Full text of DOI:10.3758/BF03206376

Memory & Cognition
001, 29 (8), 1081-1087
2
The effects of readers’ goals on inference
generation and memory for texts
PAUL VAN DEN BROEK
University of Minnesota, Minneapolis, Minnesota
ROBERT F. LORCH, JR.
University of Kentucky, Lexington, Kentucky
and
TRACY LINDERHOLM and MARY GUSTAFSON
University of Minnesota, Minneapolis, Minnesota
We investigatedthe effectsof readers’goals on inferencegenerationand memory for expository text.
College students (N = 82) read texts for the purpose of eitherstudy or entertainment. On-line inference
generation was recorded via think-aloud procedures, and off-line memory was assessedvia free recall.
Reading goalstronglyinfluencedinferentialactivity:Readerswith a study goalproduced more coherence-
building (i.e., backward/explanatory and forward/predictive) inferences, whereas readers with an en-
tertainment goal produced more associations and evaluations. These differenceswere associatedwith
superior memory for the texts in the study condition. The resultsindicatethat inference generationdur-
ing reading is partly strategic and is influenced systematically by reading purpose. We propose that
reading goals influence readers’ standards of coherence, which in turn influence the types of infer-
ences that they draw and the final memory representations that they construct.
We read texts for many different purposes. Do these pur- their goals in understanding the text (Graesser, Singer, &
poses influence the kinds of inferences that we generate, Trabasso,1994;McKoon& Ratcliff, 1992;Singer, Graesser,
and, if so, in what way do they do so? It is generallyagreed & Trabasso, 1994; van den Broek, Young, Tzeng, & Lin-
that the generation of inferences is an integral component derholm, 1999). The abilityto create inferences is affected
of reading comprehension. Inferences allow readers to by reading skill, understandingof the demands of the read-
connectthe differentparts of thetextand,hence,to perceive ing task, working memory capacity, and background
thetext as consistingofa series of connectedeventsor facts knowledgerelevant to the text topics. However, how read-
rather than isolated statements (Black & Bower, 1980; ers applytheir skillsand knowledgedependson their goals
Graesser, 1981; Kintsch, 1988; Kintsch & van Dijk, 1978; or purpose for reading. In this study, we examine the pos-
Trabasso, Secco, & van den Broek, 1984). However, infer- sible effects of readers’ goals on the production of infer-
ence making and comprehensionof a text vary across read- ences and recall of texts.
ing situationsand across readers (see, e.g., Lee-Sammons
Whitney, 1991;Lorch, Lorch, & Matthews, 1985;Lorch, is subjectto some constraints.Various factors affect the ex-
Lorch, & Mogan, 1987; Noordman, Vonk, & Kempff, tent to which readers can allocate resources to the genera-
992; van den Broek, Fletcher, & Risden, 1993; van den tion of inferences,regardless of theirreadinggoals. For ex-
Broek, Tzeng, Risden, Trabasso, & Basche, 2001; Walker ample, comprehension problems due to inefficient
Meyer, 1980;Zwaan, 1994). According to current mod- decoding of words (Perfetti, 1994), limited lexical and
The idea that readers’ goalscontrolinferentialprocesses
&
1
&
els of reading, comprehensionof a text is affected both by syntactic skills, and limited working memory capacity
readers’ abilitiesto generate appropriate inferencesand by (Just & Carpenter, 1992) may inhibit readers’ ability to
generate inferences (Oakhill, 1994). Likewise, language
proficiency and comprehension skill affect the number
and kind of inferences that readers generate (Zwaan &
Brown, 1996).
This research was supported by the Center for Cognitive Sciences at
the University of Minnesota through a grant from the National Institute
of Child Health and Human Development (HD-07151), by the Guy
Bond Endowment for Reading and Literacy, and by a Golestan fellow-
Despite these constraints, college students report using
ship at the Netherlands Institute for Advanced Study in the Humanities different strategies when the reason for reading varies
and Social Sciences to the first author. We thank Art Glenberg, Richard (Lorch, Klusewitz, & Lorch, 1995;Lorch, Lorch, & Kluse-
Marsh, and Tom Trabasso for their comments on an earlier draft of this
witz, 1993). For example, according to their self-reports,
paper. Correspondence concerning this article should be addressed to
reading in order to study involves slow reading, memo-
P. van den Broek, Department of Educational Psychology, University of
rization, and monitoringof understanding.In comparison,
readingfor entertainmentinvolvesfast reading, no reread-
Minnesota, 178 Pillsbury Drive SE, Minneapolis, MN 55455 (e-mail:
pvdbroek@umn.edu).
1
081
Copyright 2001 Psychonomic Society, Inc.

