Professional Documents
Culture Documents
DOI 10.1007/s11145-017-9757-x
Stelios A. Christodoulou1
Abstract The study is situated at the interface between reading comprehension and
critical thinking research. Its purpose was to examine the influence of reading goals
and argument quality on the comprehension and critical evaluation of argumentative
texts. Young adult readers read to comprehend or evaluate texts on two different
controversial issues. Argument quality was varied across text versions on the basis
of the hasty generalization fallacy. Text versions varied with respect to the quality
of the arguments included, but not in terms of argument content. Measures of
comprehension included main claim recall, overall recall and inferences in recall.
Text evaluation was measured with a rating task. The sample’s familiarity with the
text topics was low, and prior beliefs were relatively neutral. The results indicated
that an evaluation goal had a consistent positive effect on main claim and text recall
when compared to comprehension goal. Argument quality, however, had no main or
interactive effects on text evaluation. The findings indicate that reading to evaluate
argumentative text facilitates the representation of its content and critical argument
elements, such as the claim it promotes. However, this representation is not suffi-
cient for analyzing and critically evaluating the text’s argument line. The impli-
cations of these findings are discussed in relation to current efforts to promote
critical-analytic thinking skills in the context of reading and writing.
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1870 I.-A. N. Diakidoy et al.
Introduction
Argument Evaluation
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1871
arguments should result in their rejection and minimize, if not eliminate, any
persuasive impact they may have.
Nevertheless, critical thinking research has documented a wealth of heuristics
and prior belief biases applied in a variety of critical thinking and evaluation tasks
that eventually result in the acceptance of fallacious arguments and invalid
conclusions (West, Toplak, & Stanovich, 2008). For example, fallacious but belief-
consistent arguments are more likely to be accepted as valid than belief-inconsistent
arguments (Klaczynski et al., 1997; Macpherson & Stanovich, 2007; Stanovich &
West, 2007). From a dual-process perspective, these results are taken to reflect the
outcomes of fast and intuitive or associative, memory-based processing of
information (Evans, 2007; Sloman, 1996). Intuitive processing places fewer
demands on a limited-resource cognitive system as the evaluation of arguments
and conclusions is supported by automatically activated prior knowledge and
personal beliefs. In contrast, overcoming belief bias and identifying fallacies in
arguments requires effortful and analytic processing. This type of processing
involves the monitoring or control of any automatically activated products of
associative memory (heuristics, biases, beliefs, and world knowledge) and the
conscious, sequential application of analytic processes in the consideration of
arguments (Evans, 2007, 2012).
Although critical thinking research has explored the influence of a variety of
ability and motivational factors in argument evaluation, the relative absence of
comprehension as a potential contributor to critical argument evaluation is
noteworthy. Part of the reason is that typical argument evaluation tasks in this
line of work involve the presentation of single arguments consisting of a claim and a
supporting reason, or short and easy to comprehend argument scenarios where
protagonists explicitly state and justify opposing claims. Nevertheless, Britt and her
colleagues (Britt, Kurby, Dandotkar, & Wolfe, 2008) have documented the
difficulty that young adult readers have in representing and recalling accurately the
predicates of claims in simple two-clause arguments, when compared to their
themes. Moreover, Neuman (2003) found reading comprehension ability to predict
high school students’ ability to identify fallacies in short argument scenarios.
Therefore, comprehension may be an important contributor in argument evaluation
when that argument is embedded in longer text.
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1872 I.-A. N. Diakidoy et al.
short argument scenarios representing three different types of fallacies (false cause,
ad populum, and ad ignorantiam). Their results indicated that students who recalled
the fallacy-containing clause in verbatim or paraphrase form were more likely to
identify the fallacy and correctly reject the argument than students who had not
recalled it.
Although forming an accurate representation may seem relatively straightforward
in the case of a short argument scenario, it can become overwhelming in the case of
a longer argumentative text. As in most naturalistic interactive discourse (Rips,
1998), the argument lines in naturally-occurring longer argumentative texts tend to
be structurally more complex than in constructed brief scenarios. Claims and
counter-claims are justified on the basis of more than one reason, with textual
information representing a variety of argument elements and responses—such as
qualifiers, questions, refutations, or concessions—that give rise to sub-arguments
and possibly multiple interrelated themes (Rips, 1998; van Eemeren & Grooten-
dorst, 1999). Therefore, mentally representing the argument lines criss-crossing a
naturalistic text and their often implicit connections can be a cognitively demanding
task.
Chambliss (1995) and Larson et al. (2004) have proposed that comprehension of
argumentative text is based on the activation of an argument schema that guides
readers to identify the main claim promoted by the text as the focal point and to
organize the rest of textual information in relation to that claim. Consistent with this
proposal, Chambliss (1995) found that when text on non-controversial topics
adhered to an argument structure and explicitly signalled argument parts, high-
school readers were better able to identify these parts and to represent the gist of the
argument. In contrast, the Larson et al. (2004) findings were not as optimistic. They
employed a simple search task in order to examine college students’ ability to
identify claims and reasons. The participants were asked to read naturalistic
argumentative texts on controversial topics, to write down the main claim that the
text promoted, and to underline any reasons that supported this claim. Their results
indicated a disappointing 30% accuracy level in identifying these argument
elements.
