in religious participation is one of the central
questions in the sociology of religion, and remains contested to this day (Voas and Chaves
2016; Hout and Fischer 2014; Smith and Snell 2009; Schnabel and Bock 2017). Although
this paper does not make any claims about whether/how religion is declining in the United
States, it nonetheless contributes to this literature by developing a way to measure a
person’s experience of religion in a way that does not rely on self-reports--a perduring
limitation in this line of research. While interview responses clearly prompt self-
reflections, the concreteness with which a person describes their religious beliefs and
practices is automatic and unconscious and hence a potentially useful way of avoiding
issues of social desirability bias in the measurement of religion (Presser and Stinson 1998).
In the next section I review relevant debates about the use of interview data in
cultural analysis. I then introduce the concept of construal-level as a form of nondeclarative
culture, and make the case for why construal-level might be associated with behavior (via
psychological distance). Following this theoretical discussion, I discuss my data and
methods, present my results, and conclude with a discussion about implications for cultural
analysis.
3.3 Theoretical Background
3.3.1 Interviewing and Two Theories of Culture and Action
In cultural sociology there are two main approaches to explaining culture and
action: “toolkit theory” developed by Swidler (2013, 1986), and “strong practice theory”
47
(Lizardo and Strand 2010: 215) which draws heavily on Bourdieu (1990) and dual-process
models of cognition (Lizardo et al. 2016; Vaisey 2009). Toolkit theory, which developed
as a reaction to Parsonsian models of internalized normative values, emphasizes the
incoherence of cultural internalization and attempts to explain patterns of action in terms
of external public codes--what Swidler calls an “outside-in” model of culture and action.
By contrast, strong practice theory restores the idea of durable psychological modification
as an important motivator for action, but argues that this occurs at the level of “practical”
rather than “discursive” consciousness, thus preserving the toolkit’s rejection of Parsonsian
normatively-held declarative values as drivers of action (Strand and Lizardo 2015: 213).
Though these two explanatory approaches to culture and action do not necessarily require
particular methods, their proponents have tended to gravitate toward contrasting methods,
with proponents of the former tending to use qualitative methods, especially interviews,
and proponents of the latter gravitating toward quantitative methods, including forced-
choice surveys, experiments, and implicit association tests.
Part of the attraction of quantitative methods for strong practice theory is that they
offer a relatively direct way of measuring a person’s internal disposition toward particular
actions (Vaisey 2009; Hoffmann 2014; Miles 2015). Though this approach has been
criticized for often ignoring the question of where dispositions come from (Pugh 2013: 46),
measuring internal dispositions is no small feat and is an important part of a fully-
developed explanation of action. Even situational accounts of action rely an understanding
of the “character” of the actors (Martin 2015: 238) to make sense of why situations evoke
the responses they do. Furthermore, there is no a priori reason why the origins of internal
48
dispositions cannot be traced to specific embodied practices (Wacquant 2004; Pálsson
1994; Cornelissen 2016; Bourdieu 1984; Leschziner 2015).
It has been said that directly measuring internal dispositions is best suited for
experimental or survey methods rather than interviews (Vaisey 2009), and interviewers for
their part have supported this claim by declaring that their interests lie elsewhere (such as
more toolkit-centered approaches focused on contexts of meaning-making) (Lamont and
Swidler 2014:156). But is it truly the case that interviews cannot be used to measure
internal dispositions? The question is worth considering more carefully, because
interviews, if proven capable of measuring internal dispositions, could enhance practice-
theoretic explanations of culture and action.
If interviews could directly measure internal dispositions in ways similar to forced-
choice survey questions, they would offer a more inductive approach to measurement than
what is currently available via surveys. In various places, Vaisey has described the ability
of carefully-worded forced-choice survey questions to measure internal dispositions using
the metaphor of a police line-up. The idea is that respondents rely on their gut instincts--
their practical consciousness--to choose the best answer, and this intuition is more closely
related to action than people’s answers to open-ended interview questions. However, as
Vaisey has acknowledged, this deductive method is only accurate insofar as the right
combination of possible responses are included. If interview methods could measure
internal dispositions, by, for instance, measuring the formal properties of their responses,
then interviews would offer an additional, more inductive measurement for strong practice-
theoretic approaches by not requiring fixed choices up front.
49
Additionally, if interviews were capable of directly measuring internal dispositions,
their reputation as a “workhorse method” (Lamont and Swidler 2014) would be even more
deserving, because it would mean that interviews could effectively support toolkit and
strong practice-theoretic approaches to cultural explanation, perhaps even using the same
data. Relatedly, interview data, most fundamentally, is text data, of which 21st century
researchers are in no short supply. If interviews, as a special case of text data, could be
used to directly measure internal dispositions, then it opens the possibility that much more
data is potentially useful for studying motivation and action than has currently been
recognized.
The potential payoffs for cultural analysis make revisiting claims about the
limitations of interviewing worthwhile, but delivering on this promise requires overcoming
several challenges.
3.3.2 Challenges Facing Interview-based Measures of Experience
First, there is the well-known problem that if we ask someone why they behave a
certain way, we are likely to get “accounts” or “justifications” of behavior rather than data
on a causal relation (Martin 2017: 80; Swidler 2013). While not necessarily incorrect, there
is no reason to assume that these reflections match how we actually experience the world
(Martin 2011: 188; Margolis 1987: 76; Strand and Lizardo 2015: 213). However, this
criticism pertains only to a narrow understanding of textual analysis. Cataloging people’s
justifications and correlating it with behavior is a “rather traditional and fixed
understanding of interviewing” (Lamont and Swidler 2014: 157), and unless specifically
studying accounts (DeGloma 2014), interviewers do not need to make such naive and direct
50
questioning the focus of their study. There are many kinds of questions that could be asked
in an interview, so this critique by itself does not warrant abandoning our inquiry.
More seriously, Vaisey (2009, 1688) claims that “[t]he unstructured or
semistructured interview puts us in direct contact with discursive consciousness but gives
us little leverage on unconscious cognitive processes.” This claim, if true, would justify
skepticism of interviews for measuring internal dispositions because the theory requires
access to these kinds of processes. However, the claim is false. As Vaisey (2014: 154)
himself later admits (in response to Pugh’s (2013) critique ) “[i]t was an oversimplification
to claim that interviews ‘give us little leverage on unconscious cognitive processes.’”
Indeed, a core finding of cognitive linguistics is that linguistic production is structured via
unconscious cognitive processes (Geeraerts 2006; Gibbs and Gibbs 2017). When people
talk or type, they make use of nondeclarative associations outside conscious awareness.
Depending on a person’s previous experiences, a discursive situation will evoke certain
schemas, emotions, and representations which may be more or less shared (Wood et al.
2018).
From the perspective of cognitive linguistics, the interview can be seen as “a set of
tasks, strung together so that it feels like a conversation” Martin (2017: 74)--tasks which
activate and are structured by unconscious cognitive processes. Following this approach,
the goal is not necessarily to gather answers to “why” questions, but to ask questions which
activate nondeclarative pathways of interest to the researcher (Quinn 2005; McDonnell
2014). Recent sociological work demonstrates that interview and text methods can and do
allow access to certain nondeclarative forms of culture, including emotions (Pugh 2013;
Ecklund et al. 2017), schemas (Homan, Valentino, and Weed 2017; Farrell 2013; Ignatow
51
2009), and associations (Moore 2017; McDonnell 2014). The takeaway is that although
interview and text methods are most intuitively suited for studying declarative elements of
culture (such as propositional beliefs, values, and semantic knowledge), this does not entail
that they cannot afford access to certain elements of nondeclarative culture as well (Lizardo
2017).
The final challenge, however, is determining whether and to what extent the
cultural elements accessible via talk and text are actually associated with behavior. It is
possible, for instance, that the nondeclarative elements of culture that structure verbal
communication and reasoning are not related to the nondeclarative elements of culture that
are associated with behavior. For example, Americans make many different decisions
about marriage, even though their reasoning about marriage is structured via a shared
cultural model (Quinn 1987). Alternatively, the nondeclarative elements associated with
talk may be related to action, but only indirectly. For example, Ignatow (2009) finds that
the length of time a person participates in an online support group is associated with the
degree to which the structure of their individual posts match the structure of the group’s
overall discourse. Studies such as this demonstrate the value of text analysis for the study
of culture behavior, but do not provide a direct measure of internal disposition like Vaisey
(2009) introduced.
Forced-choice survey questions succeed at measuring internal dispositions insofar
as the question about implicit values activates the same or related nondeclarative
associations as the question about the behavior. As Vaisey (2009: 1688) notes, “Because
choosing from a fixed list of responses is akin to solving a practical problem (‘Which one
do I like?’), fixed-response survey questions may draw disproportionately on practical
52
consciousness, which has to make (as opposed to discuss) many such decisions each day”
(italics in original). If interviews are to similarly succeed, they need to do similarly. In the
next section I argue that interviews may be able to do this in certain circumstances, through
the analysis of construals.
3.3.3 Construal-Level Theory
When persons mentally represent things, they do so at a certain level of construal.
Higher-level construals are abstract and schematic, lacking incidental details that could
change without significantly changing the meaning of the object. Low-level construals, by
contrast, are concrete and contextualized, containing incidental features of an object (Trope
and Liberman 2010). For example, if we were observing a parent with their infant, a high-
level construal of the parent’s activity might be, “they are caring for their child.”
Alternatively, a low-level construal of the same activity might be, “replacing a dirty diaper
with a clean one.” In language, construal-level is manifest through more or less concrete
or abstract descriptions.
Construal operations are general cognitive capacities like perception or
categorization, but they are cultural in the sense that individuals may have similar or
dissimilar construals for particular topics or domains depending on their experiences. In
this way, individuals’ construal-levels for particular domains may be considered a form of
non-declarative culture (Lizardo 2017). For example, a Muslim and a Christian who both
discuss their religious practices using very concrete language may be said to share this
aspect of non-declarative culture, despite having different substantive beliefs and religious
habits.
53
Construal-level is associated with psychological distance, in that people tend to
construe things that are psychologically distant from them at higher levels of construal
(Trope and Liberman 2010). This association has been established experimentally for
different forms of psychological distance, including spatial distance (Fujita et al. 2006;
Henderson et al. 2006), temporal distance (how far into the past or future) (Trope and
Liberman 2003; Liberman and Trope 1998), social distance (familiarity with the social
target) (Stephan, Liberman, and Trope 2010; Nan 2007), and probability (how likely
something is to occur) (Todorov, Goren, and Trope 2007; Wakslak et al. 2006). The
relationship between psychological distance and construal-level for all of these dimensions
has also been observed in natural language use from a large sample of Twitter posts
(Snefjella and Kuperman 2015).
Building upon these findings, I make two theoretical propositions: First, in what
might be considered an extension of the social dimension of psychological distance, I
propose that individuals vary in their psychological proximity vis-a-vis particular cultural
objects, such that when asked about these objects, they respond at higher or lower levels
of construal depending on their psychological proximity to the object. Applied to my
empirical case presented below, I hypothesize that for certain adolescents, religion is a
more psychologically proximate object which is manifest via more concrete language in
their answers to interviewers’ questions about religion. Second, I propose that
psychological distance is connected to the experience of cultural objects in such a way that
it is associated with past and future patterns of behavior. Applied to my analysis of religion,
I hypothesize that adolescents for whom religion is more psychologically proximate,
54
manifested by more concrete answers to questions about religion, will be more likely to
frequently attend religious services when they are young adults than their peers.
Language concreteness may be associated with behavior under certain conditions.
Perhaps the most fundamental of these conditions is the type of interview question asked
and congruence with the measure of behavior. Knowledgeable readers may wonder why I
did not use the same interview and survey questions as Vaisey (2009), given that I am using
the same dataset and have similar theoretical aims. One reason is that the questions used in
Vaisey (2009) elicit accounts of behavior rather than descriptions of cultural objects. Based
on current research on psychological distance, an association between the way a person
discusses religion and their religious attendance is theoretically plausible, but such an
association built from accounts of behavior is theoretically unsupported.