1
082
VA N D EN BROEK, LORCH, LINDERHOLM, AND GUSTAFSON
ing, and generally less careful attention to the text. These reader’s goals would rely on some unspecified process by
reports suggest that readers may match at least some of which the reader constantly checks his or her understand-
their processing of texts to the reading situation.
ing against an abstract goal. Second, the notion of stan-
In fact, there is evidence that the pattern of inferences dards of coherence allows the integrationof two distinct—
that readers make is related to their reading purpose (e.g., and sometimes opposing—theoretical perspectives of
Lorchetal.,1985;Lorchet al.,1987;Noordman et al., 1992; reading comprehension, those emphasizing bottom-up
Schmalhofer & Glavanov, 1986; Zwaan, Magliano, & processes (e.g., McKoon & Ratcliff, 1992) and global,
Graesser, 1995). In general, this evidence suggests that top-down processes (e.g., Graesser et al., 1994), respec-
readers are sensitive to the instructionsgiven to them prior tively. We propose that bothtypesof processesoccurin text
to reading, and that their processing is oriented toward the comprehension,with the reader ’s standardsof coherence—
goal provided through those instructions.
and the ease with which they can be attained—determining
Despite the evidence that inference generation is influ- theexactcombinationofbottom-upandtop-downprocesses
enced by readers’ purpose for reading, there has been lit- at each point during reading (van den Broek et al., 1993;
tle systematic attention in the cognitive literature to the van den Broek et al., 1995). Third, the concept of stan-
connection between particular goals and the resulting in- dards of coherence connotes a limited number of stan-
ferential pattern and the mechanism by which goals affect dards that can be applied over an indefinite number of
inferencemaking.Thissituationisdue,at leastinpart, tothe reading purposes—as well as other variations in reading
lack of a theoreticalframework to guidesuch research. We situations,due to text genre, readingtask, and so on. Thus,
propose that the concept of standards of coherence may despite the myriad of different reading goals that a person
provide a productive framework for capturing how read- may face in his or her experience,the reader translates any
ing goals affect inference making in specific, systematic given goal into relevant standards of coherence, thereby
ways. In general, this conceptsuggeststhat as readers pro- operationalizing the goal. In summary, standards of co-
ceed through a text, they maintain standards of coherence herence provide both explanatory power and theoretical
that act as criteria for comprehension. These standards or parsimony for the role of reading goals in reading com-
criteria, in turn, dictate the inferential activities in which prehension.
the readers engage at each point during reading. The in-
In this study, we use a think-aloud procedure and a re-
ferential activities that are employed directly determine call task to investigate readers’ processing of expository
the level of comprehensionachieved.For different reading text in two different reading tasks. College studentsread a
situations, such as when the text genre, reading task, and/ text for the purpose of preparing for an essay examination
or reader motivation varies, readers systematically alter or for the purpose of entertainment.We selected these two
their criteria for comprehension and, hence, generate pre- purposes because students perceive reading for entertain-
dictably different patterns of inferences (van den Broek, mentandreadingfor studyasverydistinctreading“modes,”
Risden, & Husebye-Hartmann, 1995; see also Hacker, and furthermore, there is reasonable agreement within
1
998).
The notion of standards of coherence is in line with re- proach to reading (Lorch et al., 1995; Lorch et al., 1993).
centmodelsof readingcomprehension(Goldman&Varma, Thus, the two tasks should contrast sharply in the stan-
995; van den Broek et al., 1999; Zwaan et al., 1995). dards of coherence they lead readers to adopt.
These models implicitly or explicitlyposit that as readers
We chose to use a think-aloudmethodologyto study in-
each condition about what constitutes an appropriate ap-
1
proceed through a text, they attempt to maintain certain ferencing because this methodology is well suited to the
levels of coherence and engage in different inferential exploratory investigation of strategic processing. Think-
processes to do so. Thus, readers’ standards of coherence aloud methodologies can provide valuable insights into
determine to a large extent under what circumstances var- the developing representations of readers (Ericsson &
ioustypesofinferences(e.g., reinstatementsfrom prior text, Simon, 1993;Long & Bourg,1996;Pressley & Afflerbach,
backgroundknowledge inferences, forward/predictivein- 1995). Furthermore, they allow one to consider a wide
ferences, associations, etc.) are generated on line, as well array of different responses by the reader. When coupled
as the kinds of coherence that these inferences construct with measures of recall, think-aloud protocols help pre-
(
e.g., referential/anaphoric, causal, spatial, logical). Al- sent a rich view of readers’ strategies and effectiveness in
thoughit is assumed in these models that reading goals in- comprehending text.
fluence the standards that readers employ, none of the
models provides details on how goals may do so.
We investigatedthe effects of readingtask on coherence-
producing inferences by categorizing the think-aloud re-
Thus, the notion of standards of coherence is useful for sponses of readers to each sentence in each of four texts.
several reasons. First, it provides an explanatory account Three of the response categories corresponded to distinct
of precisely how goals influence the reading process. Ac- classes of inferences: backward, explanatory inferences;
cording to this account, a reading goal prompts the reader forward, predictive inferences; and lateral, associative in-
to select particular standards to maintain as he or she pro- ferences. In addition,we coded think-aloudresponses into
ceeds through the text. Without the construct of standards five other categories: paraphrases; evaluative comments;
of coherence, the claim that processing is influenced by a self-monitoring of comprehension; emotional, affective

THE EFFECTS OF READERS’ GOALS
1083
comments; and explicitrepetitionof text phrases (adapted
from Coté, Goldman, & Saul, 1998; Narvaez, van den
Broek, & Ruiz, 1999; Pressley & Afflerbach, 1995; Tra-
basso & Magliano, 1996; Zwaan & Brown, 1996).