In an attempt to provide a bridge between critical thinking and reading
comprehension research, Diakidoy et al. (2015) examined the relationship between
the comprehension of a naturally occurring argumentative text and the evaluation of
its argument line as a whole. In line with critical thinking research (Klaczynski
et al., 1997), the quality of the text’s argument line was modified on the basis of the
presence or absence of the hasty generalization fallacy, and evaluation was
indicated by a convincingness rating task. Comprehension measures included a
main claim recall task scored for accuracy and a free recall task scored for amount
of text information recalled and inferences. Argument quality was found to exert no
influence on either comprehension or evaluation. Notably, however, the evaluation
task following the reading of the text was unexpected. Therefore, it is possible that,
had students known of the evaluation requirement, they might have focused more on
the arguments and their quality allowing, thus for any influence of this variable to
emerge. In contrast, inference generation in recall (as an index of deeper
comprehension) predicted text evaluation. This result, however, indicated that the
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1873
more inferences readers generated the more positively they evaluated the text
regardless of argument quality. Moreover, the more inferences readers generated,
the more likely they were to confuse the main claim with one of its justifications. In
fact, similar to the findings of Larson et al. (2004), more than half of the college
student participants in this study were unable to recall the main claim with some
accuracy.
To summarize, the findings, thus far, suggest that readers have a notable difficulty
identifying and representing the main claim that naturalistic argumentative text
promotes, and that deeper comprehension, as indicated by inference generation,
does not facilitate the representation of important argument elements or contribute
to the critical evaluation of the argument. From a dual-process perspective, this lack
of association between comprehension and evaluation can be attributed to
differences in the underlying processes. Comprehension is conceptualized to rely
on automatic memory-based processes (Kintsch, 1988; O’Brien & Cook, 2016).
Although the possibility of more strategic comprehension processes is implicitly
acknowledged or explicitly postulated (see McNamara & Magliano, 2009), their
manifestation is considered to depend more or less on the characteristics of the
reader, the text, and the task (e.g., Graesser, 2007; Linderholm, Virtue, Tzeng, &
van den Broek, 2004; Long & Lea, 2005). In contrast, critical evaluation is
conceptualized to depend exclusively on analytical and strategic processing (Evans,
2012). Moreover, a critical aspect of deep comprehension is the generation of
inferences that fill gaps in the text and support the integration of text information
with prior knowledge. Nevertheless, inferencing with expository texts (a category to
which argumentative texts belong) in relation to that with narrative text, is more
limited (Coté, Goldman, & Saul, 1998), dependent on prior knowledge and text
support (McNamara, Kintsch, Songer, & Kintsch, 1996; Wiley & Myers, 2003), and
occasionally unproductive if it involves irrelevant associations or inaccurate prior
knowledge (Coté et al., 1998; Kendeou & van den Broek, 2005). This last
possibility is echoed in critical thinking research, where the automatic activation of
prior knowledge and belief and the inferences it supports are considered responsible
for biased evaluations (Stanovich & West, 2008), and which may underlie the
unexpected unproductive influence of inferences in the Diakidoy et al. (2015) study.
Nevertheless, a lack of association between comprehension and critical evaluation
remains particularly problematic, because without comprehension any evaluation of
a text’s argument line would be unfounded.
Reading goals
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1874 I.-A. N. Diakidoy et al.
of claims and counter-claims in relation to supporting reasons can take place only if
these elements and their interrelations are available in the mental representation that
readers form as a result of reading. However, reading for comprehension may result
in a representation that is missing key argument elements (e.g., Diakidoy et al.,
2015) or contains only parts of the argument line, depending on readers’ prior
knowledge and beliefs (e.g., Wiley, 2005). Neither of these outcomes can be
expected to provide a solid basis for critical evaluation. In contrast, it is possible to
assume that readers reading with a mind to critically evaluate (i.e., to manipulate a
representation) may be more inclined to identify and ensure the representation of
important argument elements upon which to base their evaluation than those reading
for comprehension (i.e., to construct a representation).
Research on text comprehension has shown reading goals to influence both the
processes and the outcomes of comprehension. For example, when readers are
instructed to study text, they spend more time processing central than peripheral text
information, generate more coherence building inferences and paraphrases, and
recall more than when they are instructed to read for entertainment (van den Broek
et al., 2001; Yeari et al., 2015). In contrast, the possibility of readers’ goals similarly
influencing the comprehension and/or evaluation of an argumentative text has
received little attention. In a follow-up study, Larson et al. (2004) manipulated
reading goals indirectly by asking students to generate either an additional argument
(i.e., comprehension) or a counter-argument (i.e., evaluation). Their results,
however, indicated no effects of reading goals on students’ accuracy in identifying
the argument elements. However, it is questionable whether these findings can be
taken to reflect the influence of reading goals on argumentative text comprehension
and/or evaluation. Although the ability to contribute to the argument requires
comprehension, it also reflects input from background knowledge, while generating
a single rebuttal (counter-argument) may also be the result of skimming and
locating an easy-to-rebut reason, without any prior evaluation of the argument line
as a whole. Therefore, the potential influence of reading goals on comprehension
and evaluation outcomes with naturalistic argumentative texts remains unknown.
