Accounting, Organizations and Society 35 (2010) 431–443

Contents lists available at ScienceDirect

Accounting, Organizations and Society

journal homepage: www.elsevier.com/locate/aos

The language of US corporate environmental disclosure

Charles H. Cho a,*, Robin W. Roberts b,1, Dennis M. Patten c,2
a
John Molson School of Business, Concordia University, 1455, de Maisonneuve Blvd. West, Montréal, Québec, Canada H3G 1M8
b
Kenneth G. Dixon School of Accounting, University of Central Florida, 4000 Central Florida Blvd., Orlando, FL 32816-1400, United States
c
Department of Accounting – 5520, Illinois State University, Normal, IL 61761-5520, United States

a b s t r a c t

We rely on prior work in environmental disclosure and corporate impression management

to investigate whether there are self-serving biases present in the language and verbal tone
used in corporations’ environmental disclosures. Specifically, we argue that the degree of
bias in these narratives varies systematically based on firm environmental performance,
hypothesizing that disclosures of worse environmental performers exhibit significantly
more ‘‘optimism” and less ‘‘certainty” than their better-performing counterparts. We test
our two hypotheses using a cross-sectional sample of corporate environmental disclosures
contained in US 10-K annual reports. Utilizing the content analysis software DICTION to
determine ‘‘optimism” and ‘‘certainty” scores for the disclosures, we find empirical support
for both hypotheses. Our study contributes significantly to research in environmental dis-
closure by investigating bias in the use of language and verbal tone as a tool for managing
stakeholder impressions and by finding empirical support for this role. Thus, the language
and verbal tone used in corporate environmental disclosures, in addition to their amount
and thematic content, should be considered when investigating the relation between cor-
porate disclosure and performance.
Ó 2009 Elsevier Ltd. All rights reserved.

Introduction narratives” (Jones & Shoemaker, 1994). Neu et al. (1998)

Corporate environmental information is increasingly
desired by firm stakeholders (Berthelot, Cormier, & Mag-
nan, 2003; Cormier, Gordon, & Magnan, 2004; Epstein &
Freedman, 1994) and is material to their decision-making
(Cho, Phillips, Hageman, & Patten, 2009; Milne & Patten,
2002; Neu, Warsame, & Pedwell, 1998). Because environ-
mental disclosures included in corporate accounting re-
ports (e.g., 10-K report or annual report) generally
provide relatively more verbal as opposed to numerical
information, they tend to be described as ‘‘accounting
* Corresponding author. Tel.: +1 514 848 2424x2319; fax: +1 514 848
4518.
E-mail addresses: ccho@jmsb.concordia.ca (C.H. Cho), rroberts@bus.
ucf.edu (R.W. Roberts), dmpatte@ilstu.edu (D.M. Patten).
1
Tel.: +1 407 823 6726; fax: +1 407 823 3881.
2
Tel.: +1 309 438 7857; fax: +1 309 438 8431.
argue that management often prefers accounting narra-
tives such as environmental disclosures rather than finan-
cial or other quantifiable information because disclosures
can be deliberately tailored to manage public impressions,
and, as noted by Hopwood (2009, p. 437) can be used to
‘‘increase [the company’s] legitimacy in the wider world”
and ‘‘facilitate the construction of a new and different im-
age of the company.” Thus, corporate management can use
environmental disclosures as an impression management
tool by self-servingly biasing the narrative through deci-
sions on the amount of information (quantity), the range
of topics (thematic content), and the rhetorical devices
(language and verbal tone) to be included in their disclo-
sures (Merkl-Davies & Brennan, 2007).
Based on an extensive review of research on accounting
narrative disclosures, Merkl-Davies and Brennan (2007)
developed a framework for analyzing corporate impression
management strategies. Their framework centers on the
premise that managerial self-serving motives drive

0361-3682/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2009.10.002
432 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443
attempts to bias narrative disclosures by either concealing
information about firm performance or through influenc-
ing attributions made regarding responsibility for firm per-
formance. According to Merkl-Davies and Brennan’s
impression management framework, corporate attempts
at concealment and attribution focus primarily on manip-
ulating either the presentation (language and verbal tone)
or the disclosure (quantity, thematic content, and attribu-
tion) of the information that is to be provided.
Numerous studies have examined bias in environmen-
tal disclosures. Almost all of these investigations, however,
focus either on the quantity of disclosure (e.g., Neu et al.,
1998; Patten, 1992) or the disclosures’ thematic content
(e.g., Cho & Patten, 2007; Hughes, Anderson, & Golden,
2001; Patten, 2002). Merkl-Davies and Brennan (2007)
point out that management’s ability to determine (and
bias) language use and verbal tone when crafting corporate
report disclosures also provides a powerful tool for manag-
ing impressions. This dimension of environmental disclo-
sure has not yet been examined.
The purpose of this study, then, is to draw from prior re-
search in environmental disclosure and corporate impres-
sion management to investigate whether biased language
and verbal tone is present in corporations’ environmental
disclosures. We employ findings from prior research to
support our argument that these types of narrative choices
vary systematically based on firm environmental perfor-
mance. Further, we rely specifically on the Merkl-Davies
and Brennan (2007) impression management framework
to develop our hypotheses. We expect disclosures of worse
environmental performers to exhibit significantly more
‘‘optimism” (H1) and significantly less ‘‘certainty” (H2)
than their better-performing counterparts.
We test our two hypotheses using a cross-sectional
sample of corporate environmental disclosures contained
in US 10-K financial reports. Utilizing the content analysis
software DICTION to determine ‘‘optimism” and ‘‘certainty”
scores for the disclosures, and measuring environmental
performance using environmental concern ratings from
KLD Research and Associates, Inc., we find empirical sup-
port for both hypotheses. Our study contributes signifi-
cantly to research in environmental disclosure by
investigating self-serving bias in the use of language and
verbal tone as a tool for managing stakeholder impressions
and by finding empirical support for this role. Thus, we
conclude that the language and verbal tone used in corpo-
rate environmental disclosures, in addition to their amount
and thematic content, must be considered when investi-
gating the relation between corporate disclosure and
performance.
The remainder of the paper is organized as follows.
The following section includes the conceptual develop-
ment and hypotheses to be tested in the study. The
subsequent section explains the methods and analysis,
including a discussion of the DICTION content analysis
software that was used to generate language and verbal
tone measures for ‘‘optimism” and ‘‘certainty”. The fourth
section presents the results. Conclusions, limitations, and
future research opportunities are discussed in the last
section of the paper.
Conceptual development and hypotheses
Managing impressions through biased language
Organizations adopt a number of specific impression
management strategies to attempt to maintain or enhance
their image (see, e.g., Elsbach, 1994; Elsbach & Sutton,
1992; Ginzel, Kramer, & Sutton, 1992; Livesey & Kearins,
2002). For example, several studies of annual report narra-
tives (e.g., Smith & Taffler, 2000; Sydserff & Weetman,
2002) conclude that corporations use certain language
characteristics to try to positively shape stakeholder per-
ceptions of the firm. Furthermore, some researchers claim
that managers strategically and consciously decide to pro-
vide narrative disclosures designed to manipulate the per-
ceptions and decisions of stakeholders (Clatworthy &
Jones, 2001; Yuthas, Rogers, & Dillard, 2002).
We argue that a consequence of an impression manage-
ment strategy is a resulting bias in the language and verbal
tone employed in the disclosure narrative. Aerts (1994),
drawing on the work of Schlenker (1980), examined the
presence of accounting bias in annual reports from an
impression management perspective. He argued that man-
aging impressions can be effectively accomplished with
the use of a biased accounting language because explana-
tions of organizational events and performance in annual
reports are ‘‘not simply the outcome of a straightforward
data analysis process” (1994, p. 337).
Accounting disclosures can be systematically biased be-
cause of their rationalizing capacities, their ability to avoid
responsibility assignments and their inherent ambiguity,
which constitute ‘‘interrelated performative characteristics
of accounting explanations which make them particularly
apt to confront and appease a negative performance envi-
ronment” (Aerts, 1994, p. 341). As Aerts (1994) suggests,
the coping strategies people express in their behavior can
become visible through systematic biases in their explana-
tions. In other words, the more firm performance differs
from a desired benchmark, the more management is moti-
vated to manage impressions, and the more likely it is that
narrative disclosures will be affected by a self-serving bias.
Based on their extensive review of the impression man-
agement literature, Merkl-Davies and Brennan (2007) pro-
posed a framework that divides corporate impression
management strategies into two broad categories; con-
cealment and attribution. These categories are consistent
with the proposition that disclosers can introduce bias into
the communication process through the use of inherent
ambiguity, responsibility avoidance, or both. According to
Merkl-Davies and Brennan (2007), disclosers accomplish
concealment by emphasizing good news and obfuscating
bad news. When developing disclosures, management em-
ploys thematic manipulation to emphasize good news. In
other words, they exhibit a self-serving bias in their selec-
tion of firm performance items to disclose and their disclo-
sure includes more positive than negative keywords
(Abrahamson & Park, 1994; Matsumoto, Pronk, &
Roelofsen, 2006; Rutherford, 2003). In contrast, manage-
ment employs rhetorical manipulation to obfuscate bad
news. This is accomplished through writing disclosures
 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443 433