Our main hypothesiswas that readers’ purpose for read-
ing would affect the frequency with which they made
METHOD
Participants
Participants were 82 volunteers from undergraduate psychology
classes at the University of Minnesota. Results from 10 participants
were eliminated because portions of their audio recordings were in-
audible. The data from another participant were omitted from all
coherence-producinginferences. In general, we expected analyses because of missing recall data. Of the remaining 71 partic-
that readers with the goal of studying for an exam would ipants, 50 were women and 21 were men. Their ages ranged from 17
to 51 years, with a mean age of 22 years. All participants received
seek to construct a highly coherent representation of the
course credit.
textthey read. Thus, they would generate many coherence-
producing inferences but produce few associative infer-
ences. Explanatory and predictive inferences are exam-
ples of processes that readers use to further comprehend a entific American magazine articles. The average number of words
text event, the former by connecting the currently read was 724.75 (range = 656–771), the average number of sentences was
statementto its antecedentsand the latterby identifyingits
implications. In contrast, associations typically provide
elaborative or descriptive detail that does not contribute
to a more coherent text representation. This pattern of in-
ferences for readers in the study condition would mirror
Materials
The materials consisted of four expository texts adapted from Sci-
3
6.25 (range = 32–40), and average text difficulty was at grade 11.2
(
Flesch-Kincaid Grade Level reading measure, range = 10–12).
Each text began with an introduction of a problem found in nature,
and then discussed a series of alternative explanations proposed as
answers to the problem. The topics of the texts were the origin of the
moon, the migration of sea turtles, the development of treatments
previously observed patterns for narrative (Trabasso & for meningitis, and the decline in songbird populations. The texts
Magliano, 1996; Zwaan & Brown, 1996) and expository were presented on index cards, with one typed sentence per card.
Each text had two different versions, in which 12 target sentences
texts (Coté et al., 1998).
were manipulated with the goal of affecting the amount of inferen-
We expected that readers’ use of paraphrasing, explicit
tial processing in response to each target sentence. However, these
manipulations did not influence the observed patterns of inference
generation and therefore will not be discussed further.
repetition, and self-monitoring also would reflect their
goals. One would expect readers in the study condition to
make more self-monitoring statements than readers in the
entertainment condition. Although it seems intuitively Procedure
The participants were randomly assigned to either a study or an
clear that monitoringis a direct reflectionof a standard for
comprehension,the role of paraphrasing and repeating of
text is less clear. On the one hand, it might be that para-
phrasing and repeating text indicate readers’ attempts to
understand it—for example, by extendingprocessing time
entertainment condition and were tested individually. Test sessions,
including instructions and practice, averaged 45 min; all participants
completed the task within 1 h.
The participants were seated at a table with a tape recorder and,
depending on condition, either college textbooks or magazines.
(
Coté et al., 1998; Trabasso & Magliano, 1996). On the Textbooks or magazines were displayed in an attempt to help the par-
other hand, generating inferences requires time to com- ticipants maintain either a study or an entertainment goal in reading.
Before reading the expository texts, the participants in the study
plete (Keenan, Baillet, & Brown, 1984), so paraphrasing
condition were asked to imagine that they were preparing for an
and repeating the text might actually occur in lieu of in-
essay exam, whereas the participants in the entertainment condition
were asked to imagine that they were browsing through a magazine
ferences. The relations between standards of coherence
and paraphrasing and repeating text are unclear, so we
made no specific predictions about these strategies.
In a similar sense, evaluative or affective comments
might indicate a reader’s involvementwith a text. The po-
tentialnature of evaluationsor emotionalcomments as de-
fined in our study covered a broad spectrum; they might
indicate a deep interest in the text or a more surface in-
and had come upon an article of interest. All participants received
the same instructions regarding the think-aloud task.
The procedure for the think-aloud task was based on existing
methods (Narvaez et al., 1999; Trabasso & Magliano, 1996; Zwaan
&
Brown, 1996). The participants read one of the texts out loud from
the index cards. They were asked to say out loud any thoughts that
came to mind following each sentence, before turning over the index
card and proceeding to the next sentence. Before reading their as-
volvement.In both cases, they reflected a reader’s personal signed text, the participants practiced thinking aloud by reading the
experience with a text rather than an attempt to compre- first seven sentences from one of the other experimental texts. The
experimenter reminded the participants of the think-aloud instruc-
hend a text at that specific moment. We expected, there-
tions in both the practice and experimental sessions if they did not
fore, that these comments would also reflect a reader’s
produce think-aloud responses for two consecutive sentences. The
standards of coherence:Readers in the entertainmentcon-
responses were recorded by tape recorder.