The present study extends previous research in examining the comprehension and
evaluation of argumentative text and their association as a function of reading goals
and argument quality. Reading goals were manipulated in line with the outcomes of
interest and the reasons that readers may have in naturalistic contexts. Participants
were instructed to read two argumentative texts on different controversial topics
either for comprehension or for evaluation. In order to examine the generalizability
of previous findings, one text was borrowed from the Diakidoy et al. (2015) study,
and a second text with similar structure and function but on a completely different
topic was selected. Two versions of each text were created by varying argument
quality at the level of the text (as opposed to the level of individual arguments
presented in the text) on the basis of the hasty generalization fallacy. In the low
argument quality version, justification for the main claim was shown to derive from
limited testimony and evidence, while the opposite was true in the high-argument
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1875
quality version of the same text. Comprehension measures included main claim
recall and free recall, the latter scored for text recall and inferences. Ratings of the
quality and convincingness of the texts’ argument line provided a measure of text
evaluation. Although both argumentative texts were selected to be unfamiliar and
belief neutral, measures of perceived topic knowledge and prior belief (Murphy
et al., 2005) were also obtained to examine their potential influence on text
comprehension and evaluation.
Given previous findings (Diakidoy et al., 2015; Larson et al., 2004), we expected
overall low levels of accuracy on the recall of the main claim promoted in a text.
Nevertheless, we hypothesized that reading for evaluation will direct readers’
attention to central arguments elements, as opposed to peripheral elaborations
(Yeari et al., 2015), increasing, thereby, the accuracy in main claim recall and
overall amount of text recall when compared to reading for comprehension.
However, unlike previous findings (van den Broek et al., 2001), we also
hypothesized that an evaluation goal would result in fewer general knowledge-
based inferences manifested in the recall of argumentative text. This hypothesis was
informed by considerations of the importance of an accurate representation, over an
elaborated representation, for critical evaluation (Britt et al., 2008; Glöckner &
Witteman, 2010) and of prior knowledge and belief as a potential source of bias
(Stanovich & West, 2008). Finally, we reasoned that reading to evaluate may
increase the likelihood of detecting a fallacy in the argument line, when compared
to a goal of comprehension. Given our argument quality manipulation, critical
evaluation would be manifested by negative or lower evaluations for fallacious
arguments and positive evaluations for non-fallacious arguments. Therefore, we
expected reading goals to interact with argument quality in influencing argument
line evaluations. Regardless of reading goal, however, we also hypothesized that
recall, as an index of the quality of the mental representation upon which evaluation
operates (Neuman & Weizman, 2003), would also contribute to more critical
evaluations as indicated by its interaction with argument quality.
Methods
The initial sample included 135 undergraduate students (24 males and 111 females)
from a mid-sized state university in their third (n = 84) and fourth year (n = 51) of
study. The students were enrolled in the Education (n = 59) or Psychology
(n = 76) programs, and their participation was optional in exchange for extra credit
in their required and elective psychology courses. Due to missing values on one or
more of the measures obtained, the final sample included 128 students. Although all
students read the same two texts, Reading Goal and Argument Quality were
between-subject factors. Participants were randomly assigned to a Reading Goal X
Argument Quality condition resulting in four groups (n1 = 30, n2 = 30, n3 = 33,
n4 = 35).
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1876 I.-A. N. Diakidoy et al.
Argumentative texts
The title of the text was ‘‘Moral Resistance’’ and promoted the claim that
supervisors and managers are morally justified in breaking company rules and
The manipulation of argument quality is shown in bold. The text excerpts were translated into English
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1877
Environment text
The title of the text was ‘‘People come second’’ and promoted the claim that a
substantial percentage of existing farm land should be converted into wild life parks
for environmental protection reasons (Aggelopoulos, 2010b). The text was 676
words with 63% of the clauses containing argument-related information, and the rest
presenting information related to a counter-argument (91% inter-rater agreement,
Cohen’s K = .82, p \ .01, differences resolved in conference). The high argument
quality version of this text showed the main claim to be supported by scientific
consensus regarding the hazardous environmental consequences of farming, while
the low argument quality version presented the same argument line advanced
primarily by an interested party (Table 1).