Impression Management Strategies

Required Environmental Disclosure

Managerial
impression 1. Concealment 2. Attribution
management
behavior

Impression Emphasis on good Obfuscation of bad Framing performance to

management news news mitigate blame
objective

Type of Verbal Verbal Verbal

information

Type of
Thematic Rhetorical Attribution of performance
manipulation

Focus of Bias of themes Persuasive

Internal vs. External attributions
manipulation language

Focus of Positive/negative Rhetorical features Performance attributions

analysis keywords

Hypothesized disclosure More “optimism” (H1) Less “certainty” (H2) More “optimism”, less “certainty”
expectations for (H1 and H2)
worse environmental
performers

Adapted from Merkl-Davies and Brennan (2007)

Fig. 1. Impression management strategies required environmental disclosure.

using more persuasive language and including other types
of rhetorical devices within the narrative (Sydserff & Weet-
man, 2002; Yuthas et al., 2002). This description of impres-
sion management through concealment mirrors Henry’s
(2008) conjecture that corporate communications set a po-
sitive tone when possible and introduce verbal complexity
in order to obfuscate adverse information.
Merkl-Davies and Brennan (2007, p. 126) define attribu-
tion as a ‘‘defensive framing tactic that shifts the blame for
negative outcomes away from themselves,” and in a corpo-
rate reporting context, ‘‘entails managers attributing posi-
tive organizational outcomes to internal factors
(‘‘entitlements”) and negative organizational outcomes to
external factors (‘‘excuses”).” Both Aerts (1994) and Clat-
worthy and Jones (2003), in their respective studies of cor-
porate reports, found that corporations tend to attribute
positive organizational outcomes to internal factors and
negative organizational outcomes to external factors. Aerts
(1994) also found that negative performances are more
likely to be discussed in technical accounting terms while
positive performances are presented in more straightfor-
ward, cause–effect language. The language that managers
choose to use when making corporate disclosures can also
clarify or obscure performance attributions. Language used
in corporate narratives has been found to blur attributions
(Thomas, 1997) and blur distinctions about the reasons for
negative firm performance (Jameson, 2000), perhaps by
using convoluted language (Jones, 1996). In framing the
performance attribution, corporate managers may deliber-
ately use ambivalent language in the disclosures in order
to present the company in a more positive light (Jameson,
2000).
Hypotheses for environmental disclosure
Merkl-Davies and Brennan (2007) provide a compre-
hensive schema for analyzing managerial impression man-
agement strategies in corporate narrative documents (see
p. 128). We adapt their schema to focus specifically on
how managerial impression management strategies may
be revealed through biases in the language and verbal tone
of corporate environmental disclosures, concentrating on
evidence of concealment and attribution. Our expectations
regarding corporate impression management strategies
and narrative disclosures by corporations who are worse
environmental performers are consistent with the more
434 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443
general expectations posited by Merkl-Davies and Brennan
(2007) and are presented in Fig. 1.
As shown in Fig. 1, corporate disclosures are expected to
emphasize good news by biasing narrative themes through
the use of positive keywords. Also, disclosures are expected
to attribute positive environmental performance to internal
attributions regarding corporate responsibility. The bias to-
ward reporting good news coupled with a strategy that
attributes positive performance to internal, corporate ef-
forts, draws on the notion of ‘‘optimism” as it is used in
the content analysis program DICTION. Management’s con-
centration on reporting good news sanitizes the disclosure
(Henry, 2008). Also, optimistic language can blur distinc-
tions about the corporation’s responsibility for poor perfor-
mance and attempts to present the corporation in a more
positive light (Jameson, 2000). In this context, ‘‘optimism”
refers to a language ‘‘endorsing some person, group, con-
cept, or event, or highlighting their positive entailments”
(Hart, 2001, p. 247). In Panel A of Table 1 we provide exam-
ples of corporate environmental disclosures that received
high and low DICTION ‘‘optimism” scores. We expect that
corporations that are worse environmental performers will
attempt to manage stakeholder impressions by evoking
more optimistic language in their 10-K environmental dis-
closures. We state hypothesis H1 as follows:
H1. The ‘‘optimism” exhibited in 10-K report environmen-
tal disclosures will be negatively related to firm environ-
mental performance.
As also depicted in Fig. 1, if corporate management is
implementing an impression management strategy, their
disclosures are expected to obfuscate bad news regarding
environmental performance and assign external attribu-
tions to poor environmental performance. Obfuscation is
accomplished through rhetorical plays by invoking persua-
sive language (Merkl-Davies & Brennan, 2007), weaker lan-
guage (Subramanian, Insley, & Blackwell, 1993), or using
verbal complexity to reduce stakeholders’ abilities to
clearly interpret the environmental disclosure (Henry,
2008). Framing poor corporate performance as the result
of external factors mitigates blame that stakeholders may
place on management (Thomas, 1997). Negative corporate
performance may be reported using convoluted, less cer-
tain language (Jones, 1996) in order to mask internal attri-
butions. If corporations use these techniques when
disclosing environmental performance, their disclosure
strategy reflects purposive impression management
behavior (Aerts, 2005). As used in the DICTION program,
the term ‘‘certainty” refers to language that indicates ‘‘res-
oluteness, inflexibility, completeness, and a tendency to
speak ex cathedra” (Hart, 2001, p. 246). It follows that lan-
guage designed to obfuscate bad news or mask internal
attribution would be flexible, irresolute, weak, and/or ten-
tative. In Panel B of Table 1 we provide examples of corpo-
rate environmental disclosures that received high and low
DICTION ‘‘certainty” scores.3 We expect that corporations
3
As we discuss below, we follow Demers and Vega (2009) and calculate
our ‘‘certainty” scores treating numerical terms as a positive, rather than
negative adjustment to the overall score. Our examples are based on this
adjusted ‘‘certainty” measure.
that are worse environmental performers will attempt to
manage stakeholder impressions by using less certain,
more obfuscating language in their 10-K environmental
disclosures, while better performers will invoke more ‘‘cer-
tainty” in the language and verbal tone of their disclosures.
We state hypothesis H2 as follows:
H2. The ‘‘certainty” exhibited in 10-K report environmen-
tal disclosures will be positively related to firm environ-
mental performance.
Methods and analysis
In order to test our hypotheses we obtained a represen-
tative sample of US 10-K environmental disclosures, devel-
oped empirical proxies for the ‘‘optimism” and ‘‘certainty”
constructs using DICTION, and designed an empirical mod-
el that controls for other important factors that may be re-
lated to a corporation’s use of optimistic or certain
language in its environmental disclosure. The details of
these procedures are described below.
Sample selection
To be included in this study, sample firms had to meet
the following criteria:
1. They had to be listed in the 2002 ratings of corporate
social and environmental performance compiled by
KLD Research and Analytics, Inc. (hereafter, ‘‘KLD”).
2. They had to have a fiscal year ended from June 30, 2002
to December 31, 2002, inclusive.4
3. They had to be listed on the Standards and Poor’s 500
Index for fiscal year 2002.
4. They had to have a 2002 10-K report available on the US
Securities and Exchange Commission (SEC)’s EDGAR
database, and it had to include Section 1 environmental
disclosures.5
A total of 190 firms met all five criteria and constitute
the final sample. Of these, 43 are from environmentally
sensitive industries (see Table 2 for a more specific break-
down), and sample firms range in size (based on 2002 total
asset levels) from $786 million to $370,782 million, with a
mean (median) of $23,780 million ($11,090 million). Table
2 provides descriptive statistics on sample company data
and other variables used in our analysis.
Environmental disclosure
We identify environmental disclosures in the 10-K re-
ports for the same fiscal year as the year of assessed envi-
4
A fiscal year-end subsequent to June 30, 2002 better reflects and
matches 2002 firm environmental disclosure in relation to its 2002
environmental performance.
5
It is important to note that avoidance (i.e., providing no environmental
disclosure) is also viewed as an impression management and legitimacy
tool. However, we excluded firms using this tactic from the sample because
the focus of this study was to analyze the language of environmental
disclosures that are actually provided by firms.
 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443 435

Table 1
Examples of corporate environmental disclosures.