After completing the text, the participants answered 10 written
mathematics questions as a distractor task. The distractor task was
dition would make more evaluations and affective com-
ments than readers in the study condition would.
To summarize, we expected that participantsreading to assigned to counteract any recency effects during recall of the text.
The participants then completed a written free recall task. They were
study would engage in processing aimed at understanding
instructed to write everything that they could remember from the
the internal structure of the textual information, making
text, duplicating it as closely as possible.
more forward and backward inferences and engaging in
more self-monitoring.In contrast, participantsreadingfor
entertainmentwould be more likelyto relate text events to
their own experiences, making fewer text-connecting in- from the audiotapes, and the predominant think-aloud comment as-
ferences and more associationsand evaluativecomments. sociated with each sentence in the text was scored for category of re-
Scoring
Think-aloud protocols. Think-aloud protocols were transcribed

1
084
VA N D EN BROEK, LORCH, LINDERHOLM, AND GUSTAFSON
sponse by two judges who had no knowledge of the experimental sentences of the text (means were computed instead of
condition. A comment was considered predominant if the two judges sums, because the number of sentences per text varied
considered it to constitute the essence of the utterance in response to
slightly). These data, provided in Table 1, clearly show
the sentence under consideration. The total numbers of responses
different patterns of responses in the study and entertain-
were 816 and 988 for the entertainment and study conditions, re-
ment conditions.To test the specific predictionsconcern-
spectively. The categories of response were similar to those used by
ing each response category, separate analyses of variance
included association (retrieval of information not related to text co- (ANOVAs) were computed for each of the eight response
herence; e.g., “That reminds me of a documentary I saw on sea tur- categories.EachANOVA includedthebetween-participants
Trabasso and Magliano (1996) and Zwaan and Brown (1996). They
tles.”); backward inference/explanation (retrieval of information
factors of text (four levels) and reading task (study or en-
from either prior text or background knowledge that explains the
tertainment). Tests were conducted against both parti-
current text sentence; e.g., “Oh, this is related to what I read earlier
about how magnetic field affects wave direction.”); predictive infer-
ence (forward inferences that anticipate upcoming text; e.g., “I bet
cipant variability (F ) and item variability (F ). All re-
1
2
ported effects are reliableat or beyondp = .05 unlessnoted
the turtles will use sense of smell to return to their nesting place.”); otherwise.
paraphrase (comments that capture the gist meaning of a sentence;
e.g., “So this sentence is saying that turtles use waves to get to where
they need to go.”); evaluation (opinions about the text; e.g., “These
turtles are really intelligent creatures.”); monitor comprehension (re-
flecting on one’s own understanding, e.g., “Oh, I never knew that!”);
affective response (emotional comment regarding the text; e.g., “It
makes me sad to think that hatchlings could be eaten by predators.”);
It was predicted that the readers in the study condition
would generate more coherence-producing explanatory
and predictiveinferences than would the readers in the en-
tertainmentcondition.The datain Table1 demonstratethat
both predictionswere confirmed. First, more explanatory
inferences were made in the study condition than in the
and text repetitions (verbatim repetition of all or a large proportion entertainment condition [F (1,55) = 13.86, MS = .0098;
1
e
of the text sentence). Nonresponses and responses that did not fall F (1,135) = 41.61, MS = .0255]. Second, more predictive
2
e
into any of these categories were scored as other. Interrater reliabil-
ity was 84%. Disagreements between judges were resolved by dis-
cussion. To ascertain that observed patterns were not due to consider-
ation of only predominant responses (as is customary in think-aloud
methodologies), the nondominant responses were also scored. Analy-
inferences were made in the study condition than in the
entertainment condition [F (1,55) = 6.80, MS = .0080;
1
e
F (1,135) = 21.87, MS = .0199].
2
e
The readers with a study purpose were expected to de-
ses of the total set of responses yielded results identical to those for the vote their resources to explanatory and predictive infer-
predominant responses, so only the data on the latter will be reported. ences at the expense of associative inferences, whereas
Free recall. Each text was parsed into clauses to score the partici-
pants’ written free recalls. Each participant’s recall protocol was parsed
into clauses and compared with the text analysis. A text clause was
those reading for entertainment would devote their re-
sources to connecting text information with personal ex-
periences viaassociativeinferences.Table 1 shows that the
credited as having been recalled if the participant recalled all or part of
readers in the entertainment condition produced more as-
the clause verbatim or if the gist of the text was accurately reproduced.
Two judges independently scored 52% of the protocols without knowl- sociations than did the readers in the study condition, al-
edge of the experimental condition. Interrater reliability was 84% and though the effect was marginal in the test against subject
disagreements were resolved by discussion. The remaining written re-
calls were divided among the two judges and scored separately.
variability [F (1,55) = 3.86, MS = .006, .05 , p , .1,
1
e
F (1,135) = 20.57, MS = .0107].