3.4 Data and Measurement
My data consist of 200 interview transcripts from the first wave of the National
Survey of Youth and Religion and survey data from the first and third waves. The first
wave of data was collected between 2002 and 2003 among adolescents ages 13-17, and the
third wave was collected between 2007 and 2008, when the participants were ages 18-23.
The research team selected participants with the goal of recruiting teens that “encompass a
wide diversity [including] both religious and non-religious teens from many races and
economic groups.”8 The interview transcripts were linked to the survey data allowing a
two-pronged analytic strategy.
8“NSYR Completes Personal Interviews Data Collection”
https://youthandreligion.nd.edu/announcements/nsyr-completes-personal-interviews-data-
collection/ accessed 7 December 2018.
55
My analysis of the interview data focused on a section of the interviews containing
a set of questions about religion and religious practices. Interviewers asked adolescents
whether they considered themselves religious or spiritual, and if so how; whether they
believed in a higher being, and if so, how they imagined Him/Her/It; whether they have
regular religious habits; and so on.9 Though each interviewer was given the same guide,
there was some variation in the order of questions and whether certain questions were
included or omitted.
I use an automated dictionary-based method (Stone, Dunphy, and Smith 1966) to
measure how concretely adolescents answered these questions. Through a process
analogous to sentiment analysis, (Perrin and Vaisey 2008; Young and Soroka 2012;
Eshbaugh-Soha 2010) I used a dictionary of English words annotated with concreteness
scores indicating how concrete or abstract a given word is to calculate the overall
concreteness of interview responses.
The dictionary used for the concreteness analysis of the interview text was
developed by Brysbaert et al. (2014) and features 37,058 English words and 2,896 two-
word expressions (such as “church service” or “grizzly bear”). Each word and two-word
expression in the dictionary has a concreteness score obtained by averaging all the raters’
scores for each word. Over 4,000 online participants, recruited through Amazon’s
Mechanical Turk, collectively rated the concreteness of 60,099 English words and 2, 940
two-word expressions. Most words were rated by at least 25 different participants, and
9 The complete interview guide can be found at
https://youthandreligion.nd.edu/assets/109007/personalivmethods.pdf
56
were included in the dictionary if at least 85% of raters recognized the word. Participants
were instructed to rate each word or two-word phrase on scale from 1-5, using the following
definition of concreteness10:
Some words refer to things or actions in reality, which you can experience
directly through one of the five senses. We call these words concrete words.
Other words refer to meanings that cannot be experienced directly but which
we know because the meanings can be defined by other words. These are
abstract words. Still other words fall in-between the two extremes, because
we can experience them to some extent and in addition we rely on language
to understand them. We want you to indicate how concrete the meaning of
each word is for you by using a 5-point rating scale going from abstract to
concrete.
I took several steps to maximize the use of the concreteness dictionary and
minimize bias created by missing words. First, I performed standard preprocessing on the
text data, including: sending words to lowercase; removing whitespace and punctuation;
removing common “stopwords”; and breaking apart contractions. I also transformed plural
words to their singular forms using the “Pluralize” R Package because the concreteness
dictionary omits plurals. At this point, some common words were still omitted from the
analysis because they were in a form or tense not included in the dictionary. For example,
the dictionary has an entry for “hear” but not “heard,” and “nope” but not “nah.” To
minimize bias created by these omissions, I identified words that appeared more than ten
times in discussions of religious practices that did not occur in the concreteness dictionary
and replaced them with an analogous word that did have an entry in the dictionary, if such
a word existed.
10 For the complete instructions, which includes more specific participant guidelines for rating
words, see (Brysbaert, Warriner, and Kuperman 2014, 906)
57
I assigned each document a polarity score ranging from -1 to 1 by binarizing the
concreteness dictionary at the median and dividing the difference of concrete and abstract
words in a given document by the sum of concrete and abstract words in the document (i.e.
the total number of words scored by the dictionary). In my sample, concreteness scores for
responses about religion ranged from -0.6 to 0.1.
Once I had concreteness scores for each of my adolescents, I used these as an
independent variable in a lagged ordered logit model. My dependent variable for this model
was a trichotomized ordinal variable measuring frequency of attendance of religious
services during the time of the third wave (about five years after the first wave), ranging
from “never,” to “low,” (less than once a month) to “high.” This same measure was used
for the wave one lag. I also included several wave-one covariates in the model that I
believed might be associated with both concreteness and future religious attendance. These
included: religious tradition at the time of the first wave, with Evangelical Protestant as the
reference category, parental education, a survey question in which respondents were asked
how important religion was to their daily life, and the number of their friends, out of five,
who had similar religious beliefs. Finally, I also controlled for the total number of words
respondents used when answering questions about religion.
To give an example of how the analysis works, consider the excerpts in Figure 3.1.
Both of these excerpts include responses to a question about the nature of God, but the
responses vary significantly in concreteness. Words marked as abstract by the algorithm
are colored blue, and words marked concrete are colored orange. Uncolored words are stop
words omitted from the analysis.
58
Figure 3.1: Concreteness in Discussions about God
59
3.5 Results
A coefficient plot showing the relative effect sizes is shown in Figure 3.2. The
standardized coefficients are represented as points on the plot, and the horizontal lines
extending from the points represent 95% confidence intervals. Variables that are
statistically significant at the p<.05 level have confidence intervals that do not overlap the
dashed vertical line at 0. As expected, the largest effect was for lagged religious attendance
variable from the first wave. After that, however, the effect for concreteness is statistically
indistinguishable from other variables which are known to be highly associated with
change in religious attendance--religious identity, religious friendship networks, parental
education, and subjective importance of religion. The results suggest that adolescents who
discussed religion more concretely were more likely to more frequently attend religious
services as young adults than adolescents who discussed religious less concretely.
60
Figure 3.2: Coefficient Plots
61
To facilitate the interpretation of the regression results I generated predicted
probabilities using the sample data, which are shown in Figure 3.3. The figure shows
predicted probabilities for three possible starting conditions: adolescents who never
attended religious services at the time of the first wave, adolescents who attended
infrequently, and adolescents who attended frequently. For each of these starting
conditions, the different colored lines indicate how the probabilities of ending up in the
“never,” “low,” or “high” attending categories in the third wave change based on one’s
level of concreteness during wave 1 discussions. In all three cases, the probability of being
a wave three high-attender increases as concreteness increases, and the probability of being
a wave three low-attender increases as concreteness decreases. However, the strong effect
of one’s starting point can also clearly be seen--adolescents who discuss religion very
concretely but never attend religious services still have relatively low probabilities of being
high-attenders at the time of the third wave (though they do exist in the sample).
62
Figure 3.3: Predicted Probabilities
63
3.6 Discussion and Conclusion
In this paper I set out to determine whether interview data can be used to measure
internal dispositions associated with behavior in a way that is similar to what has been
proposed by quantitative survey researchers using forced-choice questions to measure
implicit values. Previous to this study, quantitative and qualitative cultural analysts alike
had been dubious that such a measure could be generated using interview data. However,
drawing on recent insights from psycholinguistics, I proposed that under certain conditions,
interview data could produce such an indicator via language concreteness. I argued that
because language concreteness is associated with psychological distance, the way people
discuss certain cultural objects might also be related to their disposition towards those
objects--psychological proximity being an indicator of positive disposition, and
psychological distance being an indicator of negative disposition.
Using a novel approach to mixed-method, survey and interview data, I found that
adolescents who discussed religion more concretely were more likely to more frequently
attend religious services when they were young adults. The results provide initial support
to the idea that interviews are capable of accessing experience associated with behavior
more directly than previously believed.
Though I focus here on concreteness, it should be noted that concreteness is only
one of many ways to analyze the formal properties of text. I have argued that under certain
conditions, concreteness may be used as an indicator of internal dispositions toward certain
behaviors, but future analyses may find that there are additional structural properties of text
that can be extracted and similarly used. In this way, the analysis signifies a new field for
cultural analysis: cultural psycholinguistics. In addition to being theoretically promising,
64
this development adds ever more value to interview and text data, as new methods are
developed to measure the psycholinguistic properties of text.
The above analysis is an important first step but is limited in ways that need to be
recognized to enhance future research. In particular, future analyses of the link between
talk and action will benefit from more controlled interviewing procedures. While each of
the respondents included in the above analyses answered questions from the same set of
interview questions, interviewers varied somewhat in phrasing, question order, and what
questions they chose to include or omit. Because of this, the above analysis takes a coarse-
grained triangulatory approach and is unable to adequately answer the question of whether
the association between concreteness in talk about religious practices and future religious
participation is stronger for certain topics more than others. In research where one of the
goals of collecting interview data is to identify the cognitive structure of responses in
response to particular stimuli (Wood et al. 2018), it will be advantageous to exercise more
control over the interviewing process, essentially treating the interview a bit more like an
experiment and less like an unstructured conversation. Doing so should reduce
measurement error, afford greater specification of effects, and improve confidence in the
results.
Critics may accuse me of leaving “concreteness” too abstract to be useful.
However, it is important to recognize that the limitations of the current data are not inherent
to the method. If researchers take care to ask each of their respondents the same questions
about a variety of specific topics, concreteness can be easily connected to specific those
topics, and hence, those domains of experience. The extent to which concreteness is a
“black box” will depend largely on the quality of the data collection.
65
In fact, measuring language concreteness opens many possible avenues for future
research. For one, measuring concreteness in interviews may be a useful way of measuring
the extent to which religious practice has “stuck” with a person in a way that avoids issues
of social desirability bias (Presser and Stinson 1998). Additionally, the link between
concreteness in talk and experience could be investigated further by pairing text data with
participant observation or experiments. Researchers could present participants with a
stimulus or observe them in some setting (say, a religious service), and record their
descriptions. Or, researchers could compare differences in the way men and women
describe different domains of life or religious experience.
My sample included 200 respondents, which is fairly large by interview standards
but relatively small for quantitative analyses. Future analyses might increase reliability by
creatively incorporating open-end questions into survey instruments. This could increase
the number of cases in a sample, though it may do so at the richness of the qualitative data.
Whether the amount of text data that could be collected using such a method can produce
comparable results in concreteness analyses is yet to be seen.
One remaining question is how my measure generated via the concreteness of
interview responses performs in a regression model compared to a similar measure derived
from a forced-choice survey question. Does one explain more variance than the other? Do
they account for different portions of the variance? These are important questions that the
current data are unable to answer. Having said that, it is not my intention to displace survey-
based measures of internal dispositions with another. Instead, my goal has been to offer an
additional method for studying religion and culture and demonstrate a way to get even
more mileage with interview data.
66
CHAPTER 4:
A DYNAMIC THEORY OF CULTURE IN REASONING: THE CASE OF AGE-
APPROPRIATENESS
4.1 Abstract
Schema analysis in sociology has mainly focused on identifying group-based
differences in schema use but has ignored the question of how schema use and activation
varies across particular contexts. In this paper, I analyze schema activation in the reasoning
of 69 parents who were instructed to explain when it was appropriate for a person to do
each of a list of 14 behaviors. I find that the task of reasoning about age-appropriateness
activates one of three basic schematic models, providing the foundational structure for
reasoning in this domain. These models are elaborated according to the qualities of the
behavior in question. After describing the basic models and their elaborations, I identify
sources of variation and conclude with a discussion of broader implications for cultural
sociology.
4.2 Introduction
When people reason about things, their reasoning is facilitated by schemas (Cheng
and Holyoak 1985; Rumelhart 2017; Thibodeau, Hendricks, and Boroditsky 2017). A
schema is a “flexible memory structure, automatically acquired and updated from patterned
activity, composed of multi-modal neural associations” (Wood et al. 2018a, 246). A
schema lacks unit detail, applying across different situations. In this way, schemas are like
“slots” into which various components fit (Anderson 1978). For example, though people
buy many different kinds of things in many different kinds of settings from many different
people, all these situations are understood similarly because they activate the same BUY11
schema (Lewis and Durrant 2011; Rumelhart 2017).