Panel A – Examples of disclosures with high and low ‘‘optimism” scores

High ‘‘optimism” score
Andrew Corporation, 10-K Report Section 1, September 30, 2002, ‘‘optimism” score = 51.36
Andrew is committed to demonstrating the highest standard of global environmental management and achieving environmental best practices. Six
locations have been awarded certifications for ISO 14001, an international standard for environmental management systems. The company is
committed to the continual improvement of its environmental management system and practices, including resource conservation and pollution
prevention. Andrew engages in a variety of activities to comply with various federal, state and local laws and regulations involving the protection
of the environment. Compliance with such laws and regulations does not currently have a significant effect on the company’s capital
expenditures, earnings, or competitive position. In addition, the company has no knowledge of any environmental condition that might
individually or in the aggregate have a material adverse effect on its financial condition
Low ‘‘optimism” score
Maxim Integrated Products, 10-K Report Section 1, June 30, 2002, ‘‘optimism” Score = 40.21
Federal, state, and local regulations impose a variety of environmental controls on the storage, handling, discharge and disposal of certain chemicals
and gases used in semiconductor manufacturing. The Company’s facilities have been designed to comply with these regulations, and it believes
that its activities are conducted in material compliance with such regulations. There can be no assurance, however, that interpretation and
enforcement of current or future environmental regulations will not impose costly requirements upon the Company. Any failure of the Company
to control adequately the storage, use, and disposal of regulated substances could result in future liabilities
Increasing public attention has been focused on the environmental impact of electronic manufacturing operations. While the Company to date has
not experienced any materially adverse effects on its business from environmental regulations, there can be no assurance that changes in such
regulations will not have a materially adverse effect on the Company’s financial position or results of operations
Panel B – Examples of disclosures with high and low ‘‘certainty” scores
High ‘‘certainty” score
ExxonMobil Corporation, 10-K Report Section 1, December 31, 2002, ‘‘certainty” score = 120.56
ExxonMobil’s worldwide environmental costs in 2002 totaled $2343 million of which $1054 million were capital expenditures and $1289 million
were operating costs (including $400 million of site restoration and environmental provisions). These costs were mostly associated with air and
water conservation. Total costs for such activities are expected to increase to about $2.5 million in both 2003 and 2004 (with capital expenditures
representing about 50 percent of the total). The projected increase is primarily for capital projects to implement refining technology to
manufacture low-sulfur motor fuels in many parts of the world
Low ‘‘certainty” score
Fluor Corporation, 10-K Report Section 1, December 31, 2002, ‘‘certainty” score = 14.61
We believe, based upon present information available to it, that our accruals with respect to future environmental costs are adequate and any future
costs will not have a material effect on our consolidated financial position, results of operations or liquidity. Some factors, however could result
in additional expenditures or the provision of additional accruals in expectation of such expenditures. These include the imposition of more
stringent requirements under environmental laws or regulations, new developments or changes regarding site cleanup costs or the allocation of
such costs among potentially responsible parties, or a determination that we are potentially responsible for the release of hazardous substances
at sites other than those currently identified
Past and future environmental, safety and health regulations could impose on us significant additional costs that reduce our profits
We are subject to numerous environmental laws and health and safety regulations. Our projects can involve the handling of hazardous and other
highly regulated materials which, if improperly handled or disposed of, could subject us to civil and criminal liabilities. It is impossible to reliably
predict the full nature and effect of judicial, legislative or regulatory developments relating to health and safety regulations and environmental
protection regulations applicable to our operations. The applicable regulations, as well as the technology and length of time available to comply
with those regulations, continue to develop and change. In addition, past activities could also have a material impact on us. For example, when
we sold our mining business formerly conducted through St. Joe Minerals Corporation, we retained responsibility for certain non-lead related
environmental liabilities, but only to the extent that such liabilities were not covered by St. Joe’s comprehensive general liability insurance.
While we are not currently aware of any material exposure arising from our former St. Joe’s business or otherwise, the costs of complying with
rulings and regulations or satisfying any environmental remediation requirements for which we are found responsible could be substantial and
could reduce our profits

ronmental performance (i.e., 2002) because we hypothe-
size direct disclosure/performance relationships (see, e.g.,
Patten, 2002) and examine how companies manage
impressions (i.e., measured by ‘‘optimism” and ‘‘certainty”)
in relation to environmental performance. While the SEC
requires disclosure of environmental information in Sec-
tion 1 (Description of Business), Section 3 (Legal Proceed-
ings) and Section 7 (Management’s Discussion and
Analysis of Financial Condition and Results of Operations),
we focus only on Section 1 disclosure. We exclude Section 3
environmental information because, as noted by Johnson
(1993), that area specifically requires disclosure of pending
or anticipated environmental legal issues and investiga-
tions. Nelson and Pritchard (2008) report that firms with
high litigation risk disclose information using more cau-
tionary language intended to reduce the expected costs
of litigation. Our concern is that the verbal tone driven
by this specific thematic content may confound the analy-
sis of other environmental disclosure.6 We exclude Sec-
6
Only 79 of the 190 firms in our sample had Section 3 disclosures. In a
non-tabulated sensitivity test we included a one/zero indicator variable
identifying these disclosers. Results on the relation between environmental
performance and both optimism and certainty remained consistent with
those reported in the primary analysis.
436 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443

Table 2
Descriptive statistics.

n (sample size) 190

Firms from environmentally sensitive industries
Oil and gas extraction 13
Chemicals (excluding pharmaceutical) 10
Paper 8
Primary metals 6
Petroleum refining 4
Metal mining 2
Total 43

Mean Median. Std. Dev. Maximum Minimum

Firm size (2002 assets, in millions) $23,780 $11,090 $49,830 $370,782 $786
Firm profitability (2002 ROA) 1.60% 2.90% 10.32% 19.81% 66.06%
Firm capital intensity (2002 assets/2002 revenues) 2.12 1.53 1.84 0.32 14.31
Firm age (as of 2002) 45.74 36.50 35.06 150 0
KLD environmental concern ratings 0.96 0.00 1.292 4 0
Environmental disclosure ‘‘optimism” score 48.21 48.58 3.06 63.78 37.13
Environmental disclosure ‘‘certainty” score 42.78 39.75 17.95 120.56 0

