University of Southern California Dissertations and Theses

Exploring the effects of Korean subject marking and action verbs’ repetition frequency: how they influence the discourse and the memory representations of entities and events

(USC Thesis Other)

Exploring the effects of Korean subject marking and action verbs’ repetition frequency: how they influence the discourse and the memory representations of entities and events

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Exploring the effects of Korean subject marking and action verbs’ repetition frequency: How
they influence the discourse and the memory representations of entities and events

by

Lucy Kyoungsook Kim

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the Requirements for the Degree
DOCTOR OF PHILOSOPHY (LINGUISTICS)

August 2015

Copyright 2015 Lucy Kyoungsook Kim

1

Abstract
Language and memory are two of the most fundamental components of human cognition.
Individuals’ linguistic performance relies greatly on memory competence. At the same time,
linguistic expressions can also influence what comprehenders remember and how well they
remember it. Due to the close connection between these two cognitive domains, it is important to
study how linguistic descriptions can influence the representations that language users build in
their minds and how those representations are reflected in discourse and memory.
This dissertation investigated the interplay between language and memory in the areas of
entity representation denoted by nouns and event information denoted by verbs. Combining
research on nominal information and verbal information is beneficial for our understanding of
language processing as they are two vital aspects of sentence meaning.
In the nominal domain, I conducted three experiments to explore the effect of Korean
subject case marking, more specially the difference between the nominative case marker and the
topic case marker (when interpreted as marking a contrastive topic). My results show that how
subjects are interpreted (e.g., a default topic or a contrastive topic) has an effect on
comprehenders’ expectations about which entities will be mentioned in upcoming discourse.
Furthermore, the experiments also show that the order in which characters are introduced
in a story influences how well comprehenders recognize them in a memory task (i.e., a recently-
mentioned entity is recognized more accurately than an earlier-mentioned entity). However, a
follow-up off-line story continuation task revealed that the entity that comprehenders
remembered better is not the entity they talk about frequently in story continuation. Rather, the
first-mentioned character was mentioned more often in subsequent sentences. Put together, my
results point to a possible disconnect between accessibility in discourse and accessibility in
2

memory: An entity’s prominence in discourse and memory are not always parallel. This suggests
that prominence is a multifaceted concept that can be a product of many different factors
including: (i) how and where in the discourse an entity is introduced; (ii) who else is in the
discourse; and (iii) different reading goals that can affect language processing strategies.
In the verbal domain, I investigated whether and how the frequency semantics of an
action verb would influence sentence processing. My results show that comprehenders’
sensicality judgment was more accurate for the sentences containing action verbs of low intrinsic
repetition frequency (i.e., actions which typically only happen once or twice in a row) compared
to the sentences denoting high frequency actions (i.e., actions which typically happen multiple
times in a row). Moreover, in a verb recognition task, verbs of low repetition frequency were
recognized more accurately than verbs of high repetition frequency. These results add to the
existing literature that demonstrates the effect of other semantic properties that have been shown
to influence comprehenders’ linguistic and nonlinguistic behaviors. Those semantic aspects
include the speed of an action, the duration of a motion, and physical effort necessary for a
motion.
Combining all these results, this dissertation emphasizes the importance of investigating
individuals’ language processing strategies from diverse perspectives including: (i) comparing
how an entity is represented in discourse and in memory; (ii) investigating both nominal
information and verbal information in a sentence; and (iii) investigating multiple factors that
influence a linguistic concept’s accessibility in discourse and in memory.

3

Acknowledgments
I would like to express my deepest gratitude to my advisor, Dr. Elsi Kaiser, for her support and
guidance throughout this research. Her continued support and willingness to give her time so
generously has been very much appreciated.

I would also like to extend my sincere appreciation to my committee members: Dr.
Roumyana Pancheva, Dr. Andrew Simpson, and Dr. Irving Biederman for their advice during
my research.

I also gratefully acknowledge the support that I have received during my studies at the
University of Southern California including teaching assistantship opportunities from the
departments of Linguistics, Psychology, East Asian Studies Center, and East Asian Languages
and Cultures, as well as the generous fellowships from the USC Graduate School.

I would also like to offer my deepest gratitude to my family. Their love and encouragement have
been the driving force in my pursuit of this degree. Also special thanks to my friends at the
Power of Praise Church and the Linguistics Department at USC for their friendship, support, and
prayers.

Finally, my deepest appreciation is extended to my best friend, Dr. Ali M. Alghazo, who has
never stopped encouraging me during all my studies in the United States.

4

Table of Contents

Abstract…………..……………………………………………..……...………………………….1
Acknowledgments……………………………………………..……...……………..…………….3
List of Tables…………………………………………………………………………...……...….5
List of Figures……………………………………………………………...………………..…….7
Chapter 1..…………………………...………………………………………….…………...…….8
Introduction
Chapter 2..………………………………………………………………………….…………….34
Experiment 1: The effect of Korean subject marking on entities’ discourse representations
Chapter 3..………………………………………………….…………………………………….76
Experiment 2: The effect of Korean subject marking on entities’ memory representations
Chapter 4..…………………………………………...………………………………………….109
Experiment 3: Comparison between entities’ discourse and memory representations
Chapter 5..………………………………………………………………………………………143
Experiment 4: The effect of verb semantics on sentence processing and memory recall
Chapter 6..………………………………………………………………………………………178
General discussion and conclusion
References……………………………………………………………...…………………...…..187
Appendix A..……………............................................................................................................223
List of the occupation nouns used in Experiment 2
Appendix B..……………………………………………………………………………………225
List of the verb phrases used in Experiment 4

5

List of Tables

Table 1: The Givenness Hierarchy (Gundel, Hedberg, & Zacharski, 1993)…………...……… 35
Table 2: Givón’s topic continuity/predictability (1983a/b)……………………………………..36
Table 3: Ariel’s Accessibility Marking Scale (1990)…………………………………………...37
Table 4: English translation of each of the four sentences in the sample target story in (20)…..54
Table 5: Target story design in Experiment 1…………………………………………………...55
Table 6: Most commonly occurring referents of the continuation sentence subjects …………..58
Table 7: Sample continuation sentences that include Discourse Subject, Mentioned Member,
and Unmentioned Member as subjects ………………………………………………….………59
Table 8: Percentage of each of the three most common referents in Continuation Sentence 1…61
Table 9: Percentage of each of the four most common referents in Continuation Sentence 2….64
Table 10: Percentage of each of the five relevant referents in Continuation Sentence 3……….67
Table 11: Significant differences in the referent percentages between NOM- and TOP-
conditions………………………………………………………………………………………...69
Table 12: In order of most frequent referents within each continuation sentence (%) …………70
Table 13: Factors influencing discourse and memory prominence …………………………….84
Table 14: Experiment 2 target story configuration ……………………………………………..86
Table 15: Four target conditions in Experiment 2………………………………………………88
Table 16: Percentage of correct ‘YES’ responses for probed entities across the target
conditions..………………………………….……………………………………………………94
Table 17: Mean and SDs of probe word recognition RTs across the four experimental
conditions..……………………………….………………………………………………………96
Table 18: Percentage of correct responses for the comprehension questions…………………...98
6

Table 19: Factors influencing discourse and memory prominence …………………………...119
Table 20: Sample story and continuations 1, Experiment 3…………………………………...124
Table 21: Sample story and continuations 2, Experiment 3…………………………………...125
Table 22: Continuation sentences with Subject1, Subject 2, Both Subjects, and
Context Subject……..…………………………………………………………………………..127
Table 23: Percentage of the grammatical subjects in Continuation Sentence 1, Experiment 3..128
Table 24: Percentage of the grammatical subjects in Continuation Sentence 2, Experiment 3..131
Table 25: Percentage of the grammatical subjects in Continuation Sentence 3, Experiment 3..133
Table 26: Percentage of the grammatical subjects in all three continuation sentences, Exp. 3..134
Table 27: Percentage of the grammatical subjects across three continuation sentences, Exp.3.136
Table 28: Four experimental conditions of Experiment 4……………………………………..157
Table 29: Sample targets in each condition, Experiment 4……………………………………159
Table 30: Summary of predictions of Experiment 4. ………………………………………….165
Table 31: Sentence sensicality judgment accuracy rates, Experiment 4. ……….…………….166
Table 32: Verb recall accuracy rate, Experiment 4. …………………………………………...169

7

List of Figures

Figure 1: A target trial of Experiment 1 (story continuation)....................................................... 57
Figure 2: Percentage of DS, MM, and UM1 referents in Continuation Sentence 1, Exp. 1......... 62
Figure 3: Percentage of each of the four most common referents in Continuation Sentence 2.....65
Figure 4: Percentage of each of the five relevant referents in Continuation Sentence 3…...........68
Figure 5: Percentage of correct ‘YES’ responses for probe words across target conditions..…...94
Figure 6: Time taken to correctly recognize the probed protagonists, Experiment 2....................96
Figure 7: Percentage of correct responses for the comprehension questions, Experiment 2.........98
Figure 8: Grammatical subjects of Continuation Sentence 1, Experiment 3...............................129
Figure 9: Grammatical subjects of Continuation Sentence 2, Experiment 3...............................131
Figure 10: Grammatical subjects of Continuation Sentence 3, Experiment 3............................ 133
Figure 11: Grammatical subjects of all three continuation sentences combined, Experiment 3.135
Figure 12: Sentence sensicality judgment accuracy rates, Experiment 4………………………167
Figure 13: Verb recall accuracy rate, Experiment 4………………………………....................169

8

Chapter 1
Introduction

1.1 Language and memory as cognitive properties
Language is a vital property of human cognition. It is one of the defining characteristics of
humans, which sets us apart from other species (e.g., Christiansen & Kirby, 2003; Enard et al.,
2002; Gleitman & Papafragou, 2005; Hauser, Chomsky, & Fitch, 2002; McCarthy, 1933; Miller,
1962; Pinker & Jackendoff, 2005). The fact that we can express ideas that are physically absent
or ideas beyond the present time frame, including past events and future goals, exemplifies the
importance of language as a cognitive device. For example, we can talk about what we did on
our last vacation or what might happen in our lives ten years from now. Language also functions
as important communicative means (e.g., Beattie & Ellis, 2014; Jacobs, 2002; Singer, 2013). It
allows us to express and transfer our beliefs, thoughts, and intentions to others (e.g., Willems et
al., 2011). Just as individuals’ thoughts vary, so do the ways in which we linguistically express
those ideas. For that reason, understanding the cognitive processes underlying our linguistic
behaviors is a multifaceted task.
Researchers have long been investigating language from a variety of different
perspectives including the grammatical structures of language, the patterns and coordinations of
sounds in language, the meanings of words and phrases, and the communicative functions of
language, to name a few. Amongst these areas, this dissertation aims to examine the relationship
between the human sentence processing mechanism, primarily with regard to discourse
processing, and the memory representations of linguistic concepts. In particular, this dissertation
aims to explore whether the way in which individuals express ideas linguistically (e.g., whether a
9

noun is marked as a topic or not, or whether we use verbs denoting actions that repeat a few
times in a row or many times in a row) influences how information is processed, encoded, and
represented in the minds of language users. In sum, my study examines how the way in which
entity information and verb information is presented in text can influence comprehenders’
processing of discourse and their memory retention of the linguistic information.
Along with language, memory is another important component of human cognition (e.g.,
Anderson, 1994; Baddeley, 1992; Norman, 2013; Singer, 2013; Tulving, 1984). In most cases,
our memory remains intact for our daily functions. We remember where the light switches are in
our homes, how to get to work, and when to pick up the kids from school. Just as language is a
vital part of our daily activities, so is memory. Undoubtedly, memory plays an important role in
our linguistic competence (e.g., Adams & Gathercole, 2000; Baddeley, 1992; Cowan, 1996;
Gathercole, Willis, Emslie, & Baddeley, 1992; Gathercole & Baddeley, 1990; Gillam & van
Kleeck, 1996; Loftus & Loftus, 1976; McCutchen, 1996; Montgomery, 1995; Swanson,
1999). Physiologically speaking, without implicit memory or ‘unconscious remembering’, we
would not recall which articulatory muscles should coordinate to create certain sounds to
produce speech or signs. Without memory, we would not remember how the syntax of our own
language works, for example, whether adjectives come before nouns or after. The importance of
memory on linguistic performance has been widely demonstrated in the developmental literature.
For example, memory deficiency hinders efficient language development or syntactic processing
(e.g., Adams et al., 2000; Cowan, 1996; Perham et al., 2013; Weismer et al., 1999).
The relationship between language and memory has been studied extensively over many
years (e.g., Baddeley, 2003; Dillon, Chow, Wagers, Guo, Liu, & Phillips, 2014; Frackowiak,
1994; Frank, Trompenaars, & Vasishth, 2015; Gibson, 1998; Gordon, Hendrick, & Johnson,
10

2001; Lewis, Vasishth, & Van Dyke, 2006; Loftus & Palmer, 1974; Phillips, Kazanina & Abada,
2005; Phillips, Wagers, & Lau, 2011; Slevc, 2011; Ullman et al., 1997; Van Dyke & Johns,
2012). Particularly, studies have shown that working memory is a basis for language learning
and language processing (e.g., Adams & Gathercole, 2000; Engel de Abreu, Gathercole, &
Martin, 2011; Gathercole, Alloway, Willis, & Adams, 2006; Gupta, 2003; Service, 1992; Slevc,
2011; Swanson, 2003). According to some researchers, working memory is a short-lived
memory system that any information goes through before it is stored in short-term memory
1
.
Working memory allows comprehenders to simultaneously encode, store, and manipulate
information (e.g., Baddeley, 1995). The role of working memory in language development has
been demonstrated by many researchers including Adams and Gathercole (2000) who showed
that four-year-old children with better phonological memory capacities produce more elaborated
(e.g., richer vocabulary) and longer sentences compared to children with poorer phonological
memory abilities. As is evident from prior studies, memory is an essential component for
individuals’ linguistic performance.

1.2 The role of memory in language processing
How we process linguistic information, including the processing of words, phrases, and
sentences, is linked to our memory capacity
2
. The interaction between memory and language can

1
There have been different distinctions made regarding the relationship between working memory and short-term
memory (e.g., Anderson, 1990; Conway, Cowan, Bunting, Therriault, & Minkoff, 2002; Cowan, 1995; Seamon &
Kenrick, 1994). Some researchers view them as similar concepts (e.g., Anderson, 1990), and others consider
working memory to be part of short-term memory (e.g., Seamon & Kenrick, 1994). Generally, short-term memory is
considered to be a simple storage buffer where information is stored through rehearsal and chunking (i.e., grouping
of information into smaller and more manageable units). However, working memory is suggested to be more
complex in that not only does it consist of a storage component like short-term memory, but also has an attention
component that is necessary for maintaining information in memory against distractions and attention shifts (e.g.,
Baddeley & Hitch, 1974; Conway et al., 2002; Engle, Tuholski, Laughlin, & Conway, 1999; Miyake & Shah, 1999).
2
In the study of the interplay between language and memory, researchers have investigated the encoding and the
storage processes of memory formation. For example, the encoding phase of sentence meaning onto memory
11

be easily found starting at the sub-word level (e.g., see Houston & Jusczyk (2003) for infants’
memory for sound patterns), as well as at the word level. Studies demonstrate that individuals’
memory for words changes depending on the types of words. For example, Walker and Hulme
(1999) showed that people remember concrete words that refer to real tangible objects (e.g.,
house) better than abstract words that refer to ideas and concepts (e.g., risk). Individuals have
also been shown to remember words associated with either positive or negative emotions better
than neutral, non-emotional words (e.g., Ferré, Fraga, Comesaña, & Sánchez-Casas, 2014;
Kensinger & Corkin, 2003).
Going beyond the word level, memory also impacts sentence-level performance. The
information available in our focal attention (or highly accessible in memory) can change the way
we formulate sentences. For example, speakers tend to place more accessible concepts, such as
imageable and animate information, earlier in the sentence (e.g., Bock & Warren, 1985;
McDonald, Bock, & Kelly, 1993). Semantically primed information (that might be stored in
implicit memory) tends to come at the beginning of the sentence as well (e.g., Bock, 1986). In
addition, the information structure
3
of a sentence, including Topic, Focus, and Givenness,

representation has been shown to be influenced by grammatical structure (e.g., Fedorenko, Woodbury, & Gibson,
2013; Gordon, Hendrick, & Johnson, 2004; Hsiao & Gibson, 2003; Warren & Gibson, 2002). When a sentence
structure is complex, individuals tend to take longer to read it, and the accuracy of the tasks they perform with the
sentence tends to be low (e.g., sensicality judgment of the sentence or comprehension questions). However,
processing a simpler sentence takes less time, and the accuracy of the task is higher. Processing syntactically
complex sentences is believed to be more memory-intensive than processing simple sentences. There are however
other researchers who believe that sentence complexity has no effect on working memory load or processing
demand (e.g., Conway & Eagle, 1996; Halford, Maybery, O'Hare, & Grant, 1994; Towse, Hitch, & Hutton, 1998;
2002). These researchers believe that processing sentences and holding items in memory come from separate
cognitive resources. However, researchers including Gordon, Hendrick, and Levine (2002) and Fedorenko,
Woodbury, and Gibson (2013) view that comprehending sentences and holding words in memory rely on the same
memory resources.
3
Information structure is a sentence-level way of characterizing different components of a sentence with regards to
discourse-level concepts such as topic, focus, and givenness, as well as their complementary notions of comment,
background, and newness, respectively (e.g., Chafe, 1976; Krifka, 2008; Lambrecht, 1996; Roberts, 1996). For
example, a sentence may consist of an element that denotes a topic and a component that denotes a comment. Topic
is the information that conveys what a sentence is about, and comment is often the predicate information, which
expresses what is being said about the topic, as in: [The dog] topic [is chasing the cat.] comment
12

contributes to the way sentences are formulated (e.g., Büring, 1997; Lee, Gordon, & Büring,
2006). Speakers place given information that has already been introduced in the discourse earlier
in the sentence and put new information later because given information is more readily
recoverable from memory (e.g., Bock & Irwin, 1980; Quirk et al., 1972). Bock and Irwin (1980)
note that “using information that requires little processing may facilitate sentence production.
Information previously formulated for inclusion in a sentence, or readily retrieved from memory
for the preceding discourse, may be more available for production than new information” (p.
468). Because working memory is a limited-capacity system, speakers need to be optimal in
minimizing demand on working memory and not maintaining to-be-produced lexical
representations any longer than necessary (e.g., Slevc, 2011). Research has shown that memory
capacity is one of the reasons why speakers structure their sentences the way they do.
The amount of memory load also affects sentence comprehension. Clearly, there is a
limitation as to how much information one can hold in his/her working memory system for a
given task. That is why it is more difficult to process complex sentences as (1) than simpler
sentences as (2a) and (2b).

(1) The man that the woman that the child hugged kissed laughed.
(2) a. The man that the woman kissed laughed.
b. The woman that the child hugged kissed the man.
(Caplan & Waters, 1999, p. 78)

The reason complex sentences are more difficult to process than simpler ones can be attributed to
the increase in memory load (e.g., Caplan & Waters, 1999; Frank, Trompenaars, & Vasishth,
13

2015; Gordon, Hendrick, & Johnson, 2001; Just & Carpenter, 1992; King & Just, 1991; Lewis,
Vasishth, & Van Dyke, 2006; MacWhinney & Pléh, 1988; Montgomery & Evans, 2009; Phillips,
Kazanina, Abada, 2005; Phillips, Wagers, & Lau, 2011; Van Dyke & Johns, 2012). For example,
language users comprehend shorter wh-dependencies as in (3) better than longer wh-
dependencies as in (4).

(3) The fact that [the employee] who the manager hired ____ stole the office supplies worried
the executive.
(4) [The executive] who the fact that the employee stole office supplies worried ____ hired the
manger.
(Phillips, Kazanina, Abada, 2005 p. 409)
Example (3) has shorter wh-dependency than example (4) because there is a shorter liner
distance between the referent of the wh-word who (i.e., the employee) and the verb hired (which
takes the referent as its object). Example (4) has longer wh-dependency because the distance
between the referent of the wh-word who (i.e., the executive) and the verb worried is longer.
Comprehending complex sentences – e.g., sentences with longer dependencies or
structures that are otherwise syntactically or perhaps semantically more complex – requires
greater cognitive effort than simpler sentences as they put more of a burden on the working
memory system (e.g., Caplan & Waters, 1999). Increased cognitive demand often results in
slowdowns in reading time or reaction time, as well as decreased accuracy for tasks such as
comprehension questions or probe-word recognition (e.g., Koornneef & Van Berkum, 2006;
Traxler, Pickering, & McElree, 2002).
14

More generally, the role of memory in language processing has also been supported by
the notion of syntactic persistence, or structural or syntactic priming (e.g., Bock, 1986; Ferreira,
Bock, Wilson, & Cohen, 2008 on syntactic priming in people with amnesia; Jaeger & Snider,
2008). Structural priming refers to individuals’ tendency to use the same sentence structures in
clauses or sentences (e.g., Bock, 1986; Bock & Griffin, 2000). This tendency can be attributed to
temporary activation of the information in memory, found also in lexical priming (e.g., Collins &
Loftus, 1975; Hartsuiker, Bernolet, Schoonbaert, Speybroeck, & Vanderelst, 2008; Potter &
Lombardi, 1998; see Chang, Dell, & Bock, 2006 for a different view). Potter and Lombardi
(1998) note that the memory for recently encountered sentence structure is procedural and
implicit (and not declarative and explicit) because it occurs without the speaker’s awareness or
intention. Procedural memory is often associated with motor skills, used automatically without
the need for conscious control or attention. In contrast, declarative memory is for facts and
events, referring to those memories that can be explicitly remembered.
Ferreira, Bock, Wilson, and Cohen, (2008) further demonstrate that syntactic persistence
is based on procedural memory rather than declarative memory. Ferreira et al. compared patients
with anterograde amnesia (who show profound impairments in declarative memory but not in
procedural memory) with memory-unimpaired control subjects to test if amnesic patients would
also show the tendency to repeat the same syntactic structures as primed sentences. The results
indicate that patients with amnesia exhibited syntactic persistence from prime sentences to the
same extent as the control subjects did. Ferreira et al.’s results support the view that syntactic
persistence is rooted in procedural memory, and not in declarative memory.
However, the precise mechanisms that underlie structural priming are still under debate.
See Pickering, McLean, and Branigan (2013) and Pickering, Branigan, Cleland, and Stewart
15

(2000) for the memory activation account vs. Chang, Dell, and Bock (2006) for the implicit
learning account. What is important to note here is that individuals’ linguistic performance
including both production and comprehension relies heavily on adequate memory capacity.

1.3 How language influences what we remember
In the preceding section, I discussed how memory influences language production and
comprehension. Equally importantly, the reverse is true as well. Language plays a significant
role in what we remember and how we mentally process information (e.g., Loftus & Palmer,
1974). The effects of language on other cognitive activities, such as memory, visual perception,
and information processing, have been widely demonstrated in many different contexts including
human interactions (e.g., Fitzsimons & Kay, 2004; Slatcher, Vazire, & Pennebaker, 2008),
product evaluations (e.g., Buda & Zhang, 2000; Hales, Kuang, & Venkataraman, 2011; Zhang &
Schmitt, 1998), and eye witness testimonies (e.g., Fausey & Boroditsky, 2011; Loftus & Palmer
1974). Zwaan and colleagues (2002) suggest that information conveyed in linguistic expressions
has strong and immediate consequences on the mental representations that comprehenders
construct (also see Hess, Foss, & Carroll, 1995 and van Berkum, Hagoort, & Brown, 1999).
Language has been shown to be essential in our existence, and some scholars even argue that the
ways we think are shaped by the language we use (e.g., the Whorfian hypothesis)
4
.
In this section, I present existing studies that illustrate how linguistic expressions can
influence our mental representations of concepts. Language has been shown to influence not only
how we perceive the world around us but also what we remember (e.g., Loftus & Palmer, 1974;

4
The Whorfian hypothesis is not a theory that I try to prove or disprove in this dissertation. Though the theory was
viewed as exciting when it was introduced, there has been little evidence to support it for a number of years (e.g.,
Boroditsky, 2011) and is suggested to be unintelligible and/or indefensible (e.g., Gleitman & Papafragou, 2005; Li
& Gleitman, 2002).
16

von Stutterheim et al., 2012). For example, the way in which events are described influences
how and what people recall from the events (e.g., Loftus & Palmer, 1974). One of the earliest
studies on the effect of language on form perception and reconstruction was conducted by
Carmichael and his colleagues (1932). In their study, the same set of object drawings were
presented in two different word conditions. For example, a drawing of two circles connected
with a straight line was presented with the word eyeglasses in one condition and with the word
dumbbell in the other condition. When participants were asked to reconstruct the drawings as
closely and as accurately as they saw them, participants’ reconstructions of the drawings differed
between the two word conditions. For instance, the drawing looked more like eyeglasses in the
‘eyeglasses’ condition and more like a dumbbell in the ‘dumbbell’ condition. Carmichael and
colleagues concluded that comprehenders’ mental representations of the visual inputs were
influenced by the language used to describe the information.
Loftus and Palmer (1974) provide another compelling example of how linguistic
expressions can influence how comprehenders process visual information and what they
remember. In their experiment, two groups of participants viewed the same images of two cars
crashing into each other. One of the groups was asked to estimate how fast the cars were going
when they hit each other. The other group was asked how fast the cars were going when they
smashed into each other. The group, which was prompted with smashed into, estimated that the
cars were going faster than the group prompted with hit. Loftus and Palmer suggest that
linguistic choices influence the way in which comprehenders encode and remember visual
information.
Furthermore, Hales, Kuang, and Venkataraman (2011) demonstrated that language also
influences our non-linguistic behaviors. The researchers conducted a study in the field of
17

accounting, which demonstrated that the way in which financial information is described
influences individuals’ decision-making. The authors presented participants with financial
information using different language styles: (i) a vivid language with emotional and image-
provoking words (e.g., sales jumped, earnings plunged) and (ii) a bland language with less
emotional expressions (e.g., sales increased, earnings declined). The results show that vivid
language can contribute to positive investor sentiment about the company.
The research discussed thus far illustrates that language and memory are profoundly
interconnected with each other, and also that language affects both our linguistic and non-
linguistic behaviors. Consequently, studying how the two domains interact would allow us to
make progress in our understanding of human cognitive processes. In this dissertation, I examine
the relationship between language and memory by investigating how (i) noun or entity-related
information and (ii) verb or event-related information influence both the way we process texts
and how well we remember what we read.
Throughout this dissertation, I use the term entity to refer to concepts that denote
individuals in linguistic contexts (e.g., sentences or stories/discourse
5
). Entities or referents can
be denoted by a variety of different forms including names (e.g., Alex), role nouns (e.g., mother,
teacher), indefinite descriptions (e.g. a pretty woman), definite descriptions (e.g., the restaurant
owner), or pronouns (e.g., he/she), among others. In Experiments 1 through 3, I used stories
where role nouns (e.g., student, reporter) were introduced to refer to entities. Because I used role
nouns, the term entity is used interchangeably with discourse/story character or protagonist.

5
I use the term ‘discourse’ to refer to the aspects of linguistic structure above the sentence level that may consist of
a few sentences, paragraphs, dialog turns, or a long discussion. In a discourse, the information status of the content,
such as salience, focus of attention, and given/new distinction, may change dynamically (e.g., Arnold, 2003; Greene,
McKoon, & Ratcliff, 1992; Kintsch, 1974; McKoon, Ward, Ratcliff, & Sproat, 1993; Sidner, 1981; Venditti &
Hirschberg, 2003).

18

I conducted four psycholinguistic experiments, where I manipulated two main
components of sentences. The first component was entity information: I manipulated the way in
which entities are introduced in a story. I tested whether different case markers in Korean would
impact the way subjects are represented in discourse and in memory (Experiments 1, 2, and 3).
The second manipulation concerned event information. I used different verbs in sentences and
tested whether the verb semantics would impact sentence processing and later recognition of the
verbs (Experiment 4). The stimulus verbs varied in terms of the number of times their denoted
actions typically repeat in a row. I compared verbs denoting actions that generally repeat once or
twice in a row in the real world (e.g., coughing, sneezing, knocking on a door) with verbs
describing actions that usually repeat many times in a row (e.g., hiccupping, clapping, bouncing
a ball, as determined by norming). This investigation on verbs’ frequency semantics would add
to the existing literature that has demonstrated that other verb semantics, including the speed, the
duration, and the physical distance of a motion, can influence sentence processing and other non-
linguistic behaviors, such as eye gaze patterns.
In the following sections, I discuss why it is important to investigate both entity and event
information to broaden our understanding of language processing, memory retention, and the
interplay between the two. I then discuss the factors influencing mental representations of
entities and events in separate sections. First, I discuss entities and events as two fundamental
sentence constituents.

1.4 Why study entity information, as well as event information?
Studying the effects of entity and event information on language comprehension, production, and
concept retrieval is important because entities and events (i.e., nouns and verbs), are two main
19

building blocks of sentence meaning. They are the major concepts that contribute to successful
comprehension and mental representation.
When processing sentences, language users tend to remember the gist of the content
information (e.g., Bobrow, 1970; Bransford & Franks, 1971; Brewer, 1975; Mehler, 1963;
Miller, 1962; Sachs, 1967; 1974). That is, readers/listeners often forget the exact structure of the
sentence but remember what the sentence was about. Researchers suggest that comprehenders
remember semantic information of the sentence better than syntactic (e.g., Clark & Clark, 1968).
People can understand and remember what a sentence is about because they understand the basic
building blocks of a sentence (see Singer 2013 for overview). Every sentence is structured with
elementary ideas or propositions (e.g., Kintsch, 1972; Singer, 2013 and others). Each proposition
consists of a predicate (verbal information) and arguments (nominal information such as
grammatical subjects and objects). Two pieces of sentential information that greatly contribute to
comprehenders’ mental representations are the predicate information (e.g., event) and the
argument information (e.g., the entities involved in the event) (e.g., Gordon, Hendrick, &
Johnson, 2001). Understanding the connection between noun and verb phrases in a sentence is
the basis for successful sentence comprehension (e.g., Gordon, Hendrick, & Johnson, 2001), as
the overall meaning of the sentence is centered on the ‘doer’ and the ‘action’ information.
Therefore, in the pursuit of enhancing our understanding of language and memory
representations, it is necessary to include experiments tapping into how comprehenders process
entity and event information.
In the next two sections, I discuss existing work that contributes to our understanding of
the following questions:

20

I. How linguistic expressions that refer to entities influence the discourse
representations and the memory representations that people construct
II. How linguistic expressions that describe events influence how well the events
are encoded and retrieved from memory

1.5 How we build mental representations of entities
Prior research suggests that understanding the meaning of linguistic expressions may involve
mentally constructing the state of affairs described in a text (e.g., see theories of Embodied and
Grounded Cognition, Situation Models, and Simulated Cognition, e.g., Barsalou, 1999; Bergen
& Chang, 2005; Gernsbacher, 1991; Glenberg, 1997; Graesser, Singer, & Trabasso, 1994;
Pulvermüller, 1999; Zwaan & Radvansky, 1998). Both entity and event information can shape
the way in which comprehenders build the mental representations of what they read
6
. Let us first
consider how entity information is represented in readers’ minds. Integrating nominal
information into comprehenders’ mental representations is influenced by the way in which the
‘who’ information is introduced (e.g., Birch, Albrecht, & Myers, 2000; Birch & Garnsey, 1995;
Foraker & McElree, 2007). Evidence suggests that the way in which entities (or characters) are
introduced into discourse changes the characters’ accessibility in the current discourse or
subsequent continuation of the discourse, as well as the entities’ representations in
comprehenders’ memory (e.g., Arnold, 1998; Birch, Albrecht, & Myers, 2000; Chiriacescu &
von Heusinger, 2010; Gordon & Hendrick, 1998; Grosz, Weinstein, & Joshi, 1995; Gundel,
Hedberg, & Zacharski, 1993; Kaiser, 2003; Prince, 1992).

6
Gernsbacher (1991) proposes the Structure Building Framework in language comprehension. According to her
framework, “the goal of comprehension is to build a coherent, mental representation, or structure, of the information
being comprehended” (p. 217).

21

There are different factors influencing how entity information is represented. First of all,
the order in which entities are introduced in a text matters for the mental representations of those
entities (e.g., Arnold, Brown-Schmidt, & Trueswell, 2007; Carreiras, Gernsbacher, & Villa,
1995; Gernsbacher & Hargreaves, 1988; Gernsbacher, Hargreaves, & Beeman, 1989; Järvikivi,
van Gompel, Hyönä, & Bertram, 2005; Kaiser & Trueswell, 2008; Kim, Lee, & Gernsbacher,
2004; Yongming & Yao, 1995). For example, in a series of seven probe word recognition
experiments, Gernsbacher and Hargreaves (1988) showed that first-mentioned characters (e.g.,
Tina in (5)) are recognized faster than second-mentioned characters (e.g., Tina in (6)). This was
the case even when first-mentioned characters were not grammatical subjects (e.g., Lisa in (7)).

(5) Tina beat Lisa in the state tennis match.
(6) Lisa beat Tina in the state tennis match.
(7) Because of Lisa, Tina was evicted from the apartment.
(Carreiras, Gernsbacher, & Villa, 1995; Gernsbacher & Hargreaves, 1988)

The same first-mentioned advantage was found in the work by Corbett and Chang (1983), as
well as Von Ekhardt and Potter (1985). Gernsbacher and Hargreaves suggest that the first-
mentioned participant tends to be more accessible because the entity forms the foundation of the
sentence meaning, and the second-mentioned participant is represented based on this foundation.
In addition to order of mention, grammatical structure has also been shown to affect how
comprehenders interpret stories or how well the concepts in the stories are remembered (e.g.,
Birch et al., 2000; Birch & Rayner, 1997; Gernsbacher & Robertson, 2001; von Stutterheim et
al., 2012; Zimmer & Engelkamp, 1981). There is more than one way to convey a certain
22

proposition. For example, we can say Lisa ate the candy or It was Lisa who ate the candy. In
comparison to the first sentence, the second sentence is an it-cleft, and emphasizes the agent of
eating, namely Lisa. Therefore, the information which comprehenders pay attention to and
regard as important is influenced by sentence structures. In fact, information has been shown to
be remembered better when it is presented in focus structures, such as clefts (e.g., Birch,
Albrecht, & Myers, 2000; Foraker & McElree, 2007; Singer, 1976). Individuals have also been
shown to look at depictions of focused items longer than presupposed items (e.g., Zimmer &
Engelkamp, 1981). Similarly, people spend more time reading focused words than non-focused
words when their eye movements were monitored during reading (e.g., Birch & Rayner, 1997;
Vasishth, Shaher, & Srinivasan, 2012).
The studies presented in this section demonstrate that the way in which characters are
presented can influence comprehenders’ mental representations and recall of the entities. To
summarize, the factors influencing the comprehension and the mental representations of entities
include: (i) where in the sentence or discourse entities appear and (ii) how they are presented
(e.g., grammatical structures). To test entities’ representations in discourse and memory, many
studies have used focus structures, compared different grammatical roles, and manipulated order
of mention. Those studies have demonstrated that focused entities enjoy higher saliency than
non-focused ones and that subjects tend to be the focus of attention compared to objects.
It should be noted that throughout this dissertation, I use the term saliency to refer to the
degree to which entities are brought to individuals’ attention or are activated in different domains
including memory (or mental representations) or discourse. I also use prominence or
accessibility (e.g., Ariel, 1991; 2001; Givón, 1983a; Gundel, Hedberg, & Zacharski, 1993) to
refer to a similar notion: entities’ “activation” levels in memory or in discourse. Researchers
23

have studied different means to measure the saliency, prominence, or accessibility of referents.
These means include how frequently entities reappear in the discourse, what grammatical roles
they denote (e.g., subject vs. object), what forms are used to refer to them (e.g., a pronoun vs. a
noun phrase). More discussion on entities’ accessibility/prominence/saliency is presented in
Chapter 2.
In addition to order of mention and grammatical structure, the information structural roles
that entities denote may also influence referents’ discourse and memory representations. For
example, whether an entity is a discourse topic, a focus (i.e., newly introduced entity), or a
contrastive topic makes a difference as to how it is represented in the discourse and in the minds
of comprehenders. Case marking is one of the ways to mark a grammatical or a discourse
function of an entity. In Korean, there are case markers to denote a sentence subject, an object, a
(non)topic, or a focused entity. For example, Korean subjects can be marked with a nominative
marker or a topic marker. I was interested in the different interpretations that the nominative
marker and the topic marker give to subject nouns, and more broadly, to the overall discourse
structure. In Experiments 1 through 3 (conducted in Korean), story characters were introduced
using two different case markers. In the nominative condition, subjects were introduced using the
nominative case marker -i/-ka. In the contrastive topic condition, the same subjects were
introduced using the contrastive topic marker -un/-nun. The goal of this investigation was to find
out whether the different interpretations that these case markers give to subjects would affect
what readers consider to be note-worthy in discourse and how well they remember the entities.
Studying the effects of subject marking on entities’ discourse and memory representations can
inform us of other factors that can shape the structure of discourse and how entities are
represented in memory, in addition to order of mention and sentence structure.
24

1.6 How we build mental representations of events
Having considered entity information, let us now turn to the question of how event information,
denoted by verbs, could affect language processing and recall. The verb in a sentence plays a
central role for laying out the meaning of the sentence. Some scholars argue for a verb-centered
model of sentence meaning (e.g., Chafe, 1970; Fillmore 1966; 1968). Verbs act as a central
organizer that links the nominal concepts into one coherent assertion (e.g., Bird, Howard, &
Franklin, 2003; Gentner, 1981; Gentner & France, 1988; Miller & Fellbaum, 1991; Tomasello,
1992; Trueswell, Tanenhaus, & Kello, 1993). Particularly, action verbs (that describe what the
subject of the sentence is doing) are dynamic and contain rich semantic information. For
example, the same action-related event can be described using many different verbs. The event
of Jane moving from one location to another as in Jane walked can be described in more detail,
depending on the manner in which she moved – Jane dashed, Jane dawdled, Jane strode, Jane
stomped, etc. Verb semantics provides comprehenders with contextual information as they build
mental representations of the events described.
The way people process (non)linguistic information and build mental representations can
differ depending on verb meanings. Some of the aspects that can vary when thinking about
events include the speed of an action, the direction of a movement, the intensity of a movement,
or the duration of an event (e.g., Bergen, Lau, Narayan, Stojanovic, & Wheeler, 2010; Gentner,
1981; Lindsay, Scheepers, & Kamide, 2013; Matlock, 2004; Moody & Gennari, 2010;
Pulvermüller, Härle, & Hummel, 2001). These different aspects have been shown to influence
the way in which individuals process linguistic and nonlinguistic information. Gentner (1981)
suggests that verbs’ semantic information contributes to the mental representation of the sentence
meaning. She demonstrated this by showing that comprehenders’ recall of the object of the
25

sentence differed depending on the semantic richness of the verb (e.g., recall of the grammatical
object ‘clock’ in ‘Ida gave her tenants a clock’ vs. ‘Ida sold her tenants a clock’). Her
experiments showed that the object of the sentence was recalled better when they appeared with
a connective specific verb
7
, such as scrub (the bath tub) than when they appeared with a general
verb, such as clean (the bath tub). The verb scrub was regarded as a semantically more specific
verb compared to clean because it provides more specific information, namely the manner of
cleaning.
Lindsay, Scheepers, and Kamide (2013) tested whether the speed of a linguistically
described motion event (e.g., sprint vs. crawl) would influence comprehenders’ visual attention
and motor execution. Results from eye tracking experiments confirmed the effect of motion
speed on the way individuals scan visually presented scenes. Participants looked along the path
of a motion longer when the sentence contained a slow motion verb than a fast motion verb. In
contrast, participants tended to scan the path faster and spent more time looking at the final
destination of the motion when the sentence contained a fast motion verb. The authors concluded
that verb semantics, denoting the speed of motion, influences comprehenders’ visual attention.
Participants’ looking behaviors were suggested to be the results of the mental representations
that they constructed in their minds for the slow and the fast movements.
Furthermore, cross-linguistic research conducted by von Stutterheim et al. (2012)
investigated whether the presence or absence of aspectual marking (e.g., completed event vs.
ongoing event) in a language influences people’s conceptualization of events. The grammatical
notion of aspect relates to the time course of an event. An action can be ongoing as in John was
crossing the street or as complete in John has crossed the street. English is a language that marks

7
Gentner (1981) defines the connective specific verb as a verb that contains specific semantic information that
denotes how the verb’s meaning relates to its grammatical object (Gentner, 1981).
26

aspect in the verb. However, German is a language that marks aspects using independent words
or phrases. Von Stutterheim et al. compared speakers of Arabic, Russian, English, and Spanish,
languages that have grammaticalized aspectual systems, with speakers of Czech, German, and
Dutch, languages without morphologically grammaticalized imperfective or progressive aspect
(e.g., was VERB+ing). In their experiment, individuals watched six-second long video clips and
were asked to tell “what is happening” in the video. The results showed that comprehenders’
verbal descriptions of events and eye gaze at scenes were affected by whether their native
language marks ongoing events in the verb system. Speakers of languages without
grammaticalized aspect, such as Czech, German, and Dutch, tended to mention the endpoint
more frequently and fixated on the endpoint object longer compared to speakers of languages
that mark ongoing events (e.g., Arabic, Russian, English, and Spanish). Speakers of Czech,
German, and Dutch also remembered the endpoint object better than speakers of Arabic,
Russian, English, and Spanish, languages with grammaticalized aspectual markers. The results
suggest that the way in which events are described guides the focus of language users’ visual
attention.
Interestingly, research indicates that altering grammatical information may have real-life
consequences (e.g., Ansolabehere & Iyengar, 1995; Fausey & Matlock, 2011; Garramone, 1984).
Fausey and Matlock (2011) demonstrated that subtle grammar changes, such as the difference
between the past progressive tense (was + VERB + ing) and the simple past tense (VERB + ed),
influence people’s perceptions about election candidates and their willingness to reelect them.
When a candidate’s negative action was described using imperfective aspect (was + VERB +
ing, was taking hush money), people felt more strongly about not reelecting him/her than when
the negative action was described using perfective aspect (VERB + ed, took hush money).
27

To summarize, how individuals process visual scenes or sentences and how they build
mental representations of events can change depending on the verbs used to describe those
events. The last experiment of this dissertation (Experiment 4) tested the effect of verb semantics
on sentence processing and memory recall with the emphasis on a novel semantic feature that, to
the best of my knowledge, has previously not been experimentally tested from a memory
perspective in prior work. The semantic property I investigated is the frequency of motion
repetition. In addition to the speed of motion (e.g., sprint vs. crawl) or the manner of motion
(e.g., scrub vs. clean), another important feature of verb semantics has to do with how frequently
an action repeats. I use the term ‘frequency of motion/action’ to refer to the number of times a
motion or an action prototypically repeats in a row. In our physical world, we perform or observe
actions all the time. Some actions generally repeat only once or twice in a row when they occur
(e.g., coughing, sneezing), but other actions typically repeat multiple times in a row (e.g.,
hiccupping, clapping). For example, knocking on a door typically involves the knocking gesture
twice or three times (low frequency), but clapping typically involves multiple repetitions of the
hand gesture (high frequency).
Experiment 4 aimed to provide a broader understanding of the role of verb semantics on
sentence comprehension and memory retention. The goal was to see whether the expected high
or low frequency of an action described by a verb would change the way individuals process the
sentence and how well they remember having seen the verb of the sentence. I hypothesized that
processing verbs describing actions that repeat multiple times might be more difficult than
processing verbs denoting actions that repeat only once or twice in a row. This prediction
assumes that creating a mental representation of an action that repeats many times in a row might
require more cognitive resources than creating a mental representation of an action that repeats
28

only once or twice in a row. The opposite prediction would be that processing a verb of high
repetition might result in easier processing due to the multiple mental reenactments of the
denoted action than processing a verb of low repetition, which may involve only a few mental
repetitions.
The effect of motion frequency on sentence processing was measured using a sensicality
judgment task in which participants were asked to indicate whether a given sentence was
sensical, i.e., meaningful. The verb’s motion frequency effect on memory was tested using a
probe-word recognition paradigm in which participants were presented with the verbs that had
appeared in the sentences they had read, and were asked to indicate whether they recall having
seen the verbs. Experiment 4 aimed to provide new insights in our understanding of how lexical
semantics can influence sentence processing and memory retention, by testing the role of verbs’
inherent action repetition frequency.