Pre-reading measures
The Prior Knowledge and Belief questionnaire from the Diakidoy et al. (2015) study
was used to measure readers’ perceived prior knowledge and initial beliefs
regarding the controversial issues discussed in the texts. The questionnaire included
eight statements representing claims in relation to a corresponding number of
controversial issues in different domains, but only two of the statements were
directly related to the claims of the texts. Each statement was followed by a 7-point
agreement scale (1 = strongly disagree, 7 = strongly agree) and a 7-point prior
knowledge scale (1 = nothing, 7 = very much). Participants were instructed to
indicate how much they agreed with each statement and how much they thought
they knew about the issue related to the statement. Only the ratings given for the
selected text main claims were included in the analyses. Two versions of the
questionnaire with statements appearing in different order were created and
counterbalanced between participants.
Post-reading tasks
The reading of each argumentative text was followed by three tasks designed to
provide measures of text comprehension and evaluation.
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1878 I.-A. N. Diakidoy et al.
The first task was analogous to a main idea task requiring participants to recall and
write what they thought was the main claim that the previously read text promoted.
Two independent raters scored all responses in terms of the extent to which they
reflected the main claim promoted by the text. Accurate responses were given a
score of 2, responses related to supporting reasons were given a score of 1, and
responses reflecting the general topic, an irrelevant detail or a counter-claim were
given a score of 0. Inter-rater agreement was 93% for the Work Conditions text
(Cohen’s K = .85, p \ .01) and 92% for the Environment text (Cohen’s K = .86,
p \ .01). All scoring differences were resolved in conference (see Table 2 for
sample responses).
Text evaluation
The second task was a typical argument evaluation task (Klaczynski et al., 1997)
adjusted for the length and complexity of a textual argument line. The instructions
asked participants to consider the quality of the arguments presented in support of
Table 2 Sample responses to the main claim recall task in each scoring category
Responses scored 0
It is about the job situation at a time of financial crisis. The text questions whether it is right or wrong
to help people out (reader 131)
These favors are not fair and undermine businesses (reader 65)
Tompkins is making a lot of enemies in farmers (reader 1)
The text is about the environment. A rich man tries to protect it by buying land (reader 17)
Responses scored 1
All people should be able to make a decent living if they work hard (reader 51)
The text claims that if supervisors are sensitive to the needs of their workers, their productivity will
increase (reader 112)
The land belongs to the animals too, and we need to protect them (reader 43)
Farmland interferes with and pollutes water ways and reserves (reader 57)
Responses scored 2
The text supports the actions of the supervisors as morally justified on the basis on human rights
(reader 102)
It is not wrong to help out workers considering the unfairness of today’s economic situation and if
there is a need (reader 12)
The text supports the turning of farmland into wild land for social and environmental protection
reasons (reader 62)
People’s needs should not be above the needs of other animals, and certainly they should not take
precedence over the environment because in the near future there will be no environment for
humans to live in (reader 96)
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1879
the main claim as a whole and to evaluate the text’s overall convincingness
(1 = not at all convincing, 7 = very convincing).
Free recall
The final task required participants to recall all they could remember from the text
read. Recall protocols were parsed into clauses, and each clause was identified as
either representing a text-based idea (verbatim or paraphrase) or a valid inference
given text content. Two independent raters scored the recall protocols for each text
(86% agreement, Cohen’s K = .71, p \ .01 and 81% agreement, Cohen’s K = .65,
p \ .01) and resolved their scoring differences in conference. Scoring yielded an
Overall Recall score for each participant and text indicating the proportion of text
clauses recalled and a Valid Inference score reflecting the number of valid
inferences in the recall protocol.
Procedure
One month prior to the study and before signing their consent for participation,
students were informed that the general purpose of the study was to examine factors
that influence the comprehension and evaluation of argumentative texts on
contemporary controversial issues. Data collection took place in two group sessions
scheduled two weeks apart. Session 1 involved the administration of the Prior
Knowledge and Belief Questionnaire. In Session 2, half of the students were
instructed to read their assigned texts carefully in order to understand them while
the rest were instructed to read the texts in order to evaluate the quality and
convincingness of the argument line. The written instructions that both groups
received before reading each text were the following: ‘‘Please read the following
text only once but carefully in order to comprehend the point of view it presents/to
evaluate the quality and convincingness of its argument line’’. When you finish
reading you can hand in the text and proceed completing the post-reading tasks in
the order they appear in the second booklet’’. No further information about the post-
reading tasks was given at that point. After reading the first assigned text, students
completed the corresponding post-reading tasks and proceeded to the second text.
The order of text presentation was counterbalanced between students. Parts of a
filler questionnaire on study strategies were completed between text readings and
post-reading tasks. Session 2 lasted about 90 min.
Results
Overall Text Recall, Number of Valid Inferences, and Text Evaluation were main
continuous outcome measures, while Main Claim Recall was a categorical index of
comprehension. Square root transformations were applied to Overall Text Recall
and Number of Valid Inferences to correct for variance heterogeneity across goal
and argument quality conditions and to normalize their distributions, respectively.
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1880 I.-A. N. Diakidoy et al.