tion 7 disclosures because they either repeat or summarize
information disclosed in item 1 and/or item 3.
DICTION was developed by Hart, a communications re-
searcher, and focuses on the subtle power of word choice
and verbal tone (Hart, 1984). Similar to other content anal-
ysis software packages, DICTION relies on word frequency
counts, but stands out from others in a number of ways
that may be attractive for environmental accounting re-
search. First, the program relies on word counts based on
linguistic theory (see Bligh, Kohles, & Meindl, 2004). Sec-
ond, DICTION uses elements of artificial intelligence that
have been underutilized in the accounting literature. Third,
DICTION ‘‘falls within the scope of systemic linguistics, ren-
dering the approach attractive to accounting researchers
investigating impression management” (Sydserff & Weet-
man, 2002, p. 532). Fourth, the objectivity of DICTION is rel-
atively strong in relation to face validity and reliability due
to its automated nature (both for coding and quantifica-
tion) but, more importantly, the theoretical basis of the ap-
proach in linguistic semantics and its independently
attested establishment in the applied linguistics literature
(Sydserff & Weetman, 2002).
In addition to these unique benefits, DICTION allows the
flexibility of other software programs where users can cus-
tomize their own dictionaries (Short & Palmer, 2008; Syd-
serff & Weetman, 2002). DICTION deploys some 10,000
search words in 33 separate dictionaries that can be used
to analyze any given text (Short & Palmer, 2008). Based
in linguistic theory (see Bligh et al., 2004), the dictionaries
were constructed from the analysis of more than 20,000
texts, which yield a large total word corpus, and contain
no duplication (Short & Palmer, 2008; Sydserff & Weetman,
2002). The dictionaries are subsequently used for lexical
analysis (i.e., to study vocabulary and word choice)
through the analysis of five master variables: ‘‘certainty”,
‘‘optimism”, ‘‘activity”, ‘‘realism”, and ‘‘commonality”
(Hart, 2000, 2001; Short & Palmer, 2008; Sydserff & Weet-
man, 2002). DICTION’s master variables were developed
based on the rationale that ‘‘if only five questions could
be asked of a given passage, these five would provide the
most robust understanding” (Hart, 2001, p. 45). Because
DICTION can separately assess those five master variables,
it provides the ‘‘optimism” and ‘‘certainty” scores7 needed
for our hypotheses testing. A further distinctive feature of
DICTION is its use of normative values for comparative pur-
poses, including one called ‘‘corporate financial reports” that
is specifically tailored for corporate disclosures.8 Finally,
DICTION extrapolates each particular text to a 500-word
norm ‘‘equivalent” (which is the basic unit of analysis) so
that input texts of any length, such as the environmental
disclosures used in our study, can be controlled and mea-
sured consistently. Given the strengths of the DICTION pro-
gram, we use it to calculate scores of the ‘‘optimism” and
‘‘certainty” of our sample environmental disclosures.
Following Ober, Zhao, Davis, and Alexander (1999) and
Sydserff and Weetman (2002), we report the ‘‘optimism”
master variable scores without adjustment. However, as
argued by Demers and Vega (2009), DICTION’s choice to
classify numerical terms as a negative adjustment to ‘‘cer-
tainty”, while relevant in political speeches and other
forms of expository prose, is inappropriate for financial re-
port disclosures. Instead, they note that inclusion of ex
post verifiable quantitative information is actually indica-
tive of more direct and precise expression in this context,
and accordingly should be considered a positive, rather
7
The definitions of ‘‘optimism” and ‘‘certainty”, as used in DICTION and
provided by its developer (Hart) are given in the second section (‘‘Hypoth-
eses development for environmental disclosure”). The formulas for master
variables consist of subaltern variables. As such, the formula for ‘‘optimism”
is: [‘‘praise” + ‘‘satisfaction” + ‘‘insp iration”]– [‘‘blame” + ‘‘hard-
ship” + ‘‘denial”] while the one for ‘‘certainty” is: [‘‘tenacity” + ‘level-
ing” + ‘‘collectives” + ‘‘insistence”]–[‘‘numerical terms” + ‘‘ambiva-
lence” + ‘‘self-reference” + ‘‘variety”]. A typical DICTION output generally
includes the names of the variables (and subaltern variables), their
frequency, the percentage of words analyzed, the normal score range, the
standard range and whether they are out of range. DICTION is also able to
make the conversion, transfer, import and export of texts and data with
other computer programs (e.g., Excel, SPSS) easy and user-friendly (Hart,
2000, 2001).
8
We conducted the empirical tests under both the generic ‘‘all cases”
and the ‘‘corporate financial reports” normative values and, for our
particular case and sample, the results were qualitatively the same for
both hypotheses.
 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443 437

Table 3
Pearson’s correlation matrix (n = 190).

Variable EDO EDC EP SIZE EXTR CHEM PAP MET PETR MIN CAPINT ROA AGE
EDO 1.000 0.089 0.247** 0.168* 0.013 0.118 0.008 0.036 0.120 0.027 0.061 0.084 0.042
EDC 1.000 0.023 0.036 0.125 0.150* 0.084 0.084 0.243** 0.031 0.236** 0.171 0.356**
EP 1.000 0.397** 0.059 0.066 0.078 0.129 0.304** 0.046 0.018 0.015 0.117
SIZE 1.000 0.087 0.152* 0.067 0.089 0.170* 0.043 0.408** 0.008 0.024
EXTR 1.000 0.064 0.057 0.049 0.040 0.028 0.114 0.048 0.058
CHEM 1.000 0.049 0.043 0.035 0.024 0.115 0.052 0.106
PAP 1.000 0.038 0.031 0.022 0.057 0.060 0.032
MET 1.000 0.026 0.019 0.036 0.088 0.081
PETR 1.000 0.015 0.100 0.013 0.138
MIN 1.000 0.050 0.007 0.007
CAPINT 1.000 0.057 0.247**
ROA 1.000 0.134
AGE 1.000

EDOi = the 2002 environmental disclosure ‘‘optimism” score for firm i.

EDCi = the 2002 environmental disclosure ‘‘certainty” score for firm i.
EPi = the 2002 KLD environmental concern score for firm i.
SIZEi = the natural log of 2002 assets for firm i.
EXTRi = one if firm i belongs to the oil and gas extraction industry, and zero otherwise.
CHEMi = one if firm i belongs to the chemicals industry, and zero otherwise.
PAPi = one if firm i belongs to the paper industry, and zero otherwise.
METi = one if firm i belongs to the primary metals industry, and zero otherwise.
PETRi = one if firm i belongs to the petroleum refining industry, and zero otherwise.
MINi = one if firm i belongs to the metal mining industry, and zero otherwise.
CAPINTi = the 2002 capital intensity for firm i.
ROAi = the 2002 return on assets for firm i.
AGEi = the age of firm i as of 2002.
*
p < 0.05 (two-tailed).
**
p < 0.01 (two-tailed).

than a negative component of ‘‘certainty.” As such, we fol-
low Demers and Vega (2009) and compute an adjusted
‘‘certainty” score whereby the numerical terms subcompo-
nent is treated as a positive addition to the formula. For our
sample firms, environmental disclosure ‘‘optimism” scores
range from 37.13 to 63.78 and ‘‘certainty” scores ranged
from 0 to 120.56. The sample mean (median) disclosure
scores were 48.21 (48.58) and 42.78 (39.75) for ‘‘opti-
mism” and ‘‘certainty”, respectively.9
Environmental performance
Although a number of external corporate social and
environmental performance (SEP) evaluations have been
published over the past 30 years, most have been limited
to a relatively small number of companies, or have focused
on only smaller subsets of performance. In response to this
need, the independent ratings firm KLD10 has, since 1994,
9 ronmental disclosure, we use this database to identify
Applying the Demers and Vega (2009) adjustment to the certainty
scores resulted in one company with a negative value for this measure.
Because we interpreted this as evidence of little or no indication of
certainty language use we coded this observation as a zero. As a sensitivity
check, we reran all tests (1) using the negative certainty value for the firm,
and (2) deleting this company from the sample. Results in both cases were
qualitatively unchanged.
10
The professional services firm of KLD Research and Analytics, Inc. is
located at 250 Summer Street, Boston, MA 02210, USA. KLD’s social
research is distributed in SOCRATES - The Corporate Social Ratings
MonitorSM. SOCRATES is a proprietary database program that provides
access to KLD’s ratings and other data pertaining to the social records of
over 3000 publicly traded US companies (KLD Research and Analytics, Inc.,
2003).
maintained a database that appears to overcome these prob-
lems. KLD independently rates hundreds of companies
traded on US stock exchanges in terms of their social perfor-
mance across a range of dimensions related to stakeholder
concerns. The company draws upon a variety of sources to
capture relevant social performance data (Hillman & Keim,
2001; Waddock & Graves, 1997). Because the KLD database
provides a quantifiable and enhanced corporate SEP mea-
sure and preserves its independent rating system (Hillman
& Keim, 2001), the KLD data have been used extensively in
US management research on corporate social performance
issues (e.g., Waddock & Graves, 1997) and also used recently
in environmental accounting research (see, e.g., Cho & Pat-
ten, 2007; Cho, Patten, & Roberts, 2006). KLD separately as-
signs strengths and concerns across seven SEP categories11
and gives a score of zero or one for each of the strength
and concern areas included in each category.
Given the apparent benefits of the KLD ratings and be-
cause our focus is on examining the relation between firm
environmental performance and the extent of bias of envi-
environmental performance for our sample companies.
KLD analyzes corporate environmental performance based
on an extensive assessment of each firm’s environmental
management, planning and impact assessment, utilization
of resources, compliance with applicable laws and regula-
11
KLD’s social responsibility categories include community, corporate
governance, diversity, employee relations, environment, human rights, and
product (KLD Research and Analytics, Inc., 2003).
438 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443

Table 4
Results of OLS regression analysis testing the relation between environmental performance and environmental disclosure ‘‘optimism”
(EDOi = a1 + B1EPi + B2SIZEi + B3EXTRi + B4CHEMi + B5PAPi + B6METi + B7PETRi + B8MINi + B9CAPINTi + B10ROAi + B11AGEi).