1.7 Aims of this dissertation
It is evident that both language and memory are important facets of human cognition and that
they contribute to each other’s functions. In this dissertation, I investigate the relationship
between language and memory with the focus on entity information denoted by nouns and event
information denoted by verbs. I conducted four experiments in Korean. The first three
experiments are concerned with entity information, and the fourth experiment dealt with event
information. In the entity experiments, I changed the way in which characters are introduced in
stories – e.g., marked with a default nominative case or marked with a specialized topic marker. I
tested whether the information structural differences between topic-marked nouns and non-topic
marked nouns influence: (i) how well participants recall the nouns and (ii) whether participants
29

expect them to be mentioned again in subsequent discourse. In Experiment 4, I tested how verbs
that denote actions of different repetition frequencies would influence sentence processing and
memory recall. The two main research questions I explored in this research are presented in (8):

(8) The research questions for this dissertation
I. Would the way in which discourse entities are interpreted in stories influence
their prominence in story development and memory representation?
II. Would the semantics of action verbs, specifically the differences concerning
the frequency of motion repetition, influence sentence processing and how well
the event is remembered?

1.8 Overview of dissertation
In this section, I briefly summarize the content of each chapter. In the next chapter (Chapter 2), I
present Experiment 1. In this experiment, native Korean-speaking participants read stories
written in Korean. The stories were presented in two different experimental conditions. In one
condition, the target subject had a nominative case marker -i/-ka attached to it. In the other
condition, the same subject had a contrastive topic marker -un/-nun attached. The Korean topic
marker -un/-nun can have a regular topic reading, as well as a contrastive topic reading. The
stories were designed in such a way that nouns with the topic marker were interpreted as
contrastive topics (and not regular topics). The contrastive topic interpretation presupposes the
existence of other discourse entities that the noun has a contrasting value to (more on the Korean
case marking is presented in Chapter 2). Participants were instructed to write three continuation
sentences for each story. The results showed that the way participants continued the stories
30

differed depending on the case marker used on the target subjects. In particular, the first-
mentioned subject in the story, which was the global discourse topic, was highly prominent in
story continuation, especially when the target character (i.e., recently-mentioned) was marked
with the default nominative marker. However, when the target character was contrastive topic-
marked, the prominence of the discourse topic decreased. Importantly, the contrastive topic
marked entity increased the accessibility of other members in the discourse set. This indicates
that readers are sensitive to the case marking on subjects and that this grammatical change
affects how entities are represented in discourse.
Chapter 3 presents Experiment 2. In this experiment, I tested how people’s memory of
discourse characters might be influenced by the use of different case markers. A new group of
native Korean-speaking participants read stories written in Korean. Two characters (e.g., teacher
and reporter) were introduced in each story. Both protagonists were nominative-marked in one
condition and contrastive topic-marked in another condition. Because the different Korean case
markers have been shown to change the information structure of the discourse, it was predicted
that they might have different consequences on the later recognition of entities. Unlike the
predictions, the results revealed that nominative vs. contrastive topic marking did not change
how well comprehenders remembered discourse characters. For example, recall of teacher or
reporter did not change whether they were marked with the nominative marker or with the
contrastive topic marker. Even though recall of characters did not differ between the two case
marking conditions, the second-mentioned subject that had been mentioned more recently in the
story was remembered better than the first-mentioned subject, the character introduced earlier in
the story. This recency effect on character recall (also widely observed in serial recall, e.g., Bjork
& Whitten, 1974; Murdock Jr, 1962; Tzeng, 1973) motivated the next investigation in Chapter 4.
31

Chapter 4 presents Experiment 3, which employed the same target stories as Experiment
2. I was interested in whether the character that comprehenders remembered better would be the
one they talk about more in subsequent discourse. Experiment 3 differed from Experiment 1,
which was also a story continuation task. Experiment 1 included three main entities in the story:
a global topic (first-mentioned subject), members of a discourse set, and a local topic (most
recently mentioned subject). However, Experiment 3 included two entities: the first-mentioned
and the second-mentioned. While Experiment 1 examined the accessibility of the three referents,
Experiment 3 directly compared which of the two entities (first-mentioned vs. second-
mentioned) would be talked about more frequently, testing the recency effect in discourse
continuation. In Experiment 3, a new group of Korean-speaking participants read the same four-
sentence stories as Experiment 2. The task was to add three continuation sentences to each story.
The results revealed that the character that comprehenders remember better is not necessarily the
one they talk about more frequently. The results indicate that what comprehenders consider to be
more important in discourse development may not be equally prominent in their memory
representation. These results indicate that prominence in discourse and memory is not monolithic
and should not be regarded as the same.
Moving to the investigation of verb semantics, Chapter 5 reports Experiment 4, which
investigated how changes in verbal information influence what people remember. I changed the
verbal information by using different types of action verbs. The action verbs differed based on
how many times the movement of an action (denoted by the verb) repeats in a row in the real
world. In one condition, I used verbs that refer to actions which generally repeat once or twice in
a row (e.g., sneezing, coughing). In the other condition, I used verbs that refer to actions which
typically repeat many times in a row (e.g., hiccupping, clapping). Participants were asked to
32

indicate whether the sentences were sensical (i.e., meaningful). The action verbs from target
sentences were later tested for recognition. The sentence sensicality judgment and verb
recognition accuracy data suggest that the frequency of action movements influences language
processing. Specifically, verbs describing actions which typically repeat only once or twice were
processed with higher accuracy and were recognized better than verbs referring to actions which
repeat multiple times in a row. The results indicate that comprehenders are sensitive to verb
semantics and that reading about repeated actions may increase processing load.
Chapter 6 presents a summary of the results and discussion.

1.9 Contributions of the dissertation
This dissertation investigated topics that are important in understanding individuals’ linguistic
competence. The investigations can be characterized in the following way: (i) studying both
entity representation and event representation, (ii) understanding language processing in the
context of discourse and memory, and (iii) testing the effects of primacy and recency – known to
affect memory – to see how being mentioned early or late in discourse influences an entity’s
memory and discourse-level representation.
My research examines both nominal and verbal constructs, two important building blocks
of sentence meaning. Studying both entity information and event information can broaden our
understanding of language users’ linguistic behaviors. Furthermore, the semantics of nouns and
verbs might share similar or overlapping conceptualizations or cognitive processes. For this
reason, the present research emphasizes the importance of investigating both areas and
understanding comprehenders’ linguistic performance from comprehensive perspectives.
33

Additionally, this dissertation explores how entity information is represented in discourse
and in memory, as tested by a story continuation paradigm and a probe-word recognition
paradigm, respectively. Since our linguistic performance is supported by our memory ability,
how we express thoughts linguistically must be linked to how those ideas are encoded and stored
in memory, and how they are retrieved from memory. Therefore, research such as this
dissertation that connects discourse representation and memory representation can help us
understand the complex nature of the relationship between the two domains.
This dissertation also investigates the notions of primacy and recency in discourse and
memory. One might expect that entities which are prominent in discourse would also be
prominent/easily accessible in memory. However, the findings of this dissertation show that
prominence in discourse and memory are not always parallel. The results indicate that an entity’s
prominence can be affected by a variety of different factors including the discourse structure,
other entities available in the discourse, the way in which entities are introduced, where in the
discourse they are introduced, as well as other external factors, such as comprehenders’ reading
goals.
Finally, this research hopes to motivate future studies to combine investigations across
different linguistic and cognitive domains.

34

Chapter 2
Experiment 1: The effect of Korean subject marking on entities’ discourse representations

2.1 Introduction
Language comprehenders understand texts by interpreting the meanings of individual words in
sentences. However, the ways in which words are put together in sentences also impact how
information is conveyed. Experiment 1 investigates how linguistic devices, such as case markers,
affect language comprehenders’ understanding of written texts (i.e., stories or discourse) and
how those devices influence readers’ expectations of story continuation and the retrieval of
discourse concepts from memory (reported in Experiment 2).
How entities are represented in discourse is often measured by how accessible or
prominent they are in the progression of the story. The prominence, accessibility, or salience of
entities
8
has been studied widely as the concept of accessibility is related to how individuals
make connections between prior information and currently unfolding information. For example,
knowing what is prominent (or “important”) in the discourse helps speakers build a coherent
discourse by continuing to talk about the topic that is in the center of the developing discourse.
Furthermore, the prominence of a concept helps comprehenders easily interpret which
information speakers are referring back to.
Researchers have investigated which entities or referents are considered
prominent/salient as comprehenders process discourse stories (e.g., Ariel, 1988; Arnold, 1998;
Givón, 1983a/b, 1995; Grosz et al., 1995; Gundel, Hedberg, & Zacharski, 1993; Kaiser, 2010;
among others). Prominent entities are often the referents that appear frequently in subsequent

8
In this dissertation, I use the expressions of prominence, accessibility, and salience of entities interchangeably.
35

sentences. They are also more likely than non-prominent entities to appear in prominent syntactic
positions (e.g., subject) and to be introduced in a main clause (e.g., Barzilay & Lapata, 2008;
Givón, 1983).
In addition to frequency of mention and grammatical position, linguistic form also
reflects referents’ prominence levels. Various theories have been proposed to explain which
entities are assumed to be highly activated in speakers’ and/or addressees’ minds (e.g., Ariel,
1988; Givón, 1983a/b; Gundel et al., 1993). Generally, researchers believe that a highly
accessible discourse entity is the one that is in the center of the speaker/addressee’s mental
model of the discourse at a particular point in time. A well-known model of discourse
accessibility is the Givenness Hierarchy by Gundel, Hedberg, and Zacharski (1993). Gundel et
al. propose that different linguistic forms, such as determiners (e.g., this, that) and pronouns
(e.g., it), signal the memory or attention status that their intended referent is assumed to have in
the addressee’s mind. Gundel et al. identify six different levels of attention statuses for linguistic
referents (see Table 1). These levels are an implicational scale, such that each status entails (and
is therefore included by) all lower statuses, but not vice versa.

Table 1. The Givenness Hierarchy (Gundel, Hedberg, & Zacharski, 1993)
in focus activated familiar
uniquely
identifiable
referential
type
identifiable
it
that
this
this N
that N the N
indefinite this
N
a N

36

Gundel et al. claim that if an entity is in focus, it is in the center of the addressee’s
attention. A reduced referring expression (e.g., pronoun it) would be used to refer to a focused
entity. However, when a referent is merely activated, and not necessarily in focus, then it is in
the addressee’s working memory. For activated entities, more elaborated referring forms, such as
demonstrative pronouns (e.g., this/that + noun) would be used. When an entity is in the familiar
status, then it exists in the (long-term) memory of the addressee. Gundel and colleagues believe
that the cognitive status of an entity can be reliably signaled by the form that is used to refer to it.
Givón (1983a/b) also suggests that there is a high correlation between referring
expressions and the topic continuity or topic predictability of their referents (see Table 2), such
that an entity that is referred to with a zero anaphora (i.e., null form) is highly likely to be a
continuing topic.

Table 2. Givón’s topic continuity/predictability (1983a/b)

a. zero anaphora
b. unstressed/clitic/bound pronouns (‘agreement’)
c. stressed/independent pronouns
d. definite nouns
e. modified definite nouns

most continuous/accessible/predictable
topic

least predictable topic

37

Another well-known theory of prominence is Ariel’s (1990) Accessibility Theory. Ariel
argues that the forms speakers choose for referents are determined by the referents’ accessibility
levels as shown in Table 3. According to her Accessibility Marking Scale, when a referent is
highly accessible in the mind of the addressee, the form to refer to it is minimal (e.g., zero or null
pronoun). When the accessibility level of a referent is low, then the referring form is more
elaborated, such as a full name or a long description.

Table 3. Ariel’s Accessibility Marking Scale (1990)
low accessibility
1 Full name + modifier
2 Full (“namy”) name
3 Long definite description
4 Short definite description
5 Last name
6 First name
7 Distal demonstrative + modifier
8 Proximal demonstrative + modifier
9 Distal demonstrative (+ NP)
10 Proximal demonstrative (+ NP)
11 Stressed pronoun + gesture
12 Stressed pronoun
13 Unstressed pronoun
14 Cliticized pronoun
38

15 Extremely high accessibility markers (gaps, including pro,
PRO and wh-traces, reflexives, and agreement)

high accessibility

As a whole, the models highlighted above suggest that when an entity is in the center of
the discourse, it is easily accessible and is therefore mentioned frequently in a prominent
syntactic positon. Moreover, the referring expression for it would be a simple, reduced form. In
the following section, I briefly review some of the factors influencing an entity’s prominence
level in discourse and memory.

2.2 Prior work on the effect of grammatical structure in discourse processing and memory
A number of studies have shown that readers’ processing of given linguistic information is
influenced by the grammatical structures used to convey the information (e.g., Birch et al., 2000;
Birch & Rayner, 1997; Gernsbacher & Robertson, 2001; Zimmer & Engelkamp, 1981). Studies
that compared focused and de-focused elements have also revealed that grammatical changes
have different cognitive and discourse consequences. For example, focused words are often
remembered better than their non-focused counterparts (e.g., Singer, 1976; Birch et al., 2000).
The English it-cleft structure as in (9) allows speakers/writers to distinguish a particular concept
they want to focus (e.g., Hedberg, 2000).

(9) It was the king who led the troops. Focus structure
(10) The king led the troops. Neutral structure
39

Individuals have also been shown to look at depictions of focused, clefted items longer than
presupposed items (e.g., Zimmer & Engelkamp, 1981). Similarly, comprehenders tend to spend
more time reading focused words than non-focused words (e.g., Birch & Rayner, 1997; Vasishth,
Shaher, & Srinivasan, 2012).
In addition to the grammatical structure of a single sentence, research has revealed that
the relationship between sentences also influences what and how well comprehenders remember.
Black and Bern (1981) demonstrated that readers remember causally related events in narratives
better than events that were not causally related. Keenan, Baillet, and Brown (1984) also showed
that participants’ recall rates improved as the cause-effect relation between a pair of sentences
increased.
The findings highlighted above suggest that the same information can be processed and
remembered differently depending on the grammatical structure of the sentence and the
sentence’s relationship with neighboring sentences. In Experiment 1, I aimed to find out whether
other linguistic devices, such as Korean case markers, can influence how activated discourse
concepts are. Korean nouns that refer to entities accompany markers that indicate their
grammatical roles in the sentence. I compared the discourse and cognitive functions of the
NOMINATIVE (NOM) case marker -i/-ka with that of the contrastive TOPIC (TOP) marker -
un/-nun.
In Experiment 1, nouns were introduced in stories, NOM-marked in one condition and
TOP-marked in the other (e.g., student-NOM/TOP or teacher-NOM/TOP). I employed an off-
line written story continuation task to investigate how these two case markers influence the status
or prominence of the nouns in the discourse. I tested this by observing which characters
participants tended to refer to in their continuation sentences. I was interested in the functions
40

that the Korean nominative and the topic markers have on a discourse- or production-level
(Experiment 1), as well as on a cognitive- or comprehension-level (Experiment 2 reported in
Chapter 3).
In what follows, I provide an overview of Korean case markers, focusing particularly on
the interpretations of the topic marker. Then I review two existing studies that investigated the
relationship between linguistic structure and information encoding and retrieval.

2.3 Case markers in Korean
Korean has case markers that indicate the grammatical or the information structural roles of noun
phrases (Lee, 1992). As shown in (11), subjects are marked with the nominative marker -i (for
nouns that end with a consonant) or -ka (for nouns that end with a vowel). Objects are marked
with the accusative marker -ul (for nouns that end with a consonant) or -lul (for nouns that end
with a vowel).

(11) Nominative and accusative case markers in Korean
John-i Mary-lul salanghanta
John-subj. marker Mary-obj. marker to love
‘John loves Mary.’

In English, the subject and the object of a sentence are indicated generally by the order in which
they appear. For example, if a noun phrase comes before a verb, it is generally interpreted as the
subject of the sentence. When a noun phrase follows a verb, it is often the object of the sentence,
as shown in (12) where John is the subject and Mary is the object of the sentence.
41

(12) John loves Mary.
subject object

In contrast, grammatical roles in Korean are identified through the types of case markers
attached to the nouns regardless of where in the sentence they appear. Thus, the presence of case
markers allows the grammatical arguments of a sentence to move around within a sentence. In
the syntactic literature, this phenomenon is called scrambling (e.g., Mahajan, 1990; Saito, 1985).
As shown in (13), an object can come before the subject in Korean even though the canonical
word order in the language is SOV (Subject-Object-Verb). Therefore, case markers play
important roles in discourse understanding and information conveyance.

(13) Mary-lul John-i salanghanta
Mary-obj. marker John-subj. marker to love
‘Mary, John loves.’

2.3.1 Korean topic marker
In addition to the nominative and the accusative case markers, Korean nouns can also be marked
with the topic marker (TOP) -un/-nun
9
. The topic marker can occur on both subjects and objects,
and is often regarded as having three main interpretations (e.g., Han, 2001; Lee, 1992; 1999;
2003): (i) topic reading, (ii) contrastive topic reading, and (iii) contrastive focus reading as
shown in (14) (See Vermeulen (2009) for a related discussion on Japanese topic particle -wa).

9
-un is attached to noun phrases that end with a consonant, and -nun is attached to noun phrases that end with a
vowel.
42

(14)
a. topic reading
John-un Mary-lul salanghanta
John-topic marker Mary-obj. marker to love
‘As for John, he loves Mary.’

b. contrastive focus reading
John-un Mary-lul salanghanta
John- topic marker Mary-obj. marker to love
‘As for John, he loves Mary.’
[Inferred] It is only John who loves Mary when no one else does.

c. contrastive topic reading (CONT-TOP)
John-un Mary-lul salanghanta
John- topic marker Mary-obj. marker to love
‘As for John, he loves Mary.’
[Inferred] Other people in the discourse set love others.
e.g., ‘Peter loves Ann, Sue loves Jason, etc.’

On the regular topic interpretation in (14a), the presence of -un indicates that the sentence is
about the subject noun (e.g., Gundel, 1974; 1989; Neeleman, Titov, Van De Koot, & Vermeulen,
2009; Reinhart, 1982; Gregory & Michaelis, 2001), as indicated by the English paraphrase ‘As
for John…’
43

The second interpretation of -un/-nun is the contrastive focus reading in (14b). Under
this reading, the marker contrasts a noun with the entire discourse set the noun is associated with.
For example, in the noun phrase John-un when -un has the contrastive focus reading, this
marking indicates something unique about John compared to other discourse set members. This
interpretation signals a different quality about the noun that the speaker intends to focus (e.g., the
focus in (14b) is that John loves Mary when no one else does).
In addition to the regular topic interpretation and the contrastive focus interpretation, the
Korean topic marker can also have a contrastive topic interpretation as shown in (14c). Here, the
subject noun is construed as contrasting with other entities. For example, if the subject John is
contrastive topic-marked, this contrastive topic interpretation presupposes the existence of other
entities in the discourse, such as Peter or Sue (e.g., Choi, 1999). When the subject is
contrastively topic-marked, not only does the sentence provide information about John, but it
also suggests the properties of other discourse set members (e.g., If John loves Mary, how about
Peter? Peter loves Ann). Experiment 1 focuses on this contrastive topic interpretation of -un/-nun
compared with the regular topic interpretation that the nominative marker -i/-ka conveys.
The three different interpretations of the Korean topic marker can be inferred from the
discourse context or the intonation patterns in the sentence. Particularly, the difference between
the contrastive topic reading and the contrastive focus reading seems to rely greatly on
intonation. On the contrastive topic interpretation, stronger stress is given to the predicate of
John rather than the word John itself (i.e., stress on the object Mary). In contrast, on the
contrastive focus interpretation, the stress is on John itself.
Let us know compare the nominative marker -i/-ka and the topic marker -un/-nun as they
can both be attached to a subject noun. Although both -i/-ka and -un/-nun can signal subjecthood,
44

they convey different discourse properties of the noun. This difference is what I was interested
in, as well as how the two different case markers influence readers’ interpretations of the stories
they read. Consider the sentences in (15).

(15a) haksaeng-i hakkyo-ey ka-ss-ta
student-NOM school-to go-PAST-DECLATIVE
‘A/The student went to school.’
Simply stating the fact that a/the student went to school
Conveys no information about other people

(15b) haksaeng-un hakkyo-ey ka-ss-ta
student-CONT_TOP school-to go-PAST-DECLATIVE
‘The student went to school.’
With the contrastive topic interpretation, it implies that other people in the discourse set
(e.g., nurse) went somewhere else

When the nominative marker -i is used in (15a), the sentence is about the subject noun haksaeng
(‘student’) and its predicate that (s)he went to school. Upon encountering this sentence, the
mental representation that comprehenders build would probably include three elements: the
student, school, and the student going to school. However, when -un is used with the contrastive
topic reading, it implies that other people relevant in the discourse went somewhere else (e.g.,
The nurse went to the hospital). For a detailed discussion on contrastive topics from a broader
crosslinguistic perspective, see Constant (2014).
45

As discussed thus far, case markers in Korean provide important clues as to how nouns
function in sentences. This function further impacts how the whole discourse is organized.
Although there have been many theoretical studies that support the different interpretations of
the Korean topic marker (e.g., Choi, 1999; Han, 2001; Lee, 1992; 1999; 2003), there has been
little experimental research that demonstrates how the Korean contrastive topic marker might
influence entities’ discourse representations. Therefore, it is unclear whether contrastive topic
marking tends to draw language users’ attention to: (i) the entity that it is attached to, (ii) other
entities in the discourse that the contrastively topic-marked entity is associated with, or (iii) other
entities (e.g., a global discourse topic). To shed light on this, Experiment 1 aims to provide a
better understanding of the role of the Korean contrastive topic marker by analyzing whom
readers tend to talk about in discourse continuation when a subject is NOM-marked and when it
is contrastive TOP-marked.
A great deal of studies that have investigated the discourse and the memory
representation of entities manipulated factors, such as focus structure, order of mention,
grammatical roles, or verb semantics (e.g., whether the meaning of the verb is biased for subject
interpretation or object interpretation). The current investigation takes a different approach as it
looks at the prominence of entities when they are interpreted differently due to the different case
markers attached to them. The comparison between the Korean nominative marker and the
contrastive topic marker is an area where relatively little research has been done. It is important
to note that, of the three interpretations possible for -un/-nun, the contrastive topic reading was
most plausible in my experimental stories due to the discourse structure. Each story introduced a
discourse set with different members in it, and one of the set members was mentioned in the last
46

sentence of the four-sentence target stories. When this member is marked with -un/-nun, the
entity is regarded as a contrastive topic whose value has been contrasted to other set members.
Another advantage of Experiment 1 is that it explores the prominence of entities in a
larger discourse that consists of seven sentences - four stimulus sentences in the story and three
additional continuation sentences created by participants. With this rich context, Experiment 1
examines the accessibility of at least three different entities including: (i) the global discourse
topic whom the story is about; (ii) a local topic, who is mentioned last (i.e., most recently) in the
test story, marked either with NOM or TOP; and (iii) unspoken but inferred discourse set
members. Examining the effect that case marking has on the prominence of these different
entities can provide new evidence as to how entities interact in story development and how case
marking can shift the focus of discourse progression.

2.4 Prior work on how grammatical structures influence entities’ accessibility
In this section, I review two studies that are directly relevant to Experiment 1. The first study
was conducted by Chiriacescu and von Heusinger (2010), who employed a written story
continuation paradigm. This study is one of the few studies that investigated the effect of case
marking on the accessibility of entities in subsequent discourse. It has provided a foundation for
my investigation.
The second study was by Birch, Albrecht, and Myers (2000), who used an off-line story
continuation task and a series of on-line probe-word recognition tasks. Birch et al.’s work is
pertinent to my research as it tapped into both the discourse and the memory representations of
entities. The authors examined the effect of focus structures on entities’ likelihood to be
mentioned in subsequent discourse. Furthermore, they tested how well focused concepts are
47

retrieved from memory. I have pursued the same issues, the interaction between discourse
persistency and memory prominence of entities.

2.4.1 Chiriacescu and von Heusinger (2010)
Chiriacescu and von Heusinger (2010) investigated the discourse functions of pe-marked
indefinite direct objects in Romanian. In Romanian, pe-marking is generally obligatory for
definite noun phrases but optional for indefinite ones. The sentences in (16) show the optionality
of the pe-marking in indefinite objects in Romanian.

(16)
a. Petru a vizitat un prieten
Petru has visited a friend
‘Petru visited a friend’.

b. Petru l -a vizitat pe un prieten
Petru CL has visited PE a friend
‘Petru visited a friend’.
(Chiriacescu & von Heusinger, 2010, p. 299)

Both (16a) and (16b) mean ‘Petru visited a friend.’ However, the grammatical object (i.e., un
prieten) in (16a) does not have a preceding pe, but (16b) does. The researchers hypothesized that
the presence or absence of pe-marking before a direct object might signal how prominent the
object is. In their earlier work, Chiriacescu and von Heusinger (2009) observe that a pe-marked
48

object is more likely to be talked about in subsequent discourse than a non-pe-marked object. For
example, the indirect object in (16b) (i.e., PE a friend) is more likely to be mentioned
subsequently than the same object in (16a) (i.e., a friend).
Chiriacescu and von Heusinger (2010) tested this observation empirically using a web-
based story continuation experiment. Participants were presented with short stories containing
three sentences each. The first two sentences in each story provided the background, and the
third sentence contained the target direct object, pe-marked in one condition and unmarked in the
other. Participants were asked to add five continuation sentences to each story. The researchers
analyzed participants’ sentences with regards to the following three aspects: (i) referential
persistence, that is, the number of total occurrences of the subject and the object in the
continuation sentences, (ii) topic shift, that is, in which continuation sentence the target direct
object becomes the subject, and (iii) the referential forms of the direct object (e.g., being referred
to with a pronoun, full noun, or null/omission).
The analyses of participants’ continuation sentences confirmed their predictions. Pe-
marked direct objects were brought up earlier in the continuation sentences than their unmarked
counterparts. They were also more likely to be maintained in subsequent continuation sentences
than the unmarked objects, and interestingly were more so than the subject. The results also
revealed that unmarked objects were expressed with more noun modifiers than pe-marked
objects. Pe-marked objects were considered to be more activated in the minds of the
comprehenders than their unmarked counterparts. This was because participants used more
reduced referring expressions (e.g., zero anaphora or pronouns, see Ariel, 1990) in subsequent
sentences to refer back to pe-marked objects. Chiriacescu and von Heusinger concluded that
49

Romanian speakers make use of the discourse cues that pe-marking provides in discourse
interpretation and continuation.

2.4.2 Birch, Albrecht, and Myers (2000)
Birch, Albrecht, and Myers (2000) conducted an off-line story continuation task and on-line
speeded probe-word recognition tasks, which I adapted for Experiment 1 and Experiment 2. I
borrowed their research paradigms to test whether case markers are significant indicators of what
is likely to be talked about in subsequent discourse and whether different case markers influence
recall of entities. Birch et al. conducted five experiments that tested how focus constructions in
English – namely it-clefts (ex.17a) and there-sentences (ex.17c) – would influence discourse
continuation and the accessibility of focused concepts in comprehenders’ minds.

(17) Sample focus constructions employed in Birch et al.’s experiments
a. It was the king who led the troops. it-cleft
b. The king led the troops. Neutral construction

c. There was this mugger who had attacked an elderly lady. there-sentence
d. A mugger attacked an elderly lady. Neutral construction

Their target stimuli were short stories that were two to four sentences long. The final sentence in
each story contained a target concept (often a character) that was introduced by a syntactic
focusing structure (as in Example 18a) or by a neutral determiner (a, an, or the, as in Example
18b).
50

(18) An experimental item used in Birch et al.’s story continuation experiment
As Joan walked home from the subway, she saw a crowd of people near her
apartment building.
a. There was this mugger who had attacked an elderly lady.
b. A mugger had attacked an elderly lady.

Participants saw either (18a) or (18b) and were asked to write at least one sentence that was a
logical continuation of the story. The researchers measured how often participants referred to the
target noun (e.g., the ‘mugger’ in 18a and 18b). The results showed that participants made
significantly more references to target nouns that had been introduced in a focus structure, as
compared to those mentioned in a neutral structure.
Birch et al. also looked at what kinds of referential forms participants used when
referring back to the target noun (e.g., pronoun, ‘s/he’; repeated noun, ‘mugger’; or alternative
noun, e.g., ‘criminal’). They found that target nouns in the focus condition were more likely to
be referred to with pronouns than those in the neutral condition. The target nouns in the neutral
condition were more likely to be referred to with nominal expressions than pronominal
expressions, compared to those in the focus condition. In other words, when comprehenders
made reference to a focused noun, they tended to use a reduced, less specific referring expression
(e.g., pronoun). However, when they were making reference to a neutral noun, the referring
expressions were more elaborated and specific. From these findings, Birch et al. concluded that
nouns introduced in a focus structure were more activated, and were more accessible in
comprehenders’ minds.
51

Following the off-line story continuation task, Birch et al. conducted a series of four
probe-word reaction time (RT) experiments. In those experiments, participants read the same
short passages as their Experiment 1. After reading each passage that included approximately
four sentences, they were asked to indicate (by pulling a yes or a no lever) whether a probe had
appeared in the preceding sentence (e.g., mugger in 18a or 18b). The researchers tested whether
English focus constructions help facilitate the memory retrieval of the concepts introduced in the
sentences. In their first probe-word recognition experiment, the researchers compared the RTs of
the subject nouns from focus structure with those from neutral structure. Unlike their predictions,
the RT results showed that the time taken to respond to a subject noun did not differ whether it
had appeared in a focus structure or in a neutral structure. In a follow-up experiment, the
researchers included a de-emphasized structure condition in which the target noun was not in the
grammatical subject position but in de-emphasized object position (see (19) for a sample
stimulus). The results of this focus (i.e. it-cleft and there-sentence) vs. de-focus comparison
showed that the RT of a focused subject was significantly faster than that of a de-emphasized
noun. However, again there was no significant difference between the focused condition and the
neutral condition where the noun appeared with a neutral determiner, such as a, an, or the.

(19) A sample de-emphasized sentence in Birch, Albrecht, and Myers (2000)
The reporter’s question to the mayor was not answered.  mayor was the probe

Since the authors did not find a focus structure effect on probe word recognition, Birch
et al. conducted an additional experiment in which they added a delay task. In the new
experiment, after participants finished reading the final sentence of a passage, a three-digit
52

number was presented in the center of the computer monitor. Participants were asked to begin
from that number and count aloud, backward by 3s, for ten seconds. When this delay task was
added, the RT results showed a focus construction effect. Participants were faster recognizing
target concepts when they had been focused than when they had not.

2.5 Experiment 1 research questions
As discussed in the works by Chiriacescu and von Heusinger (2010) and Birch et al. (2000), how
comprehenders process discourse information is sensitive to grammatical structures. In
Experiments 1 and 2, I investigated how different case markers in Korean would influence the
progression of discourse stories and the accessibility of discourse entities in memory. The
research questions that I sought to answer are the following:

I. Would the difference between nominative case marking and contrastive topic
marking in Korean influence how comprehenders expect discourse stories to
unfold?
II. Would Korean contrastive topic marking make the subject cognitively more
accessible than nominative case marking?

In Experiment 1, using an off-line story continuation task, I tested how contrastively TOPIC-
marked subjects affect comprehenders’ expectations of the stories to unfold, as compared to
NOMINATIVE-marked subjects. I focused on the prominence of TOPIC-marked subjects
throughout discourse continuation and how they interact with other discourse entities, such as a
global discourse topic and other entities that the subject is contrasted with. Each target story
53

consisted of four sentences, and in each story a discourse set was introduced. One member of the
set was mentioned in the fourth sentence with the topic marker in one condition and with the
nominative marker in the other condition. I analyzed how participants’ continuations evolved
given that the story characters were marked with the different case markers. Because the Korean
contrastive topic marker presupposes the existence of other entities, I hypothesized that topic
marked nouns would trigger more of the discourse set members to be introduced in continuation
sentences. On the other hand, nominative-marked subjects would produce more comments on the
already introduced NOM-marked subject.

2.6 Methodology
2.6.1 Participants
Twenty four adult native speakers of Korean participated. They received $10 for their
participation. The majority of the participants were students at the University of Southern
California (N = 21), and the rest of the participants were recruited in Korea. At the time of
testing, none of the participants had lived more than five years outside Korea.

2.6.2 Materials
The experiment was a written story continuation task, a commonly used method in studies
investigating discourse comprehension and production (e.g., Arnold 1998; Chiriacescu & von
Heusinger 2010; Engelkamp & Zimmer 1983; Garrod & Sanford 1988; Gernsbacher & Shroyer
1989; Kaiser 2010; Most & Saltz, 1979; Vonk, Hustinx, & Simons 1992). Each target stimulus
was a beginning of a story that consisted of four sentences. The stories were in Korean.
Participants’ task was to write three sentences that continue the story in a natural way. There
54

were 16 target stories and 16 filler stories. All targets and fillers were normed beforehand with
six native speakers of Korean for their naturalness.
A sample target story is presented in (20) followed by the English translation in Table 4.
Table 5 shows the structure of each of the four sentences in the target stories.

(20) A sample target story in Korean
이승희 박사님은 서울에 있는 한 대학의 심리학과 교수님입 니다. 학교에는 교수님께서 지 도하시는
제자가 여러 명 있습니다. 교수 님 제자들은 ( 모두 연차가 같고 지도 교수님도 같지만), 관심이 있어
하는 연구 분야가 다들 다릅니 다. 그 중에서 부산에서 온 학생 이/ 은 아동 심리에 대해 연구를 하고
싶어합니다.

Table 4. English translation of each of the four sentences in the sample target story in (20)
sentence description
1 Dr. Seung-Hee Lee is a psychology professor at a university in Seoul.
2 He/She has a {few advisees}set at the university.
3-Set not elaborated His/Her students all have different research interests.
3-Set elaborated Though the students are all in the same cohort and have the same
mentor, they have different research interests.
4 Among them, [a student from Busan]-NOM / [a student from Busan]-
TOP wants to do research on child psychology.

55

Table 5. Target story design in Experiment 1
sentence description
1 Introduces a global discourse topic with NOM marker (e.g., Dr. Seung-Hee Lee)
2 Introduces a set that the discourse topic is associated with, (e.g., advisees)
3 States a difference between the set members, (e.g., different research interests)
4 Introduces a set member in subject position, marked with NOM or TOP marker

The first sentence in each story introduced an individual in subject position with the nominative
marker (e.g., Professor Lee-NOM). I consider this individual a global Discourse Subject (DS) (or
global discourse topic) whom the story was about (e.g., Chafe, 1976; Lambrecht, 1996; Van
Dijk, 1977). The second sentence introduced a discourse set that the global discourse subject was
associated with (e.g., a set of Professor Lee’s advisees). The third sentence provided more
information about the set. In one condition, the set was elaborated with an additional feature that
the members shared (see Table 4). In the other condition, the set was not elaborated, and the
information simply stated a different feature among the set members. The fourth sentence of a
given story introduced one member of the set and commented on this individual. I refer this
individual as the Mentioned Member (MM), which is the local subject of the story. The
Mentioned Member is the character whose case marking differed between the nominative and
the topic markers.
I manipulated two variables in the target stories. The first is the case marker on the
Sentence 4 subject noun (i.e., local subject, Mentioned Member), i.e., NOM-marked vs. TOP-
marked. When the subject of Sentence 4 was NOM-marked, it simply introduced an individual
out of the discourse set and a predicate that expresses a feature of that individual. However,
56

when the subject was TOP-marked, it elicited a contrastive topic reading, which introduced a
unique feature that the subject of Sentence 4 has, compared to other unmentioned set members.
The contrastive topic reading was available in the current design because the previous Sentence 3
stated a differing characteristic among the set members (e.g., different research interests). The
presence of -un/-nun on the local topic compares the subject with other unmentioned set
members, providing the contrastive topic reading.
The second factor manipulated was set elaboration, i.e., whether the discourse set in
Sentence 3 was elaborated with information that groups the set members together and then pulls
them apart (See Sentence 3-Set elaborated in Table 4). This variable was added to see whether an
elaborated set where the set members’ differing characteristics were explicitly mentioned would
increase the likelihood of the other unmentioned set members to be referred to in story
continuation.

2.6.3 Procedure
The story continuation task was administered using Qualtrics, a web-based survey tool.
Participants completed the study in the presence of the experimenter. Each item (story) was
presented individually, and participants continued the story by adding three sentences of their
own. It took approximately an hour for each participant to complete the experiment. Figure 1
shows a sample target trial.