Since Sewell (1992) and DiMaggio’s (1997) landmark papers, schema analysis has
been an important part of cultural analysis in sociology12. Schemas are of interest to
sociologists because, in addition to facilitating comprehension and reasoning, they are
often shared among people with similar experiences (Shore 1996). Recent sociological
work has investigated group-based differences in schematic understandings of poverty
(Steensland 2006; Homan, Valentino, and Weed 2017), medical technology (Ecklund et al.
2017), the environment (Farrell 2013; Gabriel Ignatow 2006), and controversies
surrounding science, law, and religion (Edgell and Hull 2017), among many others.
Additionally, a body of experimental work shows manipulating schema activation can
11 Following conventional usage, I designate schemas with small uppercase letters.
12 There are two main strands of schema analysis within sociology: One emerged from Sewell,
and the other from cognitive psychology. I take the latter approach. For more on this difference, see (Wood
et al. 2018).
67
68
influence reasoning and problem-solving outcomes (Homan, Valentino, and Weed 2017;
Thibodeau, Hendricks, and Boroditsky 2017).
Sociologists have identified group-based differences in schematic activation
suggesting that people with different backgrounds think about the same concepts in
different ways. However, little is known about why certain schemas are activated in a given
setting in the first place. Recently, Edgell and Hull (2017, 318) observed that schematic
structure varies across topics, suggesting that a purely group-based explanation of variation
in schematic activation is inadequate, but systematic work on understanding this domain-
driven variation has yet to be done.
Missing is a dynamic theory of culture in reasoning that explains how schema
activation varies across situations. Is schematic activation in reasoning consistent, such that
reasoning about a moral category is structurally the same regardless of the thing being
evaluated? Or alternatively, is schematic activation flexible, varying with the thing being
evaluated? Research on this front has been limited because the typical approach to schema
analysis is to focus on group-based differences in reasoning while ignoring topical
variation. In this way, the explanation of reasoning has mirrored the explanation of action
more generally, where individuals are assumed to internalize certain rules that constrain
action based on group membership (Luft 2015; Martin 2015).
In this paper, I develop a dynamic theory of culture in reasoning. The empirical
setting is parental reasoning about age-appropriateness. Rather than focusing on a specific
topic, such as reasoning about the age-appropriateness of having sex, I analyze the way
parents reason about and evaluate age-appropriateness across a series of topics. I
differentiate between the evaluative category and the object of evaluation. I show that there
69
are three basic schematic models associated with the evaluative category of age-
appropriateness. These models are schematically elaborated depending on the qualities of
the behavior being evaluated. Together, these two schematic elements constitute a
schematically-elaborated cultural model (see Figure 4.1). In schema analyses only
considering reasoning about specific objects in single settings, only the elaborated cultural
model is visible. I tease out basic models from schematic elaborations via a comparative
approach.
70
Figure 4.1: Key Concepts in a Dynamic Theory of Culture in
Reasoning
71
4.3 Theoretical Background
4.3.1 Public vs Personal Culture
The cultural analysis addressed here focuses on identifying relevant elements of
personal culture, or the culture internalized by people (Strauss and Quinn 1997). Personal
culture is distinct from public culture, or culture that is “externalized” via some material
medium outside of individual bodies (e.g. text, sounds, physical objects, etc.) (Wood et al.
2018). Personal culture can be declarative or nondeclarative. Declarative culture consists
of semantic knowledge of things like facts, explicit beliefs, and public codes, and is
described heuristically as “knowing-that.” Nondeclarative culture, on the other hand,
consists of procedural knowledge, or “knowing-how” (Lizardo 2017). When individuals
reason, they draw on bits of declarative knowledge--facts, beliefs, and autobiographical
narratives--which are strung together to support a judgment. However, this declarative
knowledge is not chosen at random but is organized via schemas. Because people can use
schemas without necessarily being able to have verbal access to their structure and
organization, schemas can be considered a form of non-declarative culture.
4.3.2 Judgment vs Reasoning
The task of evaluative reasoning, whether about age-appropriateness or some other
category, is better understood as two distinct tasks corresponding to different types of
cognition (Lizardo et al. 2016). Judgment is typically a result of automatic, “type 1”
cognitive processing. If someone was asked when it is appropriate to get a smartphone, the
judgment would be the specific time marker at which point it is appropriate to have the
device. This is the “gut instinct” discussed by Vaisey (2009).
72
Reasoning, on the other hand, happens through the slower, “type 2” cognitive
processing. In the smartphone case, reasoning would be the process of producing a
response that supports the judgment. There is no necessary relation between judgment and
reasoning. People with different judgments may nonetheless reason about their evaluations
in very similar ways. Alternatively, people with similar judgments may reason in different
ways. The extent to which reasoning and judgment correspond is ultimately an empirical
question.
Although people’s reasoning can be an ad hoc justification of a pre-existing gut
judgment, people can also take their predispositions as provisional and delay final
judgment until after a period of reasoning (Cunningham and Zelazo 2007), as seen in
cognitive experiments used to differentiate type 1 and type 2 processes where a puzzle has
an intuitive yet incorrect solution and a participant must engage in type 2 “deliberative”
thinking to reach the correct solution (Smith and DeCoster 2000). This paper is concerned
with reasoning as opposed to judgment, and I take an agnostic stance regarding the
empirical question of whether judgments precede or follow reasoning.
4.3.3 Schemas and Reasoning
Personal culture facilitates reasoning schematically through analogical mapping
(Gentner and Smith 2012; Holyoak, Holyoak, and Thagard 1996). Analogical mapping
consists of using familiar sources to understand less familiar targets. For example, the
JOURNEY schema may serve as source for understanding many different targets, such as
romantic relationships, life, and achieving goals (Lakoff and Johnson 1999; Johnson 1987).
In this way, schemas are the “building blocks” of culture in reasoning. If there is any
consistency in the way people reason, it is achieved by analogical mapping (Bar 2007).
73
A type of schemas called “image schemas” are especially useful for the analysis of
evaluative reasoning. Image schemas are simple perceptual gestalt structures. (Brad Shore
1996; Kövecses 1999; Johnson 1987; Lakoff 2008). Persons develop initial image
schemas, including schemas such as THING, PATH, CONTAINER, and OCCLUSION (Mandler
and Cánovas 2014), via repeated, early, embodied experience, mostly via the observation
of people, animals, and things in motion as well as sensorimotor interaction with objects
in space (Mandler 1992; Mandler and Cánovas 2014). Image schemas are maximally-
schematic, meaning they are sources for the widest range of targets, either in isolation or
as parts of compound schemas. As such, image schemas are well-suited for the comparative
analysis of evaluative reasoning. Image schemas provide a culturally specified yet non-
arbitrary coding scheme allowing the analyst to trace similarities within and between
individuals.
Image schemas come in different levels of complexity. Compound schemas, which
are schemas built up from multiple lower-level schemas, can be “decomposed” into image-
schematic building blocks (Kimmel 2005). For example, distinct concepts with their own
schematic structures, such as building, box, and set, are all CONTAINERS. Similarly,
metaphors such as “over the moon,” “flying high,” or “feeling up” all instantiate the UP-
DOWN image schema to convey a sense of joviality. Without awareness of the shared
image-schematic foundations between different source domains, analysts risk concluding
that different instances of reasoning are less coherent than what might actually be the case.
On the flip side, if the analyst has identified distinct image-schemas in different instances
of reasoning, they can be more confident that these actually resulted from distinct
cultural/cognitive processes.
74
Basic schematic models and schematic elaborations both make use of the same
image-schematic building blocks. For this reason, identifying schemas without comparing
across different cases tells us little about what anchors the observed structure. The
analytical task is thus to first identify the image schemas manifest in a text, and then trace
them to their anchoring elements. For the sake of clarity, I refer to image schemas and
higher-level schemas simply as “schemas.”
4.3.4 Age-Appropriateness
Age norms, or norms of age-appropriateness vary historically and across societies
(Neugarten, Moore, and Lowe 1965). With age norms, an activity perceived as
unproblematic for a healthy adult might evoke concern when it is being done by an
adolescent, and an action perceived as good and fitting for young adults might evoke
disgust when practiced by the elderly. Similarly, behaviors and life events are commonly
organized into idealized timelines, where the appropriateness of a given action or practice
depends on what actions have preceded it (for example, the timing of marriage in relation
to other markers such as education, employment, and childbearing (Kefalas et al. 2011)).
Adherence to these idealized timelines can carry social and emotional consequences (Frye
and Trinitapoli 2015).
In contemporary societies, there is a professional preoccupation with child
development and age-appropriateness. Training for K-12 instructors typically entails
courses on child development and instruction on how to choose “developmentally
appropriate” activities. Similarly, the U.S. Consumer Product Safety Commission has a
300-page document entitled “Guidelines Determining Age Appropriateness of Toys”
which provides guidelines based on seven categories of play, two to five subcategories of
75
toys, ten age groups, and 14 toy characteristics. Despite recent concern over age-
appropriateness, little research has examined the way everyday parents think and reason
about age-appropriateness. There is reason to believe, however, that parental
understandings of age-appropriateness are consequential in everyday life: disagreements
over acceptable forms of social interaction among adolescents are often the flashpoint
between parents and children (Montemayor 1983), and the behavior of children can carry
social implications for parents who take their children’s behavior as an extension of their
own self-presentation (Collett 2005).
In what follows, I analyze the schematic structure of evaluative reasoning about
age-appropriateness. I find three basic schematic models associated with the evaluative
category of age-appropriateness, which I call Behavior Exerts a Force on the Child,
Behavior Opens New Paths, and Behavior Is Dirty. I discuss each of these basic schematic
models and identify their associated schematic elaborations. These schematic elaborations
are associated with the object of evaluation, which here refers to the behavior being
evaluated.
4.4 Data and Analytic Approach
Before one can analyze culture in reasoning, one needs to “draw it out” (McDonnell
2014). A key feature of nondeclarative culture is that it cannot be accessed directly (Lizardo
2017). The best approach is for the analyst to give respondents a “productive” task that
activates nondeclarative culture and produces a tangible, analyzable product (McDonnell
2014; Martin 2017; Quinn 2005). For the present study, interviewers asked parents when
76
it was appropriate to engage in a series of practices. This request activates the
nondeclarative culture responsible for structuring responses.
The data come from interviews conducted from 2014-2015 as part of the
Intergenerational Religious Transmission Project (IRTP)13. Researchers interviewed
parents of different religious backgrounds from different regions of the United States on
topics relating to parenting and the transmission of religion. For this analysis, I used 69
interviews that had a module on age-appropriateness not included in other interviews.
Interviewers told participants they would be asked at what age it was appropriate to begin
certain practices. Interviewers then read a list of practices, one a time, allowing the parent
to respond to each one. Table 2.1 lists the practices (see chapter 2). Table A.1:
Interview Sample in the Appendix lists some demographic information of the sample. The
goal of the sample was maximizing diversity.
The resulting analysis was abductive (Timmermans and Tavory 2012). I began by
coding recurring image schemas in the text and then tried to understand how they were
organized. During this process, I developed provisional explanations and went back and
forth between them and the data, and eventually settled on an explanation that fit the data.
4.5 Basic Schematic Models and Schematic Elaborations
Parents relied extensively on declarative knowledge (especially autobiographical
memory) when they reasoned about age-appropriateness, but this declarative knowledge
was consistently structured by one of three basic schematic models. Each model represents
13 Under the direction of Christian Smith and funded by the Lilly Endowment, Inc.
77
a distinct problem the behavior in question creates for age-appropriateness. In what
follows, I identify the schematic structure of these models and describe how the models
were elaborated and adapted to the immediate situation. Though analytically distinct, the
models may or may not overlap in actual discourse. I found evidence of both.