Model explanatory power

Number of observations 190
Model F-statistic 1.869
Significance of F-statistic 0.046
Adjusted R-squared 0.048

Variable Predicted sign Parameter estimate t-Stat. Sig.* VIF

Parameter estimates
INTERCEPT None 43.914 21.443 0.000
EP (+) 0.399 1.912 0.029 1.395
SIZE (+/ ) 0.466 1.955 0.052 1.647
EXTR (+/ ) 0.491 0.557 0.578 1.055
CHEM (+/ ) 1.739 1.725 0.086 1.082
PAP (+/ ) 0.258 0.237 0.813 1.024
MET (+/ ) 0.803 0.624 0.534 1.083
PETR (+/ ) 0.869 0.538 0.591 1.148
MIN (+/ ) 0.435 0.203 0.840 1.022
CAPINT (+/ ) 0.209 1.497 0.136 1.403
ROA (+/ ) 0.024 1.109 0.269 1.035
AGE (+/ ) 0.004 0.584 0.560 1.138

EDOi = the 2002 environmental disclosure ‘‘optimism” score for firm i.

EPi = the 2002 KLD environmental concern score for firm i.
SIZEi = the natural log of 2002 assets for firm i.
EXTRi = one if firm i belongs to the oil and gas extraction industry, and zero otherwise.
CHEMi = one if firm i belongs to the chemicals industry, and zero otherwise.
PAPi = one if firm i belongs to the paper industry, and zero otherwise.
METi = one if firm i belongs to the primary metals industry, and zero otherwise.
PETRi = one if firm i belongs to the petroleum refining industry, and zero otherwise.
MINi = one if firm i belongs to the metal mining industry, and zero otherwise.
CAPINTi = the 2002 capital intensity for firm i.
ROAi = the 2002 return on assets for firm i.
AGEi = the age of firm i as of 2002.
*
Significance levels are based on a one-tailed test for the EP variable, and two-tailed for all control variables.

tions, and emissions. Concern ratings12 are generally as-
signed to companies that (1) reveal poor compliance records
with environmental laws and regulations; (2) emit hazard-
ous or toxic substances and waste in large quantities; (3) fall
behind their industry competitors in implementing preven-
tive measures to reduce environmental impact; and/or (4)
generate a significant portion of their revenues from prod-
ucts or services that negatively affect the environment
(KLD Research and Analytics, Inc., 2003). Scores for our sam-
ple firms ranged from 0 (i.e., no environmental concern) to 4
(i.e., relatively high environmental concern) with a mean
score of 0.96.13
Control variables
Prior research documents that a number of firm-specific
factors appear to influence the extent and/or thematic con-
tent of environmental disclosure provided in corporate
financial reports. In order to more carefully identify
whether differences in the language and verbal tone of
these disclosures are associated with differences in envi-
ronmental performance we control for these additional
factors. More specifically, we include control variables for
firm size, environmentally sensitive industry membership,
capital intensity, return on assets (ROA), and company age.
Each is discussed in more detail below.

12
KLD concern ratings are: (1) hazardous waste (liabilities for hazardous Firm size
waste sites exceeds $50 million), (2) regulatory problems (payment of Almost all prior studies of environmental disclosure
substantial fines or civil penalties for violations of environmental regula-
tions), (3) ozone depleting chemicals (top manufacturer of ozone depleting
indicate the extent of disclosure is significantly related to
chemicals such as HCFCs, methyl chloroform, methylene chloride, or firm size (see, e.g., Cho & Patten, 2007; Patten, 1992,
bromines), (4) substantial emissions (legal emissions of toxic chemicals 2002). Larger companies (presumably due to higher visibil-
from individual plants into the air and water are among the highest), (5) ity) tend to disclose more extensive environmental infor-
agricultural chemicals (substantial producer of agricultural chemicals, i.e.,
mation. We control for firm size using the natural log of
pesticides or chemical fertilizers), (6) climate change (derives substantial
revenues from the sale of coal or oil and its derivative fuel products, and (7) each company’s 2002 total assets.
other concern (other environmental problem such as environmental
accident (KLD Research and Analytics, Inc., 2003). Environmentally sensitive industry membership
13
A substantial number of our sample companies (n = 102) have KLD Numerous studies (e.g., Aerts & Cormier, 2009; Cho &
scores of zero. To assure that the results we report are not driven by this
sample characteristic, we include as a sensitivity test an estimation of our
Patten, 2007; Hackston & Milne, 1996; Patten, 2002) also
test models using only the 88 firms with KLD concern scores greater than document that companies in industries whose processes
zero. We report those results later in the paper. place greater stress on the natural environment appear to
 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443 439

Table 5
Results of OLS regression analysis testing the relation between environmental performance and environmental disclosure ‘‘certainty”
(EDCi = a1 + B1EPi + B2SIZEi + B3EXTRi + B4CHEMi + B5PAPi + B6METi + B7PETRi + B8MINi + B9CAPINTi + B10ROAi + B11AGEi).

Model explanatory power

Number of observations 190
Model F-statistic 5.425
Significance of F-statistic 0.000
Adjusted R-squared 0.205

Variable Predicted sign Parameter estimate t-Stat. Sig.* VIF

Parameter estimates
INTERCEPT None 17.876 1.627 0.106
EP ( ) 2.293 2.050 0.021 1.395
SIZE (+/ ) 2.482 1.941 0.054 1.647
EXTR (+/ ) 3.655 0.774 0.440 1.055
CHEM (+/ ) 11.556 2.137 0.034 1.082
PAP (+/ ) 6.441 1.101 0.272 1.024
MET (+/ ) 11.800 1.708 0.089 1.083
PETR (+/ ) 27.087 3.125 0.002 1.148
MIN (+/ ) 0.223 0.019 0.985 1.022
CAPINT (+/ ) 1.732 2.312 0.022 1.403
ROA (+/ ) 0.216 1.884 0.061 1.035
AGE (+/ ) 0.128 3.613 0.000 1.138

EDCi = the 2002 environmental disclosure ‘‘certainty” score for firm i.

EPi = the 2002 KLD environmental concern score for firm i.
SIZEi = the natural log of 2002 assets for firm i.
EXTRi = one if firm i belongs to the oil and gas extraction industry, and zero otherwise.
CHEMi = one if firm i belongs to the chemicals industry, and zero otherwise.
PAPi = one if firm i belongs to the paper industry, and zero otherwise.
METi = one if firm i belongs to the primary metals industry, and zero otherwise.
PETRi = one if firm i belongs to the petroleum refining industry, and zero otherwise.
MINi = one if firm i belongs to the metal mining industry, and zero otherwise.
CAPINTi = the 2002 capital intensity for firm i.
ROAi = the 2002 return on assets for firm i.
AGEi = the age of firm i as of 2002.
*
Significance levels are based on a one-tailed test for the EP variable, and two-tailed for all control variables.