Reading Goal and Argument Quality were main independent variables, while Text
Topic was included as a within-subject factor in some of the analyses.
Pre-reading measures of Perceived Prior Knowledge and Initial Beliefs were used
as covariates in analyses that were conducted separately for each experimental text.
The conceptual basis for this decision was the topic-specificity of these measures.
Moreover, this decision was supported by methodological considerations. Specif-
ically, Initial Beliefs for the two text topics were uncorrelated (r = -.13, p = .126),
while the Perceived Prior Knowledge measures had differential patterns of
correlation with main post-reading outcomes. It can be seen from Table 3 that
none of these measures correlate with any of the main outcomes for the Work
Conditions text. For this text, only Number of Valid Inferences in recall was
associated positively with Evaluation. In contrast, Initial Beliefs and Perceived Prior
Knowledge correlated positively with the Evaluation of the Environment text, while
Perceived Prior Knowledge was negatively associated with its Recall. Notably, the
correlation between Perceived Prior Knowledge and Recall indicated that those who
consider themselves knowledgeable of the issue recall less of the information
presented in text (Table 3).
Means and standard deviations for all pre- and post-reading measures can be seen
in Table 4. Paired t-tests indicated no significant differences between texts with
respect to these pre-reading measures (all p [ .05). Initial Beliefs were relatively
neutral (Mode = 4 for both texts), while Perceived Prior Knowledge was relatively
low (Mode = 2 for both texts). Finally, both texts received overall positive
evaluation ratings (Environment: Mode = 5; Work Conditions: Mode = 6, rating
scales 1–7).
As expected, overall levels of accuracy in this task were relatively low. For both
texts, the Main Claim promoted was accurately identified and recalled by about 55%
of the students. For the Environment text, 24.2% of the students recalled a reason
instead of the claim, while 20.3% recalled the general issue instead of the position
Table 3 Intercorrelations for initial belief, perceived prior knowledge, overall text recall, number of
valid inferences, and text evaluation for each Text
Initial belief Prior knowledge Recall Inferences Evaluation
Intercorrelations for the Environment text are presented above the diagonal. Intercorrelations for the Work
Conditions text are presented below the diagonal
* p \ .05; ** p \ .01
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1881
taken on the issue. For the Work Conditions text, 18.8% of the students recalled a
reason instead of the claim, while 25.8% recalled the general issue instead of the
position taken on the issue (Table 5).
We had hypothesized that reading for evaluation would contribute positively to
Main Claim Recall when compared to reading for comprehension. In order to test
this hypothesis, we performed multinomial logistic regressions for Main Claim
Recall with Goal as a factor, and Perceived Prior Knowledge, Overall Text Recall
and Number of Valid Inferences as the predictors. The analyses were conducted
separately for each text because of variable performance across texts resulting in
unequal cell sizes (only 8.5% missed, 5.5% confused, and 36.7% accurately recalled
the main claim of both texts). The inclusion of Overall Recall Score as a predictor
was warranted by its positive association with Main Claim Recall (for Environment:
rs = .26, p = .003; for Work Conditions: rs = .20, p = .025). Perceived Prior
Knowledge was also included on the basis of theoretical considerations and its
correlation with Overall Recall Score in the case of one text (see Table 3). Finally,
Number of Valid Inferences was included as a predictor to determine the extent to
Table 5 Numbers of
Comprehension Evaluation
participants who recalled,
goal goal
confused, and missed the main
claim in each text and reading
Environment
goal group (N = 128)
Recalled 27 44
Confused 22 9
Missed 14 12
Work Conditions
Recalled 29 42
Confused 15 9
Missed 19 14
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1882 I.-A. N. Diakidoy et al.
which its negative influence documented in the Diakidoy et al. (2015) study would
be replicated in this one.
For the Environment text, the final model was significant, v2 (8) = 19.87,
p = .011, R2 = .15 (Cox & Snell), but Reading Goal was the only significant
predictor, B = 1.19, Exp(B) = 3.31, Wald v2 = 4.14, p = .042. Students who read
for comprehension were more likely to confuse the main claim with one of the
arguments than to accurately recall it. For the Work Conditions text, the final model
was also significant, v2 (8) = 16.10, p = .041, R2 = .12 (Cox & Snell), with
Reading Goal and Perceived Prior Knowledge being marginally significant
predictors: B = 1.27, Exp(B) = 3.56, Wald v2 = 3.93, p = .047 and B = -.37,
Exp(B) = .69, Wald v2 = 3.87, p = .049, respectively. Again, students who read
for comprehension were more likely to confuse the main claim with one of the
arguments than to accurately recall it. However, all students were more likely to
confuse the main claim if they perceived their prior topic knowledge to be higher
than lower. None of the variables were significant predictors of whether students
missed the main claim entirely as opposed to accurately recalling it.