2
e
With regard to monitoring comprehension, it was ex-
RESULTS
pected that the participants studying for an exam would
monitor their comprehension more closely than would
those reading for entertainment.Contrary to this expecta-
Analyses of Think-Aloud Protocols
The response of a participant to each sentence was tion, the readers with an entertainment goal were more
noted, and these responses were then averaged across all likely to make monitoring responses than those preparing
Table 1
The Effect of Reading Purpose on Inferential Activity During Reading
Reading Goal
Entertainment
Study
Response Category
M
SE
M
SE
Explanatory inferences
Predictive inferences
Associations
Monitor
Paraphrase
Repetition
.062
.060
.072
.240
.093
.008
.132
.008
.018
.016
.014*
.018*
.033
.006
.015*
.003
.151
.117
.035
.142
.192
.025
.056
.007
.016*
.015*
.013
.017
.031*
.006*
.014
Evaluation
Affective
.003
Note—Numbers indicate the average frequency of each inference category, per participant and
per statement. The average number of “Other” responses were .325 and .275 for the entertain-
ment and study conditions, respectively. *p , .05

THE EFFECTS OF READERS’ GOALS
1085
for an exam [F (1,55) = 15.54, MS = .0105, F (1,135) = text, the readers with a study goal employed stringent
1
e
2
3
6.58, MS = .0358].
standards focused on intratextual relations, striving for
e
If we turn to paraphrasing and repetition, the partici- deep understanding and coherence in their representation
pants in the study condition produced both types of re- of the texts. By generatinginferences, these readers sought
sponses more often than did the participants in the enter- explanationsand identifiedconsequencesfor the text cur-
tainmentcondition[for paraphrasing,F (1,55)=4.89,MS = rently in their focus of attention. In addition, they fre-
1
e
.
0342, F (1,135) = 42.22, MS = .0291; for repetitions, quently repeated or paraphrased text statements. Para-
2 e
F (1,55) = 4.06, MS = .0012, F (1,135) = 16.49, MS = phrases have been associated with good comprehension
1
e
2
e
.
0041].
and, togetherwith repetitions,may reflect an effort to “buy
Finally, it was expected that evaluative and affective time”for graspingthemeaningofthetext(Cotéet al., 1998;
comments would be more common in the entertainment Smith, 1967;Trabasso & Magliano,1996). These findings
condition, reflecting the greater emphasis of that condi- indicate that readers with a study goal seek to explain and
tion on those dimensions of the reading experience. connect events within a text.
Again, Table 1 confirms that evaluative responses were
The readers in the study condition displayed little self-
more common in the entertainment condition [F (1,55) = monitoring. We had not expected this, because self-
1
1
3.00, MS = .0074, F (1,135) = 34.58, MS = .0211]. monitoringis a strategyfrequentlyused by goodcomprehen-
e 2 e
However, there were no differences between the two read- ders, particularly in study situations (Lorch et al., 1993;
ing conditions with respect to affective responses (both Pressley & Afflerbach, 1995). Closer inspection of the
Fs , 1).
monitoring responses revealed, however, that the range of
responses in this category was quite broad. For example,
they included readers’ confirmation of their understand-
Analyses of Recall Protocols
The recall datawere submittedto a between-participants ing (e.g., “I knew that”), recognition of new knowledge
ANOVA in which the factors were instructionalcondition, (e.g., “I did not know that”), reminders to remember an
text, and text version. As predicted, recall was higher in event (e.g., “I would make note of this because it might be
the study condition (M = 18.20 clauses) than in the enter- on an exam”), and detectionof inconsistencies(e.g., “That
tainment condition (M = 15.08 clauses) [F(1,65) = 4.22, does not fit with what I read before”). Given the wide va-
MS = 37.99]. No other main effects or interactions were riety of responses, results for this category should be in-
e
significant.
terpreted with caution.
In contrast to the readers with a study goal, the readers
with an entertainment goal were much less concerned
Discussion
The aim of this study was to determine whether read- with constructing a coherent representation of the text it-
ers’ goals influence the type of inferences generated dur- self but instead focused more on connectingtext events to
ingreading.Collegereaders read expositorytextswitheither their own personal experiences. These readers generated
a study or an entertainment goal. We assessed on-line in- more associativeinferences (e.g., “This reminds me of the
ferential activityvia think-aloudprocedures, and memory, movie about Apollo 13. I liked the song at the end”) and
via free recall. The think-aloudresults show that the read- evaluativecomments (e.g., “How strange!”) than did their
ers with a study goal generated more backward and pre- peers with a study goal. Both types of responses represent
dictive inferences during reading than did those with an a (momentary) withdrawal from involvement with actual
entertainment goal. The readers with a study goal also textinformation,tothepositionofa commentatorratherthan
more often paraphrased and repeated the information in a reader concerned with comprehension of the text per se.