systematically provide more extensive environmental dis-
closures. Similar to Patten (2002) and Cho and Patten
(2007) we classify the chemical (primary SIC code 28xx,
excluding pharmaceutical, code 283x), metals (33xx), min-
ing (10xx), oil exploration (13xx), paper (26xx), and petro-
leum (2911) industries as ‘‘environmentally sensitive.”
However, consistent with Cowen, Fererri, and Parker
(1987) we allow for differing environmental industry im-
pacts by including separate one/zero indicator variables
to separate firms operating in each specific environmen-
tally sensitive industry from other sample firms.14
vided by year 2002 revenues,15 profitability as year 2002
return on assets (ROA), and company age as of year 2002.
All data is gathered from COMPUSTAT.
Statistical analysis
We use ordinary least squares multiple regression to
identify the relation between environmental performance
and the different characteristics of firm environmental dis-
closure, controlling for other factors potentially influencing
language choice. Our models are stated as:

Capital intensity, profitability, and company age
Although not as consistently documented as firm size
and industry affiliation, at least in some cases, capital
intensity (e.g., Aerts & Cormier, 2009; Clarkson, Li, Richard-
son, & Vasvari, 2008; Reitenga, 2000), profitability (Bewley
& Li, 2000; Clarkson et al., 2008; Magness, 2006), and com-
pany age (Roberts, 1992) have been shown to be signifi-
cantly associated with environmental disclosure, and as
such we include these as control variables in our analysis.
We measure capital intensity as year 2002 total assets di-
(1) EDOi = a1 + B1EPi + B2SIZEi + B3EXTRi + B4CHEMi
+ B5PAPi + B6METi + B7PETRi + B8MINi + B9CAPINTi
+ B10ROAi + B11AGEi.
(2) EDC = a1 + B1EPi + B2SIZEi + B3EXTRi + B4CHEMi
+ B5PAPi + B6METi + B7PETRi + B8MINi + B9CAPINTi
+ B10ROAi + B11AGEi.
Because firms with worse environmental performance
have a higher KLD environmental concern score, we expect
to find a positive numeric relation between the EP and EDO

14
We repeat all analyses using a single environmentally sensitive
industry indicator variable (not differentiating across specific industries).
Results on the relation between environmental performance and the
‘‘optimism” and ‘‘certainty” of the disclosure language are qualitatively
similar to those reported for the separate industry models.
15
We consider this measure of capital intensity to be more appropriate
because five of our sample firms do not report a separate property, plant
and equipment (PPE) amount. Results using the more traditional PPE/total
assets capital intensity measure (with the five firms deleted) are qualita-
tively similar to those reported in the paper.
440 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443

Table 6
Results of OLS regression analysis testing the relation between environmental performance and environmental disclosure ‘‘optimism” for firms having at least
one environmental concern (EDOi = a1 + B1EPi + B2SIZEi + B3EXTRi + B4CHEMi + B5PAPi + B6METi + B7PETRi + B8MINi + B9CAPINTi + B10ROAi + B11AGEi).

Model explanatory power

Number of observations 88
Model F-statistic 1.922
Significance of F-statistic 0.049
Adjusted R-squared 0.104

Variable Predicted sign Parameter estimate t-Stat. Sig.* VIF

Parameter estimates
INTERCEPT None 48.483 17.530 0.000
EP (+) 0.718 2.237 0.014 1.434
SIZE (+/ ) 0.092 0.296 0.768 1.596
EXTR (+/ ) 1.521 1.381 0.171 1.128
CHEM (+/ ) 2.786 2.492 0.086 1.163
PAP (+/ ) 1.100 0.747 0.458 1.045
MET (+/ ) 0.509 0.376 0.708 1.164
PETR (+/ ) 1.340 0.968 0.336 1.218
MIN (+/ ) 0.809 0.450 0.654 1.054
CAPINT (+/ ) 0.068 0.315 0.754 1.361
ROA (+/ ) 0.051 0.991 0.325 1.131
AGE (+/ ) 0.009 1.252 0.214 1.147

EDOi = the 2002 environmental disclosure ‘‘optimism” score for firm i.

EPi = the 2002 KLD environmental concern score for firm i.
SIZEi = the natural log of 2002 assets for firm i,
EXTRi = one if firm i belongs to the oil and gas extraction industry, and zero otherwise.
CHEMi = one if firm i belongs to the chemicals industry, and zero otherwise.
PAPi = one if firm i belongs to the paper industry, and zero otherwise.
METi = one if firm i belongs to the primary metals industry, and zero otherwise.
PETRi = one if firm i belongs to the petroleum refining industry, and zero otherwise.
MINi = one if firm i belongs to the metal mining industry, and zero otherwise.
CAPINTi = the 2002 capital intensity for firm i.
ROAi = the 2002 return on assets for firm i.
AGEi = the age of firm i as of 2002.
*
Significance levels are based on a one-tailed test for the EP variable, and two-tailed for all control variables.

variables, and we predict a negative numeric relation be-
tween the EP variable and the EDC variable. Table 3 reports
Pearson product-moment correlation values for all of our
model variables. To assure that multicollinearity is not a
problem, we calculate and report variance inflation factors
for all independent variables.
Results
Primary tests
Table 4 presents the results of the regression analyses
testing the ‘‘optimism” hypothesis (H1). As highlighted in
the table, the model is significant (based on the model F-
statistic) and variance inflation factors are well below the
threshold value of 3.0, indicating that multicollinearity
does not appear to be a problem. With the exception of
SIZE (p = .052, two-tailed) and CHEM (p = .086, two-tailed),
the control variables do not approach statistical signifi-
cance.16 More importantly, the firm environmental perfor-
mance measure is positively associated with the
‘‘optimism” level of environmental disclosure (EDO), and
is significant at the p = .029 level, one-tailed. This supports
the argument that poorer environmental performers use a
16
For all of the tests reported in this section, we repeat analyses using
only control variables exhibiting statistical significance (at p < .10, two-
tailed). In all cases, the significance of the EP variable remains qualitatively
similar to the level reported for the full model tests.
more optimistic language tone in the wording of their
environmental disclosures.
Table 5 presents test results of the ‘‘certainty” hypothe-
sis (H2). As noted in the table, this model is also significant
(based on the model F-statistic) and the variance inflation
factors are also well below the concern threshold. Control
variables SIZE, CHEM, MET, PETR, CAPINT, ROA, and AGE
are all significantly associated with the ‘‘certainty” variable
(at p < .10, two-tailed). Firms that are larger, more profit-
able, and older tend to use more certainty in the language
of their environmental disclosures, whereas companies
with higher levels of capital intensity use exhibit lower
levels of certainty. In support of hypothesis 2, the firm
environmental performance score is significantly and neg-
atively related to environmental disclosure ‘‘certainty” le-
vel (p = .021, one-tailed). This result indicates that firms
with worse environmental performance use less certain
language in their 10-K environmental disclosures than
their better-performing counterparts.
Sensitivity tests
A substantial number of our sample firms (n = 102)
have KLD environmental concern scores of zero (0) and this
could be influencing the results we report above. If differ-
ences in environmental performance do, in fact, lead to dif-
ferences in the language and verbal tone of environmental
disclosures, we would expect that within the subset of
companies with cited environmental concerns (KLD > 0),
 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443 441

Table 7
Results of OLS regression analysis testing the relation between environmental performance and environmental disclosure ‘‘certainty” for firms having at least
one environmental concern (EDCi = a1 + B1EPi + B2SIZEi + B3EXTRi + B4CHEMi + B5PAPi + B6METi + B7PETRi + B8MINi + B9CAPINTi + B10ROAi + B11AGEi).

Model explanatory power

Number of observations 88
Model F-statistic 4.982
Significance of F-statistic 0.000
Adjusted R-squared 0.335

Variable Predicted sign Parameter estimate t-Stat. Sig.* VIF

Parameter estimates
INTERCEPT None 11.336 0.673 0.503
EP ( ) 5.195 2.657 0.005 1.434
SIZE (+/ ) 6.316 3.334 0.001 1.596
EXTR (+/ ) 0.369 0.055 0.956 1.128
CHEM (+/ ) 13.547 1.990 0.050 1.163
PAP (+/ ) 3.076 0.343 0.733 1.045
MET (+/ ) 8.275 1.004 0.319 1.164
PETR (+/ ) 26.897 3.189 0.002 1.218
MIN (+/ ) 0.881 0.080 0.936 1.054
CAPINT (+/ ) 2.140 1.636 0.106 1.361
ROA (+/ ) 0.885 2.846 0.006 1.131
AGE (+/ ) 0.097 2.146 0.035 1.147

EDCi = the 2002 environmental disclosure ‘‘certainty” score for firm i.