4.5.1 Behavior Exerts a Force on the Child
When reasoning about certain behaviors including trying alcohol, trying drugs,
viewing pornography, watching R-rated movies, and having sex, parental reasoning often
has the structure of a basic schematic model which I call Behavior Exerts a Force on the
Child. Figure 2.1 contains visual representations of the model’s two forms, following the
conventions developed by Talmy (1988). In both forms, the child is represented by a circle
and the force acting on the child is represented by a concave shape. A small dot in the circle
indicates a tendency toward rest. The positive sign indicates the locus of greater force. The
arrow above the concave shape indicates that the force is not always there but enters the
scene (i.e. when the child engages in the behavior). The line underneath the diagrams tells
the causal story. The dot here indicates rest and the greater-than sign indicates change.
78
Figure 4.2: Schematic Diagram of Behavior Exerts a Force on the Child.
79
The “inappropriate” form of this model (found on the left half of the figure) is the
story of what Talmy (1988) calls “onset causation.” Here, antagonist (in this case, the
behavior) enters the scene and exerts a force on the agonist (in this case, the child). Because
the antagonist has a greater force than the agonist, the result is that the agonist is moved
from a state of rest to a changed state. Consider, for example, the following excerpts:
[R-rated Movie] “This is what I told them, and you know I think it says this
in the bible, ‘Your eyes are the windows to your soul.’ Be careful what
you… even as an adult, you watch enough bad stuff, it can affect your sense
of empathy toward other people… and children especially, they get
overwhelmed with that... You see-you look at bad, harmful things for
entertainment, it’s gonna have a negative effect on you.” (Brandon, white,
two sons ages 17 and 12, daughter age 9)
[view porn] “Yeah, I mean, it’s very damaging to women. And I think it’s
damaging to the men that look at it. It skews their view… it skews their view
of sexuality and what women are and what’s normal and it can really ruin
the way they approach relationships for life… So I think there’s never a
place for that, at any age.” (Mary, white, two sons ages 17 and 12, daughter
age 9)
[view porn] “Oh, bad. Sets up bad expectations… it’s very destructive in
terms of what your future thoughts and desires are going to be” (Rachel,
white, two sons ages 16 and 10, daughter age 14)
[Try drugs] “No, again, unnecessary. That leads to destructive behavior.
(Tess, two daughters, ages 30 and 23)”
In each of these excerpts, the parents’ reasoning is structured by the basic schematic
model Behavior Exerts a Force on the Child. Each parent rejects the notion that the
behavior in question is ever appropriate, and their reasoning is centered around the onset
causation force dynamic. However, the structure of their reasoning varies from person to
person because they elaborate the basic model in different ways. In the first, Brandon
argues that viewing pornography damages men and women, skews their view, and
perdurably ruins the way they approach relationships. Similarly, Rachel argues that
80
viewing porn sets up bad expectations, and Tess argues that trying drugs leads to
destructive behavior. The consequences vary somewhat but an overall force dynamic
persists across cases.
In the “appropriate” form of this model (found on the right half of Figure 4.2), the
antagonist enters the scene but does not alter the agonist’s “steady state” of rest because
the counterforce of the agonist is greater than the force of the antagonist. Unlike the
inappropriate form of this model, in which the range of schematic elaborations is fairly
wide, I found three schematic elaborations that justify a behavior understood to generate a
force on the child: waiting until the child can HANDLE the force, LIMITING the AMOUNT, or
only allowing forms of the behavior where the harmful force is absent. I discuss each of
these below.
4.5.1.1 Having Enough Counterforce: “Handling”
Recall from Figure 4.2 that the appropriate, “steady state” is achieved when the
force of the child is greater than the force exerted by the behavior. The simplest way to
guarantee this outcome is to wait until the child has enough COUNTERFORCE to resist the
force generated by the behavior. This is commonly expressed verbally using the metaphor
of “handling.”
[R-rated movies] “I think it depends on the kid and what they’re able to
handle. Right? Our kid still gets scared at like PG movies (chuckles)
sometimes like really scary Harry Potter movies so I think it depends on the
kid.” (Elise, white, same-sex, 11-year-old daughter)
[view porn] “I think they can really handle it. I guess 15-16... if they’re
curious they can look at whatever they want, but their brain has to be a little
more developed” (Hazel, white, same-sex, three sons ages 13, 10, and 4).
81
[drink regularly] I know that kids binge drink in high school and it’s
something that we’ve discouraged… Discouraged doesn’t mean that we’ve
deterred it entirely, but I think that your brain isn’t really cooked for a long
time” (Judith, white, two daughters ages 27 and 23, son age 19).
The first two excerpts illustrate the “handling” metaphor in reasoning about age-
appropriateness. In the first, Elise implies that rated-R movies exert a force on children and
that the appropriate time to watch them is when the child can “handle” it. In the second,
Hazel says she thinks her sons can “handle” looking at porn, but she adds that their brains
need to be “a little more developed.” Judith says something similar when talking about
drinking regularly, arguing that before that can happen, your brain needs to be fully
“cooked.” The implication here is that there is a biological component to handling and that
engaging in certain behaviors too early may create lasting damage. Thus, handling is
related to age-appropriateness because persons of a certain age are developmentally more
likely to have the COUNTERFORCE necessary to handle things that younger persons do not.
4.5.1.2 Limiting the Amount
A second approach to ensuring an appropriate steady-state is to reduce the force
acting upon the child. This fits well when the behavior entails the consumption of a
substance. In these cases, the level of the force of the substance is often understood to be
related to the AMOUNT of the substance. In some cases, such as drugs and alcohol, this is
literal: more alcohol leads to more intoxication. In the case of media consumption, this is
metaphorical. In either case, in their reasoning about these behaviors, parents will say that
they are appropriate if the amount is limited:
[try alcohol] “any age. I mean, you can have a taste of it anytime” (Judith,
white, two daughters ages 27 and 23, son age 19).
82
[try alcohol] “I think that Fred’s given them a taste of beer, so probably 9-
10. Um, liquor, we don’t have it so I think that’s a little strong, but I think
I let Henry taste some wine the other day, so you know, you wanna taste it,
so I’d say 9-10” (Hazel, white, same-sex, three sons ages 13, 10, and 4).
[view porn] “I’m pretty confident that they wouldn’t really abuse it or use
it in a rampant way that would be inappropriate” (Hannah, white, single,
five sons ages 17 (twins), 12, 9, and 7).
[view porn] “If I learned that he and his buddies had seen online some stuff,
and as long as it wasn’t some crazy thing online or he wasn’t just obsessed
with it--if he just saw it or they looked--I’m probably gonna say something
to them about just, ‘let’s not get too obsessed with this,’ whatever. Yeah,
that’s out there, whatever” (Elijah, white, 17-year old son, 14-year old
daughter).
In the first excerpt, Judith explains that trying alcohol is appropriate at any time,
provided that it is just “a taste.” Hazel says something similar but adds a distinction
between beer, liquor, and wine. Hazel notes that liquor is “a little strong,” suggesting that
the force of it is such that the limiting strategy that applies to beer and wine is inappropriate
here. Hannah says that she thinks it is appropriate for her children to view pornography
and justifies her position by expressing her confidence that they will not “abuse” it or “use
it in a rampant way” (i.e. use it TOO MUCH) Elijah says something similar, noting that
viewing pornography is appropriate provided his son is not “obsessed” with it, and the
content is not “some crazy thing.” Presumably, these inappropriate conditions are more
likely to create lasting negative consequences.
4.5.1.3 Non-Forceful Types
The final way parents elaborate this model is by asserting that there are different
forms of the behavior in question, and certain forms are appropriate because they do not
exert the negative force.
83
[swear] “I would actually not mind if you said ‘it’s so fucking hot,’ because
it is hot, but if you’re saying it in a moment of, we’re having an argument
and you say that word, that’s when I take offense. But I have let it go
actually during one of these heated conversations, and then I come back to
it when I’ve cooled down.... ‘it’s fucking hot?’ Yeah, it’s hot, it’s so fucking
hot today, fine. But don’t say it when you’re having an angry conversation
and when you’re saying it to me in that way.” (Ana, Filipino descent, two
sons ages 16 and 9)
[view porn] So if there was like some sex positive, feminist positive porn I
think I might be open to it but I think the stuff there is pretty demeaning to
women. (Elise, white, same-sex, 11-year-old daughter)
In the first example, Ana explains that four-letter words are appropriate as long as
they are not used in anger, to harm. Ana refers to the metaphor of Anger Is a Hot Fluid in
a Container (Lakoff and Kövecses 1987; Kövecses 2012). Following this metaphor,
without heat, the forceful explosion is averted. In the third excerpt, Elise says that although
most porn is “demeaning,” implying a negative force, she is open to “feminist positive
porn,” or porn that ENABLES.
4.5.2 Behavior Opens New Paths
Some behaviors activate the Behavior Opens New Paths model. Following this
model, parents are concerned less with actual forces the behavior causes and more with
possibilities associated with the behavior. The two basic schematic forms of this model are
visually represented in Figure 4.3. In both forms, an open box represents the opening of
new possibilities, understood as possible PATHS. In the first, the emphasis is on new things
the child is able to do. For example, smartphones enable digital communication and web
browsing, dating enables varying degrees of physical contact, and drinking and attending
unsupervised parties open the possibility of doing “dumb things.” Here the child is
understood as an agent, which is understood as a form of SELF-MOTION (Lakoff and
84
Johnson 1999, 52). In the second, the emphasis is on the possible things that may happen
to a child as a result of their new abilities. Sex opens the possibility of unwanted pregnancy
and STIs, tvs and computers can bring in unwanted messages and content, and drinking at
parties can make you vulnerable to abuse.
85
Figure 4.3: Schematic Diagram of Behavior Opens New Paths
86
The problems associated with this model are dealt with through four different
schematic elaborations, depending on the specifics of the behavior in question.
Appropriateness is either established through LINKAGE, minimizing ATTRACTION,
establishing BLOCKAGE, or ensuring proper CONTAINMENT. I discuss each of these in turn.
4.5.2.1 Establishing LINKAGE
When asked about the appropriate age to get a smartphone, a number of parents,
especially those living in urban areas, said that it was necessary when the child began going
places alone.
“Middle school because they, like our kid goes to the school bus by himself,
so he has to call when the bus comes, so they can then take a bus, or a
subway, or go to the movies with a friend, and then check in. We use it just
to know where he is, so that’s why middle school” (Hazel, white, same-sex,
three sons ages 13, 10, and 4).
“The first time you’re going somewhere without your parents, when you’re
on your own. So my kid had one last year because he was walking himself
and his brother to the bus stop ten minutes away” (Lisa, white, same-sex,
three sons ages 13, 10, and 4).
For Hazel and Lisa, the question about getting a smartphone evoked memories of
their own children going places on their own, which threatened their need to know the
location of their children at all times. In this way it seems the question about smartphones
evoked a problem associated with a different behavior--going places alone--and the
smartphone was understood as a solution to the problem created by autonomous movement
outside. In these cases, the problem is solved by establishing LINKAGE between the parent
and the child.
87
4.5.2.2 Ensuring BLOCKAGE
Parents often expressed concern about the things children are able to access with
TVs, computers, social media, and smartphones. One way of dealing with this is to LIMIT
access by establishing BLOCKAGE.
[TV/Computer] “I think she does make really bad choices about things to
watch and I just want to see what she’s watching. So our television has a
block on it, and the computer does too. So when she wants to watch certain
things or go on the computer and access the internet I have to be at home so
I can unlock it. The television works but mostly only when I’m home. And
then it goes off at ten o’clock at night” (Deisha, black, single, one 14-year-
old daughter).
[Smartphone] “teenage years... But we would be the family that, it would
be in a drawer the second they walk in the door. It wouldn’t be unlimited
access” (Stan, white, two sons, 7 and 5).