the text. In contrast, the readers with an entertainmentgoal
The description of the on-line activation patterns in
more often generated monitoring and evaluative state- terms of differing standards of coherence suggests that
ments as well as associations to what they were reading readers with a study goal exert more cognitive effort than
than did those with a study goal. The recall results show do those reading for entertainment. Direct evidence for
that text memory was best for the readers with a study differinglevelsof cognitiveeffort was obtainedina follow-
goal, who remembered 20% more statements than did up study in which college readers read slightly modified
their counterparts with an entertainment goal.
versions of the same texts for either a study or an enter-
tainment purpose at their own pace and without the think-
aloud task (Brannon, 1998). Reading times differed sig-
nificantly as a function of reading goal, with the readers
The Effects of Reading Goals
on Standards of Coherence
The results demonstrate that reading goals affect both with a study goal reading considerably more slowly
readers’ inference making during reading and their subse- (239.32 msec/syllable) than did those with an entertain-
quentmemory for the text. The particularpattern of results ment goal (209.09 msec/syllable).
confirms the notion that readers employ standards of co-
Thus, the readers with a study goal adopted a strict stan-
herence, which guide their inferential activity and mem- dard of coherence, reading relatively slowly and empha-
ory representation for the text, and that readers’ goals in- sizing the need for explanationsand connectionsbetween
fluence these standards. As they proceeded through the text elements, whereas the readers with an entertainment

1
086
VA N D EN BROEK, LORCH, LINDERHOLM, AND GUSTAFSON
goal read more quickly and focused on establishing con- producing inferences will be common, although at a cost
nectionsto their personalexperience.As a consequenceof of greater effort by the reader. If the reader’s goal evokes
the different text processing activities, the readers with a a relatively relaxed standard of coherence, reading will be
study goal had better memory for text contentthan did the less effortful, but the cost will be a less coherent text rep-
readers with an entertainmentgoal. This conclusioniscon- resentation. From this point of view, the questionof which
sistentwith the speculationthatreadingfor studypurposes reading goals are associated with particular standards of
is concerned primarily with the construction of a mental coherence is more fruitful than the question of whether
representation that emphasizes within-text connections, readers’ inferential activities are strategic or not.
whereas reading for enjoymententails the constructionof
In this study, we have investigated the effects of two
a mental representation that emphasizes connections to reading goals: reading for entertainment, and reading for
one’s own background knowledge (e.g., Zwaan et al., study. These particulargoals were chosen because readers
1
995).
experience them as maximally distinctive (Lorch et al.,
987). Indeed, our results indicate that their perceptions
are quite accurate. Other reading goals are possibly quite
1
The Role of Standards of Coherence in Reading
The different patterns of inferences generated by read- distinctive as well (e.g., to obtain a specific piece of in-
ers with distinct reading goals suggest powerful top-down formation, to be able to summarize, etc.). In addition,dif-
effects on readers’ inferential processes. Apparently, spe- ferent standards of coherence may be associated with dif-
cific reading goals induce readers to allocate their efforts ferent text genres. For example, there is ample evidence
differently. The notion of standards of coherence opera- that standards of causal and referential coherence play a
tionalizes how these effects may occur and, in doing so, major role in comprehension of narrative texts (see van
provides a conceptual bridge between research on the in- den Broek, 1994, for a review). For comprehensionof dif-
fluence of reading goals in inference making and recent ferent types of text, the same standards also may be cen-
computationalmodels of text comprehension. According tral or they may be supplemented or substituted by differ-
to these models, readers attempt to maintain coherence as ent standards (Goldman & Bisanz, in press; van den
they proceed throughthe text and engage in inference gen- Broek, Virtue, Gaddy, Sung, & Tzeng, in press). Theoret-
eration in order to do so (Goldman & Varma, 1995; van ical accounts of text comprehension would benefit enor-
den Broek et al., 1999; Zwaan et al., 1995). The defini- mously from a thorough mapping of readers’ goals—and
tions of “coherence” differ among the models. For exam- other factors such as text genre, overall motivation, and
ple, some focus on referential/anaphoric coherence, oth- readingtask—and the standardsof coherence.It seems likely
ers on causal coherence, yet others on global thematic that some standards apply across a broad range of reading
coherence. We propose that readers, in fact, employ differ- situations.Probable candidatesfor such “basic” standards
ent standards—or different combinations of standards— are referential and perhaps causal coherence. Other stan-
dependingon their readinggoals. These standards, in turn, dards, such as thematic coherence, may build on the basic
lead to differences in the type and number of inferences types of coherence, and their adoption may be more situ-
that readers generate. Thus, the notion of standards of co- ation specific. Interestingly, such a view is consistentwith
herence provides both explanatory power and theoretical findings that young children gradually develop the ability
parsimony by connectingabstract goal-basedreading the- to recognize these types of coherence, from referential to
ories to current computational models of reading.
causal, and then to thematic (van den Broek, 1997, for a
Standards of coherence may capture other variations in review).
reading situations as well. For example, text genre, read-
ers’ expectations about the text, reading tasks, and so on,
may evoke specific standards and, hence, elicit specific
patterns of inference generation. Thus, the reader trans-
lates each situation into a combination of relevant stan- Brannon, K. A. (1998). The effects of reading goals on inference con-
REFERENCES
Black, J. B., & Bower, G. H. (1980). Story understanding as problem
solving. Poetics, 9, 223-250.
dards of coherence. In the end, we may observe a rela-
tively limited number of distinct reading “modes”—
characterized by different standards—which, in their var-
ious combinations, account for the myriad of reading sit-
uations that a reader may face.