EPi = the 2002 KLD environmental concern score for firm i.
SIZEi = the natural log of 2002 assets for firm i.
EXTRi = one if firm i belongs to the oil and gas extraction industry, and zero otherwise.
CHEMi = one if firm i belongs to the chemicals industry, and zero otherwise.
PAPi = one if firm i belongs to the paper industry, and zero otherwise.
METi = one if firm i belongs to the primary metals industry, and zero otherwise.
PETRi = one if firm i belongs to the petroleum refining industry, and zero otherwise.
MINi = one if firm i belongs to the metal mining industry, and zero otherwise.
CAPINTi = the 2002 capital intensity for firm i.
ROAi = the 2002 return on assets for firm i.
AGEi = the age of firm i as of 2002.
*
Significance levels are based on a one-tailed test for the EP variable, and two-tailed for all control variables.

higher concern scores (worse performance) would still be
associated with differences in ‘‘optimism” and ‘‘certainty.”
To test these relations, we repeat our analyses using those
sample companies with a KLD concern score of one or
greater (n = 88).
Tables 6 and 7 present the results of our ‘‘optimism” and
‘‘certainty” regression analyses, respectively, using only the
sample firms with cited environmental concerns. In each
case the model adjusted R-squared is higher than for the
model using the total sample, and with the exception that
SIZE in the optimism regression and CAPINT in the certainty
model are no longer statistically significant, relations for
the control variables are consistent with the results using
the total sample. Most importantly, in each case, the EP var-
iable remains significantly associated with the respective
dependent variable. EP is positively related to the ‘‘opti-
mism” score (at p = .014, one-tailed) and negatively related
to the ‘‘certainty” measure (at p = .005, one-tailed). These
results are again consistent with our hypotheses that worse
environmental performance leads to the use of more ‘‘opti-
mism” and less ‘‘certainty” in environmental disclosures.17
17
We also repeated the analyses using only firms from environmentally
sensitive industries (and deleting one of the specific industry dummy
variables). Results, not presented here, indicate that EP continued to be
positively associated with the ‘‘optimism” score and negatively related to
the ‘‘certainty” scores for this sub-sample, and both relations were
statistically significant (at p < .05, one-tailed).
Overall, the findings from both our primary analysis and
our sensitivity tests provide evidence that, as predicted,
there is a significant relationship between firm environ-
mental performance and the use of biased language and
verbal tone in 10-K report environmental disclosures.
Conclusions, limitations, and future research
Prior research reports that corporate environmental
disclosures are, like other disclosures, often used by corpo-
rations to attempt to manage stakeholder impressions
regarding environmental performance. This stream of re-
search shows that worse environmental performers tend
to use more extensive disclosures or tend to publish selec-
tive, partial disclosures as strategies for managing impres-
sions. Relying on prior research in environmental
disclosure and corporate impression management, we ar-
gued here that corporations also may attempt to manage
stakeholder impressions by self-servingly biasing the lan-
guage and verbal tone used in their environmental
disclosures.
Our results support our contention that worse environ-
mental performers use language and verbal tone to bias
the message presented in their financial report environ-
mental disclosures. We find, first, that worse environmen-
tal performance is associated with the use of more
optimistic language in our test companies’ disclosures. This
442 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443
indicates that the language and verbal tone of these disclo-
sures more strongly focus on reporting good news and
attributing positive performance to the reporting compa-
nies’ internal efforts while potentially blurring responsibil-
ity for poorer performance than the language of better
performing companies. We also find that our environmen-
tal performance measure is negatively related to the cer-
tainty scale of the disclosure. It appears, therefore, that
companies with worse environmental performance at-
tempt to mask internal attributions for their poor perfor-
mance by using convoluted and less certain language.
Our findings lend empirical support to the Merkl-Davies
and Brennan (2007) managerial impression management
framework that states corporations may use concealment
and attribution in corporate disclosures in order to present
a more favorable depiction of their performance. We con-
clude, based on our results, that corporate environmental
disclosures of poorer performing firms appear to empha-
size good news, obfuscate bad news, and slant attributions
of performance to their advantage in an attempt to manage
stakeholder impressions of their corporate environmental
performance. Our study suggests that the language and
verbal tone used in corporate disclosures, in addition to
their amount and thematic content, should be considered
when investigating the relation between corporate disclo-
sure and performance.
Our study is subject to several limitations. A key short-
coming of DICTION is its limited ability to detect homo-
graphs (Short & Palmer, 2008). Further, the quantitative
scores provided by DICTION capturing verbal tone and
language cannot portray the subtleties and complexities
that can be better analyzed by qualitative case studies
of individual firms’ environmental disclosures. In addi-
tion, such quantitative measures themselves reflect re-
search-design decisions and could be refined in future
research (Henry, 2008). It also must be noted that our
study relies on a relatively small sample of firms (espe-
cially for the sensitivity analyses focusing only on compa-
nies with at least one environmental concern) and as
such, the results should be interpreted with caution.
There are other limitations that are common in empirical
work. Our results are based on a cross-sectional sample of
public companies listed on US stock exchanges. This lim-
its the scope of our study and the extent to which these
findings would be generalizable and hold in other periods
cannot be determined. Although we relied on prior re-
search to help design our empirical model, there may be
other factors that explain these differences in the lan-
guage and verbal tone of corporate environmental disclo-
sures. Our research may be extended through longitudinal
analyses of the language and tone of environmental dis-
closures or through the development of a more complete
model of disclosure.
Acknowledgements
We would like to express our thanks to Editor Anthony
Hopwood, Editor Christopher Chapman, two reviewers,
Walter Aerts, Vicky Arnold, Sophie Audousset-Coulier,
Gaétan Breton, Peggy Dwyer, Amy Hageman, Jillian
Phillips, Vaughan Radcliffe, Steve Sutton, Antony Young,
Philippe Zarlowski, participants of the 2nd Annual Public
Interest Section Mid-Year Meeting of the American
Accounting Association in Charleston, the 29ème Congrès
de l’Association Francophone de Comptabilité in Paris,
the 2008 Critical Perspectives on Accounting Conference
in New York, the 30th European Accounting Association
in Lisbon, and participants of the research workshop held
at Florida International University for their helpful com-
ments and feedback on previous versions of this paper.
We are also grateful to Jane Namjee Cho and Ying Huang
for their valuable data assistance. Charles Cho acknowl-
edges financial support received from the Fonds Québécois
de la Recherche sur la Société et la Culture (FQRSC) and the
Social Sciences and Humanities Research Council (SSHRC)
of Canada.
References
Abrahamson, E., & Park, C. (1994). Concealment of negative organizational
outcomes: An agency theory perspective. Academy of Management
Journal, 37(5), 1302–1334.
Aerts, W. (1994). On the use of accounting logic as an explanatory
category in narrative accounting disclosures. Accounting,
Organizations and Society, 19(4/5), 337–353.
Aerts, W. (2005). Picking up the pieces: Impression management in the
retrospective attributional framing of accounting outcomes.
Accounting, Organizations and Society, 30(6), 493–517.
Aerts, W., & Cormier, D. (2009). Media legitimacy and corporate
environmental communication. Accounting, Organizations and
Society, 34(1), 1–27.
Berthelot, S., Cormier, D., & Magnan, M. (2003). Environmental disclosure
research: Review and synthesis. Journal of Accounting Literature, 22,
1–44.
Bewley, K., & Li, Y. (2000). Disclosure of environmental information by