Our hypotheses concerning the effects of Reading Goal on comprehension
measures also predicted a significant positive effect on Overall Text Recall. In order
to test this hypothesis, we first performed an overall mixed ANOVA with Reading
Goal and Argument Quality as the between-subject factors and Text as the within-
subject factor. Argument Quality was included in this analysis in order to examine
the possibility of any interactions with Reading Goal. The results indicated a
significant effect of Reading Goal only (F (1, 123) = 101.14, p = .000, g2 = .45).
The same analysis for Number of Valid Inferences yielded a significant Reading
Goal X Text interaction only (F (1, 123) = 9.13, p = .003, g2 = .07). It can be
seen from Table 6 that an evaluation goal increased recall for both texts when
compared to a comprehension goal, but decreased the amount of inference
generation only for the Environment text, t (126) = 2.03, p = .044.
In order to examine any main or interactive influences of Perceived Prior
Knowledge on comprehension measures, separate hierarchical regressions were
performed for each text. It can be seen from Table 7, that Reading Goal was a
significant positive predictor of Overall Recall Score for both texts. Perceived Prior
Knowledge, however, was a negative predictor of the recall of the Environment text
only. Similar analyses for Number of Valid Inferences indicated only Reading Goal
as a significant predictor of inference generation for the Environment text
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1883
Table 7 Hierarchical
Predictors Experimental texts
regression analyses predicting
overall recall score for each text Environment Work Conditions
2
DR b DR2 b
In all analyses for evaluation outcomes, we excluded the data of students who had
entirely missed the texts’ main claim (N = 80), as the basis for their evaluations
could not be determined. In order to test our initial hypotheses, a mixed ANOVA
with Reading Goal and Argument Quality as the between-subject factors and Text
as the within-subject factor was performed. The results indicated significant main
effects of Text (F (1, 75) = 26.50, p = .000, g2 = .26) and Reading Goal (F (1,
75) = 9.51, p = .003, g2 = .11). The effects of Argument Quality and its
interactions with Text and Reading Goal were not significant (p [ .05). The
Environment text received lower evaluation ratings than the Work Conditions text
(for the reduced sample, M = 4.71, SD = 1.25 and M = 5.51, SD = 1.05
respectively, paired t (79) = -5.48, p = .000). Moreover, it can be seen from
Table 8 that an evaluation goal had the effect of lowering the evaluation ratings for
both texts and regardless of the quality of their arguments in comparison to a
comprehension goal (Table 8).
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1884 I.-A. N. Diakidoy et al.
Discussion
The present study examined the influence of reading goals and argument quality on
comprehension and evaluation outcomes with argumentative texts in an attempt to
provide a bridge between text comprehension and critical thinking research. We
employed naturalistic persuasive text presenting a complex argument line with an
implicit main claim along with a series of justifications, counter-arguments, and
rebuttals. We manipulated argument quality on the basis of a well-known and
common informal reasoning fallacy that could be implemented in a lengthy
argument line without altering the content of individual reasons or warrants. This
decision was also motivated by the need to avoid or minimize any confounds and
biases associated with subjective ratings of argument strength and quality
Table 9 Hierarchical
Predictors Experimental texts
regression analyses predicting
text evaluation for each text Environment Work Conditions
2
DR b DR2 b
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1885
(Stanovich & West, 2007). Finally, the basis for our reading goal manipulation, as
well as the outcome measures employed, reflected theoretical and empirical
considerations of both lines of research. Overall, the results indicated significant
effect of reading goals on both comprehension and evaluation outcomes and non-
significant effects of argument quality on evaluation outcomes.
Comprehension
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1886 I.-A. N. Diakidoy et al.
Nevertheless, the present results seem to suggest that readers confident in their
knowledge of the topic were less likely to pay close attention to critical argument
elements in the text or engage in deeper processing (McNamara, et al., 1996).
The present findings are consistent with the predictions of an argument schema
hypothesis (Britt & Larson, 2003; Chambliss, 1995; Reznitskaya et al., 2007).
Explicit instructions to evaluate the argument line of a text are more likely to
activate whatever argument-relevant, structural knowledge readers may have than
general instructions to read for comprehension. This structural knowledge may have
guided readers’ attention to argument-relevant information, facilitating, thereby, the
representation and memory. Reading to evaluate, however, can also be taken to have
induced a more strategic processing of the information in text as opposed to a more
automatic and memory-based processing (Linderholm et al., 2004; O’Brien & Cook,
2016). From this perspective, an argument evaluation goal may have curtailed the
automatic activation of topic-related concepts in long-term memory and general
knowledge-based inferences (Stanovich & West, 2008) in favour of argument-
related information.
Evaluation
Based on the assumption that reading to evaluate would induce a more strategic
processing of the text content, we had initially hypothesized that reading for
evaluation would increase the likelihood of detecting a fallacy in the argument line,
when compared to reading for comprehension. Moreover, on the basis of prior
findings (Neuman & Weizman, 2003), we had hypothesized that amount of recall
would also contribute to more critical evaluations manifested by negative or lower
evaluations for fallacious arguments and positive evaluations for non-fallacious
arguments. However, the findings did not support any of the hypotheses. Although
there was a tendency for high argument quality text versions to receive more
positive evaluations than low argument quality versions, the differences were not
significant. Instead, an evaluation goal had the overall effect of lowering evaluation
ratings regardless of the quality of the argument, suggesting that this goal elicited a
more conservative but not necessarily critical approach to the text.