Standards of coherence also allow us to resolve the ap-
parent contradiction between the theoretical position that
inferential activity is predominantly automatic (Kintsch,
struction. Unpublished master’s thesis, University of Kentucky.
Coté, N. A., Goldman, S. R., & Saul, E. (1998). Students making
sense of informational text: Relations between processing and repre-
sentation. Discourse Processes, 25, 1-53.
Ericsson, K. A., & Simon, H. (1993). Protocol analysis: Verbal reports
as data. Cambridge, MA: MIT Press.
Goldman, S. R., & Bisanz, G. (in press). Toward a functional analysis
of scientific genres: Implications for understanding and learning
processes. In J. Otero, J. A. León, & A. C. Graesser (Eds.), The psy-
chology of science text comprehension. Mahwah, NJ: Erlbaum.
1
1
992; Kintsch & van Dijk, 1978; McKoon & Ratcliff, Goldman, S. R., & Varma, S. (1995). CAPping the construction–
992; Myers & O’Brien, 1998) and the theoretical posi-
integration model of discourse comprehension. In C. A. Weaver,
S. Mannes, & C. R. Fletcher (Eds.), Discourse comprehension: Es-
says in honorof Walter Kintsch (pp. 337-358).Mahwah, NJ: Erlbaum.
Graesser, A. C. (1981). Prose comprehension beyond the word. New
York: Springer-Verlag.
tion that inferentialactivitypredominantlyreflects a goal-
directed search for meaning (Graesser et al., 1994; Singer
et al., 1994; Stein & Trabasso, 1985). If the reader’s goal
evokesa relativelystrict standardof coherence,coherence- Graesser, A. C., Singer, M., & Trabasso, T. (1994). Constructing in-

THE EFFECTS OF READERS’ GOALS
1087
ferences during narrative text comprehension. Psychological Review,
01, 371-395.
Hacker, D. J. (1998). Self-regulated comprehension during normal
reading.In D. J. Hacker, J. Dunlosky, & A. C. Graesser (Eds.), Metacog-
nition in educationaltheory and practice (pp. 165-191).Mahwah, NJ:
Erlbaum.
Just, M. A., & Carpenter, P. A. (1992). A capacity theory of compre-
hension: Individual differences in working memory. Psychological
Review, 99, 122-149.
Keenan, J. M., Baillet, S. D., & Brown, P. (1984). The effects of
causal cohesion on comprehension and memory. Journal of Verbal
Learning & Verbal Behavior, 23, 115-126.
Singer,M., Graesser, A. C., & Trabasso, T. (1994). Minimal or global
inference duringreading. Journalof Memory & Language, 33, 421-441.
Smith, J. K. (1967).The responses of goodand poorreaders when asked
to read for different purposes. Reading Research Quarterly, 3, 53-83.
Stein, N. L., & Trabasso, T. (1985). The search after meaning: Com-
prehensionand comprehension monitoring.In F. J. Morrison,C. Lord,
& D. Keating (Eds.), Applied developmental psychology (Vol. 2,
pp. 33-58). San Diego: Academic Press.
Trabasso, T., & Magliano,J. (1996). Consciousunderstanding during
text comprehension. Discourse Processes, 21, 255-288.
Trabasso, T., Secco, T., & van den Broek, P. W. (1984). Causal cohe-
sion and story coherence. In H. Mandl, N. L. Stein, & T. Trabasso
(Eds.), Learning and comprehension of text (pp. 83-111). Hillsdale,
NJ: Erlbaum.
1
Kintsch, W. (1988). The role of knowledge in discourse comprehen-
sion: A construction–integration model. Psychological Review, 95,
1
63-182.
van den Broek, P. (1994). Comprehension and memory for narrative
texts: Inferences and coherence. In M. A. Gernsbacher (Ed.), Hand-
book of psycholinguistics(pp. 539-588).San Diego: Academic Press.
van den Broek, P. (1997). Discovering the cement of the universe: The
development of event comprehension from childhood to adulthood.
In P. van den Broek, P. Bauer, & T. Bourg(Eds.), Developmental spans
in event comprehension and representation: Bridging fictional and
actual events (pp. 321-342). Mahwah, NJ: Erlbaum.
Kintsch, W. (1992). How readers construct situationmodels for stories:
The role of syntactic cues and causal inferences. In A. F. Healy, S. M.
Kosslyn, & R. M. Shiffrin (Eds.), Essays in honorof WilliamK. Estes:
Vol. 2. From learning processes to cognitive processes (pp. 261-278).
Hillsdale, NJ: Erlbaum.