[Social Media] “Our son got it at ten... he has limits... the only way he can
get an Instagram account is if his dad can follow him” (Ana, Filipino
descent, two sons ages 16 and 9).
[TV in bedroom] “I’d like to control what he watches, what he sees. There’s
a lot of stuff out there, everything is easily accessible. I want to monitor
what he’s watching. We’ve already put in the port for the TV in his room,
but we have not purchased one and I don’t plan on doing that any time soon”
(Ravi, East-Indian descent, one 5-year-old son).
In some cases, BLOCKAGE is established through literal filters or lock codes. In
other cases, it established indirectly, through SIGHT. Note the way Ana and Ravi discuss
media consumption in terms of travel. Ana says that her husband FOLLOWS her son on
Instagram, and Ravi makes reference to the easily accessible “stuff out there.” In these
cases, appropriateness is established by keeping SIGHT of the child’s movements, perhaps
similarly to the parents who discussed needing to “keep in touch” with their children. The
implication here is that is that monitoring either motivates the child to block their own
behavior, or it enables quick intervention should a need arise.
88
In other cases, BLOCKAGE is established internally, and thus the behavior is
considered inappropriate until the child can exhibit self-restraint. This form applies most
often in cases where external blockage is nonsensical or not feasible.
[unsupervised parties] “I guess after high school. I would be more
comfortable because by then, I know she’s mature enough where I would
think she’s mature enough to make the right choice and conduct herself.
Because when you’re 15, 16, I’ve been there, and you wanna do whatever
older kids do, and not everybody has the same good heart as everybody and
there are kids out there just to take advantage” (Mateo, Hispanic, three
daughters ages 12, 7, and 5).
[social media] “If they are smart enough and reasonable and you know they
are not going to do something overly stupid, 16 or 17, but not before that. I
have way too many clients of mine who have some naked pictures around
and they are 12” (Maya, Indian descent, 23 year-old daughter).
[have sex] “When they can handle their birth control. This is what I tell my
daughters: when you can handle telling a man, your partner, what you like
and you’re willing to explore that with them and [you] have a language for
that and you can protect yourself sexually from transmitted diseases” (Jenn,
white, two daughters ages 24 and 22, one son age 20).
In each of these excerpts, parents say that age-appropriateness is conditional on the
child’s ability to make good decisions and avoid bad ones. Two forms of blockage are
implied: self-restraint and protection. Mateo and Maya tie self-restraint to being mature or
smart enough. “Enough” is structured by the SCALE schema which also structures “a little
bit” which was discussed previously, however in this case appropriateness is defined by a
threshold that must be met rather than a limit that must not be crossed. Jenn uses the
metaphor of “handling,” but rather than referring to counterforce as seen above, here
“handling” refers to taking necessary precautions to protect oneself from unwanted
outcomes. “Handling” entails closing off oneself to the negative potential consequences of
a behavior.
89
4.5.2.3 Minimizing ATTRACTION
In other cases, the problem of SELF-MOTION was elaborated with another schema:
ATTRACTION. Here, parents worried the behavior opened new possibilities that were
harmful or otherwise undesirable, and attractive too. Some parents dealt with this problem
by taking steps to reduce the attractiveness of the behavior.
“[Try alcohol] She’s already tried beer and wine under my control. So that
she hates it… I want her to have a healthy distance from it until she’s older.
I don’t want her to go into a party and say ‘Ooo my Mom never lets me do
this’” (Pam, white, single, 11-year-old daughter).
“[Try alcohol] I’ve let him take a sip because I don’t want it to be the Holy
Grail like ‘yeah finally I can drink!’ He’s had a taste of it and he thinks it’s
not a big deal” (Ana, Filipino descent, two sons ages 16 and 9).
[Try alcohol] “So like Jack lets Carly like sip beer from his cup now. Not to
any horrible point. Like our children are not like drunk or anything, but we
don’t want to make it seem so verboten that they want it” (Sara, white, 9
year-old son, 7 year-old daughter).
In the first excerpt, Pam imagines her daughter drinking recklessly at a party. To
avoid this possibility enabled by drinking outside the home, Pam says she wants to make
sure her daughter has “a healthy distance from it.” Counterintuitively, this “healthy
distance” is established through early exposure. However, this early exposure is carefully
controlled. The AMOUNT is reduced, as seen in previous examples, and it is CONTAINED
within Pam’s control. Ana and Sara give similar responses. Ana says she doesn’t want
drinking to be “the Holy Grail,” and Sara says she doesn’t want the behavior to be labeled
“verboten,” meaning forbidden. In all these cases, if drinking is forbidden, it becomes
especially desirable, reducing the child’s ability to partake responsibly and avoid negative
consequences.
90
4.5.2.4 Behavior is CONTAINED
Some parents elaborate the Behavior Opens New Paths model with the
CONTAINMENT schema. In these cases, appropriateness is established by limiting the
behavior to a “safe space” where children’s choices and outside influences are limited.
[Try Alcohol] “I really think in college and in a controlled environment
where they’re not gonna get in a car is probably okay” (Tess, white, two
daughters ages 30 and 23).
[Try Alcohol] “Depends on the setting. I think with my girlfriends and stuff
we tried like margaritas when we were at home when we were like 18 or
19” (Megan, white, two sons ages 25 and 22, three daughters ages 17, 14,
and 12).
[R-rated movie] “Do you remember the movie 13? I definitely showed it to
both of my daughters when they were 14 or so because it had something to
say about their lives. And I wanted to watch it with them and not out in the
world, out with a bunch of friends and thinking the behavior in the film was
worth emulating” (Judith, white, two daughters ages 27 and 23, one son age
19).
[Dating] “If I keep him in Islamic school, at least in school it’ll probably be
a little bit more controlled… the main thing is that you just gotta tell your
kid, ‘don’t make any stupid mistakes that you’ll regret and that you’ll have
to carry a burden on for the rest of your life’” (Amir, Indian Descent,
married father, two 2-year-old twin daughters, 5-year-old son).
In these excerpts the setting of the behavior takes precedence. If alcohol
consumption takes places at home or in a “controlled environment,” then the possibility of
driving while drunk is eliminated. Judith’s discussion of watching an R-rated film with her
daughters reveals interpretation as a different kind of agentic possibility, though it is
similarly constrained by containing the film-watching experience within her presence, as
opposed letting them watch it “out in the world.” Amir says that in an Islamic school, dating
would be “more controlled” and less likely to result in his sons having burdens of regret
caused by poor choices.
91
4.5.2.5 Behavior Closes off Possibilities
There is an inverted form of this cultural model which, although rare in my data, is
worth mentioning. When parents were asked about having a TV in the bedroom, some
talked about how that would prevent certain desirable states. That is, rather than opening
new possibilities, the behavior closes off possibilities.
“Never. I just feel that that’s where they’re gonna be all the time and we’re
not gonna get that family bonding like we probably should be… it’s not so
much that I’m gonna be concerned about with what they’re watching; I just
don’t want them to be there all the time” (Stephen, white, two daughters
ages 30 and 23).
“I think it kills that whole family atmosphere. Halim’s in his room watching
one show, Nina’s in her room watching another show, me and Jan are
downstairs doing our own show, let’s all just watch together, ya know”
(Khalil, Egyptian descent, two daughters ages 13 and 6, one son age 10).
In many American homes, the television has taken the role of the traditional hearth.
It is where family members gather and spend time together (Tichi 1992; Spigel 2013).
Stephen and Khalil dislike the idea of TVs in bedrooms because they believe it will close
off this possibility, depriving the family of a valued experience.
4.5.3 Behavior Is Dirty
In some cases, the reasoning task activates the Behavior Is Dirty model. According
to this model, the behavior may or may not have a designated location to which it belongs.
If there is such a place and the behavior is located there, then it is considered appropriate.
If the behavior is out of place (or has no place), then it is inappropriate and considered
dirty, disgusting, vulgar, or something similar (Douglas 2003). Schematically, a neutral
ENTITY becomes dirt by either crossing a BOUNDARY into a CONTAINER to which is does
not belong (thereby making the container “dirty”), or by making CONTACT with a landmark
92
where it is considered a foreign substance (thereby making the landmark dirty) (Lizardo
2012). The container model applies to things like movies, where a movie might be clean
except for a “dirty scene.” The landmark model applies to things like wearing makeup and
swearing, where contact is made with the individual. These two models are visually
represented in Figure 4.4.
93
Figure 4.4: Schematic Diagram of Behavior is Dirty
94
Consider the following discussions of swearing and wearing makeup:
[Swearing] “Profanity makes you look bad. It makes you look ignorant. It
makes you look not like a seasoned adult, good person when you’re saying
foul or especially when you’re saying a lot of them. You don’t need to talk
like that. And I don’t think God appreciates hearing dirty talk, bad talk.
Don’t say that stuff” (Brandon, white, two sons ages 17 and 12, one
daughter age 9).
[Wearing Makeup] “It doesn’t look nice on a small child, and it looks weird.
What is that kid from Colorado? Jonbenet Ramsey? It looks weird. I mean
it’s like small girls need to be small girls. I think when they’re old enough,
they wanna start doing makeup, I think it’s fine, but it’s like, kids I think
looks weird” (Pranab, East Indian origin, two sons 21 and 16).
Brandon describes swearing as “foul” and “dirty talk,” and says that it “makes you
look bad” in a few different ways. Similarly, Pranab says that when small children wear
makeup, it “looks weird.” These are simple aesthetic violations which are distinct from the
problems addressed in the earlier models. Brandon’s appeal to God points to the fact that
metaphorically speaking, cleanliness is next to godliness, and the problem of dirty is a
moral one.
Parents elaborate the Behavior Is Dirty model in two ways: limiting the amount of
dirt, or restricting the problematic entities to the places where they belong.
4.5.3.1 Limiting the AMOUNT
This elaboration is schematically equivalent to the elaboration of Behavior Exerts
a Force, but here the effect is to reduce the aesthetic violation rather than reduce a force.
[R-rated movie] “If some movie is a good movie, there’s some little bit
nudity in between, that’s why it’s rated R, it’s fine. But I don’t want them
to go for a dirty movie just nothing but sex going on. That I think is wrong”
(Stephen, white, two daughters, 30 and 23).
[Makeup] “Depends how they wear it. There are girls who glob it on and it
looks horrid at any age and others who are very discreet about it. So
95
probably like 14. Yeah, 9th grade.” (Dana, white, two sons ages 16 and 10,
one daughter age 14)
Stephen has a problem with sexual content in movies, but is fine as long as the
AMOUNT is limited to “some little bit, in between.” He contrasts this with a “dirty movie”
where there is “nothing but” sex. Dana similarly argues that makeup is appropriate
provided the AMOUNT is limited.
4.5.3.2 Restricted to Appropriate Location
Some parents elaborate this model by describing spaces where the behavior is
appropriate.
[Makeup] “She’s already wearing makeup but she can’t wear mascara and
eyeliner. She can wear blush and lip gloss but not in school” (Pam, white,
single, 11-year-old daughter).
[Makeup] “She’s allowed at home, play dress up thing, but not out”
(Carmen, Hispanic, single, 15-year-old son, 11-year-old daughter).
[Swear] “When you hear it on a kid you just realize ‘oh my god that sounds
so vulgar’ like even though I say it, like I’m a swearer in the house" (Elise,
white, same-sex, 11-year-old daughter).
The appropriate aesthetic location may be a physical space, such as “home,” a state,
such as “play”, or even a certain landmark, such as an adult body.
4.5.4 Summary
When parents reasoned about the age-appropriateness of the behaviors listed in
Table 2.1, their reasoning was structured by one or more of three basic cultural models.
Table 4.2 lists the three cultural models and their associated elaborations. These models
facilitate reasoning by specifying what about the behavior is relevant for the evaluative
96
task at hand. The models were not wholly determinative, however, and were elaborated in
a limited variety of ways.