57

Figure 1. A target trial of Experiment 1 (story continuation)

2.6.4 Coding
I analyzed participants’ continuation sentences focusing on their grammatical subjects.
Particularly, I checked who or what the subject of each continuation sentence referred to (i.e., the
referent of each subject). Subjects were regarded as a means of reflecting discourse prominence
(e.g., Stevenson et al., 1994; Gernsbacher & Hargreaves, 1988; 1992) because the subject is one
of the most prominent or salient concepts introduced in the sentence (e.g., Arnold, 1998; Martin
& Slevc, 2014; Slevc, 2011; Stevenson et al., 1994). Subjects are also shown to be highly topical
(e.g., Reinhart, 1982). There is a tendency that more accessible or prominent entities are more
likely to be mentioned and are also more likely to be mentioned earlier (e.g., Arnold, 2001;
Brennan, 1995; Garrod & Sanford, 1988; Kieras, 1980; Perfetti & Goldman, 1974). Therefore,
by looking at the referent of each grammatical subject, one can make inferences as to what kinds
of concepts had been more readily available in comprehenders’ minds (e.g., Arnold, 1998;
58

Garnham, 2001; Kaiser, 2010; Kehler, 2002). For this reason, I take the grammatical subjects
10

as the starting point that directs us to understanding which discourse entities might have been
most activated.
There are many different entities that can be referred to by the subject of each
continuation sentence. However, the three most commonly produced referents, which therefore
have been the focus of the analyses, are provided in Table 6.

Table 6. Most commonly occurring referents of the continuation sentence subject
referent description
a Discourse Subject (DS)
Subject of the first sentence in the story (e.g., Professor
Lee)
b Mentioned Member (MM)
Subject of the fourth sentence (e.g., a student from Busan)
whose case marking was manipulated (NOM or TOP)
c
Unmentioned Members
(UM1, UM2, & UM3)
Set members that were not mentioned in the story (e.g., a
student from Seoul; in the same set as a student from Busan,
but not mentioned in the preceding story)

In addition to these frequent referents that comprised above 80% of all data, there were other
characters that appeared in continuation sentences. However, those referents were excluded from
the analyses because of low occurrence (below 20% of all data). They include: the narrator (I),
the previous mentioned object (e.g., child psychology), the context place (e.g., university), a new

10
There are, of course, many other aspects one could systematically look into in continuation sentences (e.g.,
grammatical objects, what is the linguistic form of the subject or the object, etc.). However, for the purpose of the
present study, I focus only on who the subject of continuation sentences refers to.
59

referent (e.g., My sister too is majoring in psychology), etc. Table 7 presents a sample
continuation sentence for each of the three main subjects: Discourse Subject (DS), Mentioned
Member (MM), and Unmentioned Member (UM).

Table 7. Sample continuation sentences that include Discourse Subject, Mentioned Member, and
Unmentioned Member as subjects
Story presented
Dr. Seung-Hee Lee is a psychology professor at a university in
Seoul. He/She has a few advisees at the university. (Though the
students are all in the same cohort and have the same mentor), they
have different research interests. Among them, a student from
Busan-NOM/TOP wants to do research on child psychology.
DS continuation
그래서 박사님은 그 학생에게 그것에 관한 인턴쉽을 알아봐 주었습니다.
So the professor did some research on internship opportunities for
the student.
MM continuation
그 학생은 어릴때 부터 아이들을 좋아했습니다.
The student has always liked children since s/he was young.
UM1 continuation
다른 학생은 범죄자들의 심리를 연구하고 싶어합니다.
Another student wants to do research on criminal psychology.

2.7 Predictions
It was hypothesized that the different case markers influence comprehenders’ expectations of
how stories continue. I predicted that the subject of participants’ continuation sentences would be
different in the NOM condition and the TOP condition. More specially, I predicted that, in
comparison to the nominative marker, the topic marker -un/-nun would make the members of the
60

discourse set more salient even though not all individual members were specified or introduced.
This prediction stems from the nature of the contrastive topic marker interpretation. The story
context helped elicit the contrastive topic interpretation because each story introduced a set of
members who shared similar yet different characteristics. Therefore, having the last sentence
comment on one of the set members naturally produced the contrastive topic reading. The
contrastive topic interpretation leads readers to make associations of the currently introduced
entity with other entities that belong to the same set. As Choi (1999) points out, the Korean
contrastive topic marker entails the existence of other members. When readers were presented
with the information that a subject noun ‘X-(TOP marked) has done something,’ then this would
lead them to think about what other members Y and Z might have done. Therefore, when one of
the set members was introduced with the topic marker (eliciting the contrastive topic
interpretation), participants would be more likely to talk about other members of the set in their
continuations.
As for set elaboration (i.e., the third sentence in the story, either specified the differences
between the set members or not), I predicted that the set members would be more likely to be
addressed when a discourse set was elaborated with information that groups the members first
and then individualizes them (e.g., Though the students are all in the same cohort and have the
same mentor, they have different research interests) than when the set was not elaborated.
Participants were expected to introduce other set members more frequently when the set was
elaborated than when the set was not elaborated.
It was predicted that the proportion of the references made to the three subjects of interest
(i.e., the Discourse Subject, the Mentioned Member, and the Unmentioned Member) would
change throughout the three continuation sentences. The first continuation sentence might be
61

where the subject referent differs greatly between the NOM condition and the TOP condition.
This might be the sentence where we can see how contrastive topic marking may change the
flow of the discourse compared to nominative marking.

2.8 Results
The results of the story continuation experiment were analyzed separately by continuation
sentence: Continuation Sentence 1, Continuation Sentence 2, and Continuation Sentence 3.
Repeated measures ANOVAs showed no main effect of Set Elaboration in any of the three
continuation sentences. Therefore, I report the results of the Set-Elaborated and Set-Not-
elaborated conditions combined in each of the NOM and the TOP conditions.

2.8.1 Subject referent of Continuation Sentence 1
2.8.1.1 Discourse Subject (DS)
Table 8 presents the percentage of the three most commonly occurring referents in Continuation
Sentence 1 separately for the NOM condition and the TOP condition. Figure 2 is the graphic
representation of the distribution.

Table 8. Percentage of each of the three most common referents in Continuation Sentence 1
referent NOM TOP
occurrence (%) SD occurrence (%) SD
DS 34.9 20.8 25.5 21.0
MM 21.8 14.8 22.4 16.8
UM1 30.7 17.9 43.2 20.6
62

SUM 87.4 - 91.1 -
DS = Discourse Subject; MM = Mentioned Member; UM1 = Unmentioned Member 1

Figure 2. Percentage of DS, MM, and UM1 referents in Continuation Sentence 1, Exp. 1

As Table 8 and Figure 2 show, the most likely subject of Continuation Sentence 1 patterned
differently between the NOM condition and the TOP condition. In the NOM condition, the
Discourse Subject, whom each target story was about, was realized as the subject of
Continuation Sentence 1 most frequently (34.9%), followed by the continuations that refer to a
new unmentioned set member (UM1, 30.7%). The pattern was different in the TOP condition.
When the mentioned member in the story was contrastive topic marked, participants were most
willing to talk about an unmentioned set member in their first continuation sentence (43.2%)
than they were to continue the story about the discourse subject (25.5%).
Let us first consider the likelihood of the Discourse Subject to be mentioned. There was a
statistically significant difference between the NOM condition and the TOP condition. The
34.9
21.8
30.7
25.5
22.4
43.2
0
5
10
15
20
25
30
35
40
45
50
Discourse Subject (DS)
e.g., Prof. Lee
Mentioned Member (MM)
e.g., student from Busan
Unmentioned Member1
(UM1)
e.g., student from LA
percentage (%)
Subject of Continuation Sentence 1
NOM
TOP
63

likelihood of the Discourse Subject to appear as the subject of Continuation Sentence 1 was
greater in the NOM condition (M = 34.9%, SD = 20.8) than in the TOP condition (M = 25.5%,
SD = 21). This effect was significant in item analysis and marginal in subject analysis; t1(23) =
1.940, p1 = .065; t2(15) = 2.635, p2 < .02. Participants were more likely to mention the Discourse
Subject (i.e., global discourse topic) when the mentioned member (i.e., subject of the fourth
sentence) was NOM-marked than when it was TOP-marked.

2.8.1.2 Unmentioned Member (UM)
This section reports the results when Continuation Sentence 1 was about Unmentioned
Member 1. The likelihood that participants started their first continuation sentence with an
unmentioned, brand-new member of the set (UM1, e.g., another student in Table 7) was greater
in the TOP condition (M = 43.2%, SD = 20.4) than in the NOM condition (M = 30.7%, SD =
17.9). This difference was significant in both subject and item analyses; t1(23) = -2.261, p1 < .04,
t2(15) = -2.666, p2 < .02. Participants were more likely to continue the stories by talking about a
brand-new set member when the mentioned member was TOP-marked than when it was NOM-
marked. I performed one-sample t-tests to see whether the proportion of Unmentioned Member
continuation was greater than chance level. The likelihood of the UM1 continuation in the TOP
condition was significantly higher than chance level (set at .33); t1(23) = 2.194, p1 < .04, t2(15) =
2.582, p2< .03.

2.8.1.3 Mentioned Member (MM)
When I analyzed those trials where participants started their first continuation sentence with the
mentioned-member of the set (MM, e.g., a student from Busan in Table 7), there was no
64

significant difference between the NOM condition (M = 21.8%, SD = 14.8) and the TOP
condition (M = 22.4%, SD = 16.8); t1(23) = -.118, p1 = .907, t2(15) = -.102, p2 = .920. The
likelihood of the mentioned member to be referred to in Continuation Sentence 1 did not differ
whether it was NOM-marked or TOP-marked in the story.

2.8.2 Subject referent of Continuation Sentence 2
I now move to the results of Continuation Sentence 2. The same three most frequently occurring
referents (i.e., Discourse Subject, Mentioned Member, and Unmentioned Member) were
analyzed. Table 9 and Figure 3 summarize the percentage of each referent in Continuation
Sentence 2.

Table 9. Percentage of each of the four most common referents in Continuation Sentence 2
referent NOM TOP
occurrence (%) SD occurrence (%) SD
DS 31.2 18.4 36.9 23.4
MM 20.3 14.1 13.2 11.9
UM1 16.1 19.3 12.5 17.2
UM2 14.5 16.7 23.4 25.8
SUM 82.1 - 85.8 -
DS = Discourse Subject; MM = Mentioned Member; UM1/2 = Unmentioned Member 1/2

65

Figure 3. Percentage of each of the four most common referents in Continuation Sentence 2

In Continuation Sentence 2, the Discourse Subject (DS) continuation is predominant in both the
NOM condition and the TOP condition (31.2% and 36.9%, respectably). Another noticeable
pattern is the continuations with Unmentioned Member 2. UM2 continuation is greater in the
TOP condition (23.4%) than in the NOM condition (14.5%). Participants seem to be more
willing to introduce another unmentioned set member in Continuation Sentence 2 when the
mentioned member in the story was TOP-marked than when it was NOM-marked. When the
mentioned member was NOM-marked, the second mostly like subject of Continuation Sentence
2 after the Discourse Subject was the mentioned member itself (20.3%).

2.8.2.1 Discourse Subject
This section reports the cases where the subject of Continuation Sentence 2 was the discourse
subject. As can be seen in Figure 3, the discourse subject was the most likely subject of
31.2
20.3
16.1
14.5
36.9
13.2 12.5
23.4
0
5
10
15
20
25
30
35
40
45
Discourse Subject
(DS)
e.g., Prof. Lee
Mentioned Member
(MM)
e.g., student from
Busan
Unmentioned
Member1 (UM1)
e.g., student from LA
Unmentioned
Member2 (UM2)
e.g., another student
percentage (%)
Subject of Continuation Sentence 2
NOM
TOP
66

Continuation Sentence 2 in both the NOM condition and the TOP condition. One sample t-tests
revealed that these observations were significantly higher than chance level (defined as .25) both
in the NOM condition; t1(23) = 2.839, p1< .01; t2(15) = 2.745, p2 < .016, and in the TOP
condition; t1(23) = 3.215, p1 < .005; t2(15) = 3.945, p2 < .002. The proportion of discourse subject
continuations did not differ between the NOM condition (M = 31.2%, SD = 18.4) and the TOP
condition (M = 36.9%, SD = 23.4); t1(23) = -1.148, p1= .263, t2(15) = -1.210, p2 = .245. The
likelihood of the discourse subject to appear as the subject of Continuation Sentence 2 did not
differ whether the mentioned member was NOM-marked or TOP-marked in the story. This
pattern is different from Continuation Sentence 1 where the NOM condition had more of the
discourse subject continuations than the TOP condition.

2.8.2.2 Mentioned Member
In Continuation Sentence 2, the mentioned member continuations were more likely to occur in
the NOM condition (M = 20.3%, SD = 14.1) than in the TOP condition (M = 13%, SD = 11.9).
This difference was marginal in both the subject analysis, as well as the item analysis; t1(23) =
1.941, p1 = .065, t2(15) = 1.784, p2 = .095. Participants were more likely to comment on the
mentioned member of the set in their second continuation sentence when it was NOM-marked in
the story than when it was TOP-marked.

2.8.2.3 Unmentioned member 1 (UM1)
In Continuation Sentence 2, the Unmentioned Member 1 continuations did not differ
significantly between the NOM condition (M = 16%, SD = 19) and the TOP condition (M =
12.5%, SD = 17); t1(23) = 1.071, p1 = .295, t2(15) = 1.199, p2 = .249.
67

2.8.2.4 Unmentioned Member 2 (UM2)
Unmentioned Member 2 (UM2) refers to a newly introduced set member when there was already
a new member introduction (UM1) in Continuation Sentence 1. In Continuation Sentence 2, the
UM2 continuations were greater in the TOP condition (M = 23.4%, SD = 25.8) than in the NOM
condition (M = 14.5%, SD = 16.7). This difference was significant in subject analysis, as well as
in item analysis; t1(23) = -2.287, p1 < .04, t2(15) = -2.785, p2 < .02.

2.8.3 Subject referent of Continuation Sentence 3
In Continuation Sentence 3, none of the referent continuations (i.e., DS, MM, UM1, UM2, and
UM3) was different between the NOM condition and the TOP condition. Table 10 and Figure 4
present the percentage of the referent types in Continuation Sentence 3.

Table 10. Percentage of each of the five relevant referents in Continuation Sentence 3
referent NOM TOP
occurrence (%) SD occurrence (%) SD
DS 49.4 26.4 50.5 25.1
MM 17.1 16.4 17.1 16.4
UM1 8.8 9.3 8.8 11.3
UM2 4.6 8.0 2.6 8.2
UM3 1.5 5.6 1.5 4.2
SUM 81.4 - 80.5 -
DS = Discourse Subject; MM = Mentioned Member; UM 1/2/3 = Unmentioned Member 1/2/3

68

Figure 4. Percentage of each of the five relevant referents in Continuation Sentence 3
DS = Discourse Subject; MM = Mentioned Member; UM1/2/3 = Unmentioned Member 1/2/3

As can be seen in Figure 4, the discourse subject was the most likely subject of Continuation
Sentence 3 in both conditions, followed by the mentioned member continuations. One sample t-
tests showed that the likelihood of the discourse subject continuations was greater than chance
level (set at .20) in both the NOM condition; t1(23) = 7.83, p1 < .001; t2(15) = 11.276, p2 < .001,
and the TOP condition; t1(23) = 8.641, p1 < .001; t2(15) = 8.048, p2 < .001.
Table 11 below summarizes the three continuation sentences and presents which referents
showed significant difference between the NOM condition and the TOP condition.

49.4
17.1
8.8
4.6
1.5
50.5
17.1
8.8
2.6
1.5
0
10
20
30
40
50
60
Discourse Subject
(DS)
e.g., Prof. Lee
Mentioned
Member (MM)
e.g., student from
Busan
Unmentioned
Member1 (UM1)
e.g., student from
LA
Unmentioned
Member2 (UM2)
e.g., another
student
Unmentioned
Member3 (UM3)
e.g., first-year
student
percentage (%)
Subject of Continuation Sentence 3
NOM
TOP
69

Table 11. Significant differences in the referent percentages between NOM and TOP conditions
continuation
sentence
condition
subject referent (%)
DS MM UM1 UM2 UM3
1
NOM 35 22 31 N.A. N.A.
TOP 26 22 43 N.A. N.A.
2
NOM 31 20 16 15 N.A.
TOP 37 13 13 23 N.A.
3
NOM 49 17 9 5 2
TOP 50 17 9 3 2
Note: Statistically significant differences are highlighted.

In Continuation Sentence 1, the Discourse Subject and the Unmentioned Member continuations
showed a significant difference between the two conditions. Discourse Subject continuations
were greater in the NOM condition than in the TOP condition. However, Unmentioned Member
continuations were more frequent in the TOP condition than in the NOM condition.
In Continuation Sentence 2, Mentioned Member continuations and Unmentioned
Member 2 continuations showed a significant difference between the two conditions. Participants
were more likely to talk about the Mentioned Member in their second continuation sentence
when it was NOM-marked than when it was TOP-marked. However, participants were more
likely to introduce yet another set member (UM2) when the topic marker was used to introduce
the mentioned member than when the nominative marker was used. Continuation Sentence 3 did
not show any significant differences between the two conditions in terms of the types of subject
referents. However, the Discourse Subject was the most likely referent of the last continuation
70

sentence in both conditions. Table 12 organizes each continuation sentence in order of most
frequently occurring referents.

Table 12. In order of most frequent referents within each continuation sentence (%)
continuation
sentence
condition
from most frequent to less frequent
1 2 3 4 5
1
NOM DS (35) UM1 (31) MM (22) N.A. N.A.
TOP UM1 (43) DS (26) MM (22) N.A. N.A.
2
NOM DS (31) MM (20) UM1 (16) UM2 (15) N.A.
TOP DS (37) UM2 (23) MM (13) UM1 (13) N.A.
3
NOM DS (49) MM (17) UM1( 9) UM2 (5) UM3 (2)
TOP DS (50) MM (17) UM1 (9) UM2 (3) UM 3 (2)

As shown in Table 12, the global Discourse Subject (DS) was most frequently mentioned as the
subject across all three continuation sentences expect in Continuation Sentence 1, especially
when the mentioned member was introduced with the contrastive topic marker (the highlighted
cell). This shows that the first mentioned, topic of the story (i.e., global discourse topic) is highly
accessible in discourse development. However, as predicted, the use of the contrastive topic
marker increases the likelihood of new set member continuations. Regarding the most recently
introduced local topic (i.e., mentioned member), I initially predicted that participants would
continue to talk about the Mentioned Member in Continuation Sentence 1 when it was NOM-
marked in the story. However, the global discourse subject overrode the most recently mentioned
71

local subject continuations. This suggests that the global subject whom the story is about may
have higher accessibility than the local, recently-talked-about mentioned member.

2.9 Discussion – Experiment 1 (off-line story continuation)
In Experiment 1, I investigated the difference between nominative case marking and contrastive
topic marking in Korean, particularly concerning their effects on the prominence of different
discourse entities. I used an off-line written story continuation paradigm. Participants read four-
sentence stories and added three sentences of their own to continue each story. In each target
story, the first sentence introduced a global discourse subject whom the story was about. The
second sentence introduced a set that the global subject was associated with. The third sentence
commented on the set. In one condition the set was elaborated with extra information that
grouped the set members together first then pulled them apart with varying features. In the other
condition, the set was not elaborated. The fourth sentence introduced one member from the set
(i.e., mentioned member) with either the nominative marker -i/-ka or the topic marker -un/-nun.
Participants’ continuation sentences were analyzed in terms of who/what the grammatical subject
of each sentence referred to.
As predicted, the overall continuation patterns showed difference between the NOM
condition and the TOP condition. When the subject of the fourth sentence in the story was NOM-
marked, participants were most likely to continue the story by talking about the global discourse
subject. In contrast, participants tended to continue the story by introducing an unmentioned
member of the discourse set when the mentioned member was TOP-marked.
In the NOM condition, the prominence of the global discourse topic persisted in the
second continuation sentence where participants were still most likely to talk about the discourse
72

subject. The TOPIC-marking effect continued in the second continuation sentence as well, in that
the introduction of yet another unmentioned member was greater in the TOP condition than in
the NOM condition. Even though the discourse subject continuation increased from Continuation
Sentence 1 to Continuation Sentence 2 in the TOP condition, participants were still more likely
to introduce a new set member in this condition, compared to the NOM condition.
In Continuation Sentence 3, the Discourse Subject continuation occurred most
frequently in both conditions. The referent frequencies in the last continuation sentence did not
differ between the two conditions. It appears that participants find it most natural to end of the
story by talking about the first character introduced at the beginning of the story.
To summarize the continuation patterns, the global Discourse Subject introduced at the
start of the story has been shown to be the most prominent entity. This was especially the case in
the condition where one of the set members was introduced with the nominative marker rather
than the contrastive topic marker. As can be seen in Table 12, the Discourse Subject is the most
frequently talked about referent in the NOM condition throughout the three continuation
sentences. However, we observe that this global discourse topic prominence is reduced when a
member of the discourse set was introduced with the contrastive topic marker. Participants were
more willing to talk about unmentioned set members when one of the members was contrastively
topic-marked in the story. The results indicate that contrastive topic marking in Korean is a
strong cue to evoke the prominence of other set members that had not been introduced in the
discourse individually. Interestingly, contrastive topic marking draws comprehenders’ attention
not necessarily to its subject, but to the entities that the subject is contrasted with.
Unlike the clear case marking effect, set elaboration did not seem to influence who is
likely to be mentioned in story continuation. Initially, I had predicted that when the members of a
73

discourse set are elaborated with specific information, each inferred member would be made
more distinctive from each other, and would therefore be mentioned more frequently compared
to the no elaboration condition. However, having the additional information that grouped the
members and distinguished them apart did not affect who comprehenders tend to talk about. The
lack of an elaboration effect might be contributed to the story design. Having introduced a set
and commenting on one of the members might have been enough for comprehenders to
individualize the members in their mental construction of the story, and no additional
information about the members might have been necessary to distinguish them apart.
Based on the overall continuation patterns, we see that much of readers’ attention is
drawn to the first person introduced in the story, the global discourse topic. This observation
mirrors other findings in the literature where first-mentioned referents are privileged and are
more likely to be referred back (e.g., Arnold, 1998; Gernsbacher & Hargreaves, 1988; 1992;
Gordon et al., 1993; Hudson-D'Zmura & Tanenhaus, 1998). Whether the first-mentioned entity
is the topic of the discourse or not (because not all first-mentioned characters are topics), it does
tend to carry importance. Gernsbacher and Hargreaves (1988) propose that this might be the case
because the first-mentioned character lays a foundation for the sentence meaning or the
discourse. In Experiment 1, the first-mentioned protagonist was the topic of the story as the
whole story was built around this person. Its discourse persistence is maintained throughout the
three continuation sentences except in Continuation Sentence 1 of the TOP condition (in which
Unmentioned Member 1 was the most likely referent). Especially in the NOM condition, the
global topic was talked about most frequently throughout the entire story. The first-mentioned,
global discourse topic appears to be most prominent and accessible in the interpretation and
continuation of passages. The results support Gívon (1990)’s observation that discourse topic
74

carries much weight. However, this global topic prominence is modulated when a contrastive
topic marker is used to introduce a referent. Introduction of an entity marked with contrastive
topic briefly shifts the topichood in the story. This finding is important because the existing
theories of discourse representation might not necessarily predict that the same subjects that are
interpreted differently due to different case marking can influence discourse prominence. The
effect of subject case marking suggests that discourse saliency and topic continuity can be
affected by whether subjects are interpreted as a regular topic or a contrastive topic.
Furthermore, we see that the prominence of the global discourse topic outweighs that of
the recently mentioned local topic. Even though each target story fragment ended with a
statement about the local topic, participants were more inclined to talk about the global topic that
was introduced three sentences earlier rather than commenting on the already-mentioned set
member. The global discourse topic advantage was present even when the local topic was
marked with the nominative case. This outcome does not necessarily align with the observations
made by one of the models in discourse coherence, the Local models (e.g., Lapata, 2003;
Barzilay & Lapata, 2005; Foltz et al., 1998). In this model, adjacent sentences tend to have
similar content and have related words. This again illustrates the strong accessibility of the
discourse subject that overrides local coherence.
Experiment 1 demonstrates that the information structural role that a discourse entity
plays as indicated by case marking (e.g., contrastive topic vs. non-contrastive topic) affects the
overall discourse structure. It provides evidence that case markers in Korean that are used to
introduce entities differentiate how much attention is given to referents. Particularly, contrastive
topic marking draws comprehenders’ attention to other entities in the discourse set, even when
the members might not have all been introduced. In contrast, nominative marking on one of the
75

set members does not elicit the same level of prominence from other set members. The current
experiment focused on the discourse representation of entities. The next step of investigation
reported in the following chapter addresses the memory-level representations of discourse
entities that are case marked differently.

76

Chapter 3
Experiment 2: The effect of Korean subject marking on entities’ memory representations

3.1 Motivations for Experiment 2
The results from the off-line story continuation experiment reported in the preceding chapter
(Experiment 1) revealed that how readers incorporate upcoming stories varied depending on the
types of case markers and the different interpretations they give to entities. For example,
contrastive topic marking temporarily decreases the accessibility of the global discourse topic,
which is otherwise very prominent in story progression. When a member of a discourse set was
introduced with the Korean contrastive topic marker, the likelihood of other set members being
mentioned in subsequent sentences increased. In contrast, readers were more likely to comment
on the first-mentioned global topic when one of the set members was introduced with the
nominative marker.
The results of Experiment 1 show that Korean case markers have different discourse
consequences. First, the nominative marker preserves the prominence of the discourse topic
whose presence is well maintained throughout the discourse. Also, the case marker tends to draw
attention to the noun phrase that it is attached to. However, the contrastive topic marker draws
attention to other discourse members, temporarily overriding the prominence of the global topic.
The current on-line probe-word recognition experiment, Experiment 2, builds on the results of
Experiment 1 but, unlike Experiment 1, it probes memory representations rather than discourse-
level expectations. Experiment 2 was designed to test the memory representation of entities that
are introduced with the nominative marker or the contrastive topic marker. I was interested in
whether the different case markers would affect readers’ recall of text information including the
77

story characters. Before going into the details of the experiment, I review prior work on the
relationship between language and memory in the following section.

3.2 Prior work on the relationship between language and memory
Since Experiment 2 taps into entities’ memory-level representations, this section reviews the
research findings on the relationship between memory and language processing. First of all, it is
important to note that there are different processes involved in memory formation. Memory
research has suggested three main phases: encoding, storage, and retrieval (e.g., Cowan, 2000;
Norman & Bobrow, 1979; Ratcliff, 1978; Wickelgren, 1976). The studies discussed in this
section focus mainly on the retrieval process since they tested how well linguistic concepts are
recognized or retrieved from memory (e.g., as evident in probe-word recognition speed and
accuracy). The retrieval aspect of memory is also what my Experiment 2 focused on as I tested
the factors influencing individuals’ recognition of story characters. Nonetheless, the encoding
and the storage processes of memory formation have also been widely studied in relation to
language processing. Although all three processes are necessary in linguistic performance,
Experiment 2 will focus primarily on the retrieval process.
Because comprehenders must incorporate new information into already existing
information for successful comprehension, language processing makes demands of memory
(e.g., Baddeley, 1992; Birch & Garnsey, 1995; Caplan & Waters, 1999; Hitch, 1980; Just &
Carpenter, 1992; King & Just, 1991; MacDonald et al., 1992; Miyake et al., 1994; Myers &
O'Brien, 1998; Van Dyke & Johns, 2012). For instance, interpreting who or what a pronoun
refers to relies on reactivating earlier information in working memory (e.g., Dell, McKoon, &
Ratcliff, 1983; O'Brien, Duffy, & Myers, 1986). Comprehenders also need to make use of their
78

world knowledge stored in long term memory to appreciate and evaluate the content of linguistic
expressions.
An important linguistic aspect that has been shown to impact memory retention is lexical
semantics. Classic evidence for the role of semantics on memory comes from the distinction
between concrete and abstract words. Concrete words that refer to real tangible objects are
recalled better than abstract words that refer to ideas or concepts (e.g., Walker & Hulme, 1999).
Furthermore, Sachs (1967) studied semantic effects on recall on the sentence level. In her study,
participants decided whether the test sentence was identical to the original sentence. Individuals
were able to detect the changes both in form (i.e., syntax) and meaning (i.e., semantics) when the
test sentence was presented immediately after the original. However, their ability to detect the
change in form decreased considerably when there were intervening sentences between the
original and the test sentences. On the other hand, adults were still able to identify changes in
meaning even with the intervening sentences. Sachs suggests that the ‘surface structure’ is not
stored long, but meaning is important in the memory for linguistic information. Polisˇenska´,
Chiat, and Roy (2014) further showed that meaning plays an important role in what individuals
remember. The authors observed that four to six year-old children recalled semantically plausible
sentences better than semantically implausible sentences. The same effect was also found with
Arabic-speaking children (Wallan, Chiat, & Roy, 2011).
Other aspects of memory that have been widely observed are the primacy and the recency
effects in free serial recall. The primacy and the recency effects refer to the tendency for
individuals to better recall the first and the last items on a list compared to the items in the
middle (e.g., Bjork & Whitten, 1974; Murdock Jr, 1962; Tzeng, 1973). In an early study, Tzeng
(1973) tested individuals’ memory for words from lists. He presented participants with four lists
79

that contained 10 unrelated words in each. After the presentation of each word, participants
counted a three digit number backwards by 3s for 20 seconds. At the end of each list, participants
recalled as many of the words from the list. The results revealed that individuals’ memory was
the best for the words presented last on the lists (i.e., recency effect) followed by their memory
for the words presented first on the lists (i.e., primacy effect).
Recently, neuroimaging studies on free serial recall (not on sentence processing) have
reported that individuals recognize the most recent items faster and more accurately than the first
items, suggesting that the recency effect is stronger than the primacy effect – at least in serial
free recall (e.g., Nee & Jonides, 2008; 2011; Öztekin, Davachi, & McElree, 2010; Talmi, Grady,
Goshen-Gottstein, Moscovitch, 2005). These studies also revealed that the retrieval of the last
item from a list requires less brain activation than other items on the list, indicating ease of
access from memory.
Gernsbacher, Hargreaves, and Beeman (1989) tested the primacy and the recency effects
in the context of sentence processing. They tested how quickly comprehenders recognize having
seen sentence entities (expressed by proper names). They included two-clause target sentences as
shown in (22) and (23). Each word in the sentence was presented one by one in the center of a
screen. The researchers tested how well participants recognize having seen the names at different
time points after the offset of the sentence final word.

(22) Tina gathered the kindling, and Lisa set up the tent.
(23) Lisa set up the tent as Tina gathered the kindling.

80

Their results showed that when the test name appeared together with the last word of the
sentence, the second-mentioned name was recognized faster than the first-mentioned. However,
when the test name was presented 150 milliseconds (ms) after the sentence final word, both the
first-mentioned and the second-mentioned were recognized at the same rate. However, when the
names were tested 1400 ms and 2000 ms after the sentence, a reverse pattern was observed as the
first-mentioned was recognized faster than the second-mentioned. Gernsbacher et al. note that
the advantage of the most recent clause is short-lived, but the advantage of the first-mentioned
lasts longer. They explain the results from the perspectives of the Structure Building Framework
(Gernsbacher, 1991). They propose that comprehenders build the mental representation of each
clause as a separate substructure, but it is the first-mentioned entity that works as the foundation
for the entire context.
In addition to semantics and order of mention, grammatical structure can also alter the
prominence of information in memory. McKoon, Ratcliff, Ward, and Sproat (1993) carried out a
short-term memory test where they investigated how quickly and accurately comprehenders
respond to probe words taken from stimulus sentences that they read. Individuals were asked if
they had seen words when they had appeared in two different conditions. In one condition, the
word was in a predicate position (demanding in 24a). In the other condition, it was in a
prenominal position (demanding in 24b). Participants’ recall time was faster and the accuracy
rate was higher when the test word was in the predicate position (24a) than in the prenominal
position (24b).

(24) George is having second thoughts about his new job.
a. His critical boss is demanding at times.
81

b. His demanding boss is critical at times.

McKoon et al. also found that direct objects were retrieved from memory more accurately than
indirect objects. For example, the word magazines was recognized better when it was a direct
object as in (25) than when it was an indirect object as in (26). Their findings indicate that the
accessibility of linguistic information in memory is influenced by syntactic position.

(25) Somebody had inserted some magazines inside some newspapers late last night.
(26) Somebody had inserted some newspapers inside some magazines late last night.

Syntactic focus structure enhances memory representation as well. Gernsbacher and
Jescheniak (1995) show that concepts marked by the indefinite this or by spoken stress are
responded to more quickly in probe recognition tasks, suggesting that focused concepts are
highly activated in memory. Birch and Garnsey (1995) also revealed that comprehenders respond
to probes more quickly when the target had been focused. Moreover, Klin et al. (2004)
demonstrated that readers recognize concepts faster in a probe-word recognition task if the
concepts had been in a focus structure (e.g., in the clefted position of a wh-cleft, such as what
everyone asked about was [the dessert] focus). The researchers propose that a wh-cleft makes a
concept more salient. Similar results were also found in a study by Almor and Eimas (2008)
using an auditory lexical decision paradigm. Comprehenders recognized concepts faster when
they had been in a it-cleft or a wh-cleft. These findings show that grammatical focus structures
can be used to enhance the prominence or saliency of concepts.
82

In addition, a subject advantage is also found in memory representation. Not only
sentence subjects are highly accessible in discourse representation, they are also retained better
in memory compared to sentence objects (e.g., Clark & Card, 1969). These findings may suggest
that more prominent discourse entities are more accessible in short-term memory than less
prominent entities. What readers find prominent or noteworthy in discourse might be a reflection
of how accessible that information is in memory.
Researchers have proposed different theories explaining why certain information is
retrieved from memory better than others. A well-known theory is the ‘storage bin’ model by
Wyer and Srull (1980). According to their model, concepts that occur frequently or that have
recently been encountered are stored on top of the memory bin. When processing subsequent
information, the constructs that are on top are used first.
There is also Higgins and King (1981)’s ‘energy cell’ model, which suggests that
frequency and recent occurrence increase the energy or action potential of the constructs that are
associated with the stimuli. Therefore, frequency and recency are two of the main factors that
influence how information is recalled from memory or how people process situations.
Higgins, Bargh, and Lombardi (1985) showed how frequency and recency interact with
each other in recall. They asked individuals to categorize an ambiguous stimulus description.
Participants were divided into two groups. Group 1 was primed with more positive primes
related to the description than negatives, but the negative primes were presented most recently
(e.g., boldpositive…courageouspositive…bravepositive…carelessnegative). Group 2 was primed with
more negative descriptions than positives, but the positives occurred more recently (e.g.,
unreasonablenegative… obstinatenegative…headstrongnegative…determinedpositive). There were two
delay conditions between the prime and the stimulus presentation: (i) a short 15-second delay
83

and (ii) a long 120-second delay. The results revealed both a recency effect and a frequency
effect. There was a recency effect in the short delay condition (15 seconds). After the brief delay,
participants tended to describe the stimulus in terms of the more recently primed category.
However, after the longer delay (120 seconds), the frequency of the primes showed an effect,
such that more participants used the frequent construct to categorize the stimulus information.
To summarize, how well concepts are retrieved from memory is influenced by: (i) the
grammatical structure that is used to present the information, (ii) how often individuals
encountered the concepts, and (iii) when the encounter happened. The memory literature has
revealed a strong recency effect especially when there is no delay or a short delay between
stimulus presentation and target retrieval. Experiment 2 aims to test the availability and
accessibility of discourse entities stored in short term memory when they are either contrastive
topic marked or nominative marked. It further examines how case marking might interact with
order of mention (i.e., where in the story the entities are introduced).

3.3 Some overlapping factors that influence discourse and memory representations
Thus far, some factors influencing discourse- and memory-level prominence have been
presented. As is evident, many of the factors that influence discourse prominence can also affect
memory saliency. These factors include: semantics, syntactic structures including focus
constructions, and order and frequency of mention. First of all, the order in which items are
mentioned matters in both discourse and memory. First-mentioned entities have been shown to
be highly accessible in discourse. In a similar way, first-mentioned items are remembered better
than items in the middle. Comprehenders have also been shown to remember the last word in a
list most accurately. Grammatical role contributes to discourse and memory representations as
84

well. Subjects are more prominent than objects in both domains. Grammatical subjects tend to be
talked about more frequently in discourse continuation than grammatical objects, and they are
also recalled better from memory. Memory literature has revealed that direct objects might be
more accessible than indirect objects in memory. Syntactic structure is also important. Focus
structures enhance the prominence of linguistic information in both discourse and memory.
To summarize, Table 13 presents the three factors affecting the accessibility of entities in
discourse and memory, which are relevant to my investigation.

Table 13. Factors influencing discourse and memory prominence
discourse prominence memory prominence
order of mention first-mentioned advantage
most recently-mentioned advantage
first-mentioned advantage
grammatical role subject advantage
subject advantage
direct object advantage over indirect obj.
focus structure
focused entity is talked
about more frequently
focused entity is recognized faster and
more accurately

The factors influencing the memory representation of concepts narrow down to the following
questions: (i) when is the entity or item mentioned? (e.g., recently mentioned vs. mentioned
earlier), (ii) how frequently is it mentioned? (iii) what role does the entity play for the sentence
meaning (e.g., subject vs. object)? and (iv) has special focus or attention given to it? (e.g., using
a focus structure). In the discourse stories used in Experiment 2, there were multiple factors
influencing the structure of discourse continuation and the memory representation. First, there
85

were two different entities introduced. Second, one entity was mentioned earlier than the other.
Third, the case marking on the entities differed. Experiment 2 examined individuals’ memory for
discourse entities when all these factors were in operation. The research questions addressed in
Experiment 2 are presented in (27).

(27) Research questions for Experiment 2 (i.e., Probe-word recognition)
I. If case marking influences who is likely to be in the center of comprehenders’
mental representation of the developing discourse, would it also impact how well
entities are retrieved from memory?
II. Would there be other factors, such as order of mention, that affect the memory
for discourse entities other than case marking?

3.4 Methodology
3.4.1 Participants
Twenty four adult native speakers of Korean at the University of Southern California
participated. None had participated in Experiment 1. They received $10. Participants’ average
length of residence outside of Korea was less than five years at the time of testing.

3.4.2 Materials
The stimuli were short stories consisting four sentences each. There were 24 target stories and 36
filler stories. All stories were written in Korean. Target stories were comprised in the
configuration shown in Table 14. A sample target story in Korean and the English translation are
presented in (28) and (29), respectively.
86

Table 14. Experiment 2 target story configuration
sentence configuration sample sentence
sentence 1
Introduces two occupation
nouns in subject position
[A teacher]SUBJECT1 and [a reporter]SUBJECT2-
NOM stopped by a bookstore while waiting for
their friends to arrive.
sentence 2 Provides background setting
Both of them-TOP wanted to buy new books as
they have finished the ones they were reading.
sentence 3 Comments on SUBJECT1
[The teacher-NOM/TOP]SUBJECT1 who likes
novels finds Jiyeong Kong’s new novel
interesting.
sentence 4 Comments on SUBJECT2
[The reporter-NOM/TOP]SUBJECT2 who reads
autobiographies a lot skims through Obama’s
autobiography.

(28) Sample target story used in Experiment 2
선생님과 기자가 친구들을 기다리는 동안 서점에 들렀습니다. 두 사람 다 읽고 있던 책을 끝낸 터라
새 책을 사기로 합니다. 소설을 좋아하는 선생님이/ 은 공지영의 신작을 관심 있게 봅니다. 자서전을
많이 보는 기자가/ 는 오바마 대통령의 자서전을 훑어봅니다.

(29) English translation of the sample target story presented in (28)
A teacher and a reporter stopped by a bookstore while waiting for their friends to
arrive. Both of them wanted to buy new books as they have finished the ones they
were reading. The teacher, who likes novels, finds Jiyeong Kong’s new novel
87

interesting. The reporter, who reads autobiographies a lot, skims through Obama’s
autobiography.