These results are in line with previous findings documenting a similar lack of
argument quality effects or interactions in evaluation (Diakidoy et al., 2015;
Klaczynski et al., 1997). Nevertheless, they seem inconsistent with the findings of
Neuman and Weizman (2003) which showed that readers who recalled the fallacy-
containing clauses in short argument scenarios were more likely to identify and
reject the argument on the basis of that fallacy. Our argument quality manipulation
was based on a different fallacy, that of hasty generalization. This particular fallacy
was chosen because it allows syntactically varying the basis of the evidence or
justification associated with a claim without semantically altering any of the
premises, that is, the content of the argument line in the text. Noticing and taking it
into account for evaluation purposes would require readers to consider and question
the basis and the source of the argument, not necessarily its content. Therefore, it is
likely that the syntactic aspects upon which the argument quality manipulation was
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1887
based were bypassed, possibly as a result of attention being focused more on the
content and the implications of the argument.
In contrast to previous findings (Klaczynski et al., 1997; Stanovich & West,
2007), prior beliefs did not influence argument evaluation in this study. This
inconsistency may reflect methodological differences between previous research
and the current study, which used lengthier argumentative text, instead of short and
simple arguments, and focused on the evaluation of an extended argument line
instead of a single argument. Moreover, it must be noted, that our readers’ initial
beliefs regarding the issues addressed in the texts were relatively neutral. Therefore,
it remains possible that stronger biases either for or against the texts’ main claims
would have manifested their influence on subsequent evaluations.
Similar to research with persuasive text (Buehl et al., 2001), the evaluation-
related results highlight topic specific influences. First, the Environment text was
evaluated less positively than the Work Conditions text. Second, while perceived
prior knowledge and initial beliefs were positively associated with the evaluation of
the Environment text—as expected—that was not the case for the Work Conditions
text, whose evaluation ratings were positively associated with inference generation.
Specifically, the more inferences readers generated in recall the more positively
they evaluated this text. Considering the documented influence of personal
relevance in critical thinking research (Klaczynski et al., 1997), this association
between reader generated elaboration and evaluation may reflect the relevance of
the job situation for a group of young adults who have, search, or look forward to
finding work in today’s job market. The dire personal and financial situations
depicted in the Work Conditions text may have been too close to personal
aspirations and fears to allow for a more objective or stringent evaluation of its
argument line.
Limitations
The present results and their implications need to be considered in light of the
study’s limitations. We interpreted our comprehension-related results to indicate
that evaluation goals result in a better representation of important argument
elements. Our comprehension measures, however, provided evidence only in
relation to the representation of main claim. The extent to which our readers
were indeed able to represent argument elements and relations remains
conjectural. The free-recall task is not sufficiently targeted to provide an
adequate measure of memory for specific information, and the limited data
obtained with it (about 13% of text clauses on the average, given the
untransformed proportion scores) leaves open the possibility of more information
being represented than recalled. Similarly, we interpreted our evaluation-related
results to indicate the absence of critical evaluation. However, the fallacy
manipulation seems to have been too subtle in the context of lengthy texts that
were crafted to be persuasive in a subtle way as well. It is possible that a
content-oriented argument quality manipulation would have been more salient,
exerting, thereby, some influence on evaluations outcomes. Therefore, further
research employing a variety of fallacy types and post-reading tasks in
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1888 I.-A. N. Diakidoy et al.
Acknowledgements We would like to thank Philip Kargopoulos, George Floros, and Kalypso Iordanou
for their insightful comments concerning argument fallacies, text cohesion, and argumentation skills.
References
Aggelopoulos, G. (2010a). Moral resistance. Newspaper Ta Nea (February 26, 2010). From http://
democracy-rethymno.blogspot.com/2010/03/blog-post_03.html.
Aggelopoulos, G. (2010b). People come second. Newspaper Ta Nea (January 29, 2010). From http://
www.tanea.gr/opinions/all-opinions/article/4557833/?iid=2.
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Reading argumentative texts: comprehension and evaluation… 1889
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
1890 I.-A. N. Diakidoy et al.
Midgette, E., Haria, P., & MacArthur, C. (2008). The effects of content and audience awareness goals for
revision on the persuasive essays of fifth- and eighth-grade students. Reading & Writing: An
Interdisciplinary Journal, 21, 131–151. doi:10.10007/s11145.007-9067-9.
Murphy, K. P., Holleran, A. T., Long, F. J., & Zeruth, A. J. (2005). Examining the complex roles of
motivation and text medium in the persuasion process. Contemporary Educational Psychology, 30,
418–438. doi:10.1016/j.cedpsych.2005.05.001.