97
TABLE 4.1:
SUMMARY OF BASIC SCHEMATIC MODELS AND SCHEMATIC
ELABORATIONS
Basic
Schematic Models
Schematic Elaborations
Behavior Exerts Force Reduce Amount (substance)
Increase Counterforce
Limit to Non
-
Forceful Forms
Behavior Opens New Paths Contain Behavior
Ensure Blockage (external)
Ensure Blockage (self-
restraint)
Establish Linkage
Minimize Attraction
Behavior Is Dirty Reduce Amount (substance)
Limit to Proper Spaces
98
I do not claim that the three cultural models I identified are the only possible models
that structure reasoning about age-appropriateness, though I believe they are primary for
the topics I investigated. I did find a detailed discussion about sex that seemed to be
structured by different models:
“I think that the thinner you spread that, the less substance it has to it, and
I think that’s something that is extremely important for our moral well-
being. To have a relationship that is intimate where you, to some degree,
share some of your soul and then pass on to another--the more times you...
I think the more callous you can become, the less the experience has value.”
(Brandon, white, married father, two sons ages 17 and 12, one daughter age
9).
Brandon describes sex as a substance that is spread, as the sharing of souls, and as
a valuable experience that loses value when excessively shared. Brandon also discussed
sex in terms of force and paths, but I include these excerpts here because they suggest that
there are some behaviors, such as sex, for which there are sophisticated models that have
been developed for reasoning. These models might be more domain-specific and more
closely tied to particular groups.
4.6 Variation in Reasoning
I now turn to the question of variation. The three basic schematic models associated
with the evaluative task were available to parents, as were the many different schematic
variations. The strength of the above analysis is that it makes it possible to begin
investigating why the combined, elaborated schematic models manifest in evaluative
reasoning have the structure they do.
99
4.6.1 Variation by Behavior
The structure of evaluative reasoning depends upon what is being evaluated. There
may be a few basic cultural models associated with an evaluative category, but ultimately,
one or more of these models is activated because it fits the person’s understanding of the
thing being evaluated. Some models fit very well with certain behaviors, and others are an
odd fit. For example, Behavior Is Dirty goes hand in hand with makeup, but Behavior
Exerts a Force does not.
The basic schematic models constrain elaborations because of the behaviors
associated with them, though to a lesser extent. I have already shown that schemas such as
SCALE and AMOUNT elaborate multiple models. However, there are some elaborations,
such as LINKAGE, that only elaborate one model. The point remains, however, that these
are image schemas, and as such, they are building blocks which may be rearranged and
recombined in multiple ways. As such, we should expect them to appear in many different
kinds of places.
4.6.2 Variation by Personal Experience
Schematic structure also varies by individual experience, especially in terms of
elaborations. In my data, I found some evidence suggesting the way one elaborates a model,
or whether one elaborates at all, is related to one’s familiarity with the behavior in question.
I compared the way men and women reasoned about the age-appropriateness of makeup,
assuming women have a heightened familiarity to it. I coded each response based on
whether the parent’s reasoning about makeup manifested an elaborated model, and
compared the proportions of elaborated responses between groups. Parents elaborated
makeup by either referring to quantity (e.g. “a little bit” vs. “caked on”), type (e.g. lip gloss
100
vs. eyeliner), or setting (e.g. home vs. school). I found that elaborated models were more
common among women (73.17%) than men (36.84%). The results of this initial
comparison suggest that the relationship between personal familiarity and elaboration is at
least a plausible one.
4.6.3 Variation by Declarative Memory
Variation in schematic structure also appears related to the particular memory or
memories that the evaluative task evokes. After observing all of the discussions about
trying alcohol, I noticed a pattern: when parents talked about young children drinking, they
always elaborated their discussions with LIMITED AMOUNT. However, when parents talked
about older children drinking, they never referred to amount, and but frequently elaborated
with CONTAINMENT. I did find one mother who described her own drinking in terms of
quantity, but she was a special case because she was a physician with her own small
practice and thus always on call. This pattern persisted even within individuals: when
parents discussed both early and later drinking, they elaborated them in different ways.
4.6.4 Variation by Judgment
The interview data also suggest that some variation in schematic structure is related
to the parent’s judgment about the age-appropriateness of the behavior in question. Early
in the analysis, I examined whether schemas are associated with age but I did not find any
relation. As I noted in the introduction, reasoning and judgment are cognitive distinct tasks,
so there is no necessary reason why they should be related. That said, there was one
category of answers that was more distinct than others: the “never appropriate” answers.
Parents who responded “never” were less likely to provide any reasoning at all, and when
101
they did reason, it varied widely and included elaborations not picked up by the other
parents, as seen in the early example about the variety of justifications parents gave for
rejecting pornography. If reasoning does follow intuitive judgments, then it seems plausible
that certain judgments make certain models and elaborations more salient than others.
4.7 Discussion and Conclusion
Culture facilitates evaluative reasoning schematically by providing a structure for
thinking about the issue at hand. Previous research on culture in reasoning has focused
primarily on group-based differences in response to specific questions, and as a result, has
not developed a more general understanding of how particular instances of reasoning
manifest the structure they do. In this paper, I analyzed the way parents reasoned about the
age-appropriateness of a series of behaviors, which provided the comparative leverage to
develop a dynamic theory of culture in reasoning. I found that reasoning was structured by
two analytically distinct cultural elements: basic schematic models, associated with the
category of age-appropriateness, and schematic elaborations, associated with the behaviors
being evaluated. These two elements combined to form the complete, elaborated schematic
models. These models did not completely constitute parents’ reasoning but structured the
way parents discussed their own autobiographical experience. Additionally, I found that
the structure of elaborated schematic models varied across four dimensions: the behavior
being evaluated, the parent’s personal experience, the parent’s evaluative judgment, and
the parent’s autobiographical memory.
In sociological theory, there is a longstanding interest in understanding how internal
habits (a form of personal culture) and external situations jointly produce action (Martin
102
2015; Luft 2015; Stephen Vaisey 2008). The dynamic theory of culture in reasoning
developed here contributes to this discussion by extending the model of action developed
by Martin (2015) and Swidler (2013) to the analysis of reasoning. Martin argues that people
move through the world mostly without effort because the objects surrounding them tell
them what to do. This is possible because the objects in the world have certain qualities
which induce certain actions. These qualities are individually variable, meaning they are
always qualities to someone. As Martin (2015: 238) says, “The same pole that seems to say
'throw me' to one person, announces just as definitely 'I am too big for you to wield' to
another’.” I suggest the same principle applies to reasoning. Parents can reason about the
age-appropriateness of topics they have never thought about in this way before (as some
parents did confide) because the objects themselves tell them what to do. I argue that this
happens because the parents’ individual experience with the behaviors in question makes
certain schemas more salient for the task.
One possible objection is that the above analysis is unable to tell whether the
schematic models are activated by the task of evaluating age-appropriateness. For example,
the schematic models identified in discussions of sex could just be how people talk about
sex in general. This is where the schema analysis as a cumulative program is especially
useful. After completing my analysis, I read other work on schemas and sex and found that
when people were asked to describe sexual experience outside the context of age-
appropriateness, the schemas structuring their responses were very different (Fernández
2008; Undie, Crichton, and Zulu 2007). All reasoning is situated reasoning (Barsalou
2009), and an evaluative category creates a context which makes certain schematic models
more salient than others.
103
The above analysis identifies the basic schematic models that structure the way
parents reason about age-appropriateness, but individuals of different ages or relational
status may or may not reason in the same way. Future research is needed to determine the
extent to which these basic schematic models are general to children and adults without
children. Additionally, my analysis focused on behaviors that occur relatively early in life;
future research might include topics associated with later periods in life, such as having
children, civic participation, or retiring.
Future research might also take the elaborated schematic models discussed here and
try them out on adolescents to determine which, if any, they find most convincing or
satisfying. Findings from such a project may further our understanding of parent-child
relations and the things that tax them.
Finally, though this study focused on reasoning, not everyone engages in reasoning
when they are given an evaluative task, or they do so to varying degrees. In some cases,
people simply give their judgment and move on, and in other cases, they engage in lengthy
justification. Future research might investigate the factors that predispose a person to
reason at all, and if so, to the extent that they do.
104
APPENDIX A:
INTERVIEW SAMPLE
TABLE A.1:
INTERVIEW SAMPLE
Religion
N
Catholic
18
Protestant
10
Jewish
10
Muslim
12
Latter
-
day Saint
7
Hindu
9
Buddhist
3
None
1
Gender
Women
47
Men
22
Race
White
34
Other
26
Hispanic
6
Blac
k
3
Number of
Children (mean)
2.35
Age of Eldest Child (mean)
16.65
105
BIBLIOGRAPHY
Abend, Gabriel. 2011. “Thick Concepts and the Moral Brain.” European Journal of
Sociology. 52 (1): 143–72.
———. 2014. The Moral Background: An Inquiry into the History of Business Ethics.
Princeton University Press.
Anderson, Richard C. 1978. “Schema-Directed Processes in Language Comprehension.”
In Cognitive Psychology and Instruction, edited by Alan M. Lesgold, James W.
Pellegrino, Sipke D. Fokkema, and Robert Glaser, 67–82. Boston, MA: Springer
US.
Baayen, R. H. 2008. “Models for Lexical Richness.” In Analyzing Linguistic Data: A
Practical Introduction to Statistics, 243–58. Cambridge University Press.
Bail, Christopher A. 2012. “The Fringe Effect: Civil Society Organizations and the
Evolution of Media Discourse about Islam since the September 11th Attacks.”
American Sociological Review 77 (6): 855–79.
———. 2014. “The Cultural Environment: Measuring Culture with Big Data.” Theory
and Society 43 (3): 465–82.
Bargheer, Stefan. 2018. Moral Entanglements: Conserving Birds in Britain and
Germany. University of Chicago Press.
Bar, Moshe. 2007. “The Proactive Brain: Using Analogies and Associations to Generate
Predictions.” Trends in Cognitive Sciences 11 (7): 280–89.
Barsalou, Lawrence W. 2009. “Simulation, Situated Conceptualization, and Prediction.”
Philosophical Transactions of the Royal Society of London. Series B, Biological
Sciences 364 (1521): 1281–89.
Berger, P. L., and T. Luckmann. 1967. The Social Construction of Reality: A Treatise in
the Sociology of Knowledge. Anchor Books.
Bourdieu, Pierre. 1977. Outline of a Theory of Practice. Cambridge University Press.
106
———. 1984. Distinction: A Social Critique of the Judgement of Taste. Harvard
University Press.
———. 1990. The Logic of Practice. Stanford University Press.
Bryor Snefjella, and Victor Kuperman. 2015. “Concreteness and Psychological Distance
in Natural Language Use.” Psychological Science 26 (9): 1449–60.
Brysbaert, Marc, Amy Beth Warriner, and Victor Kuperman. 2014. “Concreteness
Ratings for 40 Thousand Generally Known English Word Lemmas.” Behavior
Research Methods 46 (3): 904–11.
Cerulo, Karen A. 2002. Culture in Mind: Toward a Sociology of Culture and Cognition.
Psychology Press.
———. 2016. “Cognition and Cultural Sociology: The Inside and Outside of Thought.”
In The SAGE Handbook of Cultural Sociology. Sage.
Chakrabarti, P., and M. Frye. 2017. “A Mixed-Methods Framework for Analyzing Text
Data: Integrating Computational Techniques with Qualitative Methods in
Demography.” Demographic Research 37(42):1351-1382.
Chao, Anne, Chun-Huo Chiu, and Lou Jost. 2010. “Phylogenetic Diversity Measures
Based on Hill Numbers.” Philosophical Transactions of the Royal Society of
London. Series B, Biological Sciences 365 (1558): 3599–3609.
Cheng, P. W., and K. J. Holyoak. 1985. “Pragmatic Reasoning Schemas.” Cognitive
Psychology 17 (4): 391–416.
Cicourel, Aaron V. 1974. Cognitive Sociology: Language and Meaning in Social
Interaction. New York, NY, US: Free Press.