The first sentence in each target story introduced two occupation nouns in subject
position (e.g., A teacher and a reporter stopped by a bookstore…)
11
. Throughout this chapter,
the character that was introduced first in the story is referred to as SUBJECT1, and the second
character is referred to as SUBJECT2. SUBJECT1 was commented on again in Sentence 3 of the
story, and SUBJECT2 was mentioned again in Sentence 4. Every target story was about two
individuals who were engaged in a similar activity (e.g., looking for books to buy, waiting to
order drinks, going to a friend’s housewarming party, etc.). This configuration made it possible
to manipulate the case marking on the protagonists so that the contrastive topic marker was
plausible.
The frequencies of target occupation nouns were controlled. I first selected the top 60
most frequently occurring occupation nouns in Korean (Kang & Kim, 2009). I then selected 48
of them and matched each pair of nouns for their frequencies and number of Korean characters
12
.
The number of Hangul (i.e., Korean) characters in the nouns ranged from two to four. I also
controlled for the semantic association of each noun pair, such that none of the nouns was
semantically associated with its pair according to The University of South Florida word
association (Nelson, McEvoy, & Schreiber, 1998). The complete list of target occupation nouns
with their frequencies is presented in Appendix A.

11
The configuration of the subject noun phrase in the first sentence was the following:
[SUBJECT1-conjunctive (i.e., ‘and’)-SUBJECT2]-nominative marker
12
The number of characters in each pair of occupation nouns matched except for two pairs, in which one of the
nouns had two characters and the other three.
88

Experiment 2 also included 36 filler stories, each consisting four sentences. There were
three filler categories: (i) stories with proper names instead of occupation titles (8 stories), (ii)
stories with other occupation nouns (16 stories), and (iii) well-known Korean folk stories and
Aesop’s fables (12 stories). About half of the stories in the third filler category (i.e., Korean folk
stories and Aesop’s fables) had their contents slightly modified from the original (e.g., a rabbit
being slow in a race and a turtle being fast). This modification was made to motivate participants
to focus on the reading. All the targets and fillers were normed beforehand with four native
speakers of Korean for their naturalness.

3.4.3 Design
I manipulated two factors in the target stories. The first factor was case marker (i.e., nominative
vs. contrastive topic) on SUBJECT1 and SUBJECT2 in the third and fourth sentences of the
story, respectively. The second variable was probe word, used in the recognition task: that is,
whether the probe word was SUBJECT1 or SUBJECT2 from the story. These two factors with
two levels each yielded the four conditions in Table 15.

Table 15. Four target conditions in Experiment 2
condition case marker for SUBJECT1 and SUBJECT2 probe word
NOM-SUBJ1 NOM SUBJECT1
NOM-SUBJ2 NOM SUBJECT2
TOP-SUBJ1 TOP SUBJECT1
TOP-SUBJ2 TOP SUBJECT2

89

The case marking on SUBJECT1 and SUBJECT2 in the target stories was always the same: both
nouns either had the nominative marker or the topic marker.
It is worth noting that the linguistic stimuli and the designs of Experiment 1 and
Experiment 2 differ. As we saw in the preceding chapter, Experiment 1 examined the discourse
representations of at least three different entities in the story: (i) a global discourse subject, (ii) a
member of a discourse set mentioned in the story, and (iii) other unmentioned members of the
set. Each story in Experiment 1 began by introducing a discourse subject and continued to
introduce a set that the discourse subject is associated with. The stories ended with a sentence
describing one of the set members (what I refer to as the mentioned member), marked with the
nominative case in one condition and with the contrastive topic case in another condition. In my
results for Experiment 1, we saw that the presence of the contrastive topic marking on the
mentioned member boosts the likelihood that participants will introduce unmentioned set
members in their continuations.
On the contrary, Experiment 2 focuses on the memory representations that
comprehenders construct for mentioned referents and thus does not directly test the unmentioned
set members. Instead, it explores the memory representations of the two topics of the story by
testing how well they were recognized when probed later (Unmentioned set members cannot be
tested this way, since they are by definition not mentioned in prior discourse). I was interested in
testing two factors that might guide the memory prominence of the entities: (i) the difference
between nominative marking and contrastive topic marking, and (ii) where in the story the
characters were commented on (i.e., sentence 3 or sentence 4).

90

3.4.4 Procedure
The experiment was conducted on a 17.2-inch monitor PC laptop computer using Paradigm
software (Perception Research Systems). For each trial, participants read a story presented on the
computer screen for 14 seconds.
13
After reading each story, they were presented with a three-
digit number and were asked to count aloud into a microphone the number backward by 3s for
15 seconds. Participants’ number counting responses were recorded. Participants had two main
tasks: (i) probe-word recognition and (ii) comprehension question. After the delay task, a probe
word appeared in the center of the screen following a warning signal ‘+’, which lasted for 700
milliseconds (ms). Participants were asked to indicate as fast and as accurately as they could
whether the word had occurred in the story they just read. They were asked to press the ‘J’ key
on the keyboard for ‘YES’ and the ‘F’ key for ‘NO’. The probe words in the target trials were
either SUBJECT1 (subject of Sentence 3 in the story) or SUBJECT2 (subject of Sentence 4).
Therefore, the correct response in each target trial was always ‘YES’. The probe words were
presented without any case markers. The number of expected ‘YES’ and ‘NO’ responses was
balanced across all items, half requiring ‘YES’ and the other half ‘NO’. There was no time limit
for the probe recognition task. Paradigm recorded participants’ yes-no responses and the time
taken to make the choice.
After participants had completed the probe-word recognition task for each item, a ‘YES-
NO’ comprehension question appeared on the monitor. The contents of the comprehension
questions were balanced, such that: (i) there were equal numbers of questions asked about each
of the four sentences in the targets and fillers, and (ii) a YES answer was expected on half of all

13
The 14-second duration of story presentation was determined after norming it with two participants, such that it
was long enough for participants to read the story but not enough for them to explicitly study the contents to prepare
for the probe recognition and the comprehension question. The comments from the 24 participants who took part in
this experiment suggest that they found this time to be sufficient to read the stories.
91

trials and a NO answer was expected on the other half. Participants were reminded which key
was for which response on the bottom of the monitor for both the probe word and the
comprehension question. There was no time limit for the comprehension questions.
The target and the filler stories were distributed among four lists using a Latin Square
design. There were also four reverse lists. Each participant saw either the NOM-marked subjects
or the TOP-marked subjects of each story and saw equal numbers of NOM-marked and TOP-
marked items during the experiment. The average time taken for each participant to complete the
experiment was approximately 50 minutes. The comprehension question in (30) was asked for
the sample story shown in (29) above.

(30) Sample comprehension question (Question from Sentence 1 requiring a ‘YES’ response)
선생님과 기자는 친구들을 기다리는 동안 서점에 갔나요?
Did the teacher and the reporter stop by a bookstore while waiting for friends?

3.5 Predictions
Experiment 1 (story continuation task) showed a case marking effect between the Korean
nominative marker and the contrastive topic marker on entities’ prominence in discourse
progression. In a discourse where there was a global discourse topic and a local topic,
participants were more likely to talk about the global discourse topic when the local topic was
NOM-marked. However, when the local topic was contrastive topic marked, the prominence of
the global discourse topic decreased, and participants were inclined to talk about other members
of the discourse set that had contrasting values to the local topic (although they were previously
unmentioned in the story). Having observed this effect, I hypothesized that the Korean
nominative marker and the topic marker might have different memory-level consequences. I
92

predicted that the different case markers would provide differing degrees of retrieval cues. More
specifically, I predicted that the use of the topic marker would enhance referent retrieval because
the subjects were made contrastive with respect to one another. This prediction is based on the
existing observations that contrastively focused referents tend to have increased saliency in both
discourse and memory (e.g., Almor, 1999; Hornby, 1974; Kaiser, 2006; 2011; Singer, 1976).
Although the previous findings are mainly about contrastively focused entities (i.e., new
information) being mentioned more frequently and recalled better, contrastive topics in this
experiment might exhibit similar patterns. This is because the contrastive topics that are
introduced in the target stories are members of a discourse set just as contrastively focused
entities would be. Therefore, the probe-word recognition would be more accurate and faster
when the subject is contrastive TOP-marked in the story than when it is NOM-marked. Similar
findings have been observed where salient entities, such as grammatical subjects, are recognized
faster than referents that are not as salient (e.g., Gernsbacher, 1990; Song & Fisher, 2005).
Regarding the comprehension question results, it was predicted that the content
information of the stories would be remembered better when the protagonists are contrastive
topic-marked than when they are nominative-marked. Because the contrastive topic marking has
been shown to make the entities in a discourse set salient, the content information surrounding
the entities might also stand out better when the characters are made contrastive to each other.
Therefore, it was predicted that participants would be better at correctly answering the
comprehension questions when the protagonists are TOP-marked in the story than when they are
NOM-marked. I predicted this to be the case even though the questions were about any of the
four sentences in the story, and not just about the last two sentences that described the
protagonists.
93

No specific predictions have been formulated for the time taken to correctly answer the
comprehension questions because the time taken to read the question and to retrieve the content
information might have varied largely across participants, washing out any possible effects.

3.6 Results
The results section is divided into two: (i) the probe-word recognition task (accuracy and RT)
and (ii) the comprehension question accuracy. I first present the probe-word recognition
accuracy and RTs, and will turn to the comprehension question accuracy
14
.

3.6.1 Probe-word accuracy
This section reports the probe-word recognition results, the degree to which participants
correctly responded to the probed characters (with ‘YES’) across the four conditions. Table 16
and Figure 5 present the accuracy rates. In the total of 576 data points, there were 35 incorrect
‘NO’ responses (6%).

14
Here, I treat the accuracy data and the reaction time data as interval measurements and analyze them with
ANOVA, although this may oversimplify aspects of the data. In future work, additional analyses – especially of the
binary accuracy data – could be conducted using logistic regression.

94

Table 16. Percentage of correct ‘YES’ responses for probed entities across the target conditions
probe word
accuracy rate (%) SD (%)
NOM-SUBJ1 90 11
NOM-SUBJ2 97 7
TOP-SUBJ1 94 11
TOP-SUBJ2 96 9

Figure 5. Percentage of correct ‘YES’ responses for probe words across four target conditions

As can be seen in Table 16 and Figure 5, the probe recognition accuracy was high in all four
conditions (above 90%). One noticeable pattern is that recognition of SUBJECT2 (96.5%
average accuracy) is overall better than recognition of SUBJECT1 (92% average accuracy)
regardless of whether the protagonists were NOM-marked or TOP-marked in the story. Different
90
97
94
96
70
75
80
85
90
95
100
NOMSUBJ1 NOMSUBJ2 TOPSUBJ1 TOPSUBJ2
percentage (%)
Probe-word recognition accuracy rate
95

from this order effect, the effect of case marking does not seem to be as clear. The average
accuracy was 93.5% when the entities were NOM-marked in the story (the average of the left
two bars in Figure 5) and 95% when they were TOP-marked (the average of the right two bars).
I conducted repeated measures ANOVAs by subjects and by items to test the effect of
case marker (e.g., NOM vs. TOP) and probe word (e.g., SUBJ1 probe vs. SUBJ2 probe) on how
accurately participants identified having seen the protagonists. The results showed a main effect
of probe word; F1(1,23) = 4.504, p1 < .05, F2(1,15)=5.068, p2 < .04. There was no main effect of
case marker and no significant interaction between case marker and probe word.
The significant effect of probe word suggests that participants responded more accurately
to the probe words when SUBJECT2 was prompted (96.5%) than when SUBJECT1 was (92%).
Contrary to the predictions however, case marker did not yield significant results. That is,
participants’ recognition of the probed subjects did not differ regardless of whether they were
NOM-marked or TOP-marked in the story.

3.6.2 Probe-word recognition reaction time (RT)
I trimmed the probe-word RT data in order to manage extreme reaction times. I first excluded
5% of the smallest and the largest RTs in order to avoid any extreme RTs affecting the overall
mean and the standard deviation calculation. This resulted in the exclusion of 58 out of 576 data
points (10%). The remaining data yielded the mean of 1526 ms and the standard deviation of 640
ms. I then excluded any RTs that were beyond the Mean RT±3 SDs (RT ≥ 3447ms and RT ≤ -
393ms) from all of the original data points
15
. This led to an exclusion of 37 data points (6% of all
data). Following a general practice in psycholinguistics (e.g., Birch et al., 2000), I also excluded

15
No RT was excluded from the shorter end of the RTs because the shortest RT observed was 647 ms. The general
practice in the word recognition RT literature is to exclude RTs smaller than 150 ms.
96

the RTs of incorrect ‘NO’ responses from the analyses (exclusion of additional 26 data points).
Table 17 shows the mean and the SDs of the probe-word recognition RTs for the remaining data
in the four target conditions. Figure 6 depicts the RT patterns.

Table 17. Mean and SDs of probe-word recognition RTs across the four experimental conditions
probe recognition
RT (ms) SD (ms)
NOM-SUBJ1 1367 292
NOM-SUBJ2 1498 384
TOP-SUBJ1 1462 400
TOP-SUBJ2 1453 359

Figure 6. Time taken to correctly recognize the probed protagonists, Experiment 2

1367
1498
1462
1453
700
800
900
1000
1100
1200
1300
1400
1500
1600
NOMSUBJ1 NOMSUBJ2 TOPSUBJ1 TOPSUBJ2
RT (ms)
Probe-word recognition RT
97

Numerically speaking, the RTs seem fairly close to each other between the conditions (with the
largest difference being 45 ms), except in the NOM-SUBJ1 condition. This is the condition
where the protagonists were NOM-marked in the story, and the probed subject was SUBJECT1.
The RT of this condition was about 100 ms shorter on average than that of the other three
conditions.
I conducted repeated measures ANOVAs by subjects and by items to test the effect of
case marker (e.g., NOM vs. TOP) and probe word (e.g., SUBJ1 probe vs. SUBJ2 probe) on the
time taken to accurately recognize the probed protagonists. The reaction time results revealed no
main effect of case marker or probe word. There was also no significant interaction between the
two factors (all ps > .16). The time taken to correctly recognize the probed characters did not
differ regardless of whether SUBJECT1 or SUBJECT2 was probed. Moreover, participants’
reaction times did not differ significantly regardless of whether the subject nouns were NOM-
marked or TOP-marked in the story.

3.6.3 Comprehension question accuracy
16

We now turn to the comprehension question accuracy. Of the total of 576 data points, 85 of them
were incorrect responses (14.75%). Table 18 shows the mean and the standard deviations of the
comprehension question accuracy rates across the four target conditions. Figure 7 illustrates the
accuracy patterns.

16
As for the comprehension questions, I only report the accuracy data and not the RT data. The time taken to
correctly answer the comprehension questions was not reported separately in the main text because comprehension
questions tend to be complex to answer, and the response times can be quite variable. In fact, a repeated measures
ANOVAs showed no main effect of CASE MAKER (NOM vs. TOP) or PROBE WORD (SUBJ1 probe vs. SUBJ2
probe) on the comprehension question RTs. The interaction between the two factors was not significant either (all ps
> .3).
98

Table 18. Percentage of correct responses for the comprehension questions

comprehension question
accuracy rate (%) SD (%)
NOM-SUBJ1 81 16
NOM-SUBJ2 88 9
TOP-SUBJ1 83 18
TOP-SUBJ2 89 14

Figure 7. Percentage of correct responses for the comprehension questions, Experiment 2

The comprehension question accuracy was fairly high overall (above 81%). From Table 18 and
Figure 7, we can observe a similar pattern as in the previous probe-word recognition accuracy.
The comprehension question accuracy is on average higher when the probed word was
SUBJECT2 than when it was SUBJECT1. That is, participants responded to the comprehension
questions more accurately when the question followed a SUBJECT2 probe than when the
81
88
83
89
60
65
70
75
80
85
90
95
NOMSUBJ1 NOMSUBJ2 TOPSUBJ1 TOPSUBJ2
percentage (%)
Comprehension question accuracy
99

question followed a SUBJECT1 probe. This may suggest that having to recognize the recently
mentioned SUBJECT2 before answering the question might have increased comprehenders’
memory for the story information. However, note that the comprehension questions were about
any of the four sentences in each story, not just from the two sentences that commented on the
protagonists. Therefore, it is interesting that participants’ answers to the comprehension
questions seem to depend on which protagonist was probed.
A repeated measures ANOVAs was conducted to test the effect of case marker (e.g.,
NOM vs. TOP) and probe word (e.g., SUBJ1 probe vs. SUBJ2 probe) on how accurately
participants responded to the comprehension questions. There was a main effect of probe word
on the comprehension question accuracy. This effect was significant in item analysis and
marginal in subject analysis; F1(1,23) = 3.682, p1 < .07, F2(1,15) = 7.87, p2 < .02. The main
effect of probe word (i.e., SUBJ1 vs. SUBJ2) indicates that participants answered the
comprehension questions more accurately in the trials where the tested probe was SUBJECT2
(88.5%) than when it was SUBJECT1 (82%).
Case marker did not have a significant effect on the comprehension question accuracy.
Whether the protagonists were TOP-marked or NOM-marked did not change the performance.
The interaction between case marker and probe word was not significant.

3.7 Discussion - Experiment 2
In Experiment 2, I employed a probe-word recognition task to test whether different case
markers in Korean would influence how well comprehenders recall the entities presented in
stories and how well they remember the story content. Participants read four-sentence stories
written in Korean in which two characters were introduced with the nominative case marker -i/-
100

ka in one condition and with the contrastive topic marker -un/-nun in the other. After reading
each story and completing a 15-second delay task, participants performed two tasks: probe word
recognition and a comprehension question. The two subject occupation nouns from each story
were presented as probe words. Participants were directed to indicate whether the probe word
had occurred in the story. Each comprehension question was about the content information taken
from one of the four sentences in the story. The results were analyzed in terms of: (i) the
accuracy and the reaction time of the probe-word recognition and (ii) the comprehension
question accuracy.
It was predicted that individuals’ memory for discourse entities would be influenced by
the way they were introduced. When two protagonists, involved in a common activity, are
introduced with the contrastive topic marker, the activity that they are engaged in and the
characters themselves might be made more distinctive from each other. This is because the
contrastive topic marker has been shown to make the members of a discourse set more salient
compared to the nominative marker. Therefore, it was predicted that comprehenders would
recognize the characters faster and more accurately when they were introduced with the
contrastive topic marker in the story than with the nominative marker. It was also hypothesized
that participants would answer the comprehension questions more accurately in the TOP
condition than in the NOM condition.
The prediction that contrastive topic-marked entities would be retrieved better from
memory was not borne out. There was no case marker effect on the recognition of entities. The
difference between contrastive topic marking and nominative marking did not change how well
participants answered the comprehension questions either. Furthermore, I predicted that TOP-
101

marked subjects would be recognized faster than their NOM-marked counterparts. However, the
reaction times showed no difference between NOM-marked subjects and TOP-marked subjects.
Though the initial hypotheses have not been confirmed, there was a clear pattern
observed throughout the results. The results indicate a strong recency effect (e.g., Baddeley &
Hitch, 1977; Postman & Phillips, 1965; Glanzer & Cunitz, 1966). The recency effect has been
shown widely in the memory literature, particularly in free serial recall. When individuals are
asked to recall the items from a list, they recall the most recent item the best. In this experiment,
participants recognized SUBJECT2 more accurately than SUBJECT1 in the probe recognition
task. SUBJECT2 is the character that was mentioned later in the story, and SUBJECT1 is the
character that was mentioned earlier. Also interestingly, the SUBJECT2 probe improved
comprehension question accuracy. Participants responded to the comprehension question more
accurately when the question followed a SUBJECT2 probe than when the question came after a
SUBJECT1 probe.
The fact that recognizing SUBJECT2 was easier than SUBJECT1 may suggest that more
recently talked-about entities (i.e., SUBJ2) has stronger presence in comprehenders’ short-term
memory. Though the recency effect has been found frequently in free recall of words, we
observed the same tendency in discourse processing and concept recall. The reason why we
observed a similar phenomenon might be due to that the retrieval of the discourse subjects from
short-term memory might have involved similar processing strategies as those involved in free
word recall.
However, the recency effect we see in probe-word recognition is contradictory to what
Gernsbacher and Hargreaves (1988) found in their probe-word recognition experiments where
first-mentioned subjects were recognized faster than second-mentioned (i.e., the primacy effect).
102

Gernsbacher and Hargreaves suggest that the first-mentioned holds a more prominent place in
memory because comprehenders lay out the foundation of the discourse with the first-mentioned
and build the subsequent discourse upon it. The discrepancy between their study and the current
Experiment 2 might be due to different factors. One factor could be the story presentation
methods. Gernsbacher and Hargreaves presented the sentences word by word instead of the
whole sentence. This method might have reinforced the integration of the characters into the
story more efficiently than my method of story presentation where all four sentences in each
story were presented at once.
Another reason why Gernsbacher and Hargreaves found the primacy effect in entity
recall might be because the two characters in their sentences denoted two different grammatical
roles as in Tina beat Lisa in the state tennis match. The first mentioned Tina is the sentence
subject and the agent of the verb beat. On the other hand, the second mentioned Lisa is the
sentence object and the recipient of the verb. Therefore, the recognition of Tina as the subject
and the agent (an active doer of the verb) might have been better than that of Lisa. However in
my experiment, both protagonists in each story were grammatical subjects as in [A teacher and a
reporter] subject stopped by a bookstore. The two characters form a single subject noun phrase of
the sentence. Further, they each were the subject of the subsequent sentences as well. Because
there is no difference in the grammatical role of the two protagonists, their presence in the
mental representation might not have varied as much as a subject and an object would have.
Moreover, because the two protagonists were introduced in one conjoined subject noun in
the first sentence, one character might not have been perceived as the topic of the story or the
foundation of the discourse structure. They might have been considered as one discourse topic
for which the story was created. Since both entities were regarded as subjects and topics,
103

participants’ recognition of them might have relied on the order of mention only and not on
discourse prominence or the information structure provided by their roles.
Furthermore, in Gernsbacher and Hargreaves’ study, the probe words appeared
immediately after the sentence final word. For example, TINA or LISA appeared 150 ms after
participants read the sentence final word. However, in my Experiment 2, there was a delay of 15
seconds during which participants were engaged in a number counting task. It could be the case
that having completed the delay task made the probe word recognition as if the probe was out of
a list of words instead of a story. By the time participants have consumed their cognitive
capability in number counting, the story contents might not have been clearly present in their
memory.
In addition to the recency effect found in the entity recall, it is also interesting that the
comprehension question accuracy increased when SUBJECT2 was probed than when
SUBJECT1 was. Note that the comprehension questions were about one of the four sentences in
the story, not just about the third or the fourth sentence that specifically described what
SUBJECT1 and SUBJECT2 were doing, respectively. One possible explanation for this recency
effect in the comprehension question accuracy is the following. Having seen SUBJECT2, which
lingers around in comprehenders’ memory longer, helps retrieve the story contents better.
SUBJECT2 provides stronger retrieval cues for content retrieval than SUBJECT1. In other
words, being cued with a more salient concept in the discourse story facilitates the retrieval of
the content information. Dillon et al. (2014)’s observation may support this explanation. The
researchers view that a recently encountered entity has high resting activation in memory. The
information that has a high resting activation may also boost the activation of the other
information that it is associated with. To apply this to the current results, SUBJECT2 would have
104

higher resting activation in memory than SUBJECT1 because it is more recently mentioned, and
thus has less temporal decay. Having to recognize highly activated SUBJECT2 may lead to the
activation of the prior content information of the story.
The opposite side of the explanation would be that having to determine whether a less
salient, earlier introduced concept (i.e., SUBJECT1) had occurred in the story blurs the memory
of story contents. To put it another way, having struggled to recognize a less salient concept
takes more of comprehenders’ cognitive capacity and thus leaves less room to tackle the
comprehension question. There might be discourse-level Attentional Blink (e.g., Broadbent &
Broadbent, 1987; Raymond, Shapiro, & Arnell, 1992) in effect here. Attentional Blink (AB) is a
phenomenon observed in Rapid Serial Visual Presentation (RSVP). When letters were presented
rapidly one after another at the same location on a screen, detecting a second target presented
short time after a first target is difficult. Having to recognize SUBJECT1 (i.e., less accessible
referent) takes so much attention that participants might have unintentionally neglected to
correctly answer the comprehension question.
Let us now turn to the discussion of the probe-word reaction times. The time to correctly
respond that SUBJECT1 and SUBJECT2 had occurred in the story did not differ significantly
regardless of which case marker was used. I predicted that the reaction time of the TOP-marked
nouns would be faster than that of the NOM-marked nouns because topic marking makes the
members in a discourse set more prominent than nominative marking (Experiment 1 results;
though, as mentioned in Section 3.4.3 (Design), Experiment 2 does not test the unmentioned
members which are the ones whose likelihood of mention was boosted in Experiment 1). The
recognition of story characters did not support this prediction. One possible explanation for this
lack of case marker effect might be due to the presence of the delay task. Having been engaged
105

in the backward number counting task, participants’ memory for the subjects and other
contextual information about them might have already faded. This explanation seems to support
my earlier claim that the probe recognition might have been more like free recall of words rather
than discourse information retrieval due to the distraction from the delay task.
Furthermore, the lack of case marker effect might have been due to the story structure.
SUBJECT1 and SUBJECT2 were made contrastive with one another in both the NOM and the
TOP conditions by the virtue of the structure of the presented discourse. Even when the
characters were marked with the nominative case, their descriptions were already contrastive.
The description of contrasting activities might have minimized the case marker effect.
It is important to acknowledge that the finding that SUBJECT2 to be recognized more
accurately than SUBJECT1, which we have so far discussed in terms of recency, may also stem
from an anticipatory effect that favors SUBJECT2, acting together with a recency effect. That is,
the comment on SUBJECT1 in sentence 3 might have invited an expectation that SUBJECT2
will be mentioned next. This expectation for the SUBJECT2 mention might have reinforced the
presence of SUBJECT2 and consequently helped make the referent easier to be recognized.

3.8 General discussion and conclusion of Experiment 1 and Experiment 2
Experiments 1 and 2 investigated the effects of the Korean nominative and the contrastive topic
case markers on discourse representation of entities and their accessibility in short-term memory.
The experiments compared how the way in which discourse entities are introduced and the roles
they play in the overall meaning would contribute to the development of the discourse and to the
memory retention of the content information. The results from the off-line story continuation
experiment revealed that comprehenders’ expectations of upcoming stories differed depending
106

on the case markers used to introduce discourse characters. Overall, the global discourse topic
showed a strong presence in discourse continuation. Crucially however, the global topic
prominence was modulated by contrastive topic marking. Contrastive topic marking made other
members in the discourse available in story continuation.
Having seen the case marker effect on discourse continuation, I predicted that it would
also affect how discourse concepts are retrieved from short-term memory. Previous studies
investigating entities’ representations in discourse and memory have indicated that the entities
that are shown to be highly accessible in discourse are also remembered better (e.g., Almor,
1999; Birch, Albrecht, & Myers, 2000; Birch & Garnsey, 1995; Foraker & McElree, 2007;
Garrod, Freudenthal, & Boyle, 1994; Gernsbacher & Hargreaves, 1998; Gundel, 1999).
Therefore, I expected that when two nouns are contrastive TOPIC-marked, their accessibility in
memory would increase compared to their NOMINATIVE-marked counterparts. This prediction
was not borne out in Experiment 2. Participants did not recognize TOP-marked referents more
accurately than NOM-marked ones. In addition, the time taken to accurately recognize discourse
characters did not differ whether they were TOPIC-marked or NOMINATIVE-marked.
The results of Experiment 1 indicate that different case markers influence how
comprehenders structure the discourse and how they build a mental model of the discourse. This
process seems to be more abstract, complex, and require more attention and active participation
from comprehenders than a recall task. In Experiment 2 with the probe-word recognition task,
the same level of attention and processing of the texts as in the story continuation experiment
might not have been required.
It is interesting to see how readers picked up on the case marker that was used to
introduce a story character and shaped their continuations based on that cue. However, this might
107

not necessarily mean contrastive TOPIC-marked entities linger around longer in short-term
memory compared to NOMINATIVE-marked entities. Structural cues, such as case markers,
may be recognized differently in language production (i.e., story continuation) and
comprehension (i.e., probe recognition) of texts in limited time.
In an attempt to reveal the discourse and the cognitive consequences of the Korean
contrastive topic marker -un/-nun, Experiment 1 and Experiment 2 indicate that the case marker
does in fact make other “unspoken” set members accessible. The marker is distinctive from the
nominative marker, such that the importance of an otherwise strongly present global discourse
topic is temporarily reduced. However, this effect might not be present to the same extent in on-
line processing of entities and their discourse interpretations.
In order to better interpret the probe-word recognition results, more studies need to be
carried out. It seems important to investigate if the presence of the delay task might have
dismissed any case marker effect. It should also be noted that the story structures were different
between Experiment 1 and Experiment 2. In Experiment 1, there was a discourse topic
introduced in the first sentence of the story, and one of the set members was commented on in
the fourth sentence. In Experiment 2, two entities (i.e., SUBJECT1 and SUBJECT2) were
introduced together in the first sentence, and they were commented on one after the other. In
order to test how memory and discourse representations interact, the next investigation may use
the same discourse structures. Experiment 3 reported in the next chapter filled this gap by using
the same target stories as Experiment 2 for an off-line story continuation task.
Experiment 2 has revealed the recency effect in entity recall; the protagonist mentioned
recently was recalled more accurately than the protagonist mentioned earlier in the story. In
order to test how the accessibility of the recently-mentioned plays out in discourse continuation,
108

Experiment 3 asked readers to add three sentences that naturally continue each story. Using the
same story design has allowed direct comparisons with regard to which entity comprehenders
remember better (Experiment 2) and whether that entity is talked about more frequently in
subsequent sentences (Experiment 3) due to the boost in the memory representation.

109

Chapter 4
Experiment 3: Comparison between entities’ discourse and memory representations

4.1 Introduction
In this chapter, I report Experiment 3, which employed an off-line story continuation task. It is a
follow-up investigation motivated by the results of Experiment 2, reported in the preceding
chapter. It is also the second story continuation experiment that I conducted after Experiment 1.
The main aim of Experiment 1 was to test the effect of different subject markings in Korean on
entities’ discourse prominence. I compared the prominence or accessibility of a protagonist when
it was introduced in a story with the contrastive topic marker and when it was introduced with
the nominative marker. In addition to the prominence of the entity whose case marking differed,
I also analyzed the prominence of other discourse participants, such as a global discourse topic or
members of a discourse set. Entities’ discourse prominence was measured by the likelihood of
the entity to be mentioned as the subject of participants’ continuation sentences.
The purpose of Experiment 3 was different from Experiment 1, which tested the effect of
case marking. Experiment 3 was carried out to directly compare its story continuation results
with the memory retrieval results from Experiment 2. The results from Experiment 2 showed that
individuals remember the character that had been mentioned more recently in the story better
than the character that had been mentioned earlier in the story (i.e., recency effect, e.g., Baddeley
& Hitch, 1977; Postman & Phillips, 1965; Glanzer & Cunitz, 1966). Using the same stories as
Experiment 2, I tested whether the recently mentioned character that readers remember better
would also be the one they talk about more frequently in their continuation sentences. Therefore,
the predictions for Experiment 3 do not follow directly from Experiment 1. Whereas the main
110

purpose of Experiment 1 was to compare the nominative marker and the contrastive topic
marker, the purpose of Experiment 3 was to compare the recency effect and the primacy effect in
discourse continuation.
Experiment 3 hoped to address how the linguistic forms we observe (e.g., spoken
utterances, sentence productions) are related to the cognitive activities that may take place in
language users’ minds. As scholars have observed, language forms are highly correlated to the
conceptual, informational, and psychological properties of interlocutors (e.g., Arnold, 2013;
Gleitman & Papafragou, 2005; MacDonald, 2013; MacDonald & Thornton, 2009; Pinker, 2007).
For example, the tendency for animate nouns to appear as a sentence subject (rather than an
object) or as a discourse topic is attributed to the fact that they are conceptually more salient than
non-animates (e.g., Clark, 1965; Bock, Loebell, & Morey, 1992; Gennari, Mirković, &
MacDonald, 2012). Furthermore, the form of a referring expression (e.g., a reduced form, such
as a pronoun vs. a modified noun phrase) reflects how accessible or activated the concept or the
entity being refer to is in the minds of the interlocutors
17
.
Because what we observe in our linguistic behavior is a reflection of other cognitive
processes, such as memory, comparing memory representations and linguistic forms would help
us better understand how the language we use might interact with our memory. By comparing
the results of Experiments 2 and 3, I aimed to test how the likelihood of an entity to be
mentioned in discourse relates to comprehenders’ memory for the entity. As some of prior work
has demonstrated, are the entities that are recognized faster or more accurately also the ones that

17
Although studies have demonstrated that the entities that are highly activated in the mind of a comprehender tend
to be referred to with a reduced referring expression (e.g., pronoun) (e.g., Arnold, 2001), there has been an
inconclusive debate about whether being referred to with a pronoun indicates higher accessibility of the entity
compared to other referents that are not referred to with a pronoun. For example, Fukumura and Van Gompel’s work
(2010) has indicated that there might not be such a correlation.
111

are more likely to be mentioned? Experiment 3 investigates this question in a larger context with
four-sentence stories and three continuation sentences to each. It might seem intuitive to think
that a particular entity is being talked about because it is most salient in the current memory
representation. However, the dynamic nature of discourse development might not always
correspond to saliency in memory.
Crucially, Experiment 2 (memory task, comprehension) and the current Experiment 3
(story continuation, production) may require different levels of cognitive processes.
Comprehenders’ primary goal in Experiment 2 was to comprehend the text and retrieve
information from it. In order for participants to recognize entities and answer comprehension
questions, the task would have required them to look backward and recall what they read, relying
primarily on short-term memory. The task might not have required readers to construct the
representation of the story beyond what had been presented in the text. In contrast, the story
continuation Experiment 3 would have required readers to not only understand the text, but also
to construct a coherent continuation of it. For that, they would have to look forward beyond what
had been presented in the text. Readers would need to connect the information available in
memory and plan a continuation with the goal of creating a coherent story. The production task
in Experiment 3 would also require readers to refer to their long term memory to create a
plausible and natural-sounding continuation of the discourse. Because the two tasks might have
required different levels of cognitive processes, we might see a discrepancy between which
subject readers talk about more frequently and which subject they remember more accurately.
As Experiment 3 focuses on the prominence, accessibility, or saliency in discourse, let us
briefly review how these concepts are measured. Researchers have used a variety of different
means to measure prominence: (i) reaction time in a comprehension-based lexical decision task,
112

(ii) likelihood of subsequent mention in a production task, (iii) likelihood of using a pronoun to
refer to an entity, (iv) likelihood of interpreting a pronoun as referring to an entity, and (iv)
memory recall of an entity. Therefore, a prominent entity is understood as the one being
frequently referred to in subsequent mention, often with a reduced linguistic form, such as a
pronoun. An entity is also considered to be prominent or salient when it is recognized fast and
accurately when probed. Among these measures, I used the likelihood of mention to test an
entity’s accessibility, more specifically, the probability of the entity being referred to again as a
subject in subsequent sentences. Therefore, the more likely for an entity to reappear as the
subject of a subsequent sentence, the higher level of prominence or accessibility the entity has in
discourse continuation, relative to other entities
18
.

4.2 Prior work on the prominence of discourse entities
In the preceding chapters, I have outlined some of the factors influencing the prominence of
entities in discourse, as well as in memory. As is evident, there are overlapping aspects between
the two representations. When an entity is highly likely to be mentioned in discourse, it is
suggested to be salient or accessible in the minds of comprehenders (e.g., Ariel, 1991; Arnold,
2010; Ferreira & Yoshita, 2003; Fukumura & Van Gompel, 2010; Kaiser, 2010; Kaiser 2011;
Rohde, & Kehler, 2014). However, it is important to note that an entity’s saliency or accessibility
changes throughout the discourse as new referents continue to be introduced and new topics are
being discussed. The complex nature of discourse therefore involves a variety of different
aspects that contribute to entities’ prominence. I will briefly summarize the factors as they are
relevant to Experiment 3.

18
In Experiment 2 with the probe-word recognition task, prominence in memory was measured by recognition
accuracy.
113

One factor that influences prominence is order of mention. The first-mentioned
participant is recognized faster and more accurately (e.g., using probe name recognition) (e.g.,
Chang, 1980; Gernsbacher & Hargreaves, 1988; Gernsbacher, Hargreaves, & Beeman, 1989;
McDonald & MacWhinney, 1995; Yongming & Yao, 1995). Gernsbacher and Hargreaves
(1988) found that the first-mentioned character was recognized more accurately even when it
was not a semantic agent of the verb as in (31) or when the participant was not a grammatical
subject in (32)

(31) Tina was beaten by Lisa in the state tennis match.
(32) Because of Tina, Lisa was evicted from the apartment.
(Examples from Gernsbacher & Hargreaves, 1988, pp. 736-738)

Gernsbacher, Hargreaves, and Beeman (1989) suggest that a first-mentioned character is more
accessible from the comprehenders’ mental representations than a second-mentioned character.
This is because the first mentioned entity may serve as the foundation for the sentence-level
structure where the subsequent information is built upon. Evidence that supports the role of the
first-mentioned character as a discourse foundation comes from reading time data. The time
taken to read the sentence initial word is generally longer than the time taken to read other words
in the sentence except for the sentence final word (e.g., Aaronson & Ferres, 1983; Aaronson &
Scarborough, 1976; Chang, 1980). Readers might require longer time to process the sentence
initial word because they are laying out the discourse foundation. Since comprehenders tend to
build the discourse structure with sentence initial information, it provides evidence to support
that the first-mentioned entity might be the one around which the discourse is developing.
114

The first-mentioned advantage may also explain how speakers form sentences. Prior
work has noted that speakers place discourse-given or more accessible information earlier in the
sentence than new information (e.g., Bock & Irwin, 1980; Davidson, Zacks, & Ferreira, 2003;
Ferreira & Yoshita, 2003; Slevc, 2011). Therefore, subjects (that tend to come sentence initially,
at least in English) are often regarded as important and prominent. Researchers propose that
placing given information earlier may allow speakers to save more time to prepare less
accessible, new information later in the sentence. Studies also show that speakers tend to mis-
recall sentences having a structure where the more accessible information is placed earlier in the
sentence.
Next, the grammatical role of an entity is equally important in who is likely to be
mentioned in subsequent discourse. Subjects have been shown to be more accessible than objects
(e.g., Arnold 2001; Brennan 1995; Fletcher 1984; Kaiser 2010; Kaiser 2011; Stevenson,
Crawley, & Kleinman, 1994; Walker, Joshi, & Prince, 1998). However, it is worth noting that in
a language like English where the canonical word order is SVO, order of mention often coincides
with subjecthood. That is, the first-mentioned entity is often the subject of a sentence, both of
which are known to boost prominence. Therefore, it is the languages with flexible word orders,
such as Finnish, Korean, and Japanese, that have allowed researchers to dissociate the effect of
order of mention and subjecthood on prominence and to study the interaction between the two
aspects (e.g., Ferreira & Yoshita, 2003; Kaiser & Trueswell, 2008).
In addition to the order in which entities are mentioned and the syntactic roles they play,
verb semantics has been shown to influence who language users pay attention to (e.g., Arnold,
2001; Au, 1986; Garvey, Caramazza, & Yates, 1975; Hartshorne, 2014; McDonald &
MacWhinney, 1995; Stevenson et al., 1994). Evidence supporting the role of verb semantics
115

comes from implicit causality verbs (e.g., Au, 1986; Brown & Fish, 1983; Fukumura & van
Gompel, 2010; Garvey & Caramazza, 1974; Garvey, Caramazza, & Yates, 1975; Guerry,
Gimenes, Caplan, & Rigalleau, 2006; Hartshorne, Sudo, & Uruwashi, 2013). Impress, admire,
and praise are examples of implicit causality verbs. The causality of an event denoted by a verb
can be attributed either to the subject or the object. For instance, the verb impress attributes the
causality to the subject. If John impresses Mary, the cause of Mary’s being impressed has to be
the result of some quality of John. In contrast, the verb admire attributes causality to the object.
If John admires Mary, it has to be something about Mary that causes John’s admiration. Studies
show that causal arguments are more salient than other arguments (e.g., Fukumura & van
Gompel, 2010; Hartshorne, 2014; Hartshorne, Sudo, & Uruwashi, 2013; McKoon & Ratcliff,
1998). For instance, when a sentence fragment is John impressed Mary because…,
comprehenders tend to continue the sentence by talking about John. However, when the sentence
fragment is John admired Mary because…, people’s focus tends to be on Mary, and their
sentence completion is generally about Mary.
Moreover, other grammatical devices, such as a focus structure, can increase the
accessibility of an entity (e.g., Almor, 1999; Almor & Eimas, 2008; Birch, Albrecht, & Myers,
2000; Kaiser, 2010; Klin, Weingartner, Guzmán, & Levine, 2004; McKoon, Ward, Ratcliff, &
Sproat, 1993). Birch, Albrecht, and Myers (2000) showed that readers made more references to
entities in story continuation if they had been mentioned in a grammatical focus structure as in
(33a) than in a neutral structure in (33b). Participants also used pronouns to refer to the target
entity more frequently when it was focused (e.g., the mayor in (33a)) than when it was not.