Canadian manufacturing companies: A voluntary disclosure
perspective. Advances in Environmental Accounting and Management,
1, 201–226.
Bligh, M. C., Kohles, J. C., & Meindl, J. R. (2004). Charting the language of
leadership: A methodological investigation of President Bush and the
crisis of 9/11. Journal of Applied Psychology, 89, 562–574.
Cho, C. H., Patten, D. M., & Roberts, R. W. (2006). Corporate political
strategy: An Examination of the relation between political
expenditures, environmental performance, and environmental
disclosure. Journal of Business Ethics, 67(2), 139–154.
Cho, C. H., & Patten, D. M. (2007). The role of environmental disclosures as
tools of legitimacy: A research note. Accounting, Organizations and
Society, 32(7–8), 639–647.
Cho, C. H., Phillips, J., Hageman, A. M., & Patten, D. M. (2009). Media
richness, user trust, and perceptions of corporate social
responsibility: An experimental investigation of visual website
disclosures. Accounting, Auditing and Accountability Journal, 22(6),
933–952.
Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2008). Revisiting
the relation between environmental performance and environmental
disclosure: An empirical analysis. Accounting, Organizations and
Society, 33(4/5), 303–327.
Clatworthy, M., & Jones, M. J. (2001). The effect of thematic structure on
the variability of annual report reliability. Accounting, Auditing and
Accountability Journal, 14(3), 311–326.
Clatworthy, M., & Jones, M. J. (2003). Financial reporting of good news and
bad news: Evidence from accounting narratives. Accounting and
Business Research, 33(3), 171–185.
Cormier, D., Gordon, I., & Magnan, M. (2004). Corporate environmental
reporting: Contrasting management’s values with reality. Journal of
Business Ethics, 49(2), 143–165.
Cowen, S. S., Fererri, L. B., & Parker, L. D. (1987). The impact of corporate
characteristics on social responsibility disclosures: A typology and
frequency-based analysis. Accounting, Organizations and Society, 12(2),
111–122.
Demers, E. A., & Vega, C. (2009). Soft information in earnings
announcement: News or Noise? 32nd Annual Congress of the
European Accounting Association Paper. Available at SSRN: <http://
ssrn.com/abstract=1152326>.
 C.H. Cho et al. / Accounting, Organizations and Society 35 (2010) 431–443
Elsbach, K. K. (1994). Managing organizational legitimacy in the California
cattle industry: The construction and effectiveness of verbal accounts.
Administrative Science Quarterly, 39, 57–88.
Elsbach, K. D., & Sutton, R. I. (1992). Acquiring organizational legitimacy
through illegitimate actions: A marriage of institutional and
impression management theories. Academy of Management Journal,
35(4), 699–738.
Epstein, M. J., & Freedman, M. (1994). Social disclosure and the
individual investor. Accounting, Auditing and Accountability Journal,
7(4), 94–109.
Ginzel, L. E., Kramer, R. M., & Sutton, R. I. (1992). Organizational
impression management as a reciprocal influence process: The
neglected role of the organizational audience. In L. L. Cummings &
B. M. Staw (Eds.). Research in Organizational Behavior (Vol. 15,
pp. 227–266). Greenwich, CT: JAI Press.
Hackston, D., & Milne, M. (1996). Some determinants of social and
environmental disclosures in New Zealand. Accounting, Auditing and
Accountability Journal, 9(1), 77–108.
Hart, R. P. (1984). Systematic analysis of political discourse: The
development of DICTION. In K. Sanders et al. (Eds.), Political
communication yearbook (pp. 1–5). Carbondale, IL: Southern Illinois
University Press.
Hart, R. P. (2000). DICTION 5.0: The text analysis program. Thousand Oaks,
CA: Scolari/Sage.
Hart, R. P. (2001). Redeveloping DICTION: Theoretical considerations. In
M. D. West (Ed.), Theory, method, and practice of computer content
analysis (pp. 43–60). Westport, CT: Greenwood Publishing Group.
Henry, E. (2008). Are investors influenced by how earnings press releases
are written? Journal of Business Communication, 45(4), 363–407.
Hillman, A. J., & Keim, G. D. (2001). Shareholder value, stakeholder
management, and social issues: What’s the bottom line? Strategic
Management Journal, 22, 125–139.
Hopwood, A. G. (2009). Accounting and the environment. Accounting,
Organizations and Society, 34(3–4), 433–439.
Hughes, S. B., Anderson, A., & Golden, S. (2001). Corporate environmental
disclosures: Are they useful in determining environmental
performance? Journal of Accounting and Public Policy, 20(3), 217–240.
Jameson, D. (2000). Telling the investment story: A narrative analysis of
shareholder reports. Journal of Business Communications, 37(1), 7–38.
Johnson, L. T. (1993). Research on environmental reporting. Accounting
Horizons, 7(3), 118–123.
Jones, M. J., & Shoemaker, P. A. (1994). Accounting narratives: A review of
empirical studies of content and readability. Journal of Accounting
Literature, 13, 141–184.
Jones, M. (1996). Readability of annual reports: Western versus Asian
evidence – A comment to contextualize. Accounting, Auditing and
Accountability Journal, 9(2), 86–91.
KLD Research and Analytics, Inc. (2003). KLD ratings data: Inclusive social
rating criteria. Cambridge, MA.
Livesey, S., & Kearins, K. (2002). Transparent and caring corporations?
Organization and Environment, 15(3), 233–258.
Magness, V. (2006). Strategic posture, financial performance and
environmental disclosure: An empirical test of legitimacy theory.
Accounting, Auditing and Accountability Journal, 19(4), 540–563.
Matsumoto, D., Pronk, M., Roelofsen, E. (2006). Do analysts mitigate
optimism by management? Working paper, University of Washington.
443
Merkl-Davies, D. M., & Brennan, N. M. (2007). Discretionary disclosure
strategies in corporate narratives: Incremental information or
impression management? Journal of Accounting Literature, 26,
116–194.
Milne, M. J., & Patten, D. M. (2002). Securing organizational legitimacy: An
experimental decision case examining the impact of environmental
disclosures. Accounting, Auditing and Accountability Journal, 15(3),
372–405.
Nelson, K. K., & Pritchard, A. C. (2008). Litigation risk and voluntary
disclosure: The use of meaningful cautionary language. 2008 American
accounting association paper. Available at SSRN: <http://ssrn.com/
abstract=998590>.
Neu, D., Warsame, H., & Pedwell, K. (1998). Managing public impressions:
Environmental disclosures in annual reports. Accounting,
Organizations and Society, 23(3), 265–282.
Ober, S., Zhao, J. J., Davis, R., & Alexander, M. W. (1999). Telling it like it is:
The use of certainty in public business discourse. Journal of Business
Communication, 36(3), 280–300.
Patten, D. M. (1992). Intra-industry environmental disclosures in
response to the Alaskan oil spill: A note on legitimacy theory.
Accounting, Organizations and Society, 17(5), 471–475.
Patten, D. M. (2002). The relation between environmental performance
and environmental disclosure: A research note. Accounting,
Organizations and Society, 27(8), 763–773.
Reitenga, A. L. (2000). Environmental regulation, capital intensity, and
cross-sectional variation in market returns. Journal of Accounting and
Public Policy, 19(2), 189–198.
Roberts, R. W. (1992). Determinants of corporate social responsibility
disclosure: An application of stakeholder theory. Accounting,
Organizations and Society, 17(6), 595–612.
Rutherford, B. A. (2003). Obfuscation, textual complexity and the role of
regulated narrative accounting disclosure in corporate governance.
Journal of Management and Governance, 7, 187–210.
Schlenker, B. R. (1980). Impression management: The self-concept social
identity, and, interpersonal relations. Belmont: Brooks-Cole.
Short, J. C., & Palmer, T. B. (2008). The application of DICTION to content
analysis research in strategic management. Organizational Research
Methods, 11(4), 727–752.
Smith, M., & Taffler, R. (2000). The chairman’s statement: A content
analysis of discretionary narrative disclosures. Accounting, Auditing
and Accountability Journal, 13(5), 624–646.
Subramanian, R., Insley, R., & Blackwell, R. (1993). Performance and
readability: A comparison of annual reports of profitable and
unprofitable corporations. Journal of Business Communication, 30,
50–61.
Sydserff, R., & Weetman, P. (2002). Developments in content analysis: A
transitivity index and DICTION scores. Accounting, Auditing and
Accountability Journal, 15(4), 523–545.
Thomas, J. (1997). Discourse in the marketplace: The making of
meaning in annual reports. Journal of Business Communication,
34(1), 126–134.
Waddock, S. A., & Graves, S. B. (1997). The corporate social performance –
financial performance link. Strategic Management Journal, 18(4),
303–319.
Yuthas, K., Rogers, R., & Dillard, J. (2002). Communicative action and
corporate annual reports. Journal of Business Ethics, 41(1–2), 141–157.