Kintsch, W., & van Dijk, T. A. (1978). Toward a model of text com-
prehension and production. Psychological Review, 85, 363-394.
Lee-Sammons,W. H., & Whitney, P. (1991). Reading perspectives and
memory for text: An individual differences analysis. Journal of Ex-
perimental Psychology: Learning, Memory, & Cognition, 17, 1074-
van den Broek, P., & Fletcher, C. R. , & Risden, K. (1993). Investi-
gations of inferential processes in reading: A theoretical and method-
ological integration. Discourse Processes, 16, 169-180.
1
082.
van den Broek, P., Risden, K., & Husebye-Hartmann, E. (1995).
The role of readers’ standards for coherence in the generation of in-
ferences during reading. In R. F. Lorch, Jr., & E. J. O’Brien (Eds.),
Sources of coherence in text comprehension (pp. 353-373). Mahwah,
NJ: Erlbaum.
Long, D. L., & Bourg, T. (1996). Thinkingaloud: Telling a story about
a story. Discourse Processes, 21, 329-339.
Lorch, R. F., Jr., Klusewitz, M. A., & Lorch, E. P. (1995). Distinc-
tions among reading situations. In R. F. Lorch, Jr., & E. J. O’Brien
(
Eds.), Sources of coherence in text comprehension (pp. 374-398). van den Broek, P., Tzeng, Y., Risden, K., Trabasso, T., & Basche, P.
Mahwah, NJ: Erlbaum.
(2001). Inferential questioning: Effects on comprehension of narra-
tive texts as a function of grade and timing. Journal of Educational
Psychology, 93, 521-529.
van den Broek, P., Virtue, S., Gaddy, M., Sung, Y., & Tzeng, Y. (in
press). Comprehension and memory of science texts: Inferential
processes and the construction of a mental representation. In J. Otero,
J. León, & A. C. Graesser (Eds.), The psychology of science text com-
prehension. Mahwah, NJ: Erlbaum.
van den Broek, P., Young, M., Tzeng, Y., & Linderholm,T. (1999).
The landscape model of reading: Inferences and the on-line construc-
tion of a memory representation. In H. van Oostendorp & S. R. Gold-
man (Eds.), The construction of mental representations during read-
ing (pp. 71-98). Mahwah, NJ: Erlbaum.
Walker, C. H., & Meyer, B. J. F. (1980). Integrating different types of
information in text. Journal of Verbal Learning & Verbal Behavior,
19, 263-275.
Lorch, R. F., Jr., Lorch, E. P., & Klusewitz, M. A. (1993). College
students’ conditional knowledge about reading. Journal of Educa-
tional Psychology, 85, 239-252.
Lorch, R. F., Jr., Lorch, E. P., & Matthews, P. D. (1985). On-line
processing of the topic structure of a text. Journal of Memory & Lan-
guage, 24, 350-362.
Lorch,R. F., Jr., Lorch,E. P., & Mogan,A. M. (1987). Effects of read-
ing task on on-line processing of a text’s topic structure. Discourse
Processes, 10, 63-80.
McKoon, G., & Ratcliff, R. (1992). Inference during reading. Psy-
chological Review, 99, 440-466.
Myers, J. L., & O’Brien, E. J. (1998). Accessing the discourse repre-
sentation during reading. Discourse Processes, 26, 131-157.
Narvaez, D., van den Broek, P., & Ruiz, A. B. (1999). The influence
of reading purpose on inference generation and comprehension in
reading. Journal of EducationalPsychology, 91, 488-496.
Noordman, L. G. M., Vonk, W., & Kempff, H. F. (1992). Causal infer-
ences during the reading of expository texts. Journal of Memory &
Language, 31, 573-590.
Zwaan, R. A. (1994). Effects of genre expectations on text comprehen-
sion. Journal of Experimental Psychology: Learning, Memory, &
Cognition, 20, 920-933.
Zwaan, R. A., & Brown, C. M. (1996). The influence of language pro-
ficiency and comprehension skill on situation-model construction.
Discourse Processes, 21, 289-327.
Oakhill, J. (1994). Children’s differences in text comprehension. In
M. A. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 821-
8
48). San Diego: Academic Press.
Zwaan, R. A., Magliano, J., & Graesser, A. C. (1995). Dimensions
of situation model constructionin narrative comprehension:An event-
indexing model. Psychological Science, 6, 292-297.
Perfetti, C. A. (1994). Psycholinguistics and reading ability. In M. A.
Gernsbacher (Ed.), Handbookof psycholinguistics(pp. 849-894).San
Diego: Academic Press.
Pressley, M., & Afflerbach, P. (1995). Verbal protocols of reading:
Thenatureofconstructively responsivereading. Mahwah, NJ: Erlbaum.
Schmalhofer,F., & Glavanov,D. (1986). Three componentsof under-
standing a programmer’s manual: Verbatim, propositional, and situa-
tional representations. Journal of Memory & Language, 25, 279-294.
(Manuscript received April 4, 2000;
revision accepted for publication August 12, 2001.)