Collett, Jessica L. 2005. “What Kind of Mother Am I? Impression Management and the
Social Construction of Motherhood.” Symbolic Interaction 28 (3): 327–47.
Cornelissen, Sharon. 2016. “Turning Distaste into Taste: Context-Specific Habitus and
the Practical Congruity of Culture.” Theory and Society 45 (6): 501–29.
Covington, Michael A., and Joe D. McFall. 2010. “Cutting the Gordian Knot: The
Moving-Average Type–Token Ratio (MATTR).” Journal of Quantitative
Linguistics 17 (2): 94–100.
Cunningham, William A., and Philip David Zelazo. 2007. “Attitudes and Evaluations: A
Social Cognitive Neuroscience Perspective.” Trends in Cognitive Sciences 11 (3):
97–104.
DeGloma, Thomas. 2014. Seeing the Light: The Social Logic of Personal Discovery.
University of Chicago Press.
107
Dewey, John. 2012. Experience and Nature. Courier Corporation.
DiMaggio, Paul. 1997. “Culture and Cognition.” Annual Review of Sociology 23 (1):
263–87.
Douglas, Professor Mary. 2003. Purity and Danger: An Analysis of Concepts of Pollution
and Taboo. Routledge.
Durkheim, Emile, and Marcel Mauss. 2009. Primitive Classification. Routledge.
Durkheim, Émile, and Joseph Ward Swain. 2008. The Elementary Forms of the Religious
Life. Courier Corporation.
Ecklund, Elaine Howard, Jared L. Peifer, Virginia White, and Esther Chan. 2017. “Moral
Schemas in Articulation and Intuition: How Religious People Evaluate Human
Reproductive Genetic Technologies.” Sociological Forum 32 (2): 277–97.
Edelmann, Achim, and John W. Mohr. 2018. “Formal Studies of Culture: Issues,
Challenges, and Current Trends.” Poetics 68 (June): 1–9.
Edgell, Penny, and Kathleen E. Hull. 2017. “Cultural Schemas of Religion, Science, and
Law in Talk About Social Controversies.” Sociological Forum 32 (2): 298–320.
Eshbaugh-Soha, Matthew. 2010. “The Tone of Local Presidential News Coverage.”
Political Communication 27 (2): 121–40.
Ewick, Patricia, and Susan Silbey. 2003. “Narrating Social Structure: Stories of
Resistance to Legal Authority.” The American Journal of Sociology 108 (6):
1328–72.
Farrell, Justin. 2013. “Environmental Activism and Moral Schemas: Cultural
Components of Differential Participation.” Environment and Behavior 45 (3):
399–423.
Feagin, Joe R. 2013. The White Racial Frame: Centuries of Racial Framing and
Counter-Framing. Routledge.
Fernández, Eliecer Crespo. 2008. “Sex-Related Euphemism and Dysphemism: An
Analysis in Terms of Conceptual Metaphor Theory.” Atlantis 30 (2): 95–110.
Frye, Margaret, and Jenny Trinitapoli. 2015. “Ideals as Anchors for Relationship
Experiences.” American Sociological Review 80 (3): 496–525.
Fujita, Kentaro, Marlone D. Henderson, Juliana Eng, Yaacov Trope, and Nira Liberman.
2006. “Spatial Distance and Mental Construal of Social Events.” Psychological
Science 17 (4): 278–82.
Geeraerts, Dirk. 2006. Cognitive Linguistics: Basic Readings. Walter de Gruyter.
108
Gentner, Dedre, and Linsey Smith. 2012. “Analogical Reasoning.” Encyclopedia of
Human Behavior 1: 130–36.
Ghaziani, Amin, and Marc J. Ventresca. 2005. “Keywords and Cultural Change: Frame
Analysis of Business Model Public Talk, 1975–2000.” Sociological Forum 20 (4):
523–59.
Gibbs, Raymond W., and Raymond W. Gibbs Jr. 2017. Metaphor Wars. Cambridge
University Press.
Goffman, Erving. 1974. Frame Analysis: An Essay on the Organization of Experience.
Harvard University Press.
Guhin, Jeffrey. 2016. “Why Worry about Evolution? Boundaries, Practices, and Moral
Salience in Sunni and Evangelical High Schools.” Sociological Theory 34 (2):
151–74.
Henderson, Marlone D., Kentaro Fujita, Yaacov Trope, and Nira Liberman. 2006.
“Transcending the‘ Here’: The Effect of Spatial Distance on Social Judgment.”
Journal of Personality and Social Psychology 91 (5): 845.
Hertz, Robert. 2013. “The Pre-Eminence of the Right Hand: A Study in Religious
Polarity.” HAU: Journal of Ethnographic Theory 3 (2): 335–57.
Hill, M. O. 1973. “Diversity and Evenness: A Unifying Notation and Its Consequences.”
Ecology 54 (2): 427–32.
Hoffmann, John P. 2014. “Religiousness, Social Networks, Moral Schemas, and
Marijuana Use: A Dynamic Dual-Process Model of Culture and Behavior.” Social
Forces 93 (1): 181–208.
Holland, Dorothy. 1987. Cultural Models in Language and Thought. Cambridge
University Press.
Holyoak, Keith J., Keith James Holyoak, and Paul Thagard. 1996. Mental Leaps:
Analogy in Creative Thought. MIT Press.
Homan, Patricia, Lauren Valentino, and Emi Weed. 2017. “Being and Becoming Poor:
How Cultural Schemas Shape Beliefs About Poverty.” Social Forces 95 (3):
1023–48.
Hout, Michael, and Claude S. Fischer. 2014. “Explaining Why More Americans Have No
Religious Preference: Political Backlash and Generational Succession, 1987--
2012.” Sociological Science 1 (9): 423–47.
Ignatow, G. 2016. “Theoretical Foundations for Digital Text Analysis.” Journal for the
Theory of Social Behaviour. 46(1):104-120.
109
Ignatow, Gabriel. 2006. “Cultural Models of Nature and Society: Reconsidering
Environmental Attitudes and Concern.” Environment and Behavior 38 (4): 441–
61.
———. 2009. “Culture and Embodied Cognition: Moral Discourses in Internet Support
Groups for Overeaters.” Social Forces 88 (2): 643–69.
Jarvis, S. 2017. “Grounding Lexical Diversity in Human Judgments.” Language Testing.
34(4): 537-553.
Jarvis, Scott. 2013. “Capturing the Diversity in Lexical Diversity.” Language Learning
63 (March): 87–106.
Jerolmack, Colin, and Shamus Khan. 2014. “Talk Is Cheap: Ethnography and the
Attitudinal Fallacy.” Sociological Methods & Research 43 (2): 178–209.
Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination,
and Reason. University of Chicago Press.
Johnson, Wendell. 1944. “Studies in Language Behavior: A Program of Research.”
Psychological Monographs 56 (2): 1–15.
Jost, Lou. 2006. “Entropy and Diversity.” Oikos 113 (2): 363–75.
———. 2007. “Partitioning Diversity into Independent Alpha and Beta Components.”
Ecology 88 (10): 2427–39.
Kazyak, Emily, Kelsy Burke, and Mathew Stange. 2018. “Logics of Freedom: Debating
Religious Freedom Laws and Gay and Lesbian Rights.” Socius 4 (January):1-18.
Kefalas, Maria J., Frank F. Furstenberg, Patrick J. Carr, and Laura Napolitano. 2011.
“‘Marriage Is More Than Being Together’: The Meaning of Marriage for Young
Adults.” Journal of Family Issues 32 (7): 845–75.
Kimmel, Michael. 2005. “Culture Regained: Situated and Compound Image Schemas.”
Pp. 285–312 in Hampe, Beate (Ed.) From Perception to Meaning: Image
Schemas in Cognitive Linguistics. De Gruyter.
Kövecses, Zoltán. 1999. “Metaphor: Does It Constitute or Reflect Cultural Models?”
Amsterdam Studies In The Theory And History Of Linguistic Science Series 4,
167–88.
Kövecses, Zoltan. 2012. Emotion Concepts. Springer.
Lakoff, George. 2008. Women, Fire, and Dangerous Things. University of Chicago
Press.
Lakoff, George, and Mark Johnson. 1999. Philosophy in the Flesh. Basic Books.
110
Lakoff, George, and Zoltán Kövecses. 1987. “The Cognitive Model of Anger Inherent in
American English.” Cultural Models in Language and Thought, 195–221.
Lamont, Michèle, and of Sociology Michele Lamont. 1992. Money, Morals, and
Manners: The Culture of the French and the American Upper-Middle Class.
University of Chicago Press.
Lamont, Michèle, and Virág Molnár. 2002. “The Study of Boundaries in the Social
Sciences.” Annual Review of Sociology 28 (1): 167–95.
Lamont, Michèle, and Ann Swidler. 2014. “Methodological Pluralism and the
Possibilities and Limits of Interviewing.” Qualitative Sociology 37 (2): 153–71.
Lee, Monica, and John Levi Martin. 2015. “Coding, Counting and Cultural Cartography.”
American Journal of Cultural Sociology 3 (1): 1–33.
Leinster, Tom, and Christina A. Cobbold. 2012. “Measuring Diversity: The Importance
of Species Similarity.” Ecology 93 (3): 477–89.
Leschziner, Vanina. 2015. At the Chef’s Table: Culinary Creativity in Elite Restaurants.
Stanford University Press.
Lewis, Penelope A., and Simon J. Durrant. 2011. “Overlapping Memory Replay during
Sleep Builds Cognitive Schemata.” Trends in Cognitive Sciences 15 (8): 343–51.
Liberman, Nira, and Yaacov Trope. 1998. “The Role of Feasibility and Desirability
Considerations in near and Distant Future Decisions: A Test of Temporal
Construal Theory.” Journal of Personality and Social Psychology 75 (1): 5.
Lizardo, O. 2012. “The Conceptual Bases of Metaphors of Dirt and Cleanliness in Moral
and Non-Moral Reasoning.” Cognitive Linguistics 23 (2): 367–93.
———. 2017. “Improving Cultural Analysis: Considering Personal Culture in Its
Declarative and Nondeclarative Modes.” American Sociological Review 82(1):88-
115.
Lizardo, Omar. 2004. “The Cognitive Origins of Bourdieu’s Habitus.” Journal for the
Theory of Social Behaviour 34 (4): 375–401.
———. 2013. “Schmaus’s Functionalist Approach to the Explanation of Social Facts: An
Assessment and Critique.” Philosophy of the Social Sciences 43 (4): 453–92.
Lizardo, Omar, Robert Mowry, Brandon Sepulvado, Dustin S. Stoltz, Marshall A. Taylor,
Justin Van Ness, and Michael Wood. 2016. “What Are Dual Process Models?
Implications for Cultural Analysis in Sociology.” Sociological Theory 34 (4):
287–310.
111
Lizardo, Omar, and Michael Strand. 2010. “Skills, Toolkits, Contexts and Institutions:
Clarifying the Relationship between Different Approaches to Cognition in
Cultural Sociology.” Poetics 38 (2): 205–28.
Luft, Aliza. 2015. “Toward a Dynamic Theory of Action at the Micro Level of Genocide:
Killing, Desistance, and Saving in 1994 Rwanda.” Sociological Theory 33 (2):
148–72.
Malti, Tina, and Marlis Buchmann. 2010. “Socialization and Individual Antecedents of
Adolescents’ and Young Adults' Moral Motivation.” Journal of Youth and
Adolescence 39 (2): 138–49.
Malvern, David, Brian Richards, Ngoni Chipere, and Pilar Durán. 2004. Lexical Diversity
and Language Development: Quantification and Assessment. Springer.
Mandler, Jean M., and Cristóbal Pagán Cánovas. 2014. “On Defining Image Schemas.”
Language and Cognition 6 (04): 510–32.
Mandler, J. M. 1992. “How to Build a Baby: II. Conceptual Primitives.” Psychological
Review 99 (4): 587–604.