116

(33) A sample story continuation item used in Birch, Albrecht, and Myers (2000)
Betty was covering local and state races. She was at City Hall for a press
conference. Minutes into the conference, an argument erupted.
a. It was [the mayor]focus who refused to answer a reporter’s question.
[Continuation _______...]
b. The mayor refused to answer a reporter’s question.
[Continuation ___________...]
(Birch, Albrecht, & Myers, 2000, p. 289)

So far, I have summarized some of the well-observed factors that influence concepts’
prominence in discourse and memory, and the likelihood of mention in subsequent discourse. A
general consensus among the studies outlined above is that when individuals decide to talk about
a certain entity, it generally reflects that the entity has some level of boosted prominence.
Therefore, the prominence or the accessibility of an entity impacts speaker’s choice of: (i) who to
talk about (or not to talk about), (ii) where in the sentence to place the information (e.g., mention
it early in the sentence or later), and (iii) which linguistic form to use to refer to the information.
As prior work has illustrated, the prominence of an entity does not stem from a single
source. It could be a result of the information structure of the discourse (e.g., topic continuity). It
could also be contributed to the interpretation of the verb. In summary, the factors affecting
prominence include: (i) where in the sentence or discourse the entity appears (e.g., the first-
mentioned advantage), (ii) whether the entity is a grammatical subject or an object, (iii) verb
semantics, and (iv) whether the entity is given and already introduced information (i.e., topic) or
is newly introduced information (i.e., focus).
117

It should be noted that a great majority of the studies testing the accessibility of discourse
entities focused primarily on how comprehenders interpret pronouns (i.e., pronoun resolution or
reference resolution) (e.g., Arnold 2001; Arnold, Eisenband, Brown-Schmidt, & Trueswell,
2000; Järvikivi, van Gompel, Hyönä, & Bertram, 2005; Kaiser 2010; Kaiser, 2011). Many of
these studies have shown that being referred to by reduced referring expressions, such as
pronouns, means that the referent is highly accessible. However, it should be taken with caution
that being referred to by a simple or reduced referring expression does not necessarily mean that
the referent is the most accessible or salient in the discourse among other available (both
mentioned and unmentioned) referents (e.g., Fukumura & van Gompel, 2010; Kaiser, 2010;
Rohde & Kehler, 2014). For example, Kaiser (2010) found that though subject pronouns
frequently refer to the preceding subject, the preceding subject is not always the most talked
about referent when participants were given an open-ended sentence completion task (i.e.,
without a pronoun prompt). She found that in such a case, comprehenders’ attention is drawn to
the focus structure or topichood of the entities in the discourse. For example, the referent whom
participants talked about the most in her experiment was the alternative to a contrastively-
focused subject, which also was the topic of the discourse.
Therefore, there are important issues to keep in mind when interpreting the prominence
of entities. One issue concerns that what or who is most accessible at a given point in discourse
cannot be determined by a single surface measurement, such as the linguistic form, used to refer
to an entity. Another issue is that prominence is influenced by a variety of different factors
including the information structure of the discourse, the kinds of words used to express ideas, the
visible availability of entities in the surrounding, or the memory limitations of the interlocutors,
to name a few. Given these issues, Experiment 3 was conducted with the aim of furthering our
118

understanding of what it means for entities to be salient in discourse and how it may relate to the
accessibility in memory. In the previous Experiment 2 where two protagonists were introduced
in a story, individuals’ memory was better for the entity that had been mentioned later in the
story (i.e., the more recently talked-about referent) than the entity that had been introduced
earlier in the story. The current experiment addresses the question of whether the entity that is
more easily retrieved from memory would be the one that individuals are willing to talk about
more frequently.

4.3 Motivations for Experiment 3
Research investigating the relationship between memory- and discourse-level representations
have focused on how frequently entities are mentioned and how well or how fast comprehenders
remember the entities. Notably however, the majority of those studies have investigated these
aspects using short narratives with one or two sentences (e.g., Birch, Albrecht, & Myers, 2000).
However, as discussed previously, entities’ prominence changes throughout discourse because
new referents and topics are introduced and integrated into the existing information. Therefore, it
can be beneficial to examine the interaction between memory and discourse in a broader context.
In the current experiment, I examined the issue using a longer discourse context by introducing
entities in four-sentence stories, asked participants to write three continuation sentences, and
tested how likely different entities are to be mentioned in the three continuation sentences. This
design provided a rich seven-sentence discourse context (see Chiriacescu & von Heusinger, 2010
for a similar design). Further, employing the same target stories as the probe-word recognition
experiment (Experiment 2) also made it possible to compare the memory results with the
discourse continuation patterns.
119

Some of the factors influencing the memory retrieval of entities have been discussed in
the preceding Chapter 3. Combined with the discussion of discourse-level representations, Table
19 summarizes the elements that have been shown to influence both discourse- and memory-
level representations.

Table 19. Factors influencing discourse and memory prominence
discourse prominence memory prominence
order of mention first-mentioned advantage
most recently-mentioned advantage
first-mentioned advantage
grammatical role subject advantage
subject advantage
direct object advantage over indirect obj.
focus structure
focused entity is talked
about more frequently
focused entity is recognized faster and
more accurately
Note: The same table is presented in Table 13 in Chapter 3.

Of those aspects outlined in Table 19, Experiment 3 focuses on order of mention. In the
experiment, there were two characters introduced in each test story (using occupation nouns),
both as subjects. They differed in the order of mention; one was introduced and commented on
before the other. The order of mention might be in effect if the story continuation results show a
discrepancy as to whom readers talk about more frequently between the two subjects.
The results of Experiment 2 indicated that comprehenders remember the character
mentioned more recently better than the character introduced earlier. This recency effect
suggests that the character mentioned recently was more activated in readers’ short-term
120

memory. If strong activation in memory results in strong activation in discourse, readers would
tend to continue the stories by talking about the recently-mentioned character. On the other hand,
if discourse representation is not a direct reflection of what is active in memory, story
continuations might not necessarily be about the recent entity. The research questions that the
current Experiment 3 explored are presented in (34).

(34) Experiment 3 research questions
I. Would participants talk about SUBJECT2 (i.e., the recently-mentioned
character) more frequently than SUBJECT1 (i.e., the first-mentioned character)
because it is more prominent in the memory representation?
II. Would participants talk about SUBJECT1 more frequently than SUBJECT2
because it was the first-mentioned character in the story that laid out the
foundation of the discourse?

4.4 Methodology
4.4.1 Participants
Thirty adult native speakers of Korean, recruited in Korea, participated in this experiment. They
received $10 for their participation.

4.4.2 Materials
Sixteen stories were selected for this experiment from the 24 target stories used in Experiment 2.
There were 16 filler stories selected from Experiment 2 as well. Each target story contained two
characters (occupation nouns such as teacher and reporter). Every target story contained four
121

sentences. The first sentence began with a subject noun phrase containing two occupation nouns
conjoined with ‘and’ (e.g., 기술자와 정치가; a mechanic and a politician). The second sentence laid
out a context where the two characters were engaged in an activity (e.g., visiting a friend’s
housewarming party). The third sentence commented on the first-mentioned character, and the
fourth sentence commented on the second-mentioned character.
All the stories were written in two different case marker conditions. The two characters
were introduced with the nominative marker in one condition, and with the contrastive topic
marker in the other condition. A sample story is presented in (35). All stories were written in
Korean. Participants’ task was to create three sentences of their own that naturally continue each
story.

(35) A sample story used in Experiment 3
기술자와 정치가가 지인의 집들이에 초대 받아 갔어요. 두 사람은 잊지 않고 집들이 선물도
챙겨왔죠. 실용적인 선물이 최고라며 기술자 가/ 는 두루마리 화장지를 잔뜩 사왔어요. 새 집은
미관이 중요하다며 정치가 가/ 는 큰 화분을 사왔어요.
[English translation]
A mechanic and a politician went to their friend’s housewarming party. They
didn’t forget to bring presents. [The mechanic-NOM/TOP]SUBJECT1 brought lots
of toilet papers, saying practical presents are the best. [The politician-
NOM/TOP]SUBJECT2 brought a large pot of plants, saying ambience is important
in a new home.

Despite its effect shown in Experiment 1, case marking did not influence how well
comprehenders remembered the story characters in Experiment 2. It also had no significant effect
122

on how readers continued the stories in the current Experiment 3. Therefore, I will not discuss
the case marking factor in the remainder of the chapter. Since case marking had no significant
effect on story continuation, the reported statistical results are based on all 16 target stories,
collapsing the NOM and the TOP conditions.

4.4.3 Procedure
Participants read the stories presented on a laptop computer using Microsoft PowerPoint. Each
PowerPoint slide contained a four-sentence story, and a blank box was provided below the story
where participants typed in three continuation sentences. Every participant saw all 16 target
items, half in the nominative condition and half in the contrastive topic condition. The targets
and fillers were distributed among two lists. To control for potential order effects, reverse lists
were created for each of the two lists for a total of four lists. Participants completed the task in
the presence of the experimenter. It took approximately an hour for participants to complete the
task.

4.5 Predictions
The focus of the analysis was which protagonist(s) participants talk about in their continuation
sentences. I analyzed whom the subject of each of the three continuation sentences referred to.
There were three main possible referents considered: SUBJECT1, SUBJECT2, and BOTH
SUBJECTS. SUBJECT1 is the character that was introduced and commented on first in the
story. SUBJECT2 is the second-mentioned character. BOTH SUBJECTS refers to SUBJECT1
and SUBJECT2 together as a single subject noun (e.g., in the form of they, the two, the teacher
123

and the reporter, etc.). The discussion of the existing literature led to the formulation of the
following two hypotheses:

Hypothesis 1: The entity that is most accessible in memory would also be the one
comprehenders talk about more frequently.
If discourse representation is a direct reflection of increased accessibility in memory,
comprehenders would continue the stories by talking about the entity that was remembered better
in Experiment 2. In this case, the recently-mentioned SUBJECT2 would appear more frequently
than SUBJECT1, who was mentioned earlier in the story. Reference to the better-remembered
SUBJECT2 might be greatest in Continuation Sentence 1 because the preceding story fragment
ends with SUBJECT2.

Hypothesis 2: Discourse representation and memory representation are not parallel.
If discourse prominence is multifaceted and does not necessarily rely on greater activation in
memory alone, participants’ continuation patterns would follow the general observations in the
literature and also resemble the results of Experiment 1 (where participants were mostly likely to
comment on the first-mentioned global discourse subject, at least in the NOM-condition). If this
prediction is borne out, the first-mentioned SUBJECT1 might appear more frequently in
continuation sentences than the second-mentioned SUBJECT2. This prediction is based on the
existing work showing that, in general, the first-mentioned entity often functions as the topic of a
discourse (e.g., Chang, 1980; Gernsbacher & Hargreaves, 1988; Gernsbacher, Hargreaves, &
Beeman, 1989; McDonald & MacWhinney, 1995; Yongming & Yao, 1995). Discourse
representation might not directly mirror memory representation because of the different levels of
124

cognitive processing involved in the recall (Experiment 2) and the continuation tasks
(Experiment 3). While Experiment 2 would have required comprehenders to retrieve the
information available in short-term memory, Experiment 3 might have required different
comprehension processes where they needed to actively build a coherent discourse.

4.6 Results of Experiment 3
I analyzed whom or what the subject of participants’ continuation sentences referred to. Results
are reported separately by the three continuation sentences. Two sample stories and continuation
sentences from participants are provided in Table 20 and Table 21. I use square brackets for
expository purposes to show the subjects that I analyzed.

Table 20. Sample story and continuations 1, Experiment 3
판사와 농부가 주일 아침 예배에 참석했습니다. 교회에서 맡은 일이 많은 두 사람은 주일 아침이 바쁩니다. 장로를
맡고 있는 판사가 대표기도를 준비합니다. 구역장을 하고 있는 농부가 새 신도 안내를 담당합니다.
Continuation sentence 1
서로서로 주어진 역할을 잘 수행하자 오늘도 완벽하게 예배가 끝났습니다.
Continuation sentence 2
판사는 농부에게 점심에 일이 없으시다면 같이 점심을 먹을 수있겠냐고
물어봅니다.
Continuation sentence 3
농부는 판사에게 환영이라며 어디에서 만나는 것이 좋을지 물어봅니다.
[English translation]
A judge and a farmer attended a Sunday church service. The two are busy with their own
responsibilities at church. [The judge] SUBJECT1, who is an elder, prepares a prayer. [The farmer]
SUBJECT2 who is a small group leader is in charge of welcoming new visitors.
Continuation sentence 1
Because [they] have completed their duties well, today’s service
ends perfectly once again.
125

Continuation sentence 2
[The judge] invites the farmer for lunch in case he didn’t have any
other plan.
Continuation sentence 3
[The farmer] welcomes the judge’s invitation and asks him where
they should meet.

Table 21. Sample story and continuations 2, Experiment 3
디자이너와 심리학자가 뉴욕 행 비행기에 올랐다. 중요한 모임에 참석하는 두 사람은 피곤하지만 아직 준비해야
할 것이 많이 있다. 바이어와 미팅이 있는 디자이너는 아직 완성 못한 작업을 잡고 고심한다. 학회에서 발표가
있는 심리학자는 파워포인트 발표를 준비한다.
Continuation sentence 1
디자이너는 종이 한장을 들고 어찌할까 계속 고민을 한다.
Continuation sentence 2
그런데 심리학자의 키보드 소리가 너무 거슬린다.
Continuation sentence 3
그는 고민하다가 심리학자에게 조금만 조용히 해줄 수 없냐고 부탁한다.
[English translation]
A fashion designer and a psychologist boarded a New York-bound flight. The two are visiting
the city for important meetings, and they still have quite a bit of work to do. [The designer]
SUBJECT1 who has a meeting with a buyer struggles with an unfinished design. [The
psychologists] SUBJECT2 who has a talk at a conference is preparing a PowerPoint presentation.
Continuation sentence 1
[The designer] continues to struggle with the piece of paper with
his/her design on it.
Continuation sentence 2 But [the typing sound] from the psychologist is too distracting.
Continuation sentence 3
After hesitating a while, [he] asks the psychologist if (he/she)
could keep it quiet.

126

4.6.1 Coding
Continuation sentences were coded in terms of who the grammatical subjects were. Subjects
were chosen for analysis because they have been suggested to play a central role in sentence
meaning (e.g., Arnold, 1998; 2001; 2008; Gernsbacher & Hargreaves, 1988; Gernsbacher,
Hargreaves, & Beeman, 1989; McCloskey, 1997; Reinhart, 1982). For example, there is a strong
tendency to place the topic in subject position (e.g., Reinhart, 1982). Also, subjects tend to be
held in the focus of the reader’s attention (e.g., Garrod & Sanford, 1988). Therefore, when
participants place a certain entity in subject position, it is considered to the referent most likely in
the center of comprehenders’ developing discourse structure.
Statistical analyses were conducted for the following four referents: SUBJECT1, the
character introduced earlier in each story (e.g., fashion designer in Table 21), SUBJECT2, who
was introduced second (e.g., psychologist in Table 21), BOTH SUBJECTS (i.e., SUBJECT1 and
SUBJECT2 as a whole (e.g., they, the two)), and the CONTEXT SUBJECT (i.e., a subject
relating to the story context, e.g., a flight attendant in Table 21). I conducted one-sample t-tests
to evaluate whether the likelihood of these referents to be mentioned (in subject position) is
greater or less than the chance level set at .25
19
. One-Sample t-tests were chosen because the
proportion of one type of subject referent depended on the proportion of the other referents. That
is, if a participant mentioned SUBJECT1 as the subject of Continuation Sentence 1, he/she could
not have made SUBJECT2 the sentence subject, for example. Therefore, comparing the
proportions across the different referent types would not have been appropriate due to their being
depend on each other.

19
The chance level was determined as .25 because there were four main subject continuations considered:
SUBJECT1, SUBJECT2, BOTH SUBJECTS (e.g., they), and CONTEXT SUBJECT.

127

4.6.2 No effects of TOP vs. NOM
I do not report the results of the story continuation task separately for the NOM and the TOP
conditions because – just as in Experiment 2 (the probe recognition task) – the likelihood-of-
mention did not differ for NOM-marked vs. TOP-marked subjects (ps > .149).
Table 22 presents sample continuation sentences, each containing reference to
SUBJECT1, SUBJECT2, BOTH SUBJECTS, and CONTEXT SUBJECT. I have added square
brackets around the subjects that I analyzed.

Table 22. Continuation sentences with Subject1, Subject 2, Both Subjects, and Context Subject
story presented
A mechanic and a politician went to their friend’s
housewarming party. They didn’t forget to bring presents. [The
mechanic-NOM/TOP]SUBJECT1 brought lots toilet papers,
saying practical presents are the best. [The politician-
NOM/TOP]SUBJECT2 brought a large pot of plants, saying
ambience is important in a new home.
SUBJECT1 continuation
기술자는 두루마리 휴지를 한쪽 구석에 세워 두었습니다.
[The mechanic] put the toilet papers in a corner.
SUBJECT2 continuation
정치가는 기술자가 사 온 화장지를 보고 성의 없는 선물이라고 말했습니다.
[The politician] looked at the toilet papers that the mechanic
brought and said they were so meaningless.
BOTH SUBJECTS
continuation
둘은 지인의 집으로 들어가며 내 선물을 더 좋아할 거라며 다툼을 벌었어요.
[The two] argued over whose present would be more
appreciated as they made their way into the house.
128

CONTEXT SUBJECT
continuation
집 주인은 두 사람을 반갑게 맞이했습니다.
[The host] welcomed the two visitors.

4.6.3 Subjects of Continuation Sentence 1
This section reports the percentage of the references made to the four subjects in the first
continuation sentence (see Table 23 and Figure 8).

Table 23. Percentage of the grammatical subjects in Continuation Sentence 1, Experiment 3
grammatical subject percentage (%)
SUBJECT1 35
SUBJECT2 15
BOTH SUBJECTS 24
CONTEXT SUBJECT 12
Other 14
Total 100

129

Figure 8. Grammatical subjects of Continuation Sentence 1, Experiment 3

When participants’ first continuation sentence was analyzed, noticeable patterns were observed
(Figure 8). The first-mentioned SUBJECT1 was more likely to be talked about in Continuation
Sentence 1 compared to the recently-mentioned SUBJECT2. SUBJECT1 was referred to 35% of
the time while SUBJECT2 was referred to less than half of the SUBJECT1 mention at 15%. This
is an interesting pattern given that SUBJECT2 was the most recently mentioned character in the
story fragment. This is also the character that comprehenders remembered better. Furthermore, in
their first continuation sentence, participants were also more inclined to talk about BOTH
SUBJECTS as a whole (e.g., they, the two) than to continue the story by talking about the
recently-mentioned second character. The subject of Continuation Sentence 1 referred to BOTH
SUBJECTS 24% of the time. BOTH SUBJECTS was the second most frequently occurring
subject in Continuation Sentence 1, following SUBJECT1.
35%
15%
24%
12%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Subj1 Subj2 Subj1&Subj2 Context subj.
Subject of Continuation Sentence 1
130

One-Sample T-tests were conducted to determine whether the proportion of each of the
four referents was different from the chance level of .25. The proportion of SUBJECT1
continuations (35%) was significantly greater than expected by chance; t1(29) = 3.516, p1 = .001,
t2(15) = 2.491, p2 = .025. In contrast, the proportion of SUBJECT2 continuations (15%) was
significantly lower than the chance level; t1(29) = -5.108, p1 < .0001, t2(15) = -5.003, p2 < .0001.
The proportion of BOTH SUBJECTS referents (24%) was not significantly different from
chance; t1(29) = -.474, p1 = .639, t2(15) = -.310, p2 = .761. Furthermore, the likelihood of the
CONTEXT SUBJECT mention (12%) was significantly lower than chance: t1(29) = -8.811, p1 <
.0001, t2(15) = -3.210, p2 = .006.
The results of Continuation Sentence 1 revealed that SUBJECT1 continuations (the
earlier-mentioned subject) were greater than expected by chance. SUBJECT2 continuations (the
recently-mentioned) and CONTEXT SUBJECT continuations were significantly fewer than
chance. BOTH SUBJECTS continuations did not differ from chance.

4.6.4 Subjects of Continuation Sentence 2
In this section, I report whom participants talked about in Continuation Sentence 2 (see Table 24
and Figure 9).

131

Table 24. Percentage of the grammatical subjects in Continuation Sentence 2, Experiment 3
grammatical subject percentage (%)
SUBJECT1 27
SUBJECT2 33
BOTH SUBJECTS 18
CONTEXT SUBJECT 10
Other 12
Total 100

Figure 9. Grammatical subjects of Continuation Sentence 2, Experiment 3

As shown in Figure 9, in Continuation Sentence 2, both SUBJECT1 mention and SUBJECT2
mention occupied substantial proportions. Numerically speaking, SUBJECT2 mention (33%)
outweighed SUBJECT1 mention (27%) this time. This is different from Continuation Sentence 1
27%
33%
18%
10%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Subj1 Subj2 Subj1&Subj2 Context subj.
Subject of Continuation Sentence 2
132

where SUBJECT1 was mentioned most frequently at 35% of the time. Furthermore, reference to
BOTH SUBJECTS decreased from 24% in Continuation Sentence 1 to 18% in Continuation
Sentence 2. This appears to be due to the increased reference to SUBJECT2.
One-Sample t-tests were conducted to test whether the four referents’ likelihood of
mention was different from the chance level of .25. SUBJECT1 continuations (27%) did not
differ significantly from the chance level; t1(29) = 1.046, p1 = .304, t2(15) = .752, p2 = .464.
However in Continuation Sentence 2, there were significantly more continuations with
SUBJECT2 (33%) than expected by chance; t1(29) = 2.937, p1 = .006, t2(15) = 2.125, p2 = .051.
The proportion of BOTH SUBJECTS continuations (18%) was significantly lower than chance;
t1(29) = -2.921, p1 = .007, t2(15) = -3.339, p2 = .004. CONTEXT SUBJECT continuations (10%)
were also fewer than expected by chance; t1(29) = -12.217, p1 < .0001, t2(15) = -5.020, p2 <
.0001.
In Continuation Sentence 2, the recently-mentioned SUBJECT2 appeared most
frequently at 33% of the time. This occurrence was greater than expected by chance. Unlike in
Continuation Sentence 1 where SUBJECT1 continuations were most frequent at 35%, the
likelihood of SUBJECT1 mention did not differ from chance in Continuation Sentence 2.

4.6.5 Subjects of Continuation Sentence 3
This section reports the reference patterns in Continuation Sentence 3 as shown in Table 25 and
Figure 10.

133

Table 25. Percentage of the grammatical subjects in Continuation Sentence 3, Experiment 3
grammatical subject percentage (%)
SUBJECT1 24
SUBJECT2 28
BOTH SUBJECTS 29
CONTEXT SUBJECT 9
Other 10
Total 100

Figure 10. Grammatical subjects of Continuation Sentence 3, Experiment 3

In Continuation Sentence 3, reference to SUBJECT1, SUBJECT2, and BOTH SUBJECTS seems
to be fairly well distributed at 24%, 28%, and 29%, respectively. One notable change in this
continuation sentence from the previous ones is the increase in BOTH SUBJECTS continuation.
24%
28%
29%
9%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Subj1 Subj2 Subj1&Subj2 Context subj.
Subject of Continuation Sentence 3
134

In Continuation Sentence 1 and 2, BOTH SUBJECTS continuations were 12% and 10%,
respectively. However, BOTH SUBJECTS as a single noun phrase was the most frequent
referent made in Continuation Sentence 3.
I conducted One-sample t-tests to evaluate whether the references made to SUBJECT1,
SUBJECT2, BOTH SUBJECTS, and CONTEXT SUBJECT were greater than the chance level
of .25. The results indicated that SUBJECT1, SUBJECT2, and BOTH SUBJECTS continuations
(24%, 28%, and 29%, respectively) did not differ significantly from chance; ps >.2. This
suggests that they were equally frequently mentioned. However, the likelihood of CONTEXT
SUBJECT continuations (9%) was significantly lower than the chance level in Continuation
Sentence 3; t1(29) = -11.604, p1 < .0001, t2(15) = -5.019, p2 < .0001.

4.6.6 Subjects of Continuation Sentences 1, 2, and 3 combined
In this section, I report the subject referents of all three continuation sentences combined. Table
26 and Figure 11 show the average percentage of the four main subjects to appear across the
three continuation sentences.
Table 26. Percentage of the grammatical subjects in all three continuation sentences, Exp. 3
grammatical subject percentage (%)
SUBJECT1 29
SUBJECT2 25
BOTH SUBJECTS 23
CONTEXT SUBJECT 10
Other 13
Total 100
135

Figure 11. Grammatical subjects of all three continuation sentences combined, Experiment 3

When all three continuation sentences were combined, SUBJECT1 occurred most frequently at
29%, followed by SUBJECT2 at 25%. BOTH SUBJECTS continuation came close to
SUBJECT2 continuation at 23%. CONTEXT SUBJECT continuation remained at its usual
proportion at 10%.
I performed One-sample t-tests to test whether the proportion of the four subject
continuations was different from the chance level of .25. When all three continuation sentences
were considered, SUBJECT1 continuations (29%) were significantly greater than expected by
chance (significant in subject analysis only); t1(29) = 2.121, p1 = .043, t2(15) = 1.556, p2 = .141.
The proportion of SUBJECT2 continuations (25%) did not differ significantly from the chance
level; t1(29) = .045, p1 = .964, t2(15) = .028, p2 = .978. BOTH SUBJECTS continuations did not
differ from the chance level either; t1(29) = -.759, p1 = .454, t2(15) = -.549, p2 = .591. CONTEXT
29%
25%
23%
10%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Subj1 Subj2 Subj1&Subj2 Context subj.
Subject of all three continuation sentences
136

SUBJECT continuations were significantly fewer than expected by chance; t1(29) = -13.515, p1 <
.0001, t2(15) = -4.506, p2 < .0001.
Table 27 below shows the percentage of the four subject referents across the three
continuation sentences. The highlighted columns represent the occurrences that are greater than
expected by chance (set at .25).

Table 27. Percentage of the grammatical subjects across three continuation sentences, Exp. 3
grammatical subject sentence 1 sentence 2 sentence 3 all 3 sentences
SUBJECT1 35 27 24 29
SUBJECT2 15 33 28 25
BOTH SUBJECTS 24 18 29 23
CONTEXT SUBJECT 12 10 9 10
Note: Highlighted columns represent the occurrences greater than the chance level of .25.

In Continuation Sentence 1, SUBJECT1 continuations were most frequent (35%), and the
occurrence was greater than expected by chance. In Continuation Sentence 2, SUBJECT2
continuations were most frequent (33%), also more than expected by chance. In Continuation
Sentence 3, no subject referent was greater than the chance level (though CONTEXT SUBJECT
continuations were significantly fewer than expected by chance). When all three continuation
sentences were considered, SUBJECT1 continuations were most frequent overall (29%), and that
was significantly higher than expected by chance.

137

4.7 Discussion of Experiment 3
Although there have been a number of studies that tested the relationship between discourse and
memory representations of entities, how the two levels interact in a broader discourse context has
not been studied widely. Experiment 3 was carried out to provide a better understanding of the
interaction between the two domains using a larger discourse context. My earlier probe-word
recognition experiment (Experiment 2 in Chapter 3) revealed that participants’ memory for a
recently-mentioned character was better than their memory for the first-mentioned character
even 15 seconds after the presentation of the story. The current Experiment 3 tested whether the
better-recalled character would also the referent that readers tend to talk about frequently in story
continuation. I formulated two hypotheses based on the observations in the literature. The first
hypothesis was built upon the observation that the character that comprehenders talk about more
frequently is also the one that they remember better (e.g., Birch, Albrecht, & Myers, 2000). In
this hypothesis, the recently-mentioned character was expected to be talked about more
frequently because it is the character that readers remembered better. On the other hand, the
second hypothesis stated that the first-mentioned would be commented on more frequently than
the second character because the first-mentioned entities have been shown to be highly
prominent and regarded as the topic of the story.
The present story continuation results suggest that the first-mentioned character might be
the most prominent entity in the discourse. Even though Experiment 2 showed that memory for
the second-mentioned entity was better, in Experiment 3, we see that there were more frequent
references made to the first-mentioned entity. This was particularly the case in the first
continuation sentence where the reference to the first-mentioned character was significantly
greater than chance. Thus, we observe an interesting asymmetry between Experiment 2 and
138

Experiment 3: memory retrieval was better for the recently-mentioned entity, but discourse
continuation was more about the first-mentioned entity.
The frequent reference made to the first-mentioned subject supports the prior evidence
suggesting that entities mentioned discourse initially tend to be considered as discourse topics
(e.g., Carpenter & Just, 1977; Kieras, 1978; 1981; Reinhart, 1982). The first-mentioned character
in each target story might have been considered as more “important” or “prominent” in story
development because it was mentioned first in the [Noun-and-Noun] subject NP (e.g., The
teacher and the reporter went to…) and was commented on first in the third sentence of the story
(the second character was commented on in the fourth sentence). When all three continuation
sentences were considered as a whole, the first-mentioned character was still talked about more
often than the second-mentioned. These results support the second hypothesis for the study,
which predicted higher accessibility and thus more frequent mention of the first-mentioned
character.
Let us further compare the results from the memory experiment (Experiment 2, Chapter
3) and the current story continuation experiment (Experiment 3). The probe-word recognition
task used in Experiment 2 showed the prominence of the recently-mentioned subject in memory,
while the story continuation task revealed the prominence of the first-mentioned subject. These
findings suggest that memory- and discourse-level representations do not always go together.
The way in which a story develops may not rely on memory representations alone. Rather,
continuing a story would require readers to connect events and maintain a certain level of
coherence throughout the text. The text coherence and inferential processes involved in reading
and subsequent writing may have required other comprehension strategies in addition to the
memory-level representations. Those strategies may have required readers to monitor back and
139

forth what has been talked about already and what needs to be talked about for the sake of the
text coherence. This process can be explained by the notion of standards of coherence proposed
by Van den Broek, Lorch, Linderholm, and Gustafson (2001). This reading model posits that
readers attempt to maintain certain levels of coherence as they process through a text and engage
in different inferential processes. Studies note that standards of coherence is influenced by the
types of goals readers are trying to achieve (e.g., reading for study or reading for entertainment;
e.g., Linderholm & van den Broek, 2002; Van den Broek et al., 2001). This distinction leads to
my next discussion.
Even though both Experiments 2 and 3 used the same target stories, the levels of mental
processes necessary for the reading goals might have been different. The production task (i.e.,
story continuation, Experiment 3) might have involved macro-level processes that require
readers to focus on understanding the entire story as a coherent text. Discourse-level
representations might demand ‘higher level’ knowledge that incorporates discourse structure,
plans, and goals into production (e.g., Fletcher, 1986). On the other hand, the probe-word
recognition task might have required micro-level processes, which emphasize understanding the
text as separate sentences. This might be why the recognition of the characters showed a recency
effect, which is found in serial free recall of items (e.g., Craik & Levy, 1976; Kintsch, 1977).
When the reading goal was to accurately recognize the characters, the more recently-introduced
entity may have been the main content in the short-term memory buffer at the time of testing.
However, this does not seem to have translated into discourse continuation. Reading for
discourse continuation may have required more global strategies where participants
comprehended the currently unfolding sentence with the anticipation of specific upcoming
140

events to be described. In contrast, reading for memory retrieval in the recognition task might
have been more local, focusing each sentence as an isolate sentence.
The comparison between entities’ memory and discourse representations has revealed
two main findings (based on Experiment 2 in Chapter 3 and Experiment 3 in this chapter). First,
we see that in a discourse context, the entity that is most activated in short-term memory might
not necessarily be the one that is most prominent in global story continuation. This outcome
contradicts what has been widely observed in the literature where the likelihood of mention often
indicates more accurate memory retrieval (e.g., Birch et al., 2000; Klin et al., 2004). However,
increased saliency in short-term memory alone cannot boost a character’s prominence in a board
discourse where readers continually adjust the processing strategies and the level of processing
resources devoted to text interpretation (e.g., Klin et al., 2004).
The combined results also suggest that the specific goals individuals have in
comprehending a text can alter the types and the levels of reading strategies and cognitive
processes to achieve the goal. Story continuation seems to have involved a top-down processing
where the coherence of a text is evaluated and achieved by making deep understanding and
connection between the sentences. In contrast, the character recognition task seems to have
required a bottom-up processing where recognition and retention of the characters in memory
could have been achieved by understanding the text as isolate sentences without making a
coherent discourse out of them.
The discrepancy found between the two experiments invites future experiments that can
help us deepen our understanding of entities’ discourse and memory level representations. For
example, one possibility for a follow-up study would be to change the way in which the two case
markers are used in the story. Instead of using the same case marker for both characters, one
141

character can be marked with the nominative marker and the other character can be marked with
the contrastive topic marker. This design is depicted in (36).

(36) Design of future follow-up experiment 1

condition 1 condition 2

SUBJECT1-NOM
SUBJECT2-TOP

SUBJECT1-TOP
SUBJECT2-NOM

Using the design in which the two case markers are used together in one condition would allow
us to test the effect of case marking more clearly. Introducing two characters in a story and
commenting on one of them in sentence 3 invites an inference that the other character would be
mentioned next. If the first mentioned character is marked with the topic marker, as in condition
2 of (36), that expectation might increase, which can increase the saliency of the second
character that is to be mentioned.
Another future experiment could involve a design where a fifth sentence is added in the
story. The fifth sentence can be about another referent as shown in (37).
(37) Design of future follow-up experiment 2
Sentence 3: comments on SUBJECT1
Sentence 4: comments on SUBJECT2
Sentence 5: comments on another referent
142

Having an additional referent might modulate the recency effect for SUBJECT2 (in a probe-
word recognition paradigm) as it is no longer the most recent entity. This design might reduce
the serial effect that we have observed. Adding a new referent can provide an opportunity to
understand possible effects of: (i) case marking, (ii) order of mention, (iii) discourse structure,
and (iv) the interaction among these factors.

143

Chapter 5
Experiment 4: The effect of verb semantics on sentence processing and memory recall

5.1 Introduction
In this chapter, I shift the focus of my investigation from entity information to event information
denoted by verbs. Experiments 1 through 3 investigated the discourse-level and the memory-
level representations of entities or individuals indicated by nouns. In addition to nominal
information, verbal or event information is also an important aspect to study in sentence
processing and memory representation since nouns and verbs are two fundamental building
blocks of a message (e.g., Gordon, Hendrick, & Johnson, 2001; McRae, Hare, Elman, & Ferretti,
2005; Wittenberg et al., 2014). In particular, verb meaning is crucial in laying out the sentence
structure (e.g., McRae, Hare, Elman, & Ferretti, 2005; for recent work, see Čech, Mačutek, &
Žabokrtský, 2011). For example, a sentence containing the verb buy involves two entities: the
individual who buys and the item being bought. Because the verb determines the structure of the
event, studying how the verb meaning contributes to sentence comprehension and recall would
help us understand the different semantic aspects that impact the mental representations
individuals construct in their minds.
In Experiment 4, I tested how the semantic properties of verbs influence the way in which
comprehenders process sentences and how well they remember the content information. The
semantic information that Experiment 4 focused on was the frequency of motion repetition.
Native Korean speakers performed a sensicality judgment task where they indicated whether
given sentences were sensical or meaningful expressions in Korean. In the sensicality judgment
paradigm, participants were asked to determine the acceptability of the sentence in terms of its
144

meaning, more specifically whether the sentence conveyed a meaningful expression in Korean.
Participants’ performance was measured in terms of accuracy and reaction times. Since all the
sentences were acceptable in grammar, participants’ judgments were based on the meaning of the
sentence.
The experimental sentences contained action verbs, such as sneezing, knocking on a door,
clapping, and bouncing a ball. The target action verbs differed in terms of their motion repetition
frequency. Some of the verbs were low repetition verbs as the actions denoted by the verbs
generally repeat once or twice in a row in the real world. These verbs included sneezing,
coughing, clearing one’s throat, knocking on a door, etc. The other group of verbs was high
repetition verbs referring to actions that typically repeat several times in a row. These verbs
included hiccupping, clapping, waving, bouncing a ball, etc. The complete list of target verbs
used in Experiment 4 is presented in Appendix B. The frequency of motion repetition is one of
the main semantic information that an action verb conveys, as many of the actions we perform or
observe in the real world often occur in succession (e.g., think about ringing a door bell or
coughing (low frequency) vs. bouncing a ball or hiccupping (high frequency)). In addition to
other physical attributes of action verbs (e.g., the speed or the duration of a motion) that have
been shown to influence the way individuals process sentences, repetition frequency may also
influence how people process sentences.
Previous studies have demonstrated that verb semantics influences the way
comprehenders perform linguistic tasks. For example, language users’ reading or speaking rates
change depending on the meaning of the verb (e.g., Fecica & O’Neill; Yao & Scheepers, 2011).
People tend to read a text more slowly if the action described by the verb denotes a slow motion
(e.g., walking), however they read a text faster when the verb denotes a fast motion (e.g., being
145

driven). In a similar way, individuals’ speaking rates slow down when the verb describes a slow
motion, and speaking rates increase when the verb denotes a fast motion. Researchers suggest
that the reason behind the change in comprehenders’ behaviors might be due to the mental
representations of the events denoted by the verb. Research has suggested that comprehending
linguistic descriptions of actions involves mental simulations of those actions (e.g., Barsalou,
1999; Bergen & Chang, 2005; Fischer & Zwaan, 2008; Kintsch, 1988; Sanford & Garrod, 1981;
Van Dijk & Kintsch 1983; Zwaan & Radvansky, 1998; Zwaan & Taylor, 2006). This line of
research is referred to as simulation-based theories of language comprehension or more broadly,
embodied cognition. In the following section, I introduce some of the studies that demonstrate
the effects of verb semantics on both linguistic and non-linguistic behaviors.
To add to the literature concerning the effect of lexical semantics, Experiment 4 aimed to
test whether a verb’s inherent repetition frequency would change the way individuals process
sentences and whether the semantic information conveyed by the verb would interact with other
aspects of the sentence, such as number adverbs. It tested: (i) whether the frequency of a motion
denoted by the verb (the difference between coughing (low frequency) vs. hiccupping (high
frequency)) would influence sentence processing, (ii) whether the way individuals process
sentences would be influenced by the specific number expression describing the frequency of the
action (e.g., coughed twice (low frequency) vs. coughed six times (high frequency)), and (iii) the
interaction between a verb’s frequency semantics and an explicit number expression noted by an
adverb.
As noted earlier, an action verb’s semantics (e.g., the duration or the speed of a motion)
has been shown to influence how fast or slow individuals read or speak, or how long they look at
visual scenes (e.g., Fecica & O’Neill; Yao & Scheepers, 2011). This observation is attributed to
146

the different representations or images comprehenders are creating in their minds, specific to the
verb meanings. A related question is then: would the same kind of effect be observed when
individuals encounter the same event described by other expressions, such as an adverb of
manner? For example, verbs can denote the speed of a motion inherently as in the difference
between walking (slow motion) vs. running (fast motion). However, the same slow or fast event
can also be described with different adverbs that denote speed as in walked fast vs. walk slowly
or ran fast vs. ran slowly. Would reading the expressions, such as ran slowly and ran fast (same
verb, different manner), alter the mental representations and would consequently change the
reader’s behavior? To the best of my knowledge, this type of research has not been conducted.
Experiment 4 tests this not in the domain of speed but in the domain of repetition
frequency. It tests: (i) the effect of verb inherent frequency semantics on sentence processing, (ii)
the effect of explicit adverbs expressing frequency (e.g., once, twice for low frequency vs. five
times, six times for high frequency), and (iii) the interaction between verb’s intrinsic frequency
semantics and explicit adverbs denoting repetition frequency.
In the following section before turning to my experiment, I review the previous studies
that demonstrated the effect of verb semantics on language processing and other non-linguistic
behaviors.