Neuman, Y. (2003). Go ahead, prove that God does not exist! On high school students’ ability to deal
with fallacious arguments. Learning and Instruction, 13, 367–380. doi:10.1016/S0959-
4752(02)00011-7.
Neuman, Y., & Weizman, E. (2003). The role of text representation in students’ ability to identify
fallacious arguments. The Quarterly Journal of Experimental Psychology, 56, 849–864.
Nussbaum, E. M. (2008). Using argumentation Vee diagrams (AVDs) for promoting argument-
counterargument integration in reflective writing. Journal of Educational Psychology, 100,
549–565. doi:10.1037/0022-0663.100.3.549.
O’Brien, E. J., & Cook, A. E. (2016). Coherence threshold and the continuity of processing: The RI-Val
model of comprehension. Discourse Processes, 53, 326–338. doi:10.1080/0163853X.2015.1123341.
Reznitskaya, A., Anderson, R. C., & Kuo, L.-J. (2007). Teaching and learning argumentation. The
Elementary School Journal, 107, 449–472. doi:10.1086/518623.
Ricco, R. B. (2007). Individual differences in the analysis of informal reasoning fallacies. Contemporary
Educational Psychology, 32, 459–484. doi:10.1016/j.cedpsych.2007.01.001.
Rips, L. J. (1998). Reasoning and conversation. Psychological Review, 105, 411–441.
Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119,
3–22.
Stanovich, K. E., & West, R. F. (2007). Natural myside bias is independent of cognitive ability. Thinking
& Reasoning, 13, 225–247. doi:10.1080/13546780600780796.
Stanovich, K. E., & West, R. F. (2008). On the failure of cognitive ability to predict myside and one-sided
thinking biases. Thinking & Reasoning, 14, 129–167. doi:10.1080/13546780701679764.
Toulmin, S. E. (1958). The use of argument. Cambridge: Cambridge University Press.
van den Broek, P., Lorch, R. F., Linderholm, T., & Gustafson, M. (2001). The effects of readers’ goals on
inference generation and memory for texts. Memory & Cognition, 29, 1081–1087.
van Eemeren, F. H., & Grootendorst, R. (1999). Developments in argumentation theory. In J. Andriessen
& P. Coirier (Eds.), Foundations of argumentative text processing (pp. 43–57). Amsterdam, NL:
Amsterdam University Press.
West, R. F., Toplak, M. E., & Stanovich, K. E. (2008). Heuristics and biases as measures of critical
thinking: Associations with cognitive ability and thinking dispositions. Journal of Educational
Psychology, 100, 930–941. doi:10.1037/a0012842.
Wiley, J. (2005). A fair and balanced look at the news: What affects memory for controversial arguments?
Journal of Memory and Language, 53, 95–109.
Wiley, J., & Myers, J. L. (2003). Availability and accessibility of information and causal inferences from
scientific text. Discourse Processes, 36, 109–129.
Yeari, M., van den Broek, P., & Oudega, M. (2015). Processing and memory of central versus peripheral
information as a function of reading goals: Evidence from eye-movements. Reading & Writing: An
Interdisciplinary Journal, 28, 1071–1097. doi:10.1007/s11145-015-9561-4.
123
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center
GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers
and authorised users (“Users”), for small-scale personal, non-commercial use provided that all
copyright, trade and service marks and other proprietary notices are maintained. By accessing,
sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of
use (“Terms”). For these purposes, Springer Nature considers academic use (by researchers and
students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and
conditions, a relevant site licence or a personal subscription. These Terms will prevail over any
conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription (to
the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of
the Creative Commons license used will apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may
also use these personal data internally within ResearchGate and Springer Nature and as agreed share
it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not otherwise
disclose your personal data outside the ResearchGate or the Springer Nature group of companies
unless we have your permission as detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial
use, it is important to note that Users may not:
1. use such content for the purpose of providing other users with access on a regular or large scale
basis or as a means to circumvent access control;
2. use such content where to do so would be considered a criminal or statutory offence in any
jurisdiction, or gives rise to civil liability, or is otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association
unless explicitly agreed to by Springer Nature in writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a
systematic database of Springer Nature journal content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a
product or service that creates revenue, royalties, rent or income from our content or its inclusion as
part of a paid for service or for other commercial gain. Springer Nature journal content cannot be
used for inter-library loans and librarians may not upload Springer Nature journal content on a large
scale into their, or any other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not
obligated to publish any information or content on this website and may remove it or features or
functionality at our sole discretion, at any time with or without notice. Springer Nature may revoke
this licence to you at any time and remove access to any copies of the Springer Nature journal content
which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or
guarantees to Users, either express or implied with respect to the Springer nature journal content and
all parties disclaim and waive any implied warranties or warranties imposed by law, including
merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published
by Springer Nature that may be licensed from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a
regular basis or in any other manner not expressly permitted by these Terms, please contact Springer
Nature at
onlineservice@springernature.com