Mannheim, Karl. 2013. Ideology and Utopia. Routledge.
Margolis, Howard. 1987. Patterns, Thinking, and Cognition: A Theory of Judgment.
University of Chicago Press.
Martin, John Levi. 2010. “Life’s a Beach but You're an Ant, and Other Unwelcome News
for the Sociology of Culture.” Poetics 38 (2): 229–44.
———. 2011. The Explanation of Social Action. Oxford University Press, USA.
———. 2015. Thinking through Theory. WW Norton, Incorporated.
———. 2017. Thinking Through Methods: A Social Science Primer. University of
Chicago Press.
McCarthy, Philip M., and Scott Jarvis. 2010. “MTLD, Vocd-D, and HD-D: A Validation
Study of Sophisticated Approaches to Lexical Diversity Assessment.” Behavior
Research Methods 42 (2): 381–92.
McDonnell, Terence E. 2014. “Drawing out Culture: Productive Methods to Measure
Cognition and Resonance.” Theory and Society 43 (3-4): 247–74.
———. 2016. Best Laid Plans: Cultural Entropy and the Unraveling of AIDS Media
Campaigns. University of Chicago Press.
Mead, George Herbert. 2015. Mind, Self, and Society: The Definitive Edition. University
of Chicago Press.
112
Miles, Andrew. 2015. “The (re) Genesis of Values: Examining the Importance of Values
for Action.” American Sociological Review 80 (4): 680–704.
Miles, Andrew, Raphaël Charron-Chénier, and Cyrus Schleifer. 2019. “Measuring
Automatic Cognition: Advancing Dual-Process Research in Sociology.”
American Sociological Review 84(2): 308-333.
Milojević, Staša. 2015. “Quantifying the Cognitive Extent of Science.” Journal of
Informetrics 9 (4): 962–73.
Mohr, John W., and Amin Ghaziani. 2014. “Problems and Prospects of Measurement in
the Study of Culture.” Theory and Society 43 (3): 225–46.
Montemayor, Raymond. 1983. “Parents and Adolescents in Conflict: All Families Some
of the Time and Some Families Most of the Time.” The Journal of Early
Adolescence 3 (1-2): 83–103.
Moore, Rick. 2017. “Fast or Slow: Sociological Implications of Measuring Dual-Process
Cognition.” Sociological Science 4: 196–223.
Myyrya, Liisa, Soile Juujärvi, and Kaija Pesso. 2010. “Empathy, Perspective Taking and
Personal Values as Predictors of Moral Schemas.” Journal of Moral Education 39
(2): 213–33.
Nan, Xiaoli. 2007. “Social Distance, Framing, and Judgment: A Construal Level
Perspective.” Human Communication Research 33 (4): 489–514.
Neugarten, B. L., J. W. Moore, and J. C. Lowe. 1965. “Age Norms, Age Constraints, And
Adult Socialization.” American Journal of Sociology 70 (May): 710–17.
Pálsson, Gísli. 1994. “Enskilment at Sea.” Man 29 (4): 901–27.
Perrin, Andrew J., and Stephen Vaisey. 2008. “Parallel Public Spheres: Distance and
Discourse in Letters to the Editor.” American Journal of Sociology 114 (3): 781–
810.
Presser, Stanley, and Linda Stinson. 1998. “Data Collection Mode and Social Desirability
Bias in Self-Reported Religious Attendance.” American Sociological Review 63
(1): 137–45.
Pugh, Allison J. 2013. “What Good Are Interviews for Thinking about Culture?
Demystifying Interpretive Analysis.” American Journal of Cultural Sociology 1
(1): 42–68.
Quinn, N. 2005. Finding Culture in Talk: A Collection of Methods. Palgrave MacMillan.
113
Quinn, Naomi. 1987. “Convergent Evidence for a Cultural Model of American
Marriage.” Pp. 173–92 in Dorothy Holland and Naomi Quinn, (Eds.) Cultural
Models in Language and Thought. Cambridge.
Robinson, Laura. 2008. “The Moral Accounting of Terrorism: Competing Interpretations
of September 11, 2001.” Qualitative Sociology 31 (3): 271–85.
Rumelhart, David E. 2017. “Schemata: The Building Blocks.” Theoretical Issues in
Reading Comprehension: Perspectives from Cognitive Psychology, Linguistics,
Artificial Intelligence and Education 11: 33.
Schalet, Amy T. 2011. Not Under My Roof: Parents, Teens, and the Culture of Sex.
University of Chicago Press.
Schnabel, Landon, and Sean Bock. 2017. “The Persistent and Exceptional Intensity of
American Religion: A Response to Recent Research.” Sociological Science 4:
686–700.
Schwartz, Barry. 1981. Vertical Classification: A Study in Structuralism and the
Sociology of Knowledge. University of Chicago Press.
Sewell, William H. 1992. “A Theory of Structure: Duality, Agency, and
Transformation.” American Journal of Sociology 98 (1): 1–29.
Shannon, C. E. 1948. “A Mathematical Theory of Communication.” Bell System
Technical Journal 27 (3): 379–423.
Shepherd, Hana, and Emily A. Marshall. 2018. “The Implicit Activation Mechanism of
Culture: A Survey Experiment on Associations with Childbearing.” Poetics 69
(August): 1–14.
Shore, Bradd. 1996. Culture in Mind: Cognition, Culture, and the Problem of Meaning.
Oxford University Press.
Silva, Graziella Moraes D. 2012. “Folk Conceptualizations of Racism and Antiracism in
Brazil and South Africa.” Ethnic and Racial Studies 35 (3): 506–22.
Simpson, E. H. 1949. “Measurement of Diversity.” Nature 163 (April): 688.
Small, Mario Luis. 2008. “Lost in Translation: How Not to Make Qualitative Research
More Scientific.” In Report from Workshop on Interdisciplinary Standards for
Systematic Qualitative Research. Washington, DC: National Science Foundation.
Smith, Christian, and Patricia Snell. 2009. Souls in Transition: The Religious and
Spiritual Lives of Emerging Adults. Oxford University Press.
114
Smith, Eliot R., and Jamie DeCoster. 2000. “Dual-Process Models in Social and
Cognitive Psychology: Conceptual Integration and Links to Underlying Memory
Systems.” Personality and Social Psychology Review 4 (2): 108–31.
Spigel, Lynn. 2013. Make Room for TV: Television and the Family Ideal in Postwar
America. University of Chicago Press.
Spillman, Lyn. 2014. “Mixed Methods and the Logic of Qualitative Inference.”
Qualitative Sociology 37 (2): 189–205.
Steensland, Brian. 2006. “Cultural Categories and the American Welfare State: The Case
of Guaranteed Income Policy.” American Journal of Sociology 111 (5): 1273–
1326.
Stephan, Elena, Nira Liberman, and Yaacov Trope. 2010. “Politeness and Psychological
Distance: A Construal Level Perspective.” Journal of Personality and Social
Psychology 98 (2): 268–80.
Stone, P. J., D. C. Dunphy, and M. S. Smith. 1966. General Inquirer: A Computer
Approach to Content Analysis. MIT Press.
Strand, Michael. 2015. “The Genesis and Structure of Moral Universalism: Social Justice
in Victorian Britain, 1834–1901.” Theory and Society 44 (6): 537–73.
Strand, Michael, and Omar Lizardo. 2015. “Beyond World Images: Belief as Embodied
Action in the World.” Sociological Theory 33 (1): 44–70.
Strauss, Claudia. 2005. “Analyzing Discourse for Cultural Complexity.” Pp. 203–42 in
Finding Culture in Talk: A Collection of Methods, Naomi Quinn (ed.) New York:
Palgrave Macmillan.
Swidler, Ann. 1986. “Culture in Action: Symbols and Strategies.” American Sociological
Review 51 (2): 273–86.
———. 2013. Talk of Love: How Culture Matters. University of Chicago Press.
Talmy, Leonard. 1988. “Force Dynamics in Language and Cognition.” Cognitive Science
12 (1): 49–100.
Tarde, Gabriel. 2010. Gabriel Tarde On Communication and Social Influence: Selected
Papers. University of Chicago Press.
Thibodeau, Paul H., Mira J. Fein, Elizabeth S. Goodbody, and Stephen J. Flusberg. 2015.
“The Depression Schema: How Labels, Features, and Causal Explanations Affect
Lay Conceptions of Depression.” Frontiers in Psychology 6 (November): 1728.
Thibodeau, Paul H., Rose K. Hendricks, and Lera Boroditsky. 2017. “How Linguistic
Metaphor Scaffolds Reasoning.” Trends in Cognitive Sciences 21 (11): 852–63.
115
Tichi, Cecelia. 1992. Electronic Hearth: Creating an American Television Culture.
Oxford University Press.
Timmermans, Stefan, and Iddo Tavory. 2012. “Theory Construction in Qualitative
Research: From Grounded Theory to Abductive Analysis.” Sociological Theory
30 (3): 167–86.
Todorov, Alexander, Amir Goren, and Yaacov Trope. 2007. “Probability as a
Psychological Distance: Construal and Preferences.” Journal of Experimental
Social Psychology 43 (3): 473–82.
Trope, Yaacov, and Nira Liberman. 2003. “Temporal Construal.” Psychological Review
110 (3): 403–21.
———. 2010. “Construal-Level Theory of Psychological Distance.” Psychological
Review 117 (2): 440–63.
Turner, Stephen P. 2018. Cognitive Science and the Social: A Primer. Routledge.
Undie, Chi-Chi, Joanna Crichton, and Eliya Zulu. 2007. “Metaphors We Love by:
Conceptualizations of Sex among Young People in Malawi.” African Journal of
Reproductive Health 11 (3): 221.
Vaisey, S., and O. Lizardo. 2010. “Can Cultural Worldviews Influence Network
Composition?” Social Forces 88(4): 1595–1618.
Vaisey, Stephen. 2008. “Socrates, Skinner, and Aristotle: Three Ways of Thinking About
Culture in Action1.” Sociological Forum 23 (3): 603–13.
———. 2009. “Motivation and Justification: A Dual-Process Model of Culture in
Action.” American Journal of Sociology 114 (6): 1675–1715.
———. 2014. “Is Interviewing Compatible with the Dual-Process Model of Culture.”
American Journal of Cultural Sociology 2 (1): 150–58.
Voas, David, and Mark Chaves. 2016. “Is the United States a Counterexample to the
Secularization Thesis?” American Journal of Sociology 121 (5): 1517–56.
Wacquant, Loïc. 2004. Body & Soul. Oxford University Press Oxford.
Wakslak, Cheryl J., Yaacov Trope, Nira Liberman, and Rotem Alony. 2006. “Seeing the
Forest When Entry Is Unlikely: Probability and the Mental Representation of
Events.” Journal of Experimental Psychology. General 135 (4): 641–53.
Warner, R. Stephen. 1978. “Toward a Redefinition of Action Theory: Paying the
Cognitive Element Its Due.” American Journal of Sociology 83 (6): 1317–49.
116
Williams, Joan C., Mary Blair-Loy, and Jennifer L. Berdahl. 2013. “Cultural Schemas,
Social Class, and the Flexibility Stigma.” The Journal of Social Issues 69 (2):
209–34.
Wood, Michael Lee, Dustin S. Stoltz, Justin Van Ness, and Marshall A. Taylor. 2018.
“Schemas and Frames.” Sociological Theory 36 (3): 244–61.
Wuthnow, Robert J. 2011. “Taking Talk Seriously: Religious Discourse as Social
Practice.” Journal for the Scientific Study of Religion 50 (1): 1–21.
Young, Lori, and Stuart Soroka. 2012. “Affective News: The Automated Coding of
Sentiment in Political Texts.” Political Communication 29 (2): 205–31.
Zerubavel, Eviatar. 1997. Social Mindscapes: An Introduction to Cognitive Sociology.
Cambridge, MA: Harvard University Press.