5.2 Prior work on the effects of verb semantics
Understanding language requires a number of different processes. One key process is retrieving
the meanings of individual words from comprehenders’ mental lexicon (e.g., Federmeier &
Kutas, 1999; Ferreira & Yoshita, 2003; Pinker, 1994; Rappaport, Levin, & Laughren, 1993;
Slevc, 2011; Ullman, 2001; Ullman et al., 1997). Comprehenders also need to understand the
147

structural relationships between phrases to compute the intended meaning using syntactic
representations (e.g., Chomsky, 1995; Ferreira & Yoshita, 2003; Gibson, 1998; Rappaport,
Levin, & Laughren, 1993; Slevc, 2011; Ullman, 2001; Ullman et al., 1997). These meaning
retrieval and structure computing processes are rapid and automatic (e.g., Fodor, 1983;
Friederici, 2002; Haung et al., 2013; Kamide, Scheepers, & Altmann, 2003; MacDonald,
Pearlmutter, & Seidenberg, 1994; Tanenhaus, Spivey-Knowlton, Eberhard, & Sedivy, 1995).
The focus of Experiment 4 is on the first component, in particular, the lexical semantics of
words. Words are stored with their inherent semantic properties. For example, the meaning of the
verb kick contains the movement of legs, and the verb pick conveys the movement of hands.
Words’ semantic properties have a strong influence on how we process language. Semantic
information allows comprehenders to build certain expectations about the information conveyed
and the grammatical structures to be used (e.g., Federmeier & Kutas, 1999; Rappaport, Levin, &
Laughren, 1993). For example, when encountering the expression picked up by hand,
comprehenders’ expectations of what’s being picked up narrow down to the objects that are light
enough to be picked up using hands (e.g., Federmeier & Kutas, 1999; Hare, McRae, & Elman,
2003; McRae, Hare, Elman, & Ferretti, 2005).
Research has shown that comprehenders’ linguistic behaviors are sensitive to words’
lexical semantics (e.g., Fecica & O’Neill, 2010; Friederici, 2002; Lau, Phillips, & Poeppel, 2008;
Lindsay, Scheepers, & Kamide, 2013; Martin & Chao, 2001; Matlock, 2004; Wittenberg et al.,
2014 among others). For example, whether a noun refers to a concrete object or an abstract
concept changes the reaction time to the word. Furthermore, neurological studies suggest that
one of the ways in which lexical items are organized and retrieved is based on meaning. Patients
148

with damage to the left prefrontal cortex show difficulty retrieving words of certain kinds, such
as names of objects belonging to a specific semantic category (e.g., Martin & Chao, 2001).
In addition to the semantic properties of nouns, language comprehension is also sensitive
to verb semantics. For example, Fecica and O’Neill (2010) showed that four to six year-old
children’s story comprehension is influenced by the speed of a movement conveyed in sentences
(e.g., walking vs. being driven). When children listened to a story where the protagonist was
described as walking to a place, their processing of subsequently related sentences was slower
relative to a story where the character was described as to be driven to the place. More
specifically, if the character was walking to a destination (which takes longer than being driven),
children took a longer time to process sentences describing the subsequent events that happened
along the way (e.g., children playing baseball, birds flying around someone’s yard) than when
the character was driven to the destination. Fecica and O’Neill concluded that children, like
adults, build mental representations of the described actions based on their world knowledge, and
their mental representations influence how they process subsequent information.
A similar study investigating the effect of action speed was conducted with adult
comprehenders (Lindsay, Scheepers, & Kamide, 2013). Lindsay et al. investigated how the speed
of an action conveyed in verbs (e.g., running vs. staggering) would influence what
comprehenders pay attention to in a visual scene and how much time they spend scanning the
scene. Their eye tracking experiment demonstrated that participants spent more time looking
along the path when they listened to sentences containing slow verbs (e.g., The hiker will mope
along the trail to the cottage) compared to sentences containing fast verbs (e.g., The hiker will
sprint along the trail to the cottage). The participants also looked at the goal earlier when
listening to sentences containing fast verbs.
149

Matlock (2004) further demonstrated that people’s behaviors are affected by the semantic
information denoting physical distance. Participants in her study spent more time reading a
sentence that depicts a longer physical distance, such as ‘The road runs through the valley,’
compared to a sentence that describes a shorter distance, such as ‘The cord runs along the wall.’
Matlock suggests that the discrepancy between the reading times might be due to the differences
in the mental representations of the depictions. A road running along a valley generally stretches
longer in physical distance compared to a cord running along the wall. Therefore, words’ lexical
semantics is suggested to influence language comprehension and individuals’ behaviors, more
generally.
Further data regarding the effects of the speed of motion in language processing comes
from Yao and Scheepers (2011). The researchers used short stories to investigate whether
comprehenders’ reading rates differ depending on how fast the protagonist was implied to be
speaking. They examined participants’ reading rates in different contextual conditions. In the
‘fast’ condition, readers were given stories in which the protagonist was implied to be speaking
at a fast rate (e.g., a boy who is nervous and shaking). In the ‘slow’ condition, participants were
given stories where the character was implied to be speaking at a slow rate (e.g., an old man on
his deathbed). The results showed a significant context effect on reading rates. That is,
participants read direct quotations faster when the protagonist was implied to be speaking fast
(e.g., nervous boy) than when the protagonist was implied to be speaking slowly (e.g., sick old
man). These studies demonstrate that words’ semantic information that denotes events, as well as
linguistic contexts, influence the behaviors of language comprehenders.
Although the current research does not include neurological testing, it is worth
highlighting the major neuropsycholinguistic findings regarding how lexical semantics
150

influences language processing. As discussed above, studies have demonstrated that words’
semantic properties influence linguistic behaviors, such as visual scanning of a scene, reading
speed, and comprehension rates. These behaviors are taken as the reflection of what is going on
in the comprehenders’ brain. Words’ semantic information has been shown to modulate brain
activations. For example, neuropsychological evidence suggests that nouns and verbs may have
different mental storage and grammatical processing mechanisms (e.g., Pinker, 1999; Ullman,
2001). Brain imaging studies have detected a greater neural activation when individuals process
verbs than nouns (e.g., Davis, Meunier, & Marslen-Wilson, 2004; Kable, Lease-Spellmeyer, &
Chatterjee, 2002; Perani et al., 1999). Some scholars suggest that processing verbs may place
greater demand in the neural system than nouns because they require more complex grammatical
operations (e.g., Shapiro & Caramazza, 2003).
Dissociation in the brain activation has been shown not only across different word
categories such as nouns and verbs, but it has also been observed within a word category.
Researchers including Moody and Gennari (2010) showed that brain activation levels differ
depending on the degree of physical effort implied by nouns. Their fMRI studies found higher
activation in the sensory-motor regions of the brain when people read sentences describing
greater physical effort (e.g., ‘The delivery man is pushing the piano’) than when they read
sentences describing relatively less physical effort (e.g., ‘The delivery man is pushing the
chair’).
Furthermore, an EEG study conducted by Pulvermüller, Härle, and Hummel (2001)
suggests that even within the category of action verbs, the orientation of actions may influence
comprehenders’ behaviors and brain reactions. The researchers found that individuals respond to
visually presented words faster if the words refer to actions performed by face (e.g., talking) than
151

other body parts, such as legs (e.g., walking). Their EEG data further showed that the ‘face-
verbs’ and the ‘leg-verbs’ elicited different patterns in neurophysiological responses.
When individuals process action-related sentences, the brain regions that are responsible
for such actions have been shown to be activated. For example, the areas in the brain dedicated
to hand motion gets activated when processing a sentence referring to a hand action, such as
‘You are giving Andy a slice of pizza’. On the other hand, processing a sentence that describes a
foot action, such as ‘You are kicking the ball across the field,’ activates the brain area
responsible for foot motion (e.g., Tettamanti et al., 2005; Aziz-Zadeh, Wilson, Rizzolatti, &
Iacoboni, 2006).
As a whole, these findings have been taken as evidence that verbal descriptions trigger
mental simulations of actions (see Fischer & Zwaan, 2008 for a review). When considering verb
semantics, it can be about a variety of different properties including: (i) action-related verbs vs.
abstract verbs (e.g., to scrub vs. to trouble, Innocenti et al., 2014); (ii) bodily orientation (e.g.,
face-related verbs vs. nonface-related verbs, to talk vs. to walk, e.g., Pulvermüller et al., 2001);
(iii) speed of motion (e.g., to walk vs. to run, e.g., Fecica & O’Neill, 2010; Lindsay et al., 2013),
or (iv) a physical distance (e.g., Matlock, 2004).
Bearing in mind these action-related semantics, Experiment 4 examined a novel semantic
property that, to the best of my knowledge, has not been investigated from the perspective of the
effect of lexical semantics on sentence processing. This property has to do with the frequency of
motion repetition. I use the term ‘frequency of repetition’ to refer to the number of times an
action generally repeats in a row. In our physical world, we observe actions take place all the
time. We knock on a door, ring a door bell, clap hands, or bounce a ball, or watch others perform
these actions. These actions repeat in succession, but some actions repeat more frequently than
152

others. For example, clapping is prototypically thought to involve several ‘hand claps’, whereas
the act of ringing a door bell does not involve as many ringing hand motions. Similarly, while
people generally sneeze once or twice in a row, they hiccup multiple times in a row.
Though there are ample studies that reveal both behavioral and neurological
consequences of different aspects of verb semantics, no research to date seems to have
investigated how the frequency of action repetition could influence language processing. To fill
this gap in the literature, I tested whether comprehenders’ linguistic performance is affected by a
verb’s repetition frequency. I used a norming study to identify verbs of high or low inherent
repetition frequencies (i.e., verbs referring to actions that generally repeat many times in a row
vs. actions that generally repeat only once or twice in a row). The effect of action frequency on
sentence processing was measured using a sensicality judgment task, as well as a verb
recognition task. Native Korean speakers read sentences written in Korean and indicated whether
the sentences were meaningful/sensical expressions in Korean. The sensicality judgment
paradigm is widely used in psycholinguistic research to explore how comprehenders interpret
linguistic information on sentence levels, as well as pragmatic levels (e.g., Bambini et al., 2013;
Gagné & Shoben, 1997; Klein & Murphy, 2001; Liedtke & Schulze, 2013; Merlo & Stevenson,
2002).
In addition to a verb’s inherent repetition frequency, I also tested whether a specific
frequency description expressed by a number adverb would influence sensicality judgment and
verb recognition. I tested both low-repetition adverbs (e.g., once/twice) and high-repetition
adverbs (e.g., five times/six times). Just as a verb’s semantics influences sentence processing, a
specific repetition frequency denoted by an adverb might also impact the way sentences are
processed. It might be the case that sentence processing is easier when an action is described to
153

be repeated once or twice (as opposed to five times or six times) regardless of whether the verb
denotes an inherently low repetition action (e.g., coughing, sneezing) or an inherently high
repetition action (e.g., hiccupping, clapping).
The number adverb can be analyzed also in terms of its matching or mismatching status
with the verb’s expected frequency. I manipulated whether the adverb matched or mismatched
the verb’s inherent frequency. Half of the target sentences included cases where the adverb
matched the action’s inherent tendency to repeat (e.g., low frequency + low adverb = coughed
twice; high frequency + high adverb = clapped six times). The other half of the targets included
sentences where the number adverb did not matched the action’s intrinsic repetition frequency
(e.g., low frequency + high adverb = coughed six times, high frequency + low adverb = clapped
twice). It was predicted that an adverb’s explicit frequency might interact with a verb’s inherent
frequency. More specifically, the sentences where the verb and the adverb mismatched in
repetition frequency would be judged less sensical or meaningful than the sentences where the
two constituents matched because the mismatching cases would be perceived as unusual. More
detailed predictions are presented in Section 5.5 below. The research questions that Experiment 4
investigated are presented in (38).

(38) Experiment 4 research questions
I. Would the repetition frequency of an action intrinsically denoted in the verb
semantics influence participants’ sensicality judgment of the sentence and the
recognition of the verb?

154

II. Would the explicit number adverb that describes the repetition frequency of the
motion in the sentence influence participants’ sensicality judgment and verb
recognition?
III. Would the potential effect of a verb’s inherent repetition frequency modulate
the effect of number adverb on the sensicality judgment and verb recognition or
vice versa?

5.3 Method
5.3.1 Norming study
Prior to the main experiment, I conducted a norming study to see how many times Korean
speakers typically think certain actions take place in a row. Forty native Korean speakers
participated in the norming study. It was a web-based survey via the Qualtrics online survey
software (Qualtrics, 2010). Participants were given 36 action verbs and were asked to indicate (in
an open-ended task) how many times each of the actions generally takes place in a row. For
example, they were asked: ‘On average, how many times would one blow his/her nose in a row?’
The verbs and the instructions in the norming study were all in Korean. Participants were
carefully instructed to provide a single natural number that best indicates the frequency of the
actions to repeat in a row. They were told to avoid expressions commonly used for quantity, such
as ‘‘many’, ‘a lot of’, or ‘a few’.
Based on the norming results, I identified a set of 24 verbs that I divided into two
frequency conditions: (i) an Expected High Frequency condition and (ii) an Expected Low
Frequency condition. The Expected High Frequency condition included verbs denoting actions
that generally repeat many times in a row, such as 박수를 치다, 딸국질을 하다. 공을 뜅기다, 손을 흔들다
155

(English: clapping, hiccupping, bouncing a ball, and waving hands). The Expected Low
Frequency condition included verbs referring to actions that typically repeat once or twice in a
row: 재채기를 하다, 기침을 하다, 초인종을 누르다, 노크를 하다 (English: sneezing, coughing, ringing a
door bell, and knocking on a door). Each condition included 12 different target actions. The
average number of repetitions per action (how many times an action takes place in a row) in the
Expected High Frequency condition was 10.7 times (ranging from 3.6 times to 36 times, SD =
6.66)
20
. The average number of repetitions per action in the Expected Low Frequency condition
was 2.2 (ranging from 1.5 times to 2.9 times, SD = .8). The average normed repetition frequency
between the High and the Low Frequency conditions differed significantly; t(22) = -2.844, p
= .009.

5.3.2 Main experiment
5.3.2.1 Participants
A new group of 32 native Korean speakers recruited in Korea participated in the main
experiment. They were paid $10 for their participation.

5.3.2.2 Materials
Participants took part in three tasks in this order: (i) a sensicality judgment task, (ii) a distractor
math task, and (iii) a verb recognition task. The accuracy data from the sensicality judgment task
(i.e., how accurately individuals judged the sentences to be meaningful) can inform us about the

20
There were three action verbs in the High Expected Frequency group whose average frequencies were higher than
the other actions in the same group. These verbs were 윗몸일 으키기를 하다, 팔굽혀펴 기를 하다, 물 장구를 치다 doing sit-ups,
doing push-ups, and paddling one’s feet in water. Their average frequencies were 36, 19, and 19, respectively. When
the average repetition of the High Expected Frequency verbs was computed excluding these three actions (since the
difference was large), the average was 5.6 times, and the standard deviation was 2.27. Thus, even without these
verbs, the High Frequency verbs are higher than the Low Frequency verbs.
156

effect that the verb and the number adverb have in the overall sentence meaning. A math task
that lasted 3.5 minutes was added to distract readers’ attention from the meaning of the sentences
they read so as to test the level of memory retention of the event described in the sentence after
some temporal decay had taken place during the task. The verb recognition task tested whether
the degree to which an event is retained in memory would change depending on the verb’s
inherent frequency, the specific number expression, or the interaction of the two.

5.3.2.3 Design
The sensicality judgment task included 24 target and 60 filler sentences in Korean. Each target
sentence included a proper name as the grammatical subject, an action verb (in the past tense),
and a number adverb (e.g., 규진이가 기침을 두 번 했다. Kyujin twice coughed ‘Kyujin coughed
twice’). See Table 29 for more examples. Verbs’ inherent (or expected) frequency (i.e., whether a
verb typically has a high or a low frequency of repetition) served as one of two independent
variables in the experiment. Half of the 24 target sentences contained actions that generally
repeat multiple times in row, and the other half contained actions that usually repeat once or
twice in a row. The number of Korean characters in the target sentences ranged from 12.8 to
14.5. The average number of characters in the Expected High Frequency condition was 14. The
average number of characters in the Expected Low Frequency condition was 13. It should be
stressed that the action repetitions depicted in the sentences have a strong bias to be interpreted
as multiple repetitions in a single event, and not multiple separate events (based on native
speaker intuitions). The complete list of all the target verb phrases used in the experiment is
presented in Appendix B.
157

In addition to verbs’ expected frequency, I also manipulated the number adverb in the
sentence. The number adverb had two levels: Low Frequency and High Frequency. The number
adverbs used in the Adverb-Low condition were ‘once’ and ‘twice’. The number adverbs in the
Adverb-High condition were ‘five times’ and ‘six times’. For examples, see Table 28. Two
number adverbs were used in each adverb condition to avoid repeating the same adverb in all the
target sentences.
Each target verb appeared with a number adverb that either matched its expected
frequency (e.g., coughing once/twice; clapping five times/six times) or mismatched the expected
frequency (e.g., coughing five times/six times; clapping once/twice). Having the two factors of
verbs’ inherent frequency and number adverb made it possible to test whether participants’
sensicality judgment and verb recognition are influenced by verb inherent semantics, explicit
number expressions, or the interaction of both. This two by two design yielded the four
conditions presented in Table 28.

Table 28. Four experimental conditions of Experiment 4
condition
action’s inherent or
expected frequency
number
adverb
sample action
Expected High-
Adverb High
high high bouncing a ball five/six times
Expected High-
Adverb Low
high low bouncing a ball once/twice
Expected Low-
Adverb High
low high knocking on a door five/six times
158

Expected Low-
Adverb Low
low low knocking on a door once/twice

In the verb-adverb frequency match condition, the target verbs were used with a number adverb
that matched their expected frequency. For example, sneezing was used with the adverb once or
twice in the match condition (e.g., 규진이가 기침을 한 번/ 두 번 했다 Kyujin coughed once/twice). In
this item, the expected repetition frequency and the adverb matched in number because sneezing
generally repeats once or twice in a row. In the same way, the verb clapping was used with five
times or six times to match its expected frequency (e.g., 규진이가 손벽을 다섯 번/ 여섯 번 쳤다 Kyujin
clapped five/six times).
In the verb-adverb frequency mismatch condition, the verbs were used with an adverb
that mismatched their expected frequency. For example, sneezing, which generally repeats once
or twice in a row, was used with five/six times (e.g., Kyujin coughed five/six times), and clapping
was used with once/twice (e.g., Kyujin clapped once/twice). The sensicality judgment and the
verb recognition performance was expected to differ between the verb-adverb matching
condition and the verb-adverb mismatching condition. Table 29 shows the four experimental
conditions with sample sentences.

159

Table 29. Sample targets in each condition, Experiment 4
verb-adverb match condition sample target sentence
verb-adverb match Expected High-Adverb High
아람이가 박수를 다섯 번 쳤다.
Aaram clapped five times.
verb-adverb mismatch Expected High-Adverb Low
아람이가 박수를 한 번 쳤다.
Aaram clapped once.
verb-adverb mismatch Expected Low-Adverb High
규진이가 기침을 여섯 번 했다.
Kyujin coughed six times.
verb-adverb match Expected Low-Adverb Low
규진이가 기침을 두 번 했다.
Kyujin coughed twice.

5.3.2.4 Procedure
5.3.2.4.1 Sensicality judgment task
The sensicality judgment task was conducted using Paradigm software (Perception Research
Systems). Each participant was seated in front of a laptop computer in a quiet room with the
experimenter present. The sentences were presented on the computer screen one at a time. Fillers
included both sensical (e.g., 희선이는 이번 달에 영화를 세 편 봤다 ‘Heeseon watched three movies this
month’), as well as nonsensical sentences (e.g., 준 수가 김치찌개에 건물을 탔다 ‘Joonsu put a building
in a kimchi stew’). Participants were instructed to read the sentences as they would naturally and
to decide at the end of each sentence whether it was a sensical or a meaningful expression in
Korean. The sensicality judgment paradigm is widely used in linguistic research (e.g., Bambini
et al.; Klein & Murphy, 2001), and is suggested to be appropriate when testing comprehenders’
meaning comprehension as it requires participants “not only to access the linguistic items but
160

also to elaborate and interpret their meanings at the level of detail that would distinguish
different senses” (Bambini et al., p. 3).
Participants pressed the ‘J’ key on the keyboard for ‘yes’ (makes sense) and ‘F’ key for
‘no’ (does not make sense). Paradigm recorded the yes-no choices and the time
21
taken to make
the choices. All target sentences were sensical sentences, requiring a ‘yes’ response. The number
of expected sensical and nonsensical sentences over the course of the full experiment (targets and
fillers) were balanced. All participants completed this task within 20 minutes.

5.3.2.4.2 Distractor task
After finishing all the targets and fillers in the sensicality judgment task, participants completed a
calculation distractor task. They were given 14 arithmetic problems to solve (e.g., (9 + 4)(8 - 3)(7
+ 2) =?). The time allotted for the distracter task was three minutes and thirty seconds.

5.3.2.4.3 Recognition task
After completing the math distractor task for 3.5 minutes, participants completed a probe-word
recognition task. On the same computer, participants saw 48 verbs, 24 from all target sentences
and 24 from filler sentences. The verbs were presented in an infinitival form without the past
tense conjugation (ex. 39) and without the number adverb. Participants were asked to indicate
whether they had seen the expression in the previous sensicality judgment task. They were told
that the verb tenses in the previous task and the recognition task would be different. They were
instructed to press the ‘J’ key for ‘yes’ even though the exact forms might be different and ‘F’

21
The reaction time results of the sensicality judgment task and the verb recognition task are not reported here
because the number of Korean characters in the High and the Low Expected Frequency conditions were different,
which might have affected the RT results. In any case, the statistical analyses of the RT data showed no significant
differences between the conditions (all ps > .3).
161

key for ‘no’. All 24 target verbs required a ‘Yes’ response. Fourteen out the 24 filler verbs
required a ‘No’ response. Therefore, 34 probe verbs required a ‘yes’ response (24 from the
targets and 10 from fillers), and 14 probe verbs required a ‘no’ response. Sample target verbs
used in the probe-word recognition task are presented in (39).

(39) Sample target verbs in the probe-word recognition task
a. 노크를 하다. knock on the door
b. 기침을 하다. cough
c. 딸국질을 하다. hiccup
d. 박수를 치다. clap

5.4 Predictions
It was predicted that participants’ sensicality judgment and recognition of verbs would differ
depending on the inherent frequency semantics that the verb expresses. It was also predicted that
the matching and mismatching status of the number adverb would modulate participants’
performance. However, it was not clear beforehand which direction these effects might go in. In
other words, would more repetitions (either based on the verb’s intrinsic meaning or the number
adverb) help or hinder comprehension (as measured by sensicality judgments) and recognition?
Below, I first sketch out two competing hypotheses concerning the effect of verbs’ inherent
frequency on sensicality judgment and verb recognition. Since there is somewhat limited existing
literature directly related to the effect of verbs’ inherent repetition frequency, the first two
hypotheses primarily predict the effect of verbs’ frequency semantics (without discussing the
effects of the adverb or the matching/mismatching number adverb). I then present a third
162

hypothesis regarding how the matching/mismatching number adverb might interact with verbs’
inherent frequency and influence sentence comprehension and verb recognition.

I. Hypothesis 1: Fewer inherent repetitions conveyed in the verb might result in easier
processing of the sentence and better recognition of the verb.
This hypothesis predicts that participants would be more accurate judging the sensicality
of the sentences containing actions that generally repeat once or twice in a row (e.g., coughing,
sneezing). In contrast, participants’ judgments might be less accurate for the sentences containing
actions that repeat multiple times in a row (e.g., hiccupping, clapping). This hypothesis was
formulated based on the effects of physical attributes on language processing. Studies have
demonstrated that physical aspects of a motion including distance and speed can alter
comprehenders’ performance (see section 5.2 for a review on existing research). Just as physical
distance or the speed of a motion depicted in language influences comprehension, the inherent
repetition frequency of a motion might also affect sentence judgment and later recognition of the
event. Reading sentences denoting actions that generally repeat once or twice might require less
of processing resources than reading sentences describing actions that typically repeat multiple
times in a row. This prediction was based on the following: (i) I assumed that fewer repetitions
make for a simpler mental representation. (ii) Prior work shows that simpler mental
representations (e.g., pushing a piano vs. pushing a chair) require less cognitive resources (e.g.,
Moody & Gennari, 2010; Shapiro & Caramazza, 2003). (iii) Building on this, I expected less of
processing load for low repetition verbs than high repetition verbs, which might result in better
sensicality judgment and better recognition of the verb.
163

II. Hypothesis 2: More inherent repetitions might result in easier processing and better
recognition.
This hypothesis is the opposite of Hypothesis 1. It predicts that processing sentences
denoting highly repeated actions would yield better sensicality judgment and verb recognition
accuracy. When sentences are about actions of multiple repetitions, such as hiccupping or
clapping, processing those sentences might result in clearer mental representations than
processing sentences of low repetition actions, such as sneezing or coughing. Research suggests
that accessing the meaning of action verbs such as smile or kick requires comprehenders to
activate the cognitive representations responsible for performing or perceiving such actions (e.g.,
Bergen, Lau, Narayan, Stojanovic, & Wheeler, 2010). Particularly, processing action-related
sentences activates the areas in the brain responsible for performing such actions (e.g., Hauk,
Johnsrude, & Pulvermüller, 2004; Pulvermüller, Härle, & Hummel, 2001; Tettamanti et al.,
2005). If comprehenders mentally reenact the motions described in the sentences, the actions that
are repeated more frequently might leave a clearer mental representation of the action. This
might lead to better sensicality judgment of the sentence and more accurate recognition of the
verb. In contrast, the mental image of the actions that generally repeat only once or twice might
be simpler than the verbs denoting actions that repeat multiple times in a row.

III. Hypothesis 3: The matching or mismatching status of the number adverb in the sentence
would modulate sensicality judgment and verb recognition accuracy.
This hypothesis concerns the matching and mismatching status of the number adverb
present in the sentence. Scholars suggest that comprehenders’ response accuracy decreases when
they experience processing difficulty or when comprehension triggers a heavy processing load
164

(e.g., Boston, Hale, Kliegl, Patil, & Vasishth, 2008; Ferreira, Christianson, & Hollingworth,
2001; Frank, 2009; Levy, 2008; Mitchell et al., 2010). Therefore, processing sentences where the
verb’s inherent repetition frequency is matched with the appropriate number adverb (e.g.,
coughing twice, hiccupping six times) was expected to be easier than processing sentences where
the verb’s expected frequency clashes with the number adverb (e.g., coughing six times,
hiccupping once). The match condition was expected to lead to better sensicality judgment and
verb recognition than the mismatch condition. This prediction was based on existing
observations that syntactic or semantic mismatch requires higher processing resources and
causes reading difficulty. For example, comprehenders have been shown to spend more time
reading phrases where the verb and its object do not match as in ‘eat justice’ compared to ‘eat
pizza’ (e.g., Boston, Hale, Kliegl, Patil, & Vasishth, 2008; Frazier & Rayner, 1982).
Furthermore, individuals’ comprehension question accuracy decreases when a sentence contains
a temporary syntactic ambiguity (e.g., garden-path sentences; Ferreira, Christianson, &
Hollingworth, 2001). In a garden-path sentence, such as While Anna dressed the baby spit up on
the bed, readers initially misinterpret the baby as the object of dressed. The accuracy of the
comprehension question about the sentence declined when there was a mismatch between
reader’s expectations of the sentence meaning and the actual meaning. Having said that, we
might also observe in Experiment 4 a mismatch or a ‘surprisal’ effect when the number adverb
mismatches the verb’s expected repetition frequency. Table 30 summarizes the predictions
formulated for the experiment.

165

Table 30. Summary of predictions of Experiment 4
Hypothesis 1
fewer physical repetition  less processing load  better sensicality
judgment and verb recognition
Hypothesis 2
more physical repetition  more simulation & clearer mental
representation  better sensicality judgment and verb recognition
Hypothesis 3
number adverb matches verb’s inherent frequency  sentence
comprehension is easy, thus accuracy would be high
number adverb mismatches verb’s inherent frequency  sentence
comprehension is difficult, thus accuracy would be low

In sum, the hypotheses predict the possibility of a main effect associated with either verb’s
inherent frequency or number adverb, individually, and the possibility of a significant interaction
between the two factors. The results might reveal a main effect of verb’s inherent frequency if
the verb meaning itself has a strong effect on the mental presentation of the described event.
However, the results might show a main effect of number adverb if the specific frequency of the
event denoted by the adverb is the main drive for the mental representation. Furthermore, results
might indicate a significant interaction between the two factors. In other words, the degree to
which the high and the low inherent frequency of the verb influences the sensicality and verb
recognition results may change depending on the high and the low frequency adverbs.

5.5 Results
The results were divided into two measures: (i) sensicality judgment accuracy and (ii) verb
recognition accuracy. I measured whether participants correctly responded that the target
166

sentences were sensical expressions (response accuracy) and whether they correctly recognized
having seen the target verbs (recognition accuracy). Response accuracy is a common indicator of
comprehension difficulty (e.g., Mitchell et al., 2010). Low response accuracy is often an
indicator of greater processing load. On the other hand, high response accuracy suggests ease of
processing.

5.5.1 Sensicality judgment accuracy rate
In this section, I report how accurately participants judged the sentences to make sense. As Table
31 and Figure 12 show, the accuracy rates were high overall (above 90% in all conditions). The
high accuracy rates confirm that the target sentences were indeed sensical and sounded natural to
participants. Though participants were highly accurate with the sensicality task, still there was a
distinct pattern between the frequency conditions. The accuracy rates of the two conditions in the
Expected High Frequency were in the low 90’s (91% and 90%), and the accuracy rates of the
two conditions in the Expected Low Frequency were in the high 90’s (98% each). Interestingly,
the low vs. high number adverb (once/twice vs. five/six times) did not seem to influence the
accuracy patterns.
Table 31. Sentence sensicality judgment accuracy rates, Experiment 4
condition
verb-adverb frequency
match status
accuracy rate (%)
Expected High - Adverb High match 91
Expected High - Adverb Low mismatch 90
Expected Low - Adverb High mismatch 98
Expected Low - Adverb Low match 98
167

Figure 12. Sentence sensicality judgment accuracy rates, Experiment 4

Repeated measures ANOVAs were conducted on participants’ sensicality judgment accuracy to
test the effects of (i) verbs’ inherent frequency, (ii) number adverb, and (iii) their interaction. A
significant main effect of verbs’ inherent frequency was found on the sensicality judgment
accuracy; F1(1, 31) = 37.933, p1 < .0001, F2(1, 44) = 2.431, p2 = .126. This effect was significant
in the by-subjects analysis but not in the by-items analysis
22
.
On the whole, participants responded more accurately to sentences containing inherently
low frequency actions, which generally repeat once or twice in a row (e.g., coughing, sneezing),
than to sentences containing inherently high frequency actions that generally repeat multiple
times in a row (e.g., clapping, hiccupping). There was no significant main effect of the number

22
All the item analyses (F 2) reported in Experiment 4 were conducted using Two-Way Between Subjects ANOVA
tests (not Repeated Measures ANOVA) with verbs’ inherent frequency and number adverb as two factors. This was
done because the low and high frequency verbs are different lexical items (e.g., coughing vs. hiccupping), so it was
not possible to rotate a given verb through all four conditions. That is, half of the verbs (12 out of 24) were in the
‘Expected-High’ condition, and the other half were in the ‘Expected-Low’ condition based on how frequently they
generally occur in a row.
91%
90%
98%
98%
70%
75%
80%
85%
90%
95%
100%
Expected High-
Adv.High
Expected High-
Adv.Low
Expected Low-
Adv.Low
Expected Low-
Adv.High
Sensicality judgment accuracy
168

adverb on the sensicality judgment accuracy; F1(1, 31) = .104, p1 = .749, F2(1, 44) = .000, p2 =
.988. Participants’ sensicality judgments did not differ whether the number adverb used in the
sentence denoted high or low repetitions. There was no significant interaction between verbs’
inherent frequency and the number adverb on the sensicality judgment; F1(1, 31) = .008, p1 =
.929, F2(1, 44) = .006, p2 = .937.
In addition to testing the effect of verbs’ inherent frequency and number adverb,
additional analyses were conducted to test whether there is a difference between when the verb
and adverb matched in frequency and when they did not (i.e., testing the matching or
mismatching status of the number adverb with the verb’s expected frequency). The results
revealed no significant main effect of the matching/mismatching status of number adverb; F1(1,
31) = .008, p1 = .929, F2(1, 44) = . 006, p2 = .937. Participants’ sensicality judgments did not
differ whether the number adverb used in the sentence matched or mismatched the verb’s
expected frequency. There was no significant interaction between verbs’ inherent frequency and
the matching/mismatching status of number adverb; F1(1, 31) = .104, p1 = .749, F2(1, 44) = .000,
p2 = .988. To summarize, the sensicality judgment results revealed that verbs’ inherent frequency
is the only factor that had a significant effect on comprehenders’ judgment on the sentence
sensicality.

5.5.2 Verb recognition accuracy rate
This section presents how accurately participants recalled having seen the target verbs used in
the original sentences. Table 32 and Figure 13 show the percentage of the correct recognition of
the verbs.

169

Table 32. Verb recognition accuracy rate, Experiment 4
condition
verb-adverb frequency
match status
accuracy rate (%)
Expected High - Adverb High match 64
Expected High - Adverb Low mismatch 76
Expected Low - Adverb High mismatch 77
Expected Low - Adverb Low match 76

Figure 13. Verb recognition accuracy rate, Experiment 4

Verb recognition accuracy across the four conditions ranged from 64% to 77%. The lowest
accuracy rate of 64% was in the Expected High-Adverb High condition where the verb’s
inherent repetition frequency and the adverb were both high. This indicates that participants
found it difficult to correctly recognize the high repetition verbs, which they had initially seen
64%
76%
77%
76%
20%
30%
40%
50%
60%
70%
80%
Expected High-
Adv.High
Expected High-
Adv.Low
Expected Low-
Adv.Low
Expected Low-
Adv.High
Verb recognition accuracy
170

with a matching high adverb (e.g., 아람이가 박수를 다섯/ 여섯 번 쳤다. Aaram clapped five/six times).
The accuracy rates from the other three conditions were compatible with each other between
76% and 77%.
I conducted Repeated measures ANOVAs on participants’ verb recognition responses to
test the effects of (i) verbs’ inherent frequency, (ii) number adverb, and (iii) their interaction. As
in the sensicality judgment responses, there was a significant main effect of verbs’ inherent
frequency on verb recognition accuracy; F1(1, 31) = 4.399, p1 = .044, F2(1, 44) = 2.720, p2 =
.106. This effect was significant in the by-subjects analysis but not in the by-items analysis. The
main effect of verbs’ inherent frequency indicates that the verb recognition accuracy in the two
Expected Low Frequency conditions is higher than that of the Expected High Frequency
conditions. In other words, remembering verbs that refer to inherently low repetition actions
(e.g., sneezing, coughing) was easier than remembering verbs that refer to inherently high
repetition actions (e.g., hiccupping, clapping). Participants showed difficulty remembering
having seen the verbs that refer to actions that generally repeat multiple times in a row in the real
world.
Repeated measures ANOVAs also revealed a main effect of number adverb on verb
recognition; F1(1, 31) = 5.246, p1 = .029, F2(1, 44) = 2.907, p2 = .095. Participants recognized the
verbs more accurately when they appeared with a low frequency adverb (once/twice) than when
they appeared with a high frequency adverb (five/six times). The interaction between verbs’
inherent frequency and number adverb on verb recognition was not significant; F1(1, 31) =
2.828, p1 = .103, F2(1, 44) = 2.278, p2= .138.
Same as the sensicality judgment results, I conducted additional analyses to test whether
the matching/mismatching status of the adverb to the verb’s expected frequency affected verb
171

recognition accuracy. Repeated measures ANOVAs revealed no main effect of the
matching/mismatching status of number adverb on verb recall; F1(1, 31) = 2.828, p1 = .103, F2(1,
44) = 2.278, p2= .138. Participants’ recognition of the verbs did not differ whether the verb and
adverb matched in frequency or not. However, there was a significant interaction between verbs’
inherent frequency and the matching/mismatching status of the number adverb on verb
recognition; F1(1, 31) = 5.246, p1 = .029, F2(1, 44) = 2.907, p2= .095.
To further test where the significant interaction stems from, I conducted planned
comparisons comparing the Expected High-Adverb High condition to the Expected High-Adverb
Low condition (the left two bars in Figure 13). I also compared the Expected Low-Adverb Low
condition to the Expected Low-Adverb High condition (the right two bars in Figure 13). The
verb recognition accuracy in the Expected High-Adverb High condition was significantly lower
than that of the Expected High-Adverb Low condition (the comparison of the two left bars in
Figure 13); t1(31) = -2.623, p1 = .013; t2(11) = -2.733, p2 = .019. This indicates that when the
adverb was of low frequency (although the verb was a high repetition action), participants’
recognition of the verb improved
23
. The verb recognition results between the Expected Low-
Adverb Low and the Expected Low-Adverb High conditions (the right two bars in Figure 13) did
not differ significantly; t1(31) = .151, p1 = .881; t2(11) = .181, p2 = .860.
The significant interaction indicates that the matching/mismatching status of the adverb
mattered in the Expected High conditions (left two bars in Figure 13), but not in the Expected
Low conditions (right two bars in Figure 13). When the verb was of high expected frequency,
using an adverb that matched its expected frequency decreased the verb recognition accuracy.

23
In fact, the verb recognition accuracy of the Expected High-Adverb High condition (64%) was significantly lower
compared to all other three conditions (76%, 77%, and 76%); ps < .039.
172

5.6 Discussion of Experiment 4
Experiment 4 set out to explore the effect of semantic properties denoted by action verbs on off-
line sentence processing. In particular, I wanted to test whether the number of times an action
typically repeats in a row would influence the way in which comprehenders process sentences
and recall having seen the verb that described the action. I divided the critical action verbs in two
conditions depending on how frequently the motion denoted by the verb typically repeats in a
row in the real world (i.e., verb’s inherent frequency). The verb’s inherent frequency was
determined by a norming study where Korean speakers indicated how many times certain actions
generally repeat in a row. The Expected Low Frequency condition included verbs referring to
actions that usually take place once or twice in a row (e.g., 재채기를 하다, 기 침을 하다, 초인종을
누르다, sneezing, coughing, ringing a doorbell). The Expected High Frequency condition
included verbs denoting actions that generally take place multiple times is a row (e.g., 딸국질을
하다, 손벽을 치다, 공을 뜅기다, hiccupping, clapping, bouncing a ball).
In addition to verbs’ inherent frequency, another variable of number adverb was added in
the design. Number adverb refers to the specific frequency expression used in the sentence: (i) a
low frequency adverb (once or twice) and (ii) a high frequency adverb (five times or six times).
The target action verbs appeared in the sentence either with a number adverb that matched their
expected frequency (e.g., ringing a door bell once/twice; clapping five/six times) or that did not
match their expected frequency (e.g., ringing a door bell five/six times; clapping once/twice).
The adverb was included so as to test whether the effects of verbs’ repetition frequency on
sentence processing is modulated by the matching or mismatching status of the number
expression used in the sentence.
173

Three hypotheses were proposed. The first hypothesis predicted that participants would
be better at processing sentences with the verbs that denote inherently fewer movement
repetitions (e.g., sneezing over hiccupping). Creating a mental representation of an action that
repeats only once or twice was believed to be easier than building a mental image of an action
that repeats multiple times in a row. Therefore, the sensicality judgment of the sentence and the
recognition of the verb were expected to be better in the low repetition condition than in the high
repetition condition. The opposite hypothesis was also proposed. This hypothesis predicted that
more repetitions would lead to a clearer mental representation of the action. If this is the case, the
sensicality judgment and the recognition of the verb were expected to be more accurate in the
high repetition condition than in the low repetition condition. A prediction regarding the
matching or mismatching status of the number adverb was also formulated. It was predicted that
comprehending sentences where the number adverb did not match the verb’s inherent frequency
would tax more processing resources than comprehending sentences where the adverb and the
verb’s expected frequency matched. Therefore, higher accuracy was expected in the verb-adverb
match condition than in the mismatch condition.

5.6.1 Discussion of the sensicality judgment results
The results from the sentence sensicality judgment task revealed a significant effect of verbs’
inherent frequency. Participants accepted the sentences containing low frequency actions more
accurately compared to the sentences containing high frequency actions. Sensicality judgments
did not differ whether the number adverb matched or mismatched the verb’s inherent frequency.
There was no significant interaction between verbs’ inherent frequency and number adverb.
174

The sensicality judgment results support the first hypothesis, which predicted that
processing sentences about inherently low repetition actions would be easier than processing
sentences denoting inherently high repetition actions. The results suggest that building a mental
representation of an action that typically repeats once or twice (e.g., coughing, sneezing) might
be simpler than building a mental representation of an action that repeats multiple times in a row
(e.g., hiccupping, waving). Creating a mental image of an action that is repeated multiple times
(e.g., waving hands) might require more cognitive resources than an action that repeats once or
twice.
The significant effect of the repetition frequency on sensicality judgment is in line with
existing research, which demonstrates that sentence processing is influenced by semantics.
Studies have shown that individuals’ linguistic behaviors are influenced by the physical
properties denoted by the verb. Some of the known physical aspects of the verb that affect
comprehenders’ reactions include the speed of a motion (e.g., walk vs. run) and the duration of
an action (e.g., traveling 100 miles vs. traveling 20 miles). The current study adds to the
literature and shows that the frequency of motion repetition influences sentence comprehension.
Interestingly, the matching and mismatching status of the number adverb in the sentence
did not influence the sensicality judgment. Whether the number adverb matched or mismatched
the verb’s expected frequency did not matter as much as the verb’s inherent frequency itself. The
verb might have played a central role in individuals’ comprehension of the sentence, and the
number adverb could have been supplementary to the overall meaning. This may indicate the
importance of verb semantics in overall sentence comprehension. The verb itself might have
provided strong enough of a cue for building the mental representation of the action described.
This hypothesis can be tested in future studies that may examine the importance of verb
175

semantics in overall sentence meaning. A future experiment might ask participants to detect
changes between original and test sentences. If number expression is secondary in overall
sentence meaning, a change in the number expression between the original and the test sentences
might not be detected as well as a slight change in the verb.

5.6.2 Discussion of the verb recognition results
Verbs’ inherent frequency was also found to influence comprehenders’ recognition of the target
verbs. The verbs referring to low frequency actions were recognized better than the verbs
referring to high frequency actions. The verb recognition results also revealed a significant
interaction between verbs’ inherent frequency and the matching/mismatching status of the
number adverb. The significant interaction stems from the decreased recognition accuracy in the
Expected High - Adverb High condition where a high frequency verb was paired with a
matching high frequency adverb, e.g., 선우가 공을 다섯/ 여섯 번 뜅겼다; Seonwu bounced the ball
five/six times. Remembering having seen a verb that naturally repeats multiple times (e.g.,
bouncing a ball) might have been more difficult when it appeared with a large number than with
a small number. The small number expression might have alleviated the cognitive load necessary
for comprehending a sentence about an action that would otherwise repeat many times typically.
In contrast, comprehending a sentence where a high frequency verb appeared with a large
number expression might have increased the processing load. However, in the Low Expected
Frequency conditions, whether the number adverb matched the verb’s expected frequency or not
did not seem to matter as much. This again supports the claim that the verb itself might have
played a central role in creating a mental representation of the described event, and this effect
176

might have been greater in the Low Expected Frequency condition than the High Expected
Frequency condition.

5.6.3 Alternative interpretations
Let us now consider possible alternative interpretations for the observation that lower inherent
repetition frequency leads to better sensicality judgment and verb recognition. Could it be that
accuracy rates decreased for high frequency actions because such actions usually do not appear
with number adverbs (i.e., adverbs seem ‘odd’)? I believe this is unlikely for two main reasons.
First, the fact that there was no significant effect of number adverb speaks against the oddness
explanation. If participants found the sentences in which high frequency actions occurred with
number adverbs unnatural, then those sentences would have been judged non-sensical. However,
the sentences of high frequency actions were still judged sensical 90% of the time. Second, the
literature on sentence processing and memory speaks also against this oddness explanation.
Comprehenders tend to remember expressions that are noticeable. If participants found those
sentences in which high frequency actions were described with specific number adverbs, this
unusual expression might have stood out, and the high frequency actions should have been
recognized better than the low frequency verbs.
In order to better understand the role of motion repetition frequency on language
processing, future studies may include the following investigations. Studies using self-paced
reading or eye-tracking-during-reading might help us clarify the current results. If
comprehenders find it unnatural to use certain verbs with number adverbs, then we might
observe slowdown in reading time when encountering those expressions. Additionally, a follow-
up experiment might also investigate sentence processing with verbs of high or low repetition
177

frequency without using number adverbs. Eliminating the number adverbs all together can zero
in on the verb semantics and may help validate the findings of the current study. If the verb
semantics has a strong effect, the same inherent repetition frequency effect should be found.
Additionally, an off-line rating experiment can be added where participants are asked to indicate
how natural the target sentences sound on a Likert-type scale. This task can help us see whether
using number adverbs with certain verbs would seem unnatural.
To conclude, Experiment 4 sheds light on how sentence comprehension and recall of
linguistic information can be influenced by a verb’s semantics, particularly an action’s inherent
bias to repeat certain number of times. Since the current investigation was the first to test the
effect of action repetition, it invites future research using different paradigms and languages to
broaden our understanding of lexical semantics on sentence processing and memory retention.

178

Chapter 6
General discussion and conclusion

6.1 Summary of the findings and implications
This dissertation aims to deepen our understanding of how the processing and memory retention
of linguistic information can be influenced by sentence forms and meanings. I have reported four
experiments that investigated entity information denoted by nouns and event information
denoted by verbs. The first three experiments focused on entity information, and tested whether
the way in which story characters are introduced into discourse (with a default nominative
marker or a topic marker) would impact their prominence in discourse and memory. The fourth
experiment dealt with event information, and investigated to what extent verb semantics can
influence sentence processing and recall of linguistic information.
Language and memory are profound components of human cognition, and they are
intricately interconnected. It is difficult to think about language without the role of memory
because in the absence of memory, one would not be able to produce sounds or signs to express
thoughts (through procedural memory) or to remember what he/she meant to say or how to say
it. Language serves memory formation as well. What individuals remember is affected by the
way in which the information is described linguistically. For example, the same visual input can
be remembered differently depending on the description or the language used to elicit the input
information from memory.
The main purpose of this dissertation was to draw us closer to understanding how
syntactic and semantic information in linguistic expressions would influence how accessible or
prominent described entities and events are in discourse and memory. The overall results of
179

Experiments 1, 2, and 3 on entity information and Experiment 4 on event information (both
investigating novel aspects) further support the existing literature and suggest that changes in
linguistic form and meaning can alter the discourse and mental representations of the denoted
entities and events.

6.2 Overview of Experiments 1, 2, and 3, exploring entities’ discourse and memory
representations
Experiment 1 (Chapter 2) was a story continuation task where participants read four-sentence
stories and added three sentences of their own that naturally continued each story. I tested
whether the way in which entities are introduced in a story would change the accessibility of the
characters in subsequent discourse. The grammar of Korean made it possible to minimally alter
the way in which characters are presented since different case markers can be attached to entities
in the language. For example, grammatical subjects can have the nominative case marker -i/-ka
attached to it. The nominative case marker signals the subject of the sentence, which is also the
default topic of the sentence (i.e., what the sentence is about). Then there is the contrastive topic
marker -un/-nun, which can also be attached to a subject. When an entity is contrastive topic-
marked, certain attributes of the character are contrasted to other discourse entities. For example,
when a subject is contrastive topic marked as in John-un loves Mary, the attribute of John loving
Mary is contrasted to the attributes of other discourse entities, such as in Bill-un loves Susan
(e.g., Choi, 1999; Han, 2001; Lee, 1992; 1999; 2003).
The story continuation experiment showed that changes as small as a case marker can
shift the prominence of different entities in discourse. First of all, the results supported the
existing observation that the global discourse topic (i.e., whom the entire text was about) is
180

highly prominent in discourse continuation, as indicated by its high likelihood of being
mentioned subsequently. However, when a member of a discourse set was contrastive topic
marked, the prominence of the global topic decreased. Instead, the contrastive topic marker
temporarily boosted the accessibility of other members of the discourse. Participants mentioned
other members of a set (initially not introduced individually in the story) more frequently in
subsequent sentences when one of the members was contrastive topic marked than when it was
nominative marked. Because the contrastive topic marker highlights a contrasting characteristic
of the entity to other discourse set members, readers’ subsequent sentences were often about
some qualities of other set members. The contrastive topic marking was strong enough to boost
the discourse set, temporarily overriding the prominence of the global discourse topic. The
subject of participants’ first two continuation sentences was frequently about other unmentioned
set members when one of the members was contrastive topic marked in the story. Overall,
Experiment 1 results demonstrate that Korean case marking, used to signal the pragmatic role of
an entity (e.g., contrastive topic vs. non-contrastive topic), can alter how much attention is paid
to that entity and to other entities in the discourse.
While Experiment 1 tested entities’ representation in discourse, Experiment 2
investigated if case marking would change the prominence of entities in memory. Individuals’
memory for story characters was tested using a probe-word recognition task. It was hypothesized
that if case marking shows an effect on who is likely to be mentioned in discourse, it might also
have an effect on which character readers tend to remember better. Specifically, I predicted that
contrastive topic marked entities would be remembered better than their nominative marked
counterparts because contrastive topic marking boosts the accessibility of individual set
members. When two characters in a story were introduced with the contrastive topic marker, the
181

information about them might have been made more distinctive by the virtue of being contrasted
to one another. Contrary to the predictions however, the difference between nominative marking
and contrastive topic marking did not change how well participants remembered the characters.
Entities were recalled equally well whether they were nominative marked or contrastive topic
marked. The lack of contrastive topic marking on entity recognition might have been due to the
story structure. Because the two characters were commented on one after another, the predicate
information about them might have been contrastive enough regardless of which case marker
was used.
In order to further test the effect of contrastive topic marking on memory representation,
future research may implement a probe-word recognition experiment using the same story
structure as Experiment 1. In the future investigation, one might test comprehenders’ memory
retention of different discourse entities including: (i) the global discourse topic whose
prominence has been shown to be high (especially in the nominative marked condition), (ii) a
local topic when it is marked differently between the nominative and the contrastive topic cases,
and (iii) the members of a discourse set. An interesting future research may also include a story
continuation task where the target story disappears from readers’ sight, which would make them
rely on what they remember from the story to continue the discourse. This design might help
reduce the discrepancy that existed between the continuation task where readers were able to re-
read the story and the memory recall task where the text was unavailable for a review.
Despite the lack of case marking effect, the results of Experiment 2 showed a compelling
recognition pattern. What mattered in entity recognition was order of mention. Of the two
characters introduced in each story, comprehenders recognized the recently-mentioned subject
better than the first-mentioned subject even 15 seconds after the presentation of the story. The
182

recency effect on the entity recognition motivated Experiment 3. The story continuation
Experiment 3 directly compared the discourse and the memory representations of entities as it
employed the same target stories as Experiment 2. I was interested in testing whether the
character that individuals remembered better (i.e., recently-mentioned subject) would be the
character that participants mentioned more frequently in subsequent sentences. The results
indicated that despite the advantage in memory, the recently-mentioned protagonist was not
mentioned most frequently. Instead, the character that was mentioned and commented on first in
the story showed an advantage in discourse continuation.
The results from Experiments 1, 2, and 3 inform us that memory representation and
discourse representation are not parallel necessarily. Though studies have indicated that certain
grammatical structures can boost the accessibility of an entity in both discourse and memory
(e.g., Birch et al., 2000; Birch & Garnsey, 1995), discourse prominence might not be a direct
reflection of what is most memorable at a given point in time when a larger discourse is taken
into consideration. In other words, memory prominence is not the only source of discourse
prominence. In a board discourse, multiple pieces of information work together to build a
coherent structure, one of which is the availability in memory. Even though individuals’ memory
for the recently-mentioned participant was better, Experiment 3 (story continuation) showed that
readers found it more natural or coherent to continue the story by talking about the first-
mentioned. Comprehenders’ continuation patterns might have been driven by their effort to make
the continuation as parallel as possible to the given story fragment. For the continuation to be
parallel, readers would have needed to talk about the first-mentioned subject first.
The discrepancy between discourse prominence and memory prominence can be
attributed to the different levels of processing involved in the production task (i.e., story
183

continuation experiment, Experiment 2) and the recognition task (i.e., probe-word recognition,
Experiment 3). In the production task, macro-level, top-down processing might have been
required where comprehenders focused on the global coherence of the text (e.g., Van den Broek
et al., 2001). This process would have required comprehenders to consider various pieces of
information in the discourse including the overall structure of the story, the participants, the
likely event to take place following the story fragment, and so on. However, in the entity
recognition task, micro-level, bottom up processing might have been required where individuals
focused on each sentence as separate instances. In this task, participants needed not to build a
continuation beyond the fragment of the information provided in the given story. Therefore, the
processing strategy might have been simpler. The asymmetry in the cognitive processes between
discourse continuation and memory retrieval might be an inherent aspect that research in this
area might have to expect.
Furthermore, the different processing levels between the two types of tasks might have
been due to the specific goals readers were trying to achieve. In the comprehension and
production task, the goal was to provide a coherent flow of the story where the prominence of
discourse entities needed to be monitored constantly. This process must have involved active and
strategic comprehension and production skills. However, the goal of the recognition task was to
understand the text and remember the content information. The processes necessary to achieve
this goal might have been more passive and local compared to those in the production task.

184

6.3 Overview of Experiment 4, testing the effects of verbs’ repetition frequency on sentence
processing and memory recall
Experiment 4 focused on another important aspect of a sentence - event information denoted by a
verb. I tested whether a physical attribute of an action verb, in particular the frequency of an
action to repeat in a row, could influence sentence processing and memory. Adult Korean
speakers judged the sensicality of sentences written in Korean in two different inherent repetition
frequency conditions: (i) the Expected Low Frequency condition where the sentence denoted an
action that generally repeats once or twice in a row in the real world, e.g., Seonwu knocked on
the door twice and (ii) the Expected High Frequency condition where the sentence denoted an
action that typically repeats multiple times in a row, e.g., Seonwu bounced the ball six times.
Participants were also tested on how well they recall having seen the target verb phrases.
The sensicality judgment and the probe-verb recognition tasks revealed an effect of
verb’s inherent motion repetition frequency. Sentences describing actions that generally repeat
once or twice in a row were judged more accurately, compared to sentences describing actions
that repeat multiple times in a row. Furthermore, verbs denoting inherently low repetition actions
were recognized more accurately than verbs denoting high repetition actions. These results add
to the existing literature demonstrating that language comprehension is influenced by lexical
semantics. Verbs’ semantic properties that have been shown to affect individuals’ (non)linguistic
behaviors include the speed and the duration of a motion. In addition to these properties, my
results suggest that actions’ inherent frequency to repeat can also influence sentence processing
and how well the event information is represented in memory.
Another noteworthy finding from Experiment 4 is the lack of a number adverb effect.
Whether the verb occurred with a number adverb that matched or mismatched its expected
185

frequency did not influence the sensicality judgment. It was predicted that the verb-adverb
matching conditions would yield better performance. However, this prediction was not borne
out. In fact, when the high frequency verbs were paired with a matching high frequency adverb,
verb recognition accuracy significantly decreased. These results suggest that verb’s inherent
repetition frequency plays a central role in building a mental representation of the described
action, regardless of the specific repetition expressed by the number adverb.
In order to better understand the effect of the repetition frequency semantics, future
research may include neuroimaging techniques and test whether inherently high or low repetition
verbs elicit different levels of brain activities.

6.4 Conclusion
To summarize, this dissertation has demonstrated that the way in which entities are presented
impacts which entity readers like to talk about in story continuation (Experiment 1). It has also
shown that order of mention affects how well entities are remembered. A recently-mentioned
subject was recognized more accurately than the first-mentioned subject (Experiment 2). In
contrast, it was the first-mentioned entity that readers tended to talk about more frequently in
subsequent sentences rather than the better-remembered recent entity (Experiment 3). These
findings suggest that saliency in memory is not the sole source of the likelihood of mention. The
prominence of an entity is different in discourse and in memory. An entity’s discourse
representation is influenced by a variety of different factors, including case marking, order of
mention, and other entities present in the discourse. Saliency in memory serves as one of the
factors.
186

Furthermore, this dissertation has provided novel evidence suggesting that processing
verbs that denote low repetition actions might require a lighter processing load and simpler
mental representation than processing verbs that refer to high repetition actions (Experiment 4).
Additionally, the findings reported in this dissertation highlight the importance of
studying both entity and event information in order to broaden our understanding of the
interaction between language comprehension and memory retention.

187

References

Aaronson, D., & Ferres, S. (1983). Lexical categories and reading tasks. Journal of Experimental
Psychology: Human Perception and Performance, 9(5), 675.
Aaronson, D., & Scarborough, H. S. (1977). Performance theories for sentence coding: Some
quantitative models. Journal of Verbal Learning and Verbal Behavior, 16(3), 277-303.
Adams, A.-M., & Gathercole, S. E. (2000). Limitations in working memory: Implications for
language development. International Journal of Language & Communication
Disorders, 35(1), 95-116.
Almor, A. (1999). Noun–phrase anaphora and focus: the informational load hypothesis.
Psychological Review, 106, 748–765.
Almor, A., & Eimas, P. D. (2008). Focus and noun phrase anaphors in spoken language
comprehension. Language and cognitive processes, 23(2), 201-225.
Anderson, J. R. (1990). Cognitive psychology and its implications. New York: Freeman.
Anderson, J. R. (1994). Learning and Memory. Wiley, New York.
Ansolabehere, A., & Iyengar, S. (1995). Going negative: How political advertisements shrink
and polarize the electorate. New York: Free Press.
Ariel, M. (1990). Accessing noun–phrase antecedents. London: Routledge.
Ariel, M. (1991). The function of accessibility in a theory of grammar. Journal of
Pragmatics, 16(5), 443-463.
Ariel, M. (2001). Accessibility theory: An overview. Text representation: Linguistic and
psycholinguistic aspects, 29-87.
188

Arnold, J. E. (1998). Reference form and discourse patterns. Doctoral dissertation, Stanford
University.
Arnold, J. E. (2001). The effect of thematic roles on pronoun use and frequency of reference
continuation. Discourse Processes, 31, 137–162.
Arnold, J. E. (2008). Reference production: Production-internal and addressee-oriented
processes. Language and Cognitive Processes, 23, 495–527.
Arnold, J. E. (2010). How speakers refer: the role of accessibility. Language and Linguistics
Compass, 4(4), 187-203.
Arnold, J. E. (2013). Information status relates to production, distribution, and comprehension.
Reply to Maryellen C. MacDonald, How language production shapes language form and
comprehension. Frontiers in Psychology, 4, Article 235, 1-3.
Arnold, J. E., Brown-Schmidt, S., & Trueswell, J. (2007). Children's use of gender and order-of-
mention during pronoun comprehension. Language and Cognitive Processes, 22(4), 527-
565.
Arnold, J. E., Eisenband, J. G., Brown-Schmidt, S., & Trueswell, J. C. (2000). The rapid use of
gender information: Evidence of the time course of pronoun resolution from
eyetracking. Cognition, 76(1), B13-B26.
Au, T.K. (1986). A verb is worth a thousand words: The causes and consequences of
interpersonal events implicit in language. Journal of Memory and Language, 25, 104-
122.
Aziz-Zadeh, L., Wilson, S. M., Rizzolatti, G., & Iacoboni, M. (2006). Congruent embodied
representations for visually presented actions and linguistic phrases describing
actions. Current biology, 16(18), 1818-1823.
189

Baddeley, A. (1992). Working memory. Science, 255(5044), 556-559.
Baddeley, A. (1992). Working memory: The interface between memory and cognition. Journal
of cognitive neuroscience, 4(3), 281-288.
Baddeley, A. (1995). Working Memory. In M. S. Gazzaniga (Ed.), The Cognitive
Neurosciences (pp. 755-764). Campidge, Mass.: The MIT Press.
Baddeley, A. (2003). Working memory and language: An overview. Journal of communication
disorders, 36(3), 189-208.
Baddeley, A. D., & Hitch, G. (1974). Working memory. In: G. A. Bower (Ed.), The psychology
of learning and motivation, vol. 8 (pp. 47–89). New York: Academic Press.
Baddeley, A. D., & Hitch, G. J. (1977). Recency reexamined. Attentiona nd Performance, 6,
647-667.
Bambini, V., Ghio, M., Moro, A., & Schumacher, P. B. (2013). Differentiating among pragmatic
uses of words through timed sensicality judgments. Frontiers in psychology, 4, 1-16.
Barsalou, L. W. (1999). Perceptual symbol systems. Behav. Brain Sci. 22, 577–660.
Barzilay, R. & Lapata, M. (2005). Modeling local coherence: an entity-based approach. In the
Proceedings of the 43rd Annual Meeting of the Association for Computational
Linguistics (ACL’05).
Barzilay, R., & Lapata, M. (2008). Modeling local coherence: An entity-based
approach. Computational Linguistics, 34(1), 1-34.
Beattie, G., & Ellis, A. (2014). The psychology of language and communication. Psychology
Press.
190

Bergen, B., & Chang, N. (2005). Embodied construction grammar in simulation-based language
understanding. Construction grammars: Cognitive grounding and theoretical
extensions, 3, 147-190.
Bergen, B., Lau, T. T. C., Narayan, S., Stojanovic, D., & Wheeler, K. (2010). Body part
representations in verbal semantics. Memory & Cognition, 38(7), 969-981.
Birch, S. L., & Garnsey, S. M. (1995) The effect of focus on memory for words in sentences.
Journal of Memory and Language, 34, 232-267.
Birch, S. L., Albrecht, J. E., & Myers, J. L. (2000). Syntactic focusing structures influence
discourse processing. Discourse Processes, 30(3), 285-304.
Bird, H., Howard, D., & Franklin, S. (2003). Verbs and nouns: The importance of being
imageable. Journal of Neurolinguistics, 16(2), 113-149.
Black, J. B., & Bern, H. (1981). Causal inference and memory for events in narratives. Journal
of Verbal Learning and Verbal Behavior, 20, 267-275.
Bobrow, S. A. (1970). Memory for words in sentences. Journal of Verbal Learning and Verbal
Behavior, 9(4), 363-372.
Bock, J. K. (1986a). Meaning, sound, and syntax: Lexical priming in sentence
production. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 12(4), 575.
Bock, J. K. (1986b). Syntactic persistence in language production. Cognitive psychology, 18(3),
355-387.
Bock, J. K., & Irwin, D. E. (1980). Syntactic effects of information availability in sentence
production. Journal of Verbal Learning and Verbal Behavior, 19, 467-484.
191

Bock, J. K., & Warren, R. K. (1985). Conceptual accessibility and syntactic structure in sentence
formulation. Cognition, 21(1), 47-67.
Bock, K., & Griffin, Z. M. (2000). The persistence of structural priming: Transient activation or
implicit learning? Journal of Experimental Psychology: General, 129(2), 177.
Bock, K., Loebell, H., & Morey, R. (1992). From conceptual roles to structural relations:
bridging the syntactic cleft. Psychological review, 99(1), 150.
Boroditsky, L. (2011). How language shapes thought. Scientific American, 304(2), 62-65.
Boston, M., Hale, J., Kliegl, R., Patil, U., & Vasishth, S. (2008). Parsing costs as predictors of
reading difficulty: An evaluation using the Potsdam Sentence Corpus. Journal of Eye
Movement Research 2(1):1, 1-12
Bransford, J. D., & Franks, J. J. (1971).The abstraction of linguistic ideas. Cognitive Psychology,
2, 331-350.
Brennan, S. E. (1995). Centering attention in discourse. Language and Cognitive Processes, 10,
137–167.
Brewer, W. F. (1975). Memory for ideas: Synonym substitution. Memory & Cognition, 3, 458-
464
Broadbent, D. E., & Broadbent, M. H. (1987). From detection to identification: response to
multiple targets in rapid serial visual presentation. Perception And Psychophysics, 4 (2),
105-113.
Brown, R., & Fish, D. (1983). The psychological causality implicit in language.
Cognition, 14(3), 237-273.
192

Buda, R., & Zhang, Y. (2000). Consumer product evaluation: the interactive effect of message
framing, presentation order, and source credibility. Journal of Product & Brand
Management, 9(4), 229-242.
Büring, D. (1997). The meaning of topic and focus: The 59th Street Bridge accent (Vol. 3).
Psychology Press.
Caplan, D., & Waters, G. S. (1999). Verbal working memory and sentence
comprehension. Behavioral and brain Sciences, 22(01), 77-94.
Caplan, D., Dede, G., Waters, G., Michaud, J., & Tripodis, Y. (2011). Effects of age, speed of
processing, and working memory on comprehension of sentences with relative clauses.
Psychology and Aging, 26(2), 439-450.
Carmichael, L.; Hogan, H. P.; Walter, A. A.. (1932). An experimental study of the effect of
language on the reproduction of visually perceived form. Journal of Experimental
Psychology 15.1, 73-86.
Carpenter, P. A. & Just, M. A. (1977). Integrative processes in comprehension. In D. LaBerge &
S. J. Samuels (Eds.). Basic processes in reading (pp. 217-242). Hillsdale, NJ: Erlbaum.
Carreiras, M., Gernsbacher, M. A., & Villa, V. (1995). The advantage of first mention in
Spanish. Psychonomic Bulletin & Review, 2(1), 124-129.
Čech, R., Mačutek, J., & Žabokrtský, Z. (2011). The role of syntax in complex networks: Local
and global importance of verbs in a syntactic dependency network. Physica A: Statistical
Mechanics and its Applications, 390(20), 3614-3623.
Chafe, W. L. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point of
view. In C. N. Li (Ed.), Subject and topic (pp. 25-56). New York: Academic Press, Inc.
193

Chang, F. R. (1980). Active memory processes in visual sentence comprehension: Clause effects
and pronominal reference. Memory & Cognition,8(1), 58-64.
Chang, F., Dell, G. S., & Bock, K. (2006). Becoming syntactic. Psychological Review, 113, 234
–272.
Chiriacescu, S., & von Heusinger, K. (2010). Discourse prominence and pe-marking in
Romanian. The International Review of Pragmatics 2(2). 298- 332.
Choi, H.-W. (1999). Optimizing structure in context. Stanford, CA: CSLI.
Chomsky, N. (1995). The minimalist program. Cambridge, MA: MIT Press.
Christiansen, M. H., & Kirby, S. (2003). Language evolution: Consensus and
controversies. Trends in cognitive sciences, 7(7), 300-307.
Clark, H. H. (1965). Some structural properties of simple active and passive sentences. Journal
of Verbal Learning and Verbal Behavior, 4(5), 365-370.
Clark, H. H., & Card, S. K. (1969). Role of semantics in remembering comparative sentences.
Journal of Experimental Psychology, 82, 545–553.
Clark, H. H., & Clark, E. V. (1968). Semantic distinctions and memory for complex
sentences. The Quarterly journal of experimental psychology, 20(2), 129-138.
Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic
processing. Psychological review, 82(6), 407.
Constant, N. (2014). Contrastive topic: Meanings and realizations. Ph.D. dissertation, University
of Massachusetts, Amherst.
Conway, A. R., Cowan, N., Bunting, M. F., Therriault, D. J., & Minkoff, S. R. (2002). A latent
variable analysis of working memory capacity, short-term memory capacity, processing
speed, and general fluid intelligence. Intelligence, 30(2), 163-183.
194

Conway, A. R. & Eagle, R. W. (1996). Individual differences in working memory capacity:
More evidence for a general capacity theory. Memory, 4(6), 577-590.
Corbett, A. T., & Chang, F. R. (1983). Pronoun disambiguation: Accessing potential
antecedents. Memory & Cognition, 11, 283–294.
Cowan, N. (1995). Attention and memory: An integrated framework. Oxford, England: Oxford
University Press.
Cowan, N. (1996). Short-term memory, working memory, and their importance in language
processing. Topics in language disorders, 17(1), 1-18.
Cowan, N. (2000). The magical number 4 in short-term memory: A reconsideration of mental
storage capacity. Behavioral and Brain Sciences, 24, 87–185.
Craik, F. I. M., & Levy, B. A. (1976). The concept of primary and memory. In W. K. Estes (Ed.),
Handbook of learning and cognitive processes (Vol. 4, pp. 133-176). Hillsdale, NJ:
Erlbaum.
Davidson, D. J., Zacks, R. T., & Ferreira, F. (2003). Age preservation of the syntactic processor
in production. Journal of Psycholinguistic Research, 32(5), 541-566.
Davis, M. H., Meunier, F., & Marslen-Wilson, W. D. (2004). Neural responses to morphological,
syntactic, and semantic properties of single words: an fMRI study. Brain and
language, 89(3), 439-449.
Dell, G. S., McKoon, G., & Ratcliff, R. (1983). The activation of antecedent information during
the processing of anaphoric reference in reading. Journal of Verbal Learning and Verbal
Behavior, 22(1), 121-132.
195

Dillon, B., Chow, W. Y., Wagers, M., Guo, T., Liu, F., & Phillips, C. (2014). The structure-
sensitivity of memory access: evidence from Mandarin Chinese. Frontiers in
psychology, 5, 1-16.
Enard, W., Przeworski, M., Fisher, S. E., Lai, C. S., Wiebe, V., Kitano, T., ... & Pääbo, S.
(2002). Molecular evolution of FOXP2, a gene involved in speech and
language. Nature, 418(6900), 869-872.
Engel de Abreu, P. M. J., Gathercole, S. E., & Martin, R. (2011). Disentangling the relationship
between working memory and language: The roles of short-term storage and cognitive
control. Learning and Individual Differences, 21(5), 569-574.
Engelkamp, J., & Zimmer, H. D. (1983). Dynamic aspects of language processing. Berlin,
Germany: Springer-Verlag.

Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. A. (1999). Working memory,
short-term memory and general fluid intelligence: A latent variable approach. Journal of
Experimental Psychology: General, 128, 309–331.
Fausey, C. M., & Boroditsky, L. (2011). Who dunnit? Cross-linguistic differences in eye-witness
memory. Psychonomic bulletin & review, 18(1), 150-157.
Fausey, C. M., & Matlock, T. (2011). Can Grammar Win Elections? Political Psychology, 32(4),
563–574.
Fecica, A. M., & O’Neill, D. K. (2010). A step at a time: Preliterate children’s simulation of
narrative movement during story comprehension. Cognition,116(3), 368-381.
Federmeier, K. D., & Kutas, M. (1999). A rose by any other name: Long-term memory structure
and sentence processing. Journal of memory and Language,41(4), 469-495.
196

Fedorenko, E., Woodbury, R. & Gibson, E. (2013). Direct evidence of memory retrieval as a
source of difficulty in long-distance structural dependencies in language. Cognitive
Science, 37, 378-394.
Ferré, P., Fraga, I., Comesaña, M., & Sánchez-Casas, R. (2014). Memory for emotional words:
The role of semantic relatedness, encoding task and affective valence. Cognition and
Emotion, (ahead-of-print), 1-10.
Ferreira, F. (2003). The misinterpretation of noncanonical sentences. Cognitive
psychology, 47(2), 164-203.
Ferreira, F., Christianson, K., & Hollingworth, A. (2001). Misinterpretations of garden-path
sentences: Implications for models of sentence processing and reanalysis. Journal of
psycholinguistic research, 30(1), 3-20.
Ferreira, V. S., & Yoshita, H. (2003). Given-new ordering effects on the production of scrambled
sentences in Japanese. Journal of psycholinguistic research, 32(6), 669-692.
Ferreira, V. S., Bock, K., Wilson, M. P., & Cohen, N. J. (2008). Memory for syntax despite
amnesia. Psychological Science, 19(9), 940-946.
Fillmore, C . J . (1966). Review of Bendix's componential analysis of general vocabulary: The
semantic structure of a set of verbs in English, Hindi, and Japanese. International Journal
of American Linguistics 32: Part II (2), Publication 41.
Fillmore, C . J . (1968). The case for case. In E. Bach & R. T. Harms (eds.), Universals in
Linguistic Theory (pp. 1-134). New York: Holt, Rinehart and Winston.
Fischer, M. H., & Zwaan, R. A. (2008). Embodied language: A review of the role of the motor
system in language comprehension. The Quarterly Journal of Experimental
Psychology, 61(6), 825-850.
197

Fitzsimons, G. M., & Kay, A. C. (2004). Language and interpersonal cognition: Causal effects of
variations in pronoun usage on perceptions of closeness. Personality and Social
Psychology Bulletin, 30(5), 547-557.
Fletcher, C. R. (1984). Markedness and topic continuity in discourse processing. Journal of
Verbal Learning and Verbal Behavior, 23, 487–493.
Fletcher, C. R. (1986). Strategies for the allocation of short-term memory during
comprehension. Journal of Memory and Language, 25(1), 43-58.
Fodor, J. A. (1983). The modularity of mind: an essay on faculty psychology. Cambridge, MA:
MIT Press.
Foltz, P., Kintsch, W., & Landauer, T. (1998). The measurement of textual coherence with latent
semantic analysis. Discourse Processes, 25(2&3), 285–307.
Foraker, S., & McElree, B. (2007). The role of prominence in pronoun resolution: Active versus
passive representations. Journal of Memory and Language, 56(3), 357-383.
Frackowiak, R. S. (1994). Functional mapping of verbal memory and language. Trends in
neurosciences, 17(3), 109-115.
Frank, S. L. (2009, July). Surprisal-based comparison between a symbolic and a connectionist
model of sentence processing. In Proceedings of the 31st annual conference of the
cognitive science society (pp. 1139-1144). Austin, TX: Cognitive Science Society.
Frank, S. L., Trompenaars, T., & Vasishth, S. (2015). Cross-linguistic differences in processing
double-embedded relative clauses: Working-memory constraints or language
statistics? Cognitive Science.
198

Fraundorf, S. H., Watson, D. G., & Benjamin, A. S. (2010). Recognition memory reveals just
how CONTRASTIVE contrastive accenting really is. Journal of memory and
language, 63(3), 367-386.
Frazier, L., & Rayner, K. (1982). Making and correcting errors during sentence comprehension:
Eye movements in the analysis of structurally ambiguous sentences. Cognitive
Psychology, 14, 178–210.
Friederici, A. (2002). Towards a neural basis of auditory sentence processing. Trends in
Cognitive Sciences, 6(2), 78–84.
Fukumura, K., & Van Gompel, R. P. (2010). Choosing anaphoric expressions: Do people take
into account likelihood of reference? Journal of Memory and Language, 62(1), 52-66.
Gagné, C. L. & Shoben, E. J. (1997). Influence of thematic relations on the comprehension of
modifier-noun combinations. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 23(1), 71-87.
Garnham, A. (2001). Mental models and the interpretation of anaphora. Hove: Psychology
Press.
Garramone, G. (1984). Voter response to negative political ads. Journalism Quarterly, 61, 250–
259.
Garrod, S. & Sanford, T. (1988). Thematic subjecthood and cognitive constraints on discourse
structure. Journal of Pragmatics 12, 519 - 534.
Garrod, S., Freudenthal, D., & Boyle, E. (1994). The role of different types of anaphor in the on-
line resolution of sentences in a discourse. Journal of Memory and Language, 33, 39–68.
Garvey, C., & Caramazza, A. (1974). Implicit causality in verbs. Linguistic inquiry, 5(3), 459-
464.
199

Garvey, C., Caramazza, A., & Yates, J. (1975). Factors influencing assignment of pronoun
antecedents. Cognition, 3, 227-243.
Gathercole, S. E., & Baddeley, A. D. (1990). Phonological memory deficits in language
disordered children: Is there a causal connection?. Journal of memory and
language, 29(3), 336-360.
Gathercole, S. E., Alloway, T. P., Willis, C., & Adams, A. M. (2006). Working memory in
children with reading disabilities. Journal of Experimental Child Psychology, 93(3), 265-
281.
Gathercole, S. E., Willis, C. S., Emslie, H., & Baddeley, A. D. (1992). Phonological memory and
vocabulary development during the early school years: A longitudinal
study. Developmental psychology, 28(5), 887.
Gennari, S. P., Mirković, J., & MacDonald, M. C. (2012). Animacy and competition in relative
clause production: a cross-linguistic investigation. Cognitive psychology, 65(2), 141-176.
Gentner, D . (1981). Some interesting differences between verbs and nouns. Cognition and Brain
Theory, 4(2), 161-178 .
Gentner, D., & France, I. M. (1988). The verb mutability effect: Studies of the combinatorial
semantics of nouns and verbs. Lexical ambiguity resolution: Perspectives from
psycholinguistics, neuropsychology, and artificial intelligence, 343-382.
Gentner, D., & Loftus, E. F. (1979). Integration of verbal and visual information as evidenced by
distortions in picture memory. American Journal of Psychology, 92(2), 363-375.
Gernsbacher, M. (1990). Language comprehension as structure building. Hillsdale, NJ: Erlbaum.
200

Gernsbacher, M. A. & Hargreaves, D. (1992). The privilege of primacy: Experimental data and
cognitive explanations. In D. L. Payne (Ed.), Pragmatics of word order flexibility (pp.
83-116). Amsterdam: John Benjamins Publishing co.
Gernsbacher, M. A. & Hargreaves, D. J. (1988). Accessing sentence participants: The advantage
of first mention. Journal of Memory and Language, 27, 699-717.
Gernsbacher, M. A. & Shroyer, S. (1989) .The cataphoric use of the indefinite ‘this’ in spoken
narratives. Memory and Cognition, 17, 536 - 540.
Gernsbacher, M. A. (1991). Cognitive processes and mechanisms in language comprehension:
The structure building framework. In G. H. Bower (Ed.), The psychology of learning and
motivation (pp. 217-263). New York: Academic Press.
Gernsbacher, M. A., & Hargreaves, D. J. (1988). Accessing sentence participants: The advantage
of first mention. Journal of Memory and Language, 27(6), 699-717.
Gernsbacher, M. A., & Jescheniak, J. D. (1995). Cataphoric devices in spoken
discourse. Cognitive psychology, 29(1), 24-58.
Gernsbacher, M. A., & Kaschak, M. P. (2003). Neuroimaging studies of language production
and comprehension. Annual Review of Psychology, 54, 91-114.
Gernsbacher, M. A., & Robertson, R. R. W. (2002). The definite article the as a cue to map
thematic information. In M. M. Louwerse & W. Van Peer, Thematics: Interdisciplinary
studies. Amsterdam/Philadelphia: John Benjamins.
Gernsbacher, M. A., Hargreaves, D. J., & Beeman, M. (1989). Building and accessing clausal
representations: The advantage of first mention versus the advantage of clause
recency. Journal of Memory and Language, 28(6), 735-755.
201

Gibson, E. (1998). Linguistic complexity: Locality of syntactic dependencies. Cognition, 68(1),
1-76.
Gillam, R. B., & van Kleeck, A. (1996). Phonological awareness training and short-term working
memory: Clinical implications. Topics in Language Disorders, 17(1), 72-81.
Givón, T. (1983b). Topic continuity in spoken English. In Givón,B48 T., (Ed.), Topic continuity
in discourse: a quantitative cross-language study (343-364). Amsterdam: John
Benjamins.
Givón, T. (1983a). Topic continuity in discourse: An introduction. In T. Givón (Ed.), Topic
continuity in discourse: A quantitative cross-language study (pp. 1–41). Amsterdam:
John Benjamins.
Givón, T. (1989). Mind, code, and context: Essays in pragmatics. Hillsdale, N.J.: L. Erlbaum
Associates.
Givón, T. (1990). Syntax: A Functional-Typological Introduction, vol. II. Philadelphia: John
Benjamins Publishing Company.
Glanzer, M., & Cunitz, A. R. (1966). Two storage mechanisms in free recall. Journal of Verbal
Learning and Verbal Behavior, 5, 351-360.
Gleitman, L., & Papafragou, A. (2005). Language and thought. In K. J. Holyoak & R. Morrison
(Eds.), Handbook of Thinking and Reasoning (pp. 633-661). Cambridge Univ. Press,
New York.
Glenberg, A. M. (1997). What memory is for: Creating meaning in the service of
action. Behavioral and brain sciences, 20(01), 41-50.
Goldberg, A. (2010). Verbs, constructions and semantic frames. Syntax, lexical semantics, and
event structure, 39-58.
202

Gordon, P. C., & Hendrick, R. (1998). The representation and processing of coreference in
discourse. Cognitive Science, 22(4), 389-424.
Gordon, P. C., Grosz, B. J. & Gilliom, L. A. (1993). Pronouns, names, and the centering of
attention in discourse. Cognitive Science, 17, 311-347.
Gordon, P. C., Hendrick, R., & Foster, K. (2000). Language comprehension and probe-list
memory. Journal of Experimental Psychology: Learning, Memory and Cognition, 26,
766–775.
Gordon, P. C., Hendrick, R., & Johnson, M. (2004). Effects of noun phrase type on sentence
complexity. Journal of Memory and Language, 51(1), 97–114.
Gordon, P. C., Hendrick, R., & Levine, W. H. (2002). Memory-load interference in syntactic
processing. Psychological Science, 13(5), 425-430.
Gordon, P.C., Hendrick, R., & Johnson, M. (2001). Memory interference during language
processing. Journal of Experimental Psychology: Learning, Memory and Cognition,
27, 1411-1423.
Graesser, A. C., Singer, M., & Trabasso, T. (1994). Constructing inferences during narrative text
comprehension. Psychological review, 101(3), 371.
Greene, S. B., McKoon, G., & Ratcliff, R. (1992). Pronoun resolution and discourse
models. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18(2),
266.
Gregory, M. & Michaelis, L. (2001) Topicalization and left-dislocation: A functional opposition
revisited. Journal of Pragmatics 33, 1665-1706.
Grosz, B. J., Weinstein, S., & Joshi, A. K. (1995). Centering: A framework for modeling the
local coherence of discourse. Computational linguistics, 21(2), 203-225.
203

Guerry, M., Gimenes, M., Caplan, D., & Rigalleau, F. (2006). How long does it take to find a
cause? An online investigation of implicit causality in sentence production. The
Quarterly journal of experimental psychology, 59(9), 1535-1555.
Gundel, J. (1974/1989). The Role of Topic and Comment in Linguistic Theory. New York:
Garland.
Gundel, J. K. (1999). On different kinds of focus. In P. Bosch & R. van der Sandt (Eds.), Focus:
Linguistic, cognitive, and computational perspectives (pp. 293-305). Cambridge:
Cambridge University Press.
Gundel, J. K., Hedberg, N., & Zacharski, R. (1993). Cognitive status and the form of referring
expressions in discourse. Language, 274-307.
Gupta, P. (2003). Examining the relationship between word learning, nonword repetition, and
immediate serial recall in adults. Quarterly Journal of Experimental Psychology Section
a-Human Experimental Psychology, 56(7), 1213-1236.
Hales, J., Kuang, X. J., & Venkataraman, S. (2011). Who believes the hype? An experimental
examination of how language affects investor judgments. Journal of Accounting
Research, 49(1), 223-255.
Halford, G. S., Maybery, M. T., O'Hare, A. W., & Grant, P. (1994). The development of memory
and processing capacity. Child Development, 65(5), 1338-1356.
Halliday, M.A.K. (1967) Notes on transitivity and theme in English (Part 2). Journal of
Linguistics 3, 199-244.
Han, C. (1998). Asymmetry in the interpretation of -(n)un in Korean. Japanese/Korean
Linguistics 7, 1-15.
204

Hare, M., McRae, K., & Elman, J. L. (2003). Sense and structure: Meaning as a determinant of
verb subcategorization preferences. Journal of Memory and Language, 48(2), 281-303.
Hartshorne, J. K. (2014). What is implicit causality? Language & Cognitive Processes, 29(7),
761-898.
Hartshorne, J. K., Sudo, Y., & Uruwashi, M. (2013). Are implicit causality pronoun resolution
biases consistent across languages and cultures? Experimental Psychology.
Hartsuiker, R. J., Bernolet, S., Schoonbaert, S., Speybroeck, S., & Vanderelst, D. (2008).
Syntactic priming persists while the lexical boost decays: Evidence from written and
spoken dialogue. Journal of Memory and Language, 58(2), 214-238.
Hauk, O., Johnsrude, I., & Pulvermüller, F. (2004). Somatotopic representation of action words
in human motor and premotor cortex. Neuron, 41(2), 301-307.
Hauser, M. D., Chomsky, N., & Fitch, W. T. (2002). The faculty of language: What is it, who
has it, and how did it evolve? Science, 298(5598), 1569-1579.
Hedberg, N. A. (2000). The referential status of clefts. Language 76, 891-920.
Higgins, E. T., Bargh, J. A., & Lombardi, W. J. (1985). Nature of priming effects on
categorization. Journal of experimental psychology: Learning, Memory, and
Cognition, 11(1), 59.
Higgins, E. T., & King, G. (1981). Accessibility of social constructs: Information-processing
consequences of individual and contextual variability. In N. Cantor & J. F. Kihlstrom
(Eds.), Personality, cognition, and social interaction (pp. 69-121). Hillsdale, NJ:
Erlbaum.
Hitch, G. J. (1980). Developing the concept of working memory. Cognitive psychology: New
directions, 154-196.
205

Hornby, P.A. (1974). Surface structure and presupposition. Journal of Verbal Learning and
Verbal Behavior, 13, 530-538.
Houston, D. M., & Jusczyk, P. W. (2003). Infants' long-term memory for the sound patterns of
words and voices. Journal of Experimental Psychology: Human Perception and
Performance, 29(6), 1143.
Hsiao, F. and Gibson, E. (2003). Processing relative clauses in Chinese. Cognition, 90(11), 3–27
Huang, Y. T., Hopfinger, J., & Gordon, P. C. (2013). Distinguishing lexical-versus discourse-
level processing using event-related potentials. Memory & cognition, 1-17.
Hudson-D’Zmura, S. & Tanenhaus, M. K. (1998). Assigning antecedents to ambiguous
pronouns: The role of the center of attention as the default assignment. In M. Walker, A.
Joshi, & E. Prince (Eds.), Centering theory in discourse (pp. 199-226). Oxford: Oxford
University Press.
Innocenti, A., De Stefani, E., Sestito, M., & Gentilucci, M. (2014). Understanding of action-
related and abstract verbs in comparison: a behavioral and TMS study. Cognitive
processing, 15(1), 85-92.
Jacobs, S. (2002). Language and interpersonal communication. In M. L. Knapp & J. A. Daly
(Eds.), Handbook of interpersonal communication (pp. 213–239). Thousand Oaks, CA:
Sage.
Jaeger, T. F., & Snider, N. (2008). Implicit learning and syntactic persistence: Surprisal and
cumulativity. In The 30th Annual Meeting of the Cognitive Science Society
(CogSci08) (p. 827).
206

Järvikivi, J., van Gompel, R. P., Hyönä, J., & Bertram, R. (2005). Ambiguous Pronoun
Resolution Contrasting the First-Mention and Subject-Preference
Accounts. Psychological Science, 16(4), 260-264.
Järvikivi, J., van Gompel, R. P., Hyönä, J., & Bertram, R. (2005). Ambiguous Pronoun
Resolution Contrasting the First-Mention and Subject-Preference
Accounts. Psychological Science, 16(4), 260-264.
Just, M. A., & Carpenter, P. A. (1992). A capacity theory of comprehension: individual
differences in working memory. Psychological review, 99(1), 122.
Kable, J. W., Lease-Spellmeyer, J., & Chatterjee, A. (2002). Neural substrates of action event
knowledge. Journal of Cognitive Neuroscience, 14(5), 795–805.
Kaiser, E. (2003). The quest for a referent: A crosslinguistic look at reference resolution. Ph.D.
dissertation, University of Pennsylvania, Philadelphia, PA.
Kaiser, E. (2006). Effects of topic and focus on salience. In Proceedings of Sinn und
Bedeutung (Vol. 10, pp. 139-154).
Kaiser, E. (2010). Investigating the Consequences of Focus on the Production and
Comprehension of Referring Expressions. International Review of Pragmatics, 2(2), 266-
297.
Kaiser, E. (2011). Focusing on pronouns: Consequences of subjecthood, pronominalisation, and
contrastive focus, Language and Cognitive Processes, 26, 1625-1666.
Kaiser, E., & Trueswell, J. C. (2008). Interpreting pronouns and demonstratives in Finnish:
Evidence for a form-specific approach to reference resolution. Language and Cognitive
Processes, 23(5), 709-748.
207

Kamide, Y., Scheepers, C., & Altmann, G. T. (2003). Integration of syntactic and semantic
information in predictive processing: Cross-linguistic evidence from German and
English. Journal of psycholinguistic research, 32(1), 37-55.
Kang, B. & Kim, H. (2009). Korean word frequencies. Seoul, Korea: Hankukmunhwa Press.
Kathryn Bock, J., & Irwin, D. E. (1980). Syntactic effects of information availability in sentence
production. Journal of verbal learning and verbal behavior, 19(4), 467-484.
Keenan, J. M., Baillet, S. D., & Brown, P. (1984). The effects of causal cohesion on
comprehension and memory. Journal of Verbal Learning and Verbal Behavior, 23, 115-
126.
Kehler, A. (2002). Coherence, reference and the theory of grammar. Stanford, CA: CSLI.
Kensinger, E. A., & Corkin, S. (2003). Memory enhancement for emotional words: Are
emotional words more vividly remembered than neutral words? Memory &
cognition, 31(8), 1169-1180.
Kieras, D. E. (1978). Good and bad structure in simple paragraphs: Effects on apparent theme,
reading time and recall. Journal of Verbal Learning and Verbal Behavior, 17, 13-28.
Kieras, D. E. (1980). Initial mention as a signal to thematic content in technical passages.
Memory and Cognition, 4, 345–353.
Kieras, D. E. (1981). Topicalization effects in cued recall of technical prose. Memory and
Cognition, 9, 541-549.
Kim, S. I., Lee, J. H., & Gernsbacher, M. A. (2004). The advantage of first mention in Korean
the temporal contributions of syntactic, semantic, and pragmatic factors. Journal of
psycholinguistic research, 33(6), 475-491.
208

King, J., & Just, M. A. (1991). Individual differences in syntactic processing: The role of
working memory. Journal of memory and language, 30(5), 580-602.
Kintsch, W. (1972). Notes on the structure of semantic memory. In E. Tulving & W. Donaldson
(Eds.), Organization of memory. New York: Academic Press.
Kintsch, W. (1974). The representation of meaning in memory. Hillsdale, NJ: Erlbaum.
Kintsch, W. (1977). Memory and Cognition. New York: Wiley.
Kintsch, W. (1988). The role of knowledge in discourse comprehension: a construction-
integration model. Psychological review, 95(2), 163.
Klein, D. E., & Murphy, G. L. (2001). The representation of polysemous words. Journal of
Memory & Language, 45(2), 259–282.
Klin, C. M., Weingartner, K. M., Guzmán, A. E., & Levine, W. H. (2004). Readers’ sensitivity to
linguistic cues in narratives: How salience influences anaphor resolution. Memory &
Cognition, 32(3), 511-522.
Koornneef, A. W., & Van Berkum, J. J. (2006). On the use of verb-based implicit causality in
sentence comprehension: Evidence from self-paced reading and eye tracking. Journal of
Memory and Language, 54(4), 445-465.
Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica, 55(3-4),
243-276.
Lambrecht, K. (1996). Information structure and sentence form: Topic, focus, and the mental
representations of discourse referents (Vol. 71). Cambridge university press.
Lapata, M. (2003). Probabilistic text structuring: Experiments with sentence ordering. In
Proceedings of the annual meeting of ACL, 2003.
209

Lau, E. F., Phillips, C., & Poeppel, D. (2008). A cortical network for semantics:(de) constructing
the N400. Nature Reviews Neuroscience, 9(12), 920-933.
Lee, C. (1992). Definite/specific and case marking in Korean. Theoretical issues in Korean
linguistics, 325-341.
Lee, C. (1999). Contrastive topic: a locus of the interface–evidence from Korean and
English. The semantics/pragmatics interface from different points of view, 1, 317-342.
Lee, C. (2003). Contrastive topic and/or contrastive focus. Japanese/Korean Linguistics, 12,
352-364.
Lee, C., Gordon, M. K., & Büring, D. (Eds.). (2006). Topic and focus: Cross-linguistic
perspectives on meaning and intonation (Vol. 82). Springer Science & Business Media.
Levy, R. (2008). Expectation-based syntactic comprehension. Cognition, 106, 1126–1177.
Lewis, R. L., Vasishth, S., & Van Dyke, J. A. (2006). Computational principles of working
memory in sentence comprehension. Trends in cognitive sciences,10(10), 447-454.
Li, P., & Gleitman, L. (2002). Turning the tables: Language and spatial
reasoning. Cognition, 83(3), 265-294.
Liedtke, F., & Schulze, C. (Eds.). (2013). Beyond words: content, context, and inference (Vol.
15). Walter de Gruyter.
Linderholm, T., & van den Broek, P. (2002). The effects of reading purpose and working
memory capacity on the processing of expository text. Journal of Educational
Psychology, 94(4), 778.
Lindsay, S., Scheepers, C., & Kamide, Y. (2013). To dash or to dawdle: Verb-associated speed
of motion influences eye movements during spoken sentence comprehension. PloS
one, 8(6), e67187.
210

Loftus, E. F., & Palmer, J. C. (1974). Reconstruction of automobile destruction: An example of
the interaction between language and memory. Journal of verbal learning and verbal
behavior, 13(5), 585-589.
Loftus, G. R., & Loftus, E. F. (1976). Human memory: The processing of information.
Psychology Press.
Long, D. L., Johns, C. L., & Jonathan, E. (2012). A memory-retrieval view of discourse
representation: The recollection and familiarity of text ideas. Language and Cognitive
Processes, 27(6), 821-843.
MacDonald, M. C. (2013). How language production shapes language form and
comprehension. Frontiers in Psychology, 4, Article 226, 1-16.
MacDonald, M. C., & Thornton, R. (2009). When language comprehension reflects production
constraints: Resolving ambiguities with the help of past experience. Memory &
cognition, 37(8), 1177-1186.
MacDonald, M. C., Pearlmutter, N. J., & Seidenberg, M. S. (1994). The lexical nature of
syntactic ambiguity resolution. Psychological review, 101(4), 676.
MacWhinney, B., & Pléh, C. (1988). The processing of restrictive relative clauses in
Hungarian. Cognition, 29(2), 95-141.
Mahajan, A. (1990). The A/A-Bar Distinction and Movement Theory. PhD dissertation, MIT.
Martin, A., & Chao, L. L. (2001). Semantic memory and the brain: structure and
processes. Current opinion in neurobiology, 11(2), 194-201.
Martin, R.C., & Slevc, L.R. (2014). Language production and working memory. In Goldrick, M.,
Ferreira, V., & Miozzo, M. (Eds.), The Oxford Handbook of Language Production (pp.
120-131). Oxford University Press.
211

Matlock, T. (2004). Fictive motion as cognitive simulation. Memory & Cognition, 32, 1389–
1400.
McCarthy, D. (1933). Language development.
McCloskey, J. (1997). Subjecthood and subject positions (pp. 197-235). Springer Netherlands.
McCutchen, D. (1996). A capacity theory of writing: Working memory in
composition. Educational Psychology Review, 8(3), 299-325.
McDonald, J. L., & MacWhinney, B. (1995). The time course of anaphor resolution: Effects of
implicit verb causality and gender. Journal of Memory and Language, 34(4), 543-566.
McDonald, J. L., Bock, K., & Kelly, M. H. (1993). Word and world order: Semantic,
phonological, and metrical determinants of serial position. Cognitive Psychology, 25(2),
188-230.
McElree, B. (2006). Accessing recent events. Psychology of Learning and Motivation, 46, 155–
200.
McKoon, G., & Ratcliff, R. (1998). Memory-based language processing: Psycholinguistic
research in the 1990s. Annual Review of Psychology, 49(1), 25-42.
McKoon, G., Ratcliff, R., Ward, G., & Sproat, R. (1993). Syntactic prominence effects on
discourse processes. Journal of Memory and Language, 32(5), 593-607.
McKoon, G., Ward, G., Ratcliff, R., & Sproat, R. (1993). Morphosyntactic and pragmatic factors
affecting the accessibility of discourse entities. Journal of Memory and Language, 32(1),
56-75.
McKoon, G., Ward, G., Ratcliff, R., & Sproat, R. (1993). Morphosyntactic and pragmatic factors
affecting the accessibility of discourse entities. Journal of Memory and Language, 32(1),
56-75.
212

McRae, K., Hare, M., Elman, J. L., & Ferretti, T. (2005). A basis for generating expectancies for
verbs from nouns. Memory & Cognition, 33(7), 1174-1184.
Mehler, J. (1963). Some effects of grammatical transformations on the recall of English
sentences. Journal of Verbal Learning and Verbal Behavior, 2(4), 346-351.
Merlo, P., & Stevenson, S. (Eds.). (2002). The lexical basis of sentence processing: Formal,
computational and experimental issues (Vol. 4). John Benjamins Publishing.
Miller, G. A. (1962). Some psychological studies of grammar. American Psychologist, 17(11),
748-762.
Miller, G. A., & Fellbaum, C. (1991). Semantic networks of English. Cognition, 41, 197–229.
Miller, L. M., & Roodenrys, S. (2009). The interaction of word frequency and concreteness in
immediate serial recall. Memory & Cognition, 37(6), 850–865
Mitchell, J., Lapata, M., Demberg, V., & Keller, F. (2010). Syntactic and semantic factors in
processing difficulty: An integrated measure. Proceedings of the 48th Annual Meeting of
the Association for Computational Linguistics, 196–206.
Miyake, A., & Shah, P. (1999). Models of working memory: Mechanisms of active maintenance
and executive control. New York: Cambridge University Press.
Montgomery, J. W. (1995). Examination of phonological working memory in speciŽ caly
languageimpaired children. Applied Psycholinguistics, 16, 355–378.
Montgomery, J. W., & Evans, J. L. (2009). Complex sentence comprehension and working
memory in children with specific language impairment. Journal of Speech, Language,
and Hearing Research, 52(2), 269-288.
Moody, C. L., & Gennari, S. P. (2010). Effects of implied physical effort in sensory-motor and
pre-frontal cortex during language comprehension. Neuroimage, 49(1), 782-793.
213

Most, R. B., & Saltz, E. (1979). Information structure in sentences: New information. Language
& Speech, 22, 89–95.
Murdock Jr, B. B. (1962). The serial position effect of free recall. Journal of experimental
psychology, 64(5), 482.
Myers, J. L., & O'Brien, E. J. (1998). Accessing the discourse representation during
reading. Discourse processes, 26(2-3), 131-157.
Nee, D. E., & Jonides, J. (2008). Neural correlates of access to short-term memory. Proceedings
of the National Academy of Sciences, 105(37), 14228-14233.
Neeleman, A., Titov, E., Van De Koot, H., & Vermeulen, R. (2009). A syntactic typology of
topic, focus and contrast. Alternatives to cartography, 15-52.
Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University of South Florida word
association, rhyme, and word fragment norms. http://www.usf.edu/FreeAssociation/.
Norman, D. A. (Ed.). (2013). Models of human memory. Elsevier.
Norman, D. A., & Bobrow, D. G. (1979). Descriptions: An intermediate stage in memory
retrieval. Cognitive Psychology, 11(1), 107-123.
O'Brien, E. J., Duffy, S. A., & Myers, J. L. (1986). Anaphoric inference during reading. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 12(3), 346.
Öztekin, I., Davachi, L., & McElree, B. (2010). Are representations in working memory distinct
from representations in long-term memory? Neural evidence in support of a single
store. Psychological Science.
Papagno, C., Cecchetto, C., Reati, F., & Bello, L. (2007). Processing of syntactically complex
sentences relies on verbal short-term memory: Evidence from a short-term memory
patient. Cognitive Neuropsychology, 24(4), 292 - 311.
214

Perani, D., Cappa, S. F., Schnur, T., Tettamanti, M., Collina, S., Rosa, M. M., & Fazio, F.
(1999). The neural correlates of verb and noun processing. A PET study. Brain, 122(Pt
12), 2337–2344.
Perfetti, C. A., & Goldman, S. R. (1974). Thematization and sentence retrieval. Journal of
Verbal Learning and Verbal Behavior, 13, 70–79.
Perham, N. Whelpley, C., Hodgetts, H. (2013). Impaired memory for syntactical information in
poor readers. Memory, 21(2), 182-188.
Phillips, C., Kazanina, N., Abada, S. H., (2005). ERP effects of the processing of syntactic long-
distance dependencies. Cognitive Brain Research, 22 (3), 407–428.
Phillips, C., Wagers, M. W., & Lau, E. F. (2011). Grammatical illusions and selective fallibility
in real-time language comprehension. In Runner, J. (Ed.), Experiments at the Interfaces.
Syntax & Semantics, 37 (pp. 147–180). Emerald Group Publishing Limited.
Pickering, M. J., Branigan, H. P., Cleland, A. A., & Stewart, A. J. (2000). Activation of syntactic
information during language production. Journal of Psycholinguistic Research, 29(2),
205-216.
Pickering, M. J., McLean, J. F., & Branigan, H. P. (2013). Persistent structural priming and
frequency effects during comprehension. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 39(3), 890.
Pinker, S. (1994). The language instinct. New York: William Morrow
Pinker, S. (1999). Words and rules: The ingredients of language. Basic Books.
Pinker, S. (2007). The stuff of thought: Language as a window into human nature. Penguin.
Pinker, S., & Jackendoff, R. (2005). The faculty of language: what's special about
it? Cognition, 95(2), 201-236.
215

Polišenská, K., Chiat, S., & Roy, P. (2014). Sentence repetition: what does the task
measure?. International Journal of Language & Communication Disorders, 50(1), 106-
118.
Postman, L., & Phillips, L. W. (1965). Short-term temporal changes in free recall. Quarterly
Journal of Experimental Psychology, 17, 132-138.
Potter, M. C. (1999). Understanding sentences and scenes: The role of conceptual short-term
memory. In V. Coltheart (Ed.), Fleeting memories (pp. 13–46). Cambridge, MA: MIT
Press.
Potter, M. C., & Lombardi, L. (1998). Syntactic priming in immediate recall of
sentences. Journal of Memory and Language, 38(3), 265-282.
Prince, E. F. (1992). The ZPG letter: Subjects, definiteness, and information-status. Discourse
description: diverse analyses of a fund raising text, 295-325.
Pulvermüller, F. (1999). Words in the brain's language. Behavioral and brain sciences, 22(02),
253-279.
Pulvermüller, F., Härle, M., & Hummel, F. (2001). Walking or talking?: Behavioral and
neurophysiological correlates of action verb processing. Brain and language, 78(2), 143-
168.
Qualtrics. http://www.qualtrics.com/academic-solutions/usc
Quirk, R., Greenbaum, S., Leech, G. N., & Svartvik, J. (1972). A grammar of contemporary
English. New York: Seminar Press.
Rappaport, M., Levin, B., & Laughren, M. (1993). Levels of lexical representation (pp. 37-54).
Springer Netherlands.
Ratcliff, R. (1978). A theory of memory retrieval. Psychological review, 85(2), 59-108.
216

Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporary suppression of visual
processing in an RSVP task: An attentional blink? Journal of Experimental Psychology:
Human Perception and Performance, 18(3), 849.
Reinhart, T. (1982). Pragmatics and Linguistics.-An analysis of sentence topics. Bloomington:
Indiana University Linguistics Club.
Roberts, C. (1996). Information structure in discourse: Towards an integrated formal theory of
pragmatics. Working Papers in Linguistics-Ohio State University Department of
Linguistics, 91-136.
Roger L. (2013). Memory and surprisal in human sentence comprehension. In Roger P. G. van
Gompel (Ed.), Sentence Processing (pp. 78–114). Psychology Press, Hove.
Rohde, H., & Kehler, A. (2014). Grammatical and information-structural influences on pronoun
production. Language, Cognition and Neuroscience, 29(8), 912-927.
Sachs, J. S. (1967). Recopition memory for syntactic and semantic aspects of connected
discourse. Perception & Psychophysics, 2(9), 437-442.
Sachs, J. S. (1974). Memory in reading and listening to discourse. Memory & Cognition, 2, 95-
100.
Saito, M. (1985). Some Asymmetries in Japanese and their Theoretical Implications. Ph.D
dissertation, MIT.
Sanford, A. J., & Garrod, S. C. (1981). Understanding written language: Explorations of
comprehension beyond the sentence. New York: Wiley.
Savin, H. B., & Perchonock, E. (1965). Grammatical structure and the immediate recall of
English sentences. Journal of Verbal Learning and Verbal Behavior 4, 348-353.
Seamon, J. G., & Kenrick, D. T. (1994). Psychology. Englewood Cliffs, NJ: Prentice Hall.
217

Service, E. (1992). Phonology, working memory, and foreign-language learning. Quarterly
Journal of Experimental Psychology Section a-Human Experimental Psychology, 45(1),
21-50.
Shapiro, K., & Caramazza, A. (2003). The representation of grammatical categories in the brain.
Trends in Cognitive Science, 7(5), 201–206.
Sidner, C. L. (1981). Focusing for interpretation of pronouns. Computational Linguistics, 7(4),
217-231.
Singer, M. (1976). Thematic structure and the integration of linguistic information. Journal of
Verbal Learning and Verbal Behavior, 15, 549-558.
Singer, M. (2013). Psychology of Language (PLE: Psycholinguistics): An Introduction to
Sentence and Discourse Processes. Psychology Press.
Slatcher, R. B., Vazire, S., & Pennebaker, J. W. (2008). Am “I” more important than “we”?
Couples’ word use in instant messages. Personal Relationships,15(4), 407-424.
Slevc, L. R. (2011). Saying what's on your mind: Working memory effects on sentence
production. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 37(6), 1503.
Song, H. J., & Fisher, C. (2005). Who’s “she”? Discourse prominence influences preschoolers’
comprehension of pronouns. Journal of Memory and Language, 52(1), 29-57.
Song, H. J., & Fisher, C. (2007). Discourse prominence effects on 2.5-year-old children's
interpretation of pronouns. Lingua, 117(11), 1959-1987.
Stevenson, R. J., Crawley, R. A., & Kleinman, D. (1994). Thematic roles, focus and the
representation of events. Language and Cognitive Processes, 9(4), 519–548.
218

Swanson, H. L. (1999). Reading comprehension and working memory in learning-disabled
readers: Is the phonological loop more important than the executive system?. Journal of
Experimental Child Psychology, 72(1), 1-31.
Swanson, H. L. (2003). Age-related differences in learning disabled and skilled readers' working
memory. Journal of Experimental Child Psychology, 85(1), 1-31.
Talmi, D., Grady, C. L., Goshen-Gottstein, Y., Moscovitch, M. (2005). Neuroimaging the serial
position curve: A test of single-store versus dual-store models. Psychological Sciences,
16(9), 716-723.
Tanenhaus, M. K., Spivey-Knowlton, M. J., Eberhard, K. M., & Sedivy, J. C. (1995). Integration
of visual and linguistic information in spoken language
comprehension. Science, 268(5217), 1632-1634.
Tettamanti, M., Buccino, G., Saccuman, M. C., Gallese, V., Danna, M., Scifo, P., ... & Perani, D.
(2005). Listening to action-related sentences activates fronto-parietal motor
circuits. Journal of cognitive neuroscience, 17(2), 273-281.
Tomasello, M. (1992). First verbs: A case study of early grammatical development. Cambridge
University Press.
Towse, J. N., Hitch, G. J., & Hutton, U. (1998). A reevaluation of working memory capacity in
children. Journal of memory and language, 39(2), 195-217.
Towse, J. N., Hitch, G. J., & Hutton, U. (2002). On the nature of the relationship between
processing activity and item retention in children. Journal of Experimental Child
Psychology, 82(2), 156-184.
219

Traxler, M. J., Pickering, M. J., & McElree, B. (2002). Coercion in sentence processing:
Evidence from eye-movements and self-paced reading. Journal of Memory and
Language, 47(4), 530-547.
Trueswell, J. C., Tanenhaus, M. K., & Kello, C. (1993). Verb-specific constraints in sentence
processing: separating effects of lexical preference from garden-paths. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 19(3), 528-553.
Tulving, E. (1984). Precis of elements of episodic memory. Behavioral and Brain
Sciences, 7(02), 223-238.
Tzeng, O. J. (1973). Positive recency effect in a delayed free recall. Journal of Verbal Learning
and Verbal Behavior, 12(4), 436-439.
Ullman, M. T. (2001). A neurocognitive perspective on language: The declarative/procedural
model. Nature reviews neuroscience, 2(10), 717-726.
Ullman, M. T., Corkin, S., Coppola, M., Hickok, G., Growdon, J. H., Koroshetz, W. J., & Pinker,
S. (1997). A neural dissociation within language: Evidence that the mental dictionary is
part of declarative memory, and that grammatical rules are processed by the procedural
system. Journal of cognitive neuroscience,9(2), 266-276.
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(8), 1081-
1087.
Van Dijk, T. A. (1977). Sentence topic and discourse topic. Papers in slavic philology, 1, 49-61.
Van Dijk, T. A., & Kintsch, W. (1983). Strategies of discourse comprehension. New York:
Academic Press.
220

Van Dyke, J. A., & Johns, C. L. (2012). Memory interference as a determinant of language
comprehension. Language and linguistics compass, 6(4), 193-211.).
Van Dyke, J. A., & Johns, C. L. (2012). Memory interference as a determinant of language
comprehension. Language and linguistics compass, 6(4), 193-211.
Vasishth, S., Shaher, R., & Srinivasan, N. (2012). The role of clefting, word order and given-new
ordering in sentence comprehension: Evidence from Hindi. The Mind Research
Repository (beta), (1).
Venditti, J. J., & Hirschberg, J. (2003). Intonation and discourse processing. In Proceedings of
the international congress of phonetic sciences (pp. 315-318). Saarbrücken,, Germany:
University of Saarbrücken.
Vermeulen, R. (2009). Topics, contrast and contrastive topics in Japanese. MIT Working Papers
in Linguistics, 58, 361-372.
Von Ekhardt, B., & Potter, M. C. (1985). Clauses and the semantic representation of
words. Memory & Cognition, 13, 371–376.
Von Stutterheim, C., Andermann, M., Carroll, M., Flecken, M., & Schmiedtová, B. (2012). How
grammaticized concepts shape event conceptualization in language production: Insights
from linguistic analysis, eye tracking data, and memory performance. Linguistics, 50 (4),
833-867.
Vonk, W., Hustinx, L. G., & Simons, W. H. (1992). The use of referential expressions in
structuring discourse. Language and Cognitive Processes, 7, 301–333.
Walker, I., & Hulme, C. (1999). Concrete words are easier to recall than abstract words:
Evidence for a semantic contribution to short-term serial recall. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 25(5), 1256–1271.
221

Walker, M., Joshi, A., & Prince, E. (1998). Centering in naturally occurring discourse: An
overview. In M. Walker, A. Joshi, & E. Prince (Eds.), Centering theory in discourse.
New York: Oxford University Press.
Wallan, A., Chiat, S., & Roy, P. (2011). Evaluation of an Arabic Sentence-Repetition Test for
Preschool Children. Poster presented at the ASHA Convention 2011. San Diego.
Warren, T. & Gibson, E. (2002). The influence of referential processing on sentence complexity.
Cognition, 85(1), 79–112.
Weismer, S. E., Evans, J., & Hesketh, L. J. (1999). An examination of verbal working memory
capacity in children with specific language impairment. Journal of Speech, Language,
and Hearing Research, 42, 1249-1260.
Wickelgren, W. A. (1976). Memory storage dynamics. In W. K. Estes (Ed.), Handbook of
learning and cognitive processes (Vol. 4). Potomac, Md.: Erlbaum.
Willems, R. M., Benn, Y., Hagoort, P., Toni, I., & Varley, R. (2011). Communicating without a
functioning language system: implications for the role of language in
mentalizing. Neuropsychologia, 49(11), 3130-3135.
Wittenberg, E., Jackendoff, R., Kuperberg, G., Paczynski, M., Snedeker, J., Wiese, H., &
Wittenberg, E. (2014). The processing and representation of light verb
constructions. Structuring the Argument: Multidisciplinary research on verb argument
structure, 10, 61-80.
Wyer, R. S., & Srull, T. K. (1980). The processing of social stimulus information: A conceptual
integration. Person memory: The cognitive basis of social perception, 227-300.
Yao, B., & Scheepers, C. (2011). Contextual modulation of reading rate for direct versus indirect
speech quotations. Cognition, 121(3), 447-453.
222

Yongming, C., & Yao, C. (1995). The advantage of first mention and the accessibility of various
constituents in Chinese sentences. Psychological science, 1.
Zhang, S., & Schmitt, B. (1998). Language-dependent classification: The mental representation
of classifiers in cognition, memory, and ad evaluations. Journal of Experimental
Psychology: Applied, 4(4), 375.
Zimmerh, H. D., & Engelkamjp, J. (1981). The given-new structure of cleft sentences and their
influence on picture viewing. Psychological Research, 43, 375-389.
Zwaan, R. A. (1996). Processing narrative time shifts. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 22(5), 1196–1207.
Zwaan, R. A., & Radvansky, G. A. (1998). Situation models in language comprehension and
memory. Psychological bulletin, 123(2), 162.
Zwaan, R. A., & Taylor, L. J. (2006). Seeing, acting, understanding: motor resonance in
language comprehension. Journal of Experimental Psychology: General, 135(1), 1.
Zwaan, R. A., Magliano, J. P., & Graesser, A. C. (1995). Dimensions of situation model
construction in narrative comprehension. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 21(2), 386.
Zwaan, R. A., Stanfield, R. A., & Yaxley, R. H. (2002). Language comprehenders mentally
represent the shapes of objects. Psychological Science, 13 (2), 168-171.

223

Appendix A
List of the occupation nouns used in Experiment 2
pair
occupation
noun in Korean
English translation
raw frequency per
15 million
frequency order
1
선생님
teacher 10781 187
기자
reporter 9795 209
2
작가
writer 5253 427
의사
doctor 3214 751
3
모델
model 1771 1404
스님
Buddhist monk 1295 1866
4
배우
Actor 1440 1697
변호사
lawyer 1275 1891
5
목사
priest 1151 2055
군인
soldier 1005 2287
6
검사
prosecutor 953 2400
가수
singer 898 2544
7
화가
painter 797 2833
약사
pharmacist 757 2952
8
국회의원
congressman 678 3239
( 학원) 강사
(after school)
instructor
482 4225
9
판사
judge 670 3279
농부
farmer 499 4114
10
소설가
novelist 605 3550
경찰관
police officer 496 4138
11
기술자
mechanic 347 5476
정치가
politician 327 5704
12
간호사
nurse 315 5861
외교관
diplomat 268 6579
13
경비원
guard 190 8463
요리사
cook 112 12103
14
디자이너
designer 181 8716
심리학자
psychologist 108 12398
15
한의사
acupuncturist 172 9039
건축가
architect 154 9749
16
작곡가
composer 149 9989
웨이터
waiter 111 12178
17
승무원
flight attendant 102 12888
만화가
cartoonist 100 13061
18
리포터
reporter 74 15784
회계사
CPA 59 18144
19
개그맨
comedian 49 20274
224

소방관
fire fighter 37 23831
20
미용사
hair dresser 40 22839
배달원
deliverer 32 25830
21
성우
voice actor 57 18533
댄서
dancer 19 34020
22
마술사
magician 56 18738
경호원
body guard 55 18965
23
사진사
photographer 55 18965
번역가
translator 27 28318
24
영양사
nutritionist 24 30125
청소부
genitor 59 18144

225

Appendix B
List of the verb phrases used in Experiment 4

target phrase
Korean English
ave.
freq.
SD telicity
face
verb?
low
expected
frequency
군침을 삼키다
swallows nervously 1.49 0.69
telic F
트림을 하다
burps 1.54 0.90
telic F
초인종을 누르다
rings a door bell 1.59 0.64
atelic NF
하품을 하다
yawns 1.73 0.99
telic F
눈을 깜빡이다
blinks 1.81 0.66
telic F
코를 풀다
blows nose 2.30 0.85
atelic F
노크를 하다
knocks on a door 2.35 0.54
atelic NF
재채기를 하다
sneezes 2.38 1.01
telic F
기침을 하다
coughs 2.65 1.18
telic F
고양이가 야옹하고
울다
a cat meows 2.68 1.53
telic F
고개를 끄덕이다
nods 2.84 1.21
telic F
목청을 가다듬다
clears throat 2.92 3.62
atelic F
high
expected
frequency
공을 뜅기다
bounces a ball 3.65 1.81
atelic NF
접시를 문질러 닦다
scrubs a plate 4.03 2.67
atelic NF
손벽을 치다
박수를 치다
claps
(used twice in two different
forms)
4.14 1.73
telic NF
킬킬하고 웃다
chuckles 4.41 2.50
telic F
자신을 등을 밀다
scrubs one's own back 4.59 3.43
atelic NF
개가 꼬리를 흔들다
a dog wags tail 6.24 5.35
atelic NF
딸국질을 하다
hiccups 8.73 8.08
telic F
음식을 씹다
chews food 9.24 5.89
atelic F
물장구를 치다
kicks water 18.57 9.00
telic NF
윗몸일으키기를 하다
do sit-ups 35.90 18.54
atelic NF
팔굽혀펴기를 하다
do push-ups 18.70 21.03
atelic NF
F = face-oriented verb; NF = non face-oriented verb

Abstract (if available)

Abstract Language and memory are two of the most fundamental components of human cognition. Individuals’ linguistic performance relies greatly on memory competence. At the same time, linguistic expressions can also influence what comprehenders remember and how well they remember it. Due to the close connection between these two cognitive domains, it is important to study how linguistic descriptions can influence the representations that language users build in their minds and how those representations are reflected in discourse and memory. ❧ This dissertation investigated the interplay between language and memory in the areas of entity representation denoted by nouns and event information denoted by verbs. Combining research on nominal information and verbal information is beneficial for our understanding of language processing as they are two vital aspects of sentence meaning. ❧ In the nominal domain, I conducted three experiments to explore the effect of Korean subject case marking, more specially the difference between the nominative case marker and the topic case marker (when interpreted as marking a contrastive topic). My results show that how subjects are interpreted (e.g., a default topic or a contrastive topic) has an effect on comprehenders’ expectations about which entities will be mentioned in upcoming discourse. ❧ Furthermore, the experiments also show that the order in which characters are introduced in a story influences how well comprehenders recognize them in a memory task (i.e., a recently-mentioned entity is recognized more accurately than an earlier-mentioned entity). However, a follow-up off-line story continuation task revealed that the entity that comprehenders remembered better is not the entity they talk about frequently in story continuation. Rather, the first-mentioned character was mentioned more often in subsequent sentences. Put together, my results point to a possible disconnect between accessibility in discourse and accessibility in memory: An entity’s prominence in discourse and memory are not always parallel. This suggests that prominence is a multifaceted concept that can be a product of many different factors including: (i) how and where in the discourse an entity is introduced

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Processing the dynamicity of events in language

PDF

When things are left unsaid: existential and anaphoric implicit objects in discourse

PDF

Narrowing the focus: experimental studies on exhaustivity and contrast

PDF

Discourse-level processing of nominal possessive constructions

PDF

Towards the unity of movement: implications from verb movement in Cantonese

PDF

Knowledge and use of evidentials in L1 and L2 Korean

PDF

The notion of topic-comment constructions and the meaning of the Korean topic marker '-(n)un'

PDF

Interpretation of pronominal forms across languages

PDF

Towards a correlational law of language: three factors constraining judgement variation

Asset Metadata

Creator Kim, Lucy Kyoungsook (author)

Core Title Exploring the effects of Korean subject marking and action verbs’ repetition frequency: how they influence the discourse and the memory representations of entities and events

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Linguistics

Publication Date 07/28/2015

Defense Date 05/21/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag action verb’s repetition frequency,entity’s discourse representation,entity’s memory representation,Korean subject case marking,memory recall,OAI-PMH Harvest,sensicality judgment,sentence processing,story continuation,verb semantics

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Kaiser, Elsi (committee chair), Biederman, Irving (committee member), Pancheva, Roumyana (committee member), Simpson, Andrew A. (committee member)

Creator Email kyoungsk@usc.edu,lucykimmy@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-609369

Unique identifier UC11301295

Identifier etd-KimLucyKyo-3713.pdf (filename),usctheses-c3-609369 (legacy record id)

Legacy Identifier etd-KimLucyKyo-3713.pdf

Dmrecord 609369

Document Type Dissertation

Format application/pdf (imt)

Rights Kim, Lucy Kyoungsook

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA