Citation
Verbal and visuo-spatial processing demands in writing

Material Information

Title:
Verbal and visuo-spatial processing demands in writing
Creator:
Lea, Joseph D., 1970-
Publication Date:
Language:
English
Physical Description:
v, 85 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Business executives ( jstor )
Cognitive models ( jstor )
Cognitive psychology ( jstor )
Experimental psychology ( jstor )
Memory ( jstor )
Modeling ( jstor )
Phonology ( jstor )
Working memory ( jstor )
Writing exercises ( jstor )
Writing processes ( jstor )
Dissertations, Academic -- Psychology -- UF ( lcsh )
Psychology thesis, Ph. D ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1998.
Bibliography:
Includes bibliographical references (leaves 77-84).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Joseph D. Lea.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
029548063 ( ALEPH )
41456389 ( OCLC )

Downloads

This item has the following downloads:


Full Text








VERBAL AND VISUO-SPATIAL PROCESSING DEMANDS IN WRITING


By
JOSEPH D. LEA




















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA
1998














ACKNOWLEDGEMENTS


I would like to thank my chair, Dr. Michael Levy, for all of the time and effort he

has given to this project and for all of the support he has given me over the years. I would

also like to thank my other committee members, Dr. Ira Fischler, Dr. Stephen Blessing,

Dr. Bart Weitz, and Dr. Shari Ellis for their time and suggestions which have greatly

improved this project. In addition, I would like to thank Dr. Lise Abrams for attending

the defense, and for her many helpful comments and suggestions. I would also like to

take this opportunity to thank my classmates Pam Marek, Brian Howland, Nancy

Lincoln, Michele Simmons, Ryan West, Katherine White, Glenn Gunzelman, and Dean

Sabatinelli for their camaraderie. Special thanks to IBM's Santa Teresa Laboratory for

the incentive to finish this project on schedule. Finally, I would like to thank my wife,

Elena Schuhmann, my parents, and my grandfather for all of their support and

encouragement.














TABLE OF CONTENTS




AKNOWLEDGEMENTS .................................................................. ii

A B ST R A C T ................................................................................. v

CHAPTERS

1 INTRODUCTION .................................................................. 1
Resources and. Structures .............................................. 4
Capacity Theories ....................................................... 5
Structural Theories ..................................................... 8
Multiple Resource Theories ............................................ 12
Working Memory ....................................................... 15
Working Memory in Writing .......................................... 20
Adapted Working Memory Model .................................. 29
Sum m ary ........................................................... .... 32

2 EXPERIMENT 1 ........................................................ 33
M eth o d .................................................................... 3 5
Results and Discussion ................................................. 36

3 EXPERIMENT 2 ....................................................... 39
Experiment 2A ..................................... 39
M etho d ................................................................... 44
Results and Discussion ................................................. 45
EXPERIMENT 2B ..................................................... 48
M ethod .................................................................... 50
Results and Discussion ................................................. 52
Combined Analysis of Experiments 2A and 2B ................... 54

4 EXPERIMENT 3 ........................................................ 58
M etho d .................................................................... 59
Results and Discussion ................................................. 61

5 GENERAL DISCUSSION ........................................... 65










APPENDIX
REFERENCES
BIOGRAPHICAL
SKETCH


.......... I ... I ......................................................
... ...... ...... ... .................. ....... ............

........................................................................














Abstract of Dissertation Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

VERBAL AND VISUO-SPATIAL PROCESSING DEMANDS IN WRITING

By

Joseph D. Lea III

August, 1998

Chairperson: C. Michael Levy
Major Department: Psychology

Writing research has recently begun to consider the role that limitations in

cognitive capacity play in the writing process, thus obligating writing researchers to

examine existing theories of cognitive limitations. This dissertation reviews four major

classes of cognitive processing models that might serve as theoretical bases for extension

to written language production. Four experiments assess the merits of a

reconceptualization of Kellogg's (1996) model of working memory and writing by

emphasizing the role of resource pools rather than binary functioning cognitive

structures. Empirical findings support the recent trend toward multicomponent classes of

working memory or resource models, particularly those that distinguish between visuo-

spatial and verbal/phonological processing. Results also indicate that there is a significant

visuo-spatial component in the planning stages of writing.














CHAPTER 1
INTRODUCTION

Writing research has recently begun to consider the role that limitations in

cognitive capacity play in the writing process. A common assumption is that writers have a

limited cognitive capacity, the size and/or efficiency of which varies between individuals.

Research on cognitive capacity and writing is relatively new compared to studies of

capacity limitations in reading, and to spoken language production and comprehension

(see Gathercole and Baddeley, 1993 for an overview), obligating writing researchers to

develop and refine appropriate models necessary for theoretical development and testing.

Fortunately, the nature of human processing and storage limitations has long interested

psychologists working within an information processing framework, and accounts of

other language production tasks have led to the development of constructs such as

working memory, resource capacity, and mental workload.

In part, the contribution of this dissertation lies in the extension of debates in the

working memory and attention literature to the nearly untouched domain of writing. As

both a thinking and language production task, writing research may challenge these

existing models. Four approaches to cognitive limitations are introduced below, followed

by a critical analysis of information processing theory development, and rationale for the

formation of a hybrid model that the employs elements of these models with the greatest

explanatory power for written language production tasks.









" Structural theories were popularized in the 1950s with experimental investigations of

the dichotic listening task which revealed that attention was severely limited (Cherry,

1953; Broadbent, 1958; Moray, 1959). Several classes of theoretical models were

generated to localize the point in information processing where the bottleneck occurs.

Early-selection theories claimed that this bottleneck is in the early processes such as

perception, and late-selection theories pointed to stages of decision making and

response selection as the limiting stages of information processing.

" Early human factors research in the measurement of mental workload led to the

conceptualization of the human operator possessing a pool of limited-capacity

resources (Knowles, 1963), This resource capacity approach offers the simplest

framework for interpreting the demands placed on the writer by postulating an

undifferentiated pool of mental resources (Kahneman, 1973). Performance on any

given task declines only when the aggregate of task demands exceed the capacity of

resources.

" An alternate conception of this resource economy makes a distinction in the type of

resources available at any given moment. This "multiple resource" approach posits that

there may be various types of resources and that different tasks may require different

types of resources in various compositions. The most widely supported division of

resources is a neurologically plausible division between visuo-spatial and verbal

resources (Wickens, 1984; Friedman & Poison, 1981). Friedman and Polson (1981)

proposed a framework with the intention of extending multiple resource models to our

understanding of cerebral specialization phenomena. They suggest that the physically

distinct cerebral hemispheres can be seen as independent resource systems, and that









the left and right hemispheres together form a system of two mutually inaccessible and

finite pools of resources.

* Baddeley & Hitch's (1974) model of working memory is a multi-component system

for both the storage and the processing of verbal and visuo-spatial information. It has

three primary components: a limited-capacity central executive and two peripheral

slave systems. One of these slave systems, the phonological loop, handles verbal and

auditory information while the other, the visuo-spatial scratchpad, is specialized for

processing visual and spatial information.

Most approaches to cognitive processing limitations recognize a division in

processing for verbal and visuo-spatial tasks, owing to the vast literature on hemispheric

differences. Since the doctrine of cerebral localization in the late 1700's, particular

cognitive functions have been assigned to specific regions of the brain. Generally, visuo-

spatial processing has been localized to the right hemisphere while the left hemisphere is

believed to play a stronger role in verbal tasks. Popular multiple resource models

(Wickens, 1984; Friedman & Polson, 1981) and Baddeley and Hitch's (1974) working

memory model build on this neurological separation of visuo-spatial and verbal

processing.

As a writer plans ideas, translates ideas into prose, types or handwrites sentences,

and monitors all of these activities, many demands are placed on temporary storage and

processing capacity. This dissertation seeks to provide an account of the division of labor

between visuo-spatial and verbal processing demands and capacity limitations as they

influence the writing process. Special treatment will be given to the wholly untouched

question of how visuo-spatial resources constrain writing. The balance of this chapter









critically reviews the literature regarding the models introduced above, focusing on their

assumptions about the nature of human cognitive limitations. The review concludes with a

summary and critique of existing literature, followed by a proposal for an adapted theory

of working memory which is tested in Experiment 1. Experiments 2a and 2b compare the

involvement of phonological and visuo-spatial processing in essay writing, and Experiment

3 investigates the role that writing topic plays in the resource demands of writing.

Resources and Structures

Two hypothetical constructs have greatly influenced theory development

addressing cognitive limitations. First, we will consider the notion of limited processing

resources that are demanded by non-automatic tasks. According to resource theories,

performance declines only when the aggregate of task demands exceed the available

resources. For example, when someone uses a word processor to translate an idea into

text, we might expect that simultaneously generating new ideas and editing grammatical

and spelling mistakes would cause a disruption in the rate at which text is generated.

Juggling that writing task with talking on the telephone or listening to the radio would

further drive down performance. A second construct proposed to explain processing

efficiency is referred to as structure. According to a structural view (Pashler &Johnston,

1989), when people attempt to simultaneously engage in competing mental operations,

incompatible stages or processes are initiated, which results in interference between tasks.

In other words, two tasks compete when they each simultaneously require a single

common process. For example, we would expect interference between two tasks such as

conversing on the telephone while simultaneously listening to a radio interview, more so

than the dual-task combination than carrying on a conversation while riding a bike,






5


presumably because similar processes are demanded by producing and comprehending

audible speech.

Two classes of theoretical models of attention have emerged from consideration of

these hypothetical variables. These are capacity or resource theories, described by

Kahneman (1973), and structural theories which began with Broadbent (1958) and

recently have been elaborated by Pashler and his associates (Pashler & Johnston, 1989;

Pashler & Carrier, 1995). After considering structural and capacity theories in more

detail, we will examine multiple resource models and Baddeley's working memory model

as each offer a rapprochement of structural and capacity theories. Further, Experiment 1

will reveal empirical findings suggesting that writing research may be best guided by a

model that combines the strengths of both working memory and multiple resource

theories.

Capacity Theories

How is it that we may become less conscious-rather than become entirely unaware

- of some things when we become more conscious of others? Capacity theories seek to

explain this common experience by suggesting that some resource pool is approaching its

limitation. Thus, when a primary task demands more of these resources, fewer are

available for a concurrent secondary task. Whereas structural theories assume that

structures are dedicated to one task at a time, the resource view assumes that capacity can

be allocated between separate activities in graded quantity.

Nearly all of the original paradigms investigating the nature of resources (or mental

capacity or effort) involve presenting two tasks to an individual to process simultaneously,

or so-called dual-task conditions. Resource theories in large part were created to explain








instances where two processes were simultaneously activated. Perhaps the earliest task

(although not a traditional dual-task) was the Stroop task where reading processes were

observed to compete with color naming processes for a common pathway. Although

theorizing about pathways has evolved into structural theories of attention, the basic

question about mechanisms of interference is the same. In the 1950s British psychologists

Broadbent, Moray, and others began working with the dichotic listening task, to test

various structural models of the selection of attention; for example, late versus early

selection. The probe technique emerged shortly thereafter, where participants respond to a

discrete stimulus while performing some primary task. A participant's baseline reaction

time was subtracted from his or her reaction time to a tone presented during performance

of a primary task in order to compute an interference reaction time, which was assumed to

be an index of mental effort.

Kahneman (1973) brought unity to the concept of capacity as an intervening

variable in dual-task performance. His ideas and synthesis of the research at that time

facilitated the evolution of the resource metaphor from an intuition to a quantitative theory

with testable predictions. Since the early 1970s, capacity theories have continued to

influence basic and applied research. Capacity theories are especially important in applied

research, where workload measurement is investigated. Capacity theories have been

applied to aircraft cockpit design (Rolfe, 1971), the failure detection of autopilot systems

(Wickens & Kessel, 1980), target tracking systems used by military pilots (Fracker &

Wickens, 1989), and in the design of visual displays requiring the integration of

information from several sources (Yee, Hunt, & Pellegrino, 1991; Boles & Wickens,

1987). The notion of resources has also played a role in the more basic pursuits of








learning, memory, and writing research. Researchers studying automaticity have also

benefited from the resource conceptualization of attention. As a task becomes more

automatic, fewer resources are needed for its successful completion (Laberge, 1973).

Despite the explanatory power over dual-task performance and success in applied

settings, capacity theories have received some criticism. Navon (1984) argued that most

effects interpreted in terms of resources can also be accommodated well by theories that

do not assume any limit on resources and by theories in which the resources construct

does not exist at all. Navon describes a diverse range of phenomenon that are typically

discussed in terms of resources, and takes issue with how necessary resource terminology

is for dealing with these phenomena. Manipulations of motivation level through payoff

schemes may be interpreted in the resource framework as regulating the amount of

resources allocated to the task. Navon argues that resources are not necessary to explain

the effect of motivation though, and he suggests that processing may be modulated

without any change in the supply of resources. In manipulations of task difficulty, the

more difficult a task, the more resources are said to be needed by that task. Navon

suggests that other factors such as poor luminance contrast might impair the quality of

sensory data available to a process independently of resources. Manipulations of task

complexity are typically explained by the notion that more complex tasks require more

resources. Navon argues that independent of resources, more complex tasks may require

more steps or mental operations, thus presenting more opportunity for errors. Dual-task

deficits are typically thought to degrade the performance of the primary task by curtailing

the amount of resources in its service. Navon points to other possible sources of

interference such as structural interference, some form of cross talk similar to that in









parallel phone lines, or interfering patterns of activation in which inhibition necessary for

one task disrupts activation on concurrent task.

In defense of resource theories, some researchers (Herdman & Friedman, 1985;

Wickens, 1976) claim that distinguishing between resource and structure theories may be

accomplished by manipulation of task priority in dual-task conditions. If performance can

be traded off from primary to secondary tasks and vice versa, then their interpretation is

that the two tasks interfere because the are consuming common resources. If the two tasks

compete for a common structure then the expectation is that performance will not trade

off between the two tasks in a gradual manner. Navon (1984) counters by arguing that

even single structural mechanisms could be timeshared between tasks by some schedule of

queuing or alternation.

Despite continuing debate, the resources construct continues to influence

information processing theory development and human factors concerns, and is at the very

least a useful metaphor. It seems likely that resource theories will continue to flourish as

long as continue to be useful in applied psychology.

Structural Theories

Early studies of the limitations of attention often involved the dichotic presentation

of verbal material. Subjects typically wore stereo headphones, and different messages were

simultaneously presented to each ear. Participants attended to one of the two messages

and repeated it aloud immediately after hearing it (shadowing). This technique ensured

that the participants focused on one of the two messages. Experimental investigations of

the dichotic listening task revealed that attention was severely limited (Cherry, 1953;

Broadbent, 1958; Moray, 1959). Several classes of theoretical models were generated to








localize the point in information processing where the bottleneck occurred. Early-selection

theories claimed that the bottleneck was in the early processes such as perception, and

late-selection theories pointed to stages of decision making and response selection as the

limiting stages of information processing.

Another paradigm was also often used in the investigation of the limitations of

attention. The psychological refractory period (PRP) paradigm (Bertelson, 1967; Welford,

1967), also known as the overlapping tasks paradigm (Pashler, 1984), involves the

presentation of two stimuli (S I and S2) in rapid succession. Participants make a response

to each stimulus (RI and R2) as quickly as possible. Using this paradigm, many

researchers (Kantowitz, 1974; Pashler, 1984; Pashler & Johnston, 1989; Fagot & Pashler,

1992) concluded, as did the late-selection theorists, that limitations in processing were

localized in the response initiation phase.

A subsequent extension of this structural model by the late-selection theorists

postulated that there is not a single stage or mental operation that acts as the source of

interference. Instead, a limited-capacity central processor, when engaged by one task, is

unavailable to a second task requiring the same operation. Thus, the performance of the

second task will necessarily deteriorate. By suggesting that there are a number of

operations that require the exclusive attention of the limited-capacity central processor in

order to proceed, this view permits more than one bottleneck in the information

processing system. Restrictions in the ability to carry on multiple mental computations

may occur at a variety of stages, including perceptual identification, decision and response

selection, response initiation and execution (Pashler & Johnston, 1989), and motor

mechanisms (Kahneman, 1973).









Broadbent (1982) criticized studies of continuous performance that rely on

accuracy measures because they cannot discriminate between the predictions of capacity

theory regarding simultaneous mental processing of both tasks and the predictions of

structural models regarding strategy switching. Pashler and Johnston (1989) incorporated

Broadbent's suggestions when they investigated dual-task interference with the PRP. They

argued that this paradigm potentially provides much more detailed information about the

time course of dual-task interference than is obtained in continuous dual-task studies.

Pashler and Johnston divided models of dual-task interference into two categories:

1. capacity sharing models discussed earlier, and 2. "postponement models" which are

structural in nature. In postponement models, some bottleneck stage or process cannot

operate simultaneously for each of two overlapping tasks. As a result, processing of this

stage in the second task is literally postponed. This produces the relatively straightforward

hypothesis that as stimulus onset asynchrony (SOA) between S I and S2 is reduced, there

should come a point at which any further reduction in the SOA produces a corresponding

increase in the duration of R2. However, even when this prediction is confirmed, it does

not seem to contradict predictions from the capacity theory.

Kahneman (1973) addressed the notion that structural theories and a limited

capacity theory may account for the same findings. He observed that the assumptions of

structural theories are more restricting than the limited capacity models, and that

predictions, especially for the PRP, have typically failed to be confirmed. Pashler and

Johnston concede that the PRP paradigm has been studied somewhat less of late perhaps

reflecting Kahneman's early concerns.








A major finding that has frequently been observed with the overlapping tasks

paradigm is a slowing of the Rirelative to single-task performance (Kahneman, 1973).

Similar findings in discrete movement research are thought to indicate the increased effort

needed to compile a more complex movement (Henry & Rogers, 1960). R2 slowing is

readily accounted for with the capacity models because both tasks are assumed to be

performed with depleted allocations of capacity. Postponement models, in contrast, do not

predict this slowing directly. It has been suggested that RI slowing may result from a

grouping strategy in which the subject essentially treats S 1 and S2 as a compound stimulus

and selects a corresponding compound response (Pashler & Johnston, 1989). Other

strategies have been proposed as well, but in any case, RI slowing by itself is not thought

to be especially diagnostic of the underlying causes of dual-task interference.

Because both capacity and structural models can provide a plausible account of the

basic results, more analytic tests that make distinctive, non-obvious predictions are

necessary. Pashler (1984) used a chronometric approach to develop a method for testing

postponement models inspired by additive factors logic (Steinberg, 1969). By

manipulating task factors that increased or decreased the duration of selected stages of

processing and looking for effects that were additive (or underadditive), Pashler was able

to test hypotheses about the locus of the singe-channel bottleneck. Pashler reported a

preliminary study using this methodology that found underadditive effects that were

consistent with the predictions from a postponement model, and could not be explained

with capacity theory. This study supported the late-selection class of theories, suggesting

that a bottleneck exists either in the decision or response selection stages or in both stages.








Postponement models predict that effects at stages prior to the bottleneck should

become progressively smaller at shorter SOAs. As SOA shortens, the wait for the

processor is thought to become progressively longer, and the proportion of trials on which

delay in early stimulus processing of S2 will have any effect on RT2 progressively

decreases. Although Pashler (1984) only used one SOA and could not test this prediction,

Pashler and Johnston's findings (1989) lend stronger support to the postponement model

with a bottleneck at the stages of decision making and/or response selection. In particular,

results indicating underadditivity present a problem for capacity theory as capacity models

predict that dual-task slowing of RT2 should interact overadditively with factors that

increase the difficulty of the second task.

There seems to be a general acknowledgement in the dual-task literature that

structural theories have very little empirical support outside of isolated laboratory tasks.

Structural theories have played a large role in studies of cognitive limitations, particularly

in their focus on mental stages. It is interesting in a historical sense that structural

assumptions have been carried over into the Baddeley and Hitch (1974) model of working

memory that will be discussed in detail later. Baddeley and Hitch, after all, were trained in

Broadbent's lab where much of the early structural theorizing and testing took place.

Multiple Resource Theories

A search for a compromise to the structure/capacity debate has led directly to

multiple resource models (Navon & Gopher, 1979) where resources are declared to reside

within structures or pools. Until multiple resource models emerged, interference was

thought to depend exclusively on either the extent that two tasks draw on a common

resource pool or structural limitations where some stage of processing cannot proceed in









parallel with another. Neither of these accounts explains all of the phenomena related to

interference though. The central capacity notion cannot explain why some secondary

tasks interfere more with one primary task than another equally difficult primary task. For

example, vocal responses interfere more than spatial responses with recall of a sentence,

but less than spatial responses with recall of a line diagram (Brooks, 1968). Pure structural

models are also inadequate because processes that share similar stages or mechanisms may

interfere with each other, but they seldom block each other completely (Navon & Gopher,

1979). For example, when Triesman and Davies (1973) presented participants with stimuli

simultaneously to the same modality, performance was impaired, but not entirely impeded.

Multiple resource theory posits that there may be various types of resources and

different tasks may require different types of resources in various compositions resources

(Norman & Bobrow, 1975; Navon & Gopher, 1979; Wickens, 1984). On the one hand

this approach is structural in the sense that it identifies the source of interference as

overlapping mechanisms. On the other hand, multiple resource theory is a capacity

approach, because it does not assume that a mechanism can be accessed and used by only

one process at a time. Rather, mechanisms have a capacity that can be shared by several

processes until those processes demand more resources than the capacity allows. Thus,

multiple resource theory is a marriage of capacity and structural models where structure is

important, but each structure is limited by its own capacity.

Wickens (1984) proposed a multiple resource model that accounts for a great deal

of the empirical findings from dual-task studies. This model assumes that the extent to

which two tasks share common resource pools determines how much they will interfere in

a dual-task paradigm. In particular, the resource architecture of Wickens's model contains








three dimensions derived from traditional psychological dichotomies: (a) Based on the

general finding that it is easier to time-share an auditory and visual task than two auditory

or two visual tasks, Wickens proposed a separation of resource pools specialized for

dealing with different processing modalities. (b) The processing codes dimension

distinguishes information that is spatial or analog from that which is verbal or linguistic.

(c) Wickens proposed a separation of processing stages based on the finding that tasks

demanding either response processes or cognitive/perceptual processes will interfere with

each other to a greater extent than will a perceptual and a response task.

Friedman and Polson (1981) proposed a framework with the intention of extending

the resource metaphor to our understanding of cerebral specialization phenomena.

Essentially a special case of a multiple resources model of limited-capacity, they suggest

that the physically distinct cerebral hemispheres can be seen as independent resource

systems. Specifically, the left and right hemispheres together form a system of two

mutually inaccessible and finite pools of resources. This conceptualization differs from

other multiple resource models in that it is based, at least grossly, on the physical structure

of the brain. In some respects this anatomical basis goes beyond the intent of earlier

multiple resource theories, while at the same time taking a step back to the most clearly

supported resource differentiation of all, the distinction of processing which is coded

verbally from visuo-spatially coded processing. Ultimately however, their theory, like

other multiple resource models, seeks to provide insight into the mechanisms that might be

responsible for those patterns of task interference not easily explained by models

postulating a single pool of undifferentiated resources.








Although the existence of multiple resource pools appeals to many researchers

(Wickens, 1984; Friedman & Poison, 1981; Norman & Bobrow, 1975), others are

skeptical (Navon, 1984). Navon and Gopher (1979) noted that multiple resource theory

might leave researchers disconcerted by the prospect of devaluation of the precious time-

honored concept of attention and that the proliferation of resources might seem strange

or threatening. Those who favored the more conservative central capacity theory argued

that multiple resource theory is not logically falsifiable and that it is impossible to know

how many resource pools there are. Therefore, new resource pools could be postulated to

account for any pattern of results. Multiple resource theory, as a class of models, may not

need to be falsifiable to be worthwhile. Rather than struggling to find a critical test for the

entire approach, researchers might do better by focusing on generating and testing specific

multiple resource models that delineate the composition of resources.

Working Memory

Like the multiple resource approach, Baddeley and Hitch's (1974) model of

working memory is another hybrid of structural and capacity models. Instead of resources

within structures, this model implies both resource and structural components. The

model's hybrid nature often goes overlooked, perhaps because working memory has been

primarily connected with higher level language tasks such as speech production,

vocabulary acquisition, and speech comprehension rather than the typical lower level

cognitive studies of processing limits where the capacity/structure debate flourished.

Working memory has recently become a popular construct in writing research because

speech production, vocabulary acquisition and speech comprehension presumably share

some characteristics with written language production.








The Baddeley and Hitch (1974) model evolved when neurological and

experimental data did not fit assumptions concerning the functioning of the unitary short-

term store in the modal model (Atkinson & Shiffren, 1968). In the modal model, the short-

term store was positioned as the route information must take to gain access into a long-

term store, but neurological evidence indicated that poor performance on an auditory

memory span task was not linked to a long-term learning deficit (Shallice & Warrington,

1970). Nor did an impairment in the short-term store interfere with comprehension and

production of speech (Vallar & Baddeley, 1984). Second, the short-term store was

considered to play a major role in retrieval from the long-term store. However, a digit

span task performed concurrently with the retrieval phase of a free recall task, did not

depress accuracy of retrieval (Baddeley, et al., 1984a). Neither did rehearsal of a 6-digit

number influence recall of paired associates (Baddeley, Thomson, & Buchanan, 1975).

Third, a key postulate of the modal model was that the probability of information being

transferred to long-term store was heightened by increased rehearsal. Yet, the time spent

rehearsing target items interspersed at varying intervals in a longer list of words was

unrelated to recall (Craik & Watkins, 1973). Nor did frequent rehearsal prior to the start

of a free recall experiment improve recall of the rehearsed items (Tulving, 1966). Finally,

another tenet of the modal model was the recency effect in free recall tasks was derived

from information that remained in the short-term store. This was inconsistent with findings

that the recency effect remained even when participants counted backwards by three's for

20 seconds after each word (Tzeng, 1973).

Baddeley's model is a multi-component system for both the storage and the

processing of verbal and visuo-spatial tasks. It has three primary components: a limited-








capacity central executive and two peripheral slave systems. One of these slave systems

handles verbal and auditory information (the phonological loop). The other is specialized

for visual and spatial information (the visuo-spatial scratchpad) These slave systems do

not represent pools of limited resources that can act in parallel like the central executive.

Instead, they operate serially, in the same sense that a computer's printer may receive

multiple print jobs, but can only perform one print job at a time.

The phonological loop is assumed to be most important in the production,

comprehension, and development of language. Architecturally, it has two subcomponents:

(a) a phonological store that is believed to hold phonological information for

approximately 2 seconds and (b) an articulatory mechanism that is implicated in the

transfer of written verbal material to the phonological loop. Because the working memory

model emerged from research on verbal tasks, the phonological loop has received much

more study than either the central executive or the visuo-spatial scratchpad. There are five

empirical sources of support for the phonological loop. First, evidence for the articulatory

control process comes from the word length effect: memory span is smaller for long words

than that for short words (Baddeley, Thomson, & Buchanan, 1975). This is assumed to

occur because rehearsal takes longer for longer words than for short words, allowing

more decay to occur before the next rehearsal cycle. It is typically found that people

remember as many words as they can read in 2 seconds (Baddeley, Thomson, &

Buchanan, 1975). Second, performance is disrupted when participants vocalize some

predefined pattern of speech (e.g., tee tah, tee tah ) while they simultaneously perform

some primary task. This articulatory suppression is claimed to engage the phonological

store and block its ability to participate in accomplishing the primary task. Disruptions









have been observed in a variety of tasks, including vocabulary acquisition and serial

learning (Baddeley, et al., 1984b). Third, performance degradation occurs when

participants attempt to perform a primary task while irrelevant speech is presented that

they are instructed to ignore. This procedure has been shown to disrupt serial recall of

visually presented lists (Salame' & Baddeley, 1982). Fourth, acoustic or phonological

similarity effects are common where the dissimilar sounding items are recalled better than

similar sounding items is interpreted as evidence that the phonological store is speech-

based (Conrad & Hull, 1964; Baddeley, 1966). Words or nonwords that sound alike

interfere with each other in memory span tests more than semantically-related words that

differ in their acoustic properties. Fifth, neuropsychological patients frequently have

specific phonological loop deficits, but suffer no general cognitive impairment (Baddeley,

Papagno, & Vallar, 1988).

The visuo-spatial scratchpad has been less researched than the phonological loop,

but the notion that spatial and verbal processes may each draw on functionally separate

resources is well supported (Wickens, 1984). Where phonological similarity effects have

been observed in tasks believed to require the phonological loop, visual similarity effects

have been observed in visuo-spatial tasks. It is common for stimuli that look alike to

interfere with each other. Hue and Ericsson (1988) observed this phenomenon with

Chinese characters in a study using participants who presumably had no experience with

the Chinese language. Frick (1988) elaborated on the visual similarity effect, arguing on

the basis of the frequency of errors like "P" being mistaken for "R" that images in visuo-

spatial working memory are unparsed. Further evidence for the existence of a separate

visuo-spatial and verbal storage systems comes from neuropsychological studies where








patients with right posterior lesions can be markedly impaired on tests of memory span for

movements to different spatial locations, despite having normal auditory-verbal memory

spans (De Renzi & Nichelli, 1975).

A separate visuo-spatial processing system is also claimed to be involved in

planning and executing spatial tasks. For example, Japanese abacus experts can perform

complex calculations without the aid of the abacus, and appear to do so by simulating the

apparatus using visuo-spatial working memory (Hatano & Osawa, 1983). Researchers

have also implicated the visuo-spatial scratchpad in keeping track of changes in the visual

perceptual world over time (Kahneman, Triesman, & Gibbs, 1992), maintaining

orientation in space and directing spatial movement (Thomson, 1983), and comprehending

certain types of verbal information (Manin & Johnson-Laird, 1982).

The central executive component of Baddeley's model is described as a limited-

capacity mechanism responsible for coordinating tasks and managing the two slave

systems. It operates in a fundamentally different way than the two slave systems, because

it may be actively involved in meeting the demands of two or more concurrent tasks.

Accordingly, performance declines only when the collective task demands exceed the

capacity of resources.

Baddeley acknowledges (1992) that embarrassingly little direct research has

focused on the central executive and that it is somewhat of a catch-all component

responsible for a wide range of attentional duties. Nevertheless, one source of evidence

derives from the study of Alzheimer's disease, a disorder typically associated with deficits

in tasks believed to depend primarily on central executive functioning. In one study

(Baddeley, 1993), Alzheimer's patients were trained to do a visual task and a verbal task.








They then performed both tasks together in a dual-task paradigm. As the disease

progressed, single-task performance remained unchanged, but dual-task performance

dropped markedly. This effect was interpreted as support for the modularity of the central

executive, reasoning that time-sharing ability is accomplished entirely by the central

executive.

The Baddeley model has competitors other than single and multiple resource

models (Kellogg, 1996). Martin, Shelton, and Yaffee (1994) postulate that verbal working

memory may be further partitioned into subsystems for phonological and semantic

representations. Hirst and Kalmar (1987) indicate that working memory might also be

characterized to explain interference as a function of the semantic similarity of processing.

Klapp and Netick (1988) suggest that working memory resources may be further divided

between processing and storage. Ericsson and Kintsch (1995) challenge the utility of

postulating resource limits and raise the point that experts in a domain circumvent short-

term capacity limitations through the structures of long-term working memory.

Working Memory in Writing

Kellogg (1996) proposed a model of writing that details how specific writing

processes rely on the main components of Baddeley's working memory model. In this

proposal, Kellogg distinguishes among three language production processes: formulating,

executing, and monitoring. Each of these involves two basic level processes. Formulation

consists of planning ideas and translating them into sentences that may later be

handwritten or typed. Execution includes programming (controlling motor movements)

and executing (typing, handwriting, or dictation). Monitoring includes reading and

editing. The model is very clear that writers do not necessarily progress serially from








formulating text to executing it, and then to monitoring what they have written. Instead,

the model supports simultaneous activation of each of these processes, provided that the

demands placed on the central executive do not exceed its capacity limitations. Thus, the

model anticipates that the typing of a word or phrase may take place simultaneously with

the formulation of new material or reading previously written material. This is possible

only when execution is well-practiced and can proceed virtually automatically, so that

central executive resources are not needed. In contrast, the basic processes of formulation

and monitoring are much less likely to ever become automated.

Kellogg considers the demands that the basic processes of writing make on

working memory. He hypothesizes that formulation places the heaviest burden on working

memory, particularly on the central executive, but also on both slave components.

Planning demands the resources of the visuo-spatial scratchpad particularly when writers

plan by visualizing ideas, organizational schemes, supporting graphics, appearances of the

orthography and layout. Creating ideas (Shepard, 1978) and recalling them from long-

term memory (Paivio, 1986) can invoke visual imagery. Gathercole and Baddeley (1993)

claim that planning also requires central executive resources in all its facets such as

generating ideas, trying out various organization schemes, or debating the appropriate

tone for a given audience. When formulating, writers covertly talk to themselves in the

form of inner speech or pre-text as they are generating sentences (Baddeley & Lewis,

1981), thus activating the phonological loop. In particular, phonological representations of

the words and sentences are stored in the phonological loop. To connect the phonological

loop with formulation tasks Kellogg notes cases of apraxic and dyspraxic patients who

make phonemic and other linguistic mistakes in speech and who also are impaired in








memory span tests and fail to show phonological similarity and word length effects

(Waters, Rochon, & Caplan, 1992). Translating, unlike conversational speech production

(Bock, 1982), demands central executive resources when the writer struggles to find just

the right words and sentence structures. Long pauses and high degrees of expended

cognitive effort suggest the involvement of the central executive in these cases (Kellogg,

1996) as well as the positive correlation of memory span with the ability to select lexical

items for use in a sentence (Daneman & Greene, 1986). Kellogg adds that the

involvement of the central executive in translation may vary with the demands of the task

at hand. A writer composing polished final draft prose on the first attempt may require

central executive resources to a greater extent than a writer composing a more

conversational email to a friend.

Typing and handwriting (execution) are believed to make minimal demands on the

central executive when highly practiced. In young children however, Bourdin & Fayol

(1994) found that handwriting demands more capacity than speaking, consistent with the

notion that novel activities of all kinds require the central executive to control the schemas

used in motor output (Kellogg, 1996).

Gathercole and Baddeley (1993) claim that reading, one of the basic processes of

monitoring, requires the resources of both the phonological loop and the central executive.

Kellogg suggests that the most significant demand of monitoring stems from editing, not

reading, because editing takes so many forms, ranging from the detection of a motor

programming effort to a revision in the organization of ideas in a text (Kellogg, 1996).








Relevant Working Memory and Writing Research

Owing in part to its newness, little research has been reported that explicitly tests

the implications of Kellogg's model. One prediction of the Kellogg model suggests that

the phonological loop plays a strong role in monitoring. Jones, Miles, and Page (1990)

introduced an irrelevant speech manipulation to explore phonological loop involvement in

proofreading. Although proofreading is clearly not the same as editing one's own writing,

it does involve some aspects of the monitoring process as described by Kellogg (1996).

Jones, et al. compared participants' ability to detect two types of errors, contextual

(grammatical and word choice) and non-contextual (spelling and typographical errors),

while they heard either meaningful or reversed speech. The irrelevant speech consisted of

a taped lecture either played forward (meaningful speech) or backward (reversed speech).

The detection rate for non-contextual errors was disrupted in the meaningful speech

condition, but this effect did not extend to contextual errors. No effects were observed for

the reversed irrelevant speech manipulation, nor were there differences between reversed

irrelevant speech and silence in a second experiment, lending support to the idea that a

semantic component is important in the irrelevant speech effect. These results are

consistent with Kellogg's (1996) predictions that the phonological loop is involved in the

detection of spelling and typographical errors, but inconsistent with the prediction that

irrelevant speech would hinder the detection of grammatical and word choice errors.

Investigations of irrelevant speech effects on reading comprehension also failed to

support the role of the phonological loop as a temporary store for sentence

representations (Martin, Wogalter, & Forlanon, 1988). In the initial investigation,

comprehension scores declined in two conditions involving irrelevant speech compared to








a silent control. The irrelevant speech consisted of either words in a meaningful order or

the same words randomly scrambled. Comprehension in the two speech conditions was

not reliably different. In a second experiment, Martin et al. (1988) manipulated

instrumentation (present or not) and lyrics (sung, spoken, or none) within subjects.

Findings indicated that the presence of verbal material adversely affected comprehension,

while instrumentation did not.

Madigan and Linton (1996) asked participants to compose two essays on a word

processor, one while hearing irrelevant speech and the other in silence. Prior to the writing

session, participants completed a reading span task designed by Daneman and Carpenter

(1980). This reading span test is used to assess capacity by incorporating a processing and

a storage component. Participants read aloud an increasingly larger set of sentences. After

each set, they recalled the last word of each sentence. Madigan and Linton based the

working memory span on the number of last words correctly recalled. Those who scored

in the top third of the sample were classified as high capacity, while those in the bottom

third were classified as low capacity.

Irrelevant speech reliably reduced the holistic quality of the essays, also causing a

decline in syntactic complexity (the proportion of subordinate clauses that occur before

the main verb) among high span writers. Writing fluency, as measured by word production

times and pauses within selected clauses, were unaffected by the irrelevant speech

manipulation. The investigators argue that the reduction in holistic quality is inconsistent

with assumptions of the Kellogg model related to the relatively minor influence of the

phonological loop when compared to the central executive, on writing quality. Kellogg








(1996), however, disagrees, contending that "the small effect size is consistent with the

hypothesis that irrelevant speech disrupts only translating and reading" (p 69).

Ransdell, Levy and Kellogg (1977) used irrelevant speech and memory load

manipulations comparing performance of writers with high and low writing span in a series

of experiments. They hypothesized that irrelevant speech, linked only to translating and

reading, would have a relatively small effect on fluency and quality compared to a memory

load relying on central executive resources. They also expected that the number of pauses

would decline when the central executive rather than the phonological loop was affected

by the manipulation.

In each of three experiments, participants composed two 10-minute essays. In the

first experiment, one essay was written while hearing irrelevant speech, the other in

silence. In the second experiment, they wrote with or without a six digit concurrent load.

In the final investigation, participants wrote one essay in the presence of unattended

irrelevant speech, and one essay while they were instructed to attended to the irrelevant

speech, periodically responding "yes" or "no" to questions related to either phonological,

spatial or semantic characteristics of the message. Because the semantic task was

positioned as more heavily dependent on central executive capacity, it was expected that

decrements in performance would be greatest in this condition.

Results of the first experiment demonstrated that irrelevant speech reliably reduced

fluency, and an increase in the number of pauses longer than five seconds, but did not

influence the percentage of pauses at clause boundaries. Essay quality was similar in the

irrelevant speech and silent conditions, an unexpected finding. The researchers suggested

two possible explanations for the immunity of essay quality to the speech manipulation.








First, they considered the possibility that the phonological loop is activated only for

complex sentences. Second, they raised the possibility that the articulatory loop was more

intimately involved in the writing task than the phonological loop, and that irrelevant

speech derived its influence via corrupting material in the store.

In the second experiment, memory load also resulted in a reliable decrement in

fluency and in sentence length. Only the fluency of low span writers was adversely affected

by the concurrent task. Concurrent load also altered the temporal patterns of composing.

When remembering and recalling digits, participants paused longer and more often. High

span writers also paused less frequently at clause boundaries. Furthermore, unlike

exposure to irrelevant speech, the concurrent task also reliably reduced essay quality.

In the third experiment, attending and responding to irrelevant speech led to a

reliable decrement in writing fluency, sentence length and pause frequency compared to

baseline, but pause length and location were similar in both conditions. Additionally,

overall writing quality did not decline in the attended speech condition, although scores of

high span writers with "good sentence" instructions were adversely affected in the

organization and development subgroup. The type of question answered in the attended

speech condition (phonological, spatial, or semantic) did not affect any dependent

measure. Of the three secondary tasks, accuracy was higher in the semantic condition than

in the spatial or phonological conditions which were equivalent.

The most consistent result throughout the three experiments was the expected

decrement in fluency as a result of the experimental manipulation. Neither unattended nor

attended speech resulted in a quality decrement. In contrast, the six-digit concurrent load

did adversely affect quality relative to a control condition. The six-digit load was also the








only task that lengthened pause duration and reduced the proportion of pauses at clause

boundaries. Here, there is a trade-off between quality and fluency that involves an increase

in the number of pauses rather than in their duration.

To test Kellogg's (1996) predictions of the phonological loop involvement in

translating ideas into sentences (formulation) and reading text (monitoring), Marek and

Levy (1996) designed a series of computer-based tasks to simulate these primary writing

processes. These writing subtasks included creating sentences from groups of words

(formulation), copying an existing text (execution) and proofreading (monitoring). Writers

first performed each of the three tasks in silence, then repeated each task while listening to

computer-generated speech. Irrelevant speech during the formulation task reliably reduced

both the number of target words used in the sentences and sentence quality, a measure

based on a combination of meaningfulness and technical quality. Irrelevant speech did not

influence performance measures related to execution or monitoring as predicted by the

model but did lend support to the role of the phonological loop in formulation.

One possible explanation for the null result found for monitoring relates to the

proportion of editing versus reading in the tasks. In the proofreading task, twenty errors

were embedded into approximately 150 words of text. Participants were asked not only to

detect the errors, but to identify them as well, within a 5-minute time period. Beyond the

obvious possibility that the results reflect a difference between proofreading someone

else's work and reviewing one's own writing, the task was not designed to enable

recording of the amount of time spent reading versus the amount of time spent editing.

Nor did the researchers record or control how much of the selection was actually read.









The detection rate for errors ranged from 5% to 56%, but whether this primarily illustrates

differing ability to identify the mistakes or differences in reading fluency remain unknown.

Another prediction of the Kellogg model suggests that loading the central

executive should detract from writing fluency and quality the most, because the central

executive plays a part in all three writing processes. Jeffery and Underwood (1995)

conducted an experiment in which participants created sentences from information

presented, with either a 0, 3, or 6 digit preload. If this construction task appropriately

reflects formulation, the Kellogg model predicts that loading the central executive should

disrupt its performance. Rather than an overall disruption, Jefferey and Underwood found

a fluency-quality tradeoff. Although the preload actually reduced sentence initiation time,

quality of the sentences, as measured by the level of idea coordination, also dropped in the

preload conditions.

In a second experiment, participants copied rather than constructed sentences,

again with or without a preload. In this context, sentence initiation time under a 6-digit

preload was higher than in the non-preload condition. Because semantic processing is

required only in the construction task, Jefferey and Underwood concluded that the

decrease in sentence initiation time in their first experiment was attributable to a lack of

available resources for semantic processing in the preload conditions. More generally,

differences between preload and control conditions suggest that semantic processing

during writing involves the central executive.

Jefferey and Underwood positioned the primary task in these experiments as

constructing rather than formulating sentences. However, other writing researchers

(Kellogg, 1996; Ransdell & Levy, 1996) have suggested that constructing sentences








represents one aspect of formulation in writing. Debate surrounding inconsistencies at this

level underscores the difficulty involved in studying the specific components of a complex

task such as writing. It raises questions concerning the relationship of the part to the

whole, and calls attention to the problem of developing a satisfactory operational

definition for a specific component that has been separated from its task context. Although

investigations of writing at the subcomponent level are important, great difficulty

surrounds this practice and some aspects of writing may not be profitably reduced to this

level. Thus, this dissertation investigates essay writing as a complete task rather than

breaking it down into formulation, execution, and monitoring.

Adapted Working Memory Model

The balance of this chapter closely examines the assumptions of Baddeley's

working memory model and proposes an adaptation that brings working memory into

closer alignment with the attention literature while simultaneously preserving its

orientation toward complex language tasks. These adaptations to the working memory

model have subtle implications on the dual-task methods employed by this dissertation and

other studies of this nature. Before addressing these methodological implications,

however, a closer analysis of working memory is needed.

There are several interesting assumptions that are commonly made about the

processing capabilities of the two slave systems. First, in comparisons of visuo-spatial and

phonological loads, the structural assumption implies that the relative difficulty of the

secondary tasks is irrelevant because any secondary task, regardless of difficulty, will

engage the visuo-spatial scratchpad or phonological loop entirely, or not at all. Therefore,

a phenomenologically easy visuo-spatial task, such as identifying in which ear a sound is








played through headphones, should load the visuo-spatial scratchpad as much as a more

difficult spatial task, such as mentally rotating a three dimensional object. If we assume,

instead, that the slave systems represent pools of resources, then the difficulty of the

secondary tasks becomes a variable of interest. This variable should be experimentally

controlled because of the common finding that increasing the difficulty of a secondary task

almost always results in a performance degradation in the primary task (Wickens, 1984;

Navon & Gopher, 1979).

If we assume the slave systems are resource pools, then when we compare the

relative effects of visuo-spatial and phonological load tasks on a primary task, it becomes

important to equalize the overall difficulty of the loading tasks. Consider the implications

of mistakenly assuming the slave systems are structures. If we disregard the difficulty of

loading tasks, a disproportionately more difficult phonological task would disrupt the

primary task of writing more than the visuo-spatial secondary task, possibly leading to the

conclusion that the visuo-spatial scratchpad plays a smaller role in writing than it actually

does. Thus, when comparing the contributions made by the two slave systems to a primary

task under dual-task conditions, it may be more risky to assume that these components are

structures rather than resource pools.

Another implication of Baddeley's theoretical assumptions arises when comparing

effects across studies. Under the assumption of structural slave systems, differences in

secondary task structure (including characteristics such as the frequency of response

selection and execution as well as input and output modalities) will not affect the extent to

which that task engages its targeted component. Resource theorists argue that just as task

difficulty is likely to influence primary and secondary task interference, the frequency of








response selection and the input and response modality are likely to affect the degree to

which they interfere with a primary task.

Overall, a structural assumption allows for much easier secondary task selection,

but the results may be inaccurate if the visuo-spatial scratchpad and phonological loop are

in fact pools of resources rather than structures. The more rigorous secondary task

selection methods associated with a resource assumption are acceptable regardless of

whether the visuo-spatial scratchpad and phonological loop act as structures or resources.

In light of these considerations, a new model of working memory was proposed

that posits two mutually inaccessible pools of resources for verbal and visuo-spatial

processing and storage such as Friedman and Poison's (1981) multiple resource model.

The current model differs from Friedman and Poison (1981) in its inclusion of Baddeley's

notion of a limited capacity central executive system that acts for a wide range of

attentional duties including coordinating information between the visuo-spatial and verbal

systems.

Experiment 1 seeks empirical support for this adapted working memory model, by

testing the hypothesis that the phonological loop may be partially disrupted within subjects

rather than disrupted entirely or not at all.

Summary

As a writer plans ideas, translates ideas into prose, types or handwrites sentences,

and monitors all of these activities, many demands are placed on the writer's temporary

storage and processing capacity. Investigations of writing at this level are fairly new, thus

obligating writing researchers to turn to existing theories of cognitive limitations for a

framework in which these demands may be assessed.








Theory development surrounding processing limitations has typically employed

two hypothetical constructs, resources and structures. Early accounts of processing

demands were explained by either resource (Kahneman, 1973) or structural (Broadbent,

1958) models, but more recently elaborate models have evolved that embody both

resource and structural aspects. The two most prevalent of such models are multiple

resource models (Norman & Bobrow, 1975; Wickens, 1984; Friedman & Poison, 1981)

and Baddeley's working memory model (Baddeley & Hitch, 1974). Both propose

divisions in visuo-spatial and verbal processing, generally in alignment with the

hemispheric asymmetry literature.

Kellogg (1996) proposed a model of writing that details how specific writing

processes rely on the main components of Baddeley's working memory model. Before

testing the specific claims of Kellogg's model, a closer examination of working memory as

a model of cognitive limitations is necessary. In particular, the assumptions of the

structural nature of the phonological loop and visuo-spatial scratchpad seem outdated.

This dissertation proposes an adaptation to the working memory model whereby

the phonological loop and visuo-spatial scratchpad operate as pools of mutually

inaccessible resources for verbal and visuo-spatial processing and storage similar to

Friedman and Poison's multiple resource model (1981). This hybrid model of working

memory and multiple resources is tested in Experiment 1. Experiments 2a and 2b consider

the division of labor between visuo-spatial and verbal processing and storage components,

and Experiment 3 addresses the role of writing topic in the resource demands of writing.














CHAPTER 2
EXPERIMENT 1

If the phonological loop and visuo-spatial scratchpad of Baddeley's working

memory model are engaged in a binary, all-or-none fashion, it follows that in a dual-task

paradigm, secondary tasks of varying difficulty will have the same loading effect. This

assumption is comparatively easy to test. For example, in a dual-task experiment, if two

phonological or visuo-spatial secondary tasks of varying difficulty interfere equally with a

primary task, then the all-or-none processing assumption is supported. Conversely, if the

two phonological or visuo-spatial secondary tasks vary in the relative degree to which they

disrupt primary task performance, then graded processing is supported.

Baddeley's working memory model predicts no differences in performance related

to type of irrelevant speech, because speech sounds, regardless of meaning, are believed to

have obligatory access to the phonological store. Those sounds are believed to engage the

phonological loop entirely due to its structural nature. Inconsistent with these predictions,

Jones, Miles, and Page (1990) found that meaningful speech interfered more with the

detection of spelling and typographical errors in a proofreading task than the same speech

played backward. However, in a second experiment, Jones et al. found no difference

between the effect of reversed speech and a silent control on error detection rates,

suggesting that reversed speech may not engage the phonological loop at all. It could be

argued, however, that there may be so little phonemic similarity between forward and

backward speech that it may not be reasonable to claim that speech played backward is








really "speech." Further, Levy and Marek (1998) report similar patterns of result in a

writing study when computer-generated speech was "natural" and when the individual

words in each sentence were randomly rearranged, but no changes made to their phonemic

attributes.

A recent study using magnetic resonance imaging revealed that irrelevant speech

played backward is processed in different areas of the brain than is forward speech, raising

questions whether the phonological loop acts alone in processing speech. Baddeley's

working memory model again predicts no differences in performance related to type of

irrelevant speech, because speech sounds, regardless of meaning, are believed to have

obligatory access to the phonological store, and are believed to engage the phonological

loop entirely due to its structural nature.

Experiment 1 compares two types of irrelevant speech to test the hypothesis that

the phonological loop can be differentially loaded in a dual-task paradigm. One recording

was made of a passage read by the experimenter, and the other featured the same passage

generated using a text-to-speech computer program. The Baddeley model predicts that

accuracy on a primary memory task would be best in the absence of irrelevant speech.

When either the human-generated or the computer-generated speech is delivered while

participants perform another (primary) task, performance on the latter should decline, but

the performance decrements should be equal for the two types of irrelevant speech. This

would be so even if the materials delivered in a computer-synthesized speech generated a

lesser processing load (because it lacked nuances, had reduced inflections, and so forth)

than the speech produced by a human voice. In contrast, our modified model, which








anticipates such secondary tasks producing differential processing loads, would predict

differential outcomes on the primary task.

Method

Participants

Twenty-six undergraduate students from the University of Florida participated in

this experiment in partial fulfillment of a course requirement. One participant was excluded

from the data because he or she stopped recording the recalled consonants. Participants,

were neither vision nor hearing impaired.

Materials

Sets of five to nine randomly generated consonants were printed on transparencies

and presentated on an overhead projector. Consonant sets contained no duplicate letters,

and no two sets were identical. Participants recorded their responses on printed answer

sheets during the free recall phase of the experiment.

Two tapes of irrelevant speech were recorded. One featured a children's fable read

by the experimenter. The other featured the same story generated via computer running a

text-to-speech program. The content and acoustic energy level of the two taped

presentations was the same.

Procedure

The to-be-remembered consonant sets were shown to the participants for 10

seconds each. Participants then had 20-seconds to recall the consonants. This sequence

was repeated using 15 different sets of consonants for the baseline (no irrelevant speech)

and each of the experimental conditions. During the experimental conditions, participants








heard one of two types of irrelevant speech during the presentation portion of each trial.

On alternate trials, the irrelevant speech was human-generated or generated via computer.

Results and Discussion

The patterns in the data are quite compelling (see Figure 1). For nearly all

participants, the human-generated speech disrupted letter recall more than computer-

generated speech. A within-subjects analysis of variance showed that both human-

generated irrelevant speech and computer-generated irrelevant speech interfered with

participants recall of consonants (F (1,23) = 27.65, p<.001 and F (1,23) = 15.72, p<.001

respectively). Recall accuracy was lower in the presence of human-generated speech than

the computer-generated speech (F (1,23) = 33.85, p<.001). This pattern of results matches

their subjective reports that the computer-generated speech was easier to disregard. If the

phonological loop is engaged in an all-or-none fashion, it follows that the irrelevant

computer-generated speech should have the same effect as irrelevant human-generated

speech. These results are more consistent with the notion that the phonological loop is

comprised of a pool of resources that can be engaged in a graded fashion. Further testing

is needed to determine if the visuo-spatial scratchpad also operates as a resource pool

rather than a structure.










100
90 \
85 "

7Sumn-Genrazd \.

70 -' Computer-C.nerated \
65
60
5 6 7 8 9
Memory Set Size


Figure 1. Retention of consonant strings as a function of memory set size and irrelevant
speech condition.


Regardless of whether we assume that the visuo-spatial scratchpad operates as a

pool of resources like the phonological loop, because the phonological loop operates as a

resource pool, it is necessary to adopt the following criteria for selecting secondary tasks

when making comparisons between these two working memory components. First,

secondary tasks should be equally demanding of attentional resources. Equivalency may be

accomplished by comparing secondary task performance in isolation, or by placing each

secondary task in a probe-tone paradigm in which subjects perform the load task as a

primary task and are interrupted occasionally by a tone to which they respond as quickly

as possible. Reaction times in probe tone paradigms are typically taken as an index of the

mental effort demanded by the primary task. Second, the loading tasks should not change

in the degree to which they draw on controlled processes throughout the experiment. If a

load can become automated over the course of the dual-task conditions, then by the most

common definitions of automaticity, it no longer draws on attentional resources, thus








allowing participants to work on the primary task as if they were working under single-

task conditions. Third, the tasks should have similar structural composition, requiring the

same mode of input and output. Fourth, response selection and execution should occur

equally often in the tasks. Finally, secondary tasks should require different strategies for

their successful performance. For example, a visuo-spatial task should not be easily

performed by re-coding visuo-spatial stimuli into phonological stimuli. Thus, secondary

tasks should be constructed so that performance differences will emerge if participants use

the same strategies for each task. The secondary task selection process for Experiment 2B

serves to illustrate these selection criterion.

Once appropriate secondary tasks are selected, the rationale for dual-task studies is

relatively straightforward. By selectively engaging one component of working memory in

a secondary task, the primary task may then only draw on the remaining components.

Thus, using a dual-task paradigm, experiment 2A investigates the contribution of verbal

resources to writing in a dual-task paradigm, while experiment 2B investigates the

contribution of visuo-spatial resources to writing. Given the adaptations we have made to

the working memory model, comparisons of the relative involvement of verbal and visuo-

spatial processing can be made across these two experiments by adhering to the secondary

task selection criteria described above.














CHAPTER 3
EXPERIMENT 2

Experiment 2A

Most resource models (Wickens, 1984; Friedman & Polson, 1981) and working

memory theories (Baddeley & Hitch, 1974) acknowledge a division in processing for

verbal and visuo-spatial tasks, in alignment with the literature on hemispheric differences.

Generally, visuo-spatial processing is localized to the right hemisphere while verbal

processing is identified with the left hemisphere.

Brooks (1968) demonstrated a clear dissociation of verbal and visuo-spatial

processing in an experiment were participants were presented with a block capital letter

with the bottom left hand comer marked with a star. Participants were asked to look away

from the letter and, holding it in memory, to go around the letter clockwise from the star

responding "yes" if the comer in question pointed outward or "no" if the comer pointed

inward. Hence for the block letter F, the correct responses would be "yes, yes, yes, yes, no

, no, yes, yes, no." Brooks also developed a verbal task where participants were

presented with a sentence like, "A bird in hand is not in the bush". Participants were asked

to hold this sentence in memory, and then successively categorize each word as either a

noun (in which case participants would respond "yes"), or a verb ("no"). Hence, for that

particular sentence the correct sequence would be "no, yes, no, yes, no, no, no, no, yes".

Brooks used two methods of responding, either spoken or manual (this involved pointing

to a series of yeses or nos scattered irregularly down the response sheet). He observed a








clear interaction between type of memory task and mode of response, with the visual letter

task being performed more accurately when the response was spoken than when it

involved pointing, while the reverse was the case for the sentence task. This finding

demonstrates the interference of tasks that share similar processes, and points to separate

systems for verbal and visuo-spatial memory.

Subsequent selective interference effects have been observed: (a) Concurrent

visuo-spatial pursuit tracking disrupted performance on Brook's (1968) spatial task, but

not on a corresponding abstract task (Baddeley, 1975); (b) A verbal task interfered with

right hand dowel balancing (Kinsbourne & Hicks, 1978); and (c) a spatial task interfered

with left hand dowel balancing (McFarland and Ashton, 1978). Also, in the perceptual

domain, Moscovitch and Klein (1980) observed that recognition performance was more

impaired when two spatial targets were presented simultaneously (a face and a random

polygon), than when a spatial and a verbal target were presented simultaneously.

EEG research supports the intuition that writing, like other verbal tasks, is

supported by the left hemisphere. Gallin & Ornstein (1972) set out to select areas of the

brain that the split-brain and brain lesion research had shown were involved in different

tasks, and to record the brain's electrical activity while a person was doing these tasks. In

general, they found that ordinary people, doing ordinary activities, turned on and off the

two sides of their brain appropriately. Specifically, Gallin & Ornstein (1972) reported that

the right hemisphere showed more alpha activity than the left while writing a letter and the

left hemisphere showed more beta activity. While the person was arranging blocks, the left

hemisphere showed more alpha than the right side, and the right hemisphere showed more

beta waves. So when people write, they turn on the left hemisphere and turn off the right








side of the brain. While arranging blocks in space, they turn on the right side and turn off

the left hemisphere.

Baddeley (1993) investigated the working memory demands of chess by selectively

disrupting components of working memory in a dual-task paradigm. Two groups of

expertise were compared: weak club players formed the basic group and an advanced

group was comprised of expert chess players with Elo gradings ranging from 130 to 240.

Participants were required to memorize and recall chess positions as a primary task, while

performing one of three secondary tasks or a control condition where no concurrent

activity was performed. To disrupt the phonological loop, participants were given a

secondary task that required them to utter the word "the" at the rate of one per second

(articulatory suppression). The visuo-spatial scratchpad was disrupted by requiring

participants to tap the keypad of a computer at a one-second rate following a fixed order.

Random letter generation, again at a one-second rate, was required to disrupt the central

executive component of working memory. Overall, performance in the two groups

differed greatly, but the pattern of interference was equivalent, with articulatory

suppression had little effect, while both finger tapping, and random letter generation

produced greater disruption. These findings suggest that verbal processes play little or no

role in memory for chess positions compared to the role of visuo-spatial or central

executive processes.

Several criticism can be made of Baddeley's dual-task methods in the study of

chess. Because participants performance was not measured on any of the secondary tasks

there is no evidence that participants performed these tasks as they were instructed.

Participants may simply have slowed or even stopped performing the secondary task as









primary task demands increased. Additionally, it is possible that participant's performance

of these tasks became automatic, thus requiring little or no resources. These secondary

tasks implicitly assume all-or-none processing of the phonological loop, visuo-spatial

scratchpad and the central executive because no attempts were made to ensure that these

tasks are equally demanding.

The purpose of the present experiment was similar to Baddeley's (1993)

investigation of the working memory requirements of chess, but instead of chess, this

study assess the contributions of the phonological loop to essay writing in a dual-task

paradigm. Writing researchers have addressed this issue by employing irrelevant speech

manipulations while measuring essay quality and fluency (Ransdell & Levy, 1998;

Ransdell, Levy, & Kellogg, 1997), proofreading ability (Jones, Miles, Page, 1990),

temporary storage of sentence representations (Martin, Wogalter, & Forlanon, 1988), and

creating sentences from groups of words, proofreading, and copying text (Marek & Levy,

1996).

Because the long-term research goal was to compare phonological loop

(experiment 2A) and visuo-spatial scratchpad (experiment 2B) contributions to writing, a

new secondary task was needed which, unlike irrelevant speech, required measurable

responses from the participant. A signal detection paradigm (Green & Swets, 1966)

paradigm was implemented because it would allow for easy comparison of performance to

a similar visuo-spatial secondary task. Signal detection is a well-established method used

by engineers and by psychophysicists for estimating an observer's discriminative capacity

or sensitivity to a signal independent of the observer's response bias. The measure of

sensitivity is a pure number represented by the symbol, d' which is based on the








participants' number of hits and false alarms. The greater the value of d', the better the

individual's performance.

Additional efforts were necessary to assure that this secondary task did not

become automatic. Dual-task conditions were designed so the data could be parted into

five-minute epochs for which individual d' scores could be calculated. Thus, patterns of

d' scores could be inspected qualitatively on an individual basis and participants who

showed dramatic changes in secondary task performance over time could be eliminated

from the study. No participants were excluded from this study based on this information.

Figure 2 illustrates the average secondary task performance and the percentage of writing

as a function of these five-minute epochs.




30

25

~20



10

1.4 1.43 2.31.6


1-5 5-10 10-15 15-20
Timn (5-minute apoca)
Figure 2. Average percentage written and dual task performance as a function of time.








Method

Participants

Forty-one University of Florida general psychology students participated in this

study. All participants reported to be proficient touch typists (avg. typing speed was 32

wpm) and had experience interacting with a computer using a mouse.

Materials and Apparatus

Participants wrote essays on 2 open-ended topics, "The perfect job" and "The

greatest high" using a Windows-based word processor developed especially for this

research effort. The program supported basic functions of text entry, cutting and pasting,

and mouse control at the point of insertion. The program also enabled the presentation of

the secondary task stimuli (directional symbols) in a large (120 point) sans serif font

displayed to the left of the text composition window. The characters subtended

approximately 3 degrees of visual angle and were readily identifiable from the writers'

peripheral vision.

Procedure

Participants first completed a 2-minute typing test in which they were asked to

transcribe as quickly as possible text that appeared on their screens. Next, they spent 20

minutes writing an essay on one of the two topics. Participants were given verbal notice 5

minutes before their time expired.

Participants next performed a task designed to load the phonological loop under

single-task conditions for 5 minutes. This task was designed to load the phonological loop,

but not the visuo-spatial scratchpad. During this task, a single letter or a single digit was

shown for 5 seconds. At the end of this period, the visible stimulus was erased and








replaced by another single character. The characters typically alternated between letters

and digits. The participant's task was to indicate by pressing the mouse button -

whenever the category (letter or digit) remained the same from one presentation to the

next (i.e. two consecutive letters or numbers). A response was required, on the average,

twice per minute on a variable interval schedule. Consecutive letters and consecutive digits

occurred equally often. If participants responded correctly a green circle appeared instead

of the next character, but if they missed an opportunity to respond correctly a red circle

appeared instead of the next character. False alarms did not disrupt the sequence of

character presentation, and no feedback was given to indicate a false alarm had been

made.

In the final, dual-task phase of the experiment, participants wrote for 20 minutes

on whichever topic they had not already addressed earlier under single-task conditions. At

the same time, they simultaneously performed the phonological loading task that they had

just finished. Instructions for the loading task, stimuli presentation rate, and method of

responding were identical to the single-task conditions. Participants were instructed to

perform both tasks to the best of their ability. To maintain interest in the task, the research

promised that the person who performed the best overall would win $100.

Results and Discussion

Document quality was assessed on a 6-point scale using the essay-sort method

(Madigan, Johnson, & Linton, 1994) where two independent raters sorted essays first into

3 categories of quality, and then rated essays as high or low quality within category (inter-

rater reliability = 88%). Document fluency was measured by the number of words








generated per minute. A signal detection paradigm (Green & Swets, 1966) was employed

in the phonological secondary task.

As illustrated in Figure 3, performance in the present experiment deteriorated

significantly from single to dual-task conditions on all measures. Words produced per

minute dropped 21% (1 (42) = 7.19, p<.001) and the quality of the essays declined 12.3%

(1 (42) = 2.71, p=.01) with the addition of the phonological loop secondary task. Most

compelling was a 60% decrement in d' associated with the addition of the secondary task

(t(42) = 8.91, p <.001).





Quality



Fluency (WPM)



2nd task (d


-80 -60 -40 -20 0 20 40 60 80
Percent Change
Figure 3. Decrement in quality, fluency, and secondary task d' from single to dual-task
phonological load conditions.


Kellogg's (1996) predictions for the involvement of the phonological loop in

writing were that this component would play a strong role in monitoring, particularly in

the detection of spelling and typographical errors. Also, the phonological loop should









have a minor influence on writing quality relative to the central executive, but more

influence on quality than the visuo-spatial scratchpad.

Decrements in quality associated with the phonological secondary task seem to

conflict with Levy, Ransdell & Kellogg (1997) who failed to find an effect of irrelevant

speech on essay quality. Inconsistencies here can be explained by the greater demands of

the letter matching task due to response selection and execution stages that were required.

It is important to keep in mind that the primary purpose of this experiment is the eventual

comparison to the effects of a visuo-spatial loading condition, and therefore the response

selection and execution stages are necessary, and will algebraically cancel each other out

when that comparison is made.

Researchers using an irrelevant speech manipulation have observed disruptions in

the detection rate for non-contextual errors (Jones, Miles, and Page, 1990), diminished

comprehension (Martin, Wogalter, & Forlanon, 1988), lower holistic quality of essays and

diminished syntactic complexity as measured by the proportion of subordinate clauses that

occur before the main verb (Madigan & Linton, 1996).

The present experiment supports Ransdell, Levy & Kellogg's (in press) finding

that irrelevant speech reliably reduced fluency as measured by words per minute. Madigan

and Linton (1996), however, found no effect of irrelevant speech on fluency as measured

by word production times and pauses within selected clauses. This calls into question the

relationship between different measures of writing fluency.

Further questions involve how a visuo-spatial secondary task will effect essay

quality, fluency, and d' scores compared to the phonological secondary task observed in








this experiment 2B describes the selection and effects of a visuo-spatial secondary task,

followed by the combined analysis of experiments 2a and 2b.

Experiment 2B

Because of the impressive amount of research on the effects of irrelevant speech

on the variety of tasks discussed earlier, the focus of considerable recent work in writing

has also focused on variables that might influence the engagement of the phonological

loop. The role of visuo-spatial processing in writing has received very little attention

compared to that of the phonological loop (Dinet & Passerault, 1998; Levy, White, &

Lea, 1998).

Kellogg (1996) claimed that visuo-spatial resources are involved in planning,

particularly when writers plan by visualizing ideas, organizational schemes, supporting

graphics, appearances of the orthography and layout. In one of the only studies

investigating the role of the visuo-spatial scratchpad in writing, Ransdell, Levy, & Kellogg

(1997) had participants write one essay in the presence of unattended irrelevant speech,

and one essay while they were instructed to attend to the irrelevant speech periodically

responding "yes" or "no" to questions related to either phonological, spatial or semantic

characteristics of the message. The type of question answered in the attended speech

condition did not affect any dependent measure. One interpretation of this lack of findings

is that the combination of making judgements requiring a response on irrelevant speech

may have simultaneously loaded the phonological loop (with irrelevant speech), the central

executive (by requiring response selection and execution) and then either the phonological

loop (phonological judgements) or the central executive (semantic judgements) again or

the visuo-spatial scratchpad (with spatial judgements).








In the present experiment, we use the methodology underlying Experiment 2A to

focus on how processing in the visuo-spatial scratchpad influences writing. The

hypothesis was that a visuo-spatial task, like the phonological task in the previous

experiment, would cause a significant decrement in writing quality and fluency.

Additionally, when comparing the dual-task performance of a visuo-spatial task and

writing to dual-task performance of a phonological task and writing (Experiment 2A),

dual-task performance would be better in the visuo-spatial condition.

In order for performance here to be compared to Experiment 2A, a secondary task was

needed to engage the visuo-spatial scratchpad in a signal detection paradigm where no

significant differences exist between the visuo-spatial and phonological single task d'

scores. Additionally, this visuo-spatial secondary task would also need to be designed so

that d' scores could be calculated for each 5-minute epoch, to assess whether that task

became automated. Figure 4 illustrates the average secondary task performance and the

average percentage of writing as a function of these five-minute epochs.






















1.2 1.91""

1-5 5-10 10-15
Time (5-minute epochs)


1.83

15-20


Figure 4. Average percentage written and dual-task performance as a function of time.


Stimulus presentation modality and rate and the rate of responding needed to be

equivalent for the visuo-spatial and the phonological secondary tasks. Additionally, the

visuo-spatial secondary task had to be constructed so that performance differences would

emerge if participants encoded the stimuli verbally as they did in the phonological loop

secondary task.

Method

Participants

Forty-four University of Florida general psychology students participated in this

study. All participants reported to be proficient typists (avg. typing speed was 34 wpm)

and had experience interacting with a computer using a mouse.








Apparatus and Materials

Participants wrote essays on 2 open-ended topics, "The perfect job" and "The

greatest high". They used a Windows-based word processor written especially for this

research effort. The program supported basic functions of text entry, cutting and pasting,

and mouse control at the point of insertion. The program also enabled the presentation of

the secondary task stimuli (directional symbols) in a large (120 point) sans serif font

displayed to the left of the text composition window. The characters subtended

approximately 3 degrees of visual angle and were readily identifiable from the writers'

peripheral vision.

Procedure

Participants first completed a 2-minute typing test in which they were asked to

transcribe as quickly as possible text that appeared on their screens. Participants then

wrote an essay for 20 minutes on one of the two topics. Half wrote on one topic, while the

other half wrote on the other. Participants were given notice 5 minutes before their time

expired.

Participants next performed a visuo-spatial task for 5 minutes. This task was

designed to load the visuo-spatial scratchpad but not the phonological loop. Rather than

letters and digits, the computer displayed equal-sized arrows that pointed up, down, left,

or right. The participants' task was to maintain in memory a representation of an x, y

coordinate system and an imaginary placeholder at the 0,0 coordinate. They were

instructed to move the imaginary placeholder one standard, but arbitrary, unit of distance

in the direction of the arrow presented. They were further instructed to click the mouse

button any time the placeholder moved away from either axis and then returned to either









axis. A new arrow was presented every 5 seconds and, on the average, a response was

required twice per minute. Crossings of the x- and y-axis occurred equally often. If

participants responded correctly a green circle appeared instead of the next arrow, but if

they missed an opportunity to respond correctly a red circle appeared instead of the next

arrow. False alarms did not disrupt the sequence of arrow presentation, and no feedback

was given to indicate a false alarm had been made.

In the final, dual-task phase of the experiment, participants wrote for 20 minutes

on whichever topic they had not already addressed earlier as the primary task. At the same

time, they simultaneously performed the secondary loading task that they had just finished.

Instructions for the loading task, stimuli presentation rate, and method of responding were

identical to the single-task conditions. Participants were instructed to perform both tasks

to the best of their ability. Again, to maintain interest in the task, the researcher promised

that the person who performed the best overall would win $100.

Results and Discussion

Like experiment 2A, document quality was assessed on a 6-point scale using the

essay-sort method (Madigan, Johnson, & Linton, 1994) where two independent raters

sorted essays first into 3 categories of quality, and then rated essays as high or low quality

within category (inter-rater reliability = 84%). Document fluency was measured by the

number of words generated per minute. A signal detection paradigm (Green & Swets,

1966) was employed in the phonological secondary task.

The secondary task met all of the criteria established for secondary task selection,

most important of which was that there was no significant difference between visuo-spatial

and phonological d' scores under single task conditions (3.50 and 4.09 respectively). If the








visuo-spatial task had been re-coded into a verbal task then a phonological representation

of the presented arrows (up, up, right, down, up, left, etc) would need to be maintained in

working memory. If this had been the case then we would have expected performance to

be much worse compared to the phonological secondary task because participants would

need to hold more information (up to 10 chunks) in memory than they would on the verbal

secondary task where only the last character needed to be remembered. If they instead

adopted a visuo-spatial coding strategy and stored the location of a mental placeholder as

they were instructed, then only one chunk needed to be maintained in memory and single-

task performance would be equivalent to the phonological secondary task. No significant

differences were found between the phonological d' (4.09) and the visuo-spatial d' (3.5)

under single task conditions.

Results indicated that much like experiment 2, performance in the present

experiment deteriorated significantly from single to dual-task conditions on all measures

(see Figure 5). Words produced per minute dropped 13% (t (39) = 3.17, p<.O1) and the

quality of the essays declined 13.6% (t (39) = 2.95, 12<.01) with the addition of the

phonological loop secondary task. A 44% decrement was observed in d' comparing single

to dual-task conditions ((39) = 6.01, p <.001).











Quality



Fluency (WPM)



2nd task (d')


-80 -60 -40 -20 0 20 40 60 80
Percent Change

Figure 5. Decrement in quality, fluency, and secondary task d' from single to dual-task
visuo-spatial load conditions.


Combined Analysis of Experiments 2A and 2B

As Figure 6. shows, selectively engaging either component of working memory

reliably deteriorated fluency, quality and secondary task performance. Table 3 lists means

and standard deviations for fluency, quality and secondary task d' scores under single and

dual-task conditions. Writing fluency slowed significantly overall, F(1, 81) = 49.06, P

<.001, and as Kellogg's model anticipates, there was a significant interaction between the

component of working memory that was loaded and the testing conditions (either single or

dual-task), indicating that the phonological secondary task disrupted fluency more than the

visuo-spatial secondary task, F(1,8 1) = 3.99, p .04. Interestingly, the d' measured during

the phonological secondary task declined much more than during the visuo-spatial

scratchpad, as reflected in the interaction between loading task and the testing conditions,

F(1, 81) = 5.79, p < .01. The decrement in d' associated with the dual-task for the visuo-









spatial group was 44%. For the phonological group, the d' fell 60%. Although quality was

not affected more by one working memory load than another, both fluency and scores

deteriorated more in the pairing of the phonological task with writing than in the pairing of

the visuo-spatial task and writing. Despite the equal degradation in writing quality, these

results suggest greater difficulty in timesharing writing with a phonological task than a

with visuo-spatial task which is consistent with Kellogg's (1996) predictions.

Table 3.

Means and standard errors for d', writing speed, and writing quality scores for the single
and dual visuo-spatial and phonological tasks.

Measure Visuo-spatial Visuo-spatial Phonological Phonological
... in1. Task Dual-task Single Task Dual-task
......................................... .............. .......................................... W ................p u -tak................................


Mean
SD
WPM
Mean
SD
Quality
Mean
SD


3.50 1.96 4.09
1.72 0.85 1.73

14.7 12.8 16.2
0.79 0.70 0.62

3.16 2.73 3.18
1.04 .95 0.92


1.64
0.67

12.8
0.57

2.79
0.82












Quality JNvs-P



Fluency (WPM)



2nd task (d')


-80 -60 -40 -20 0 20 40 60 80
Percent Change
Figure 6. Decrement in quality, fluency, and secondary task d' from single to dual-task
conditions separated by resource loading condition.


These findings support the recent trend toward multi-component models of

working memory, and in particular those that distinguish between visuo-spatial and verbal

resources. It is clear that these results do not support the undifferentiated class of models

that dominated early writing process research. Undifferentiated models would predict that

equally demanding secondary tasks would be equally interfering with the writing primary

task. Our phonological secondary task clearly interfered more than the visuo-spatial

secondary task despite no single-task performance differences (in fact, trends in the single-

task data indicated that the visuo-spatial task was slightly more difficult than the

phonological task although differences were not significant).

In the future this line of research may be extended to the subprocess level of

writing where we can investigate the demands that formulating, executing and monitoring

place on the visuo-spatial scratchpad and phonological loop. These future investigations






57


will employ methods based on a resource capacity assumption of phonological loop and

visuo-spatial scratchpad processing, and will include the methods for selecting secondary

tasks that we specified earlier.














CHAPTER 4
EXPERIMENT 3

Experiments 2a and 2b lend support to the notion that visuo-spatial processes are

active in writing, because of the size of decrements in writing quality and fluency observed

in the presence of the visuo-spatial secondary task. However, questions about the specific

role of visuo-spatial processes in writing remain unanswered. Kellogg (1996) suggests that

planning may draw on visuo-spatial processes when ideas, organizational schemes, or

appearances of the orthography and layout are visualized. The current experiment tests

Kellogg's claims by manipulating the writing topic so that one of two similar essays

requires more visuo-spatial processing than the other. Under dual-task conditions designed

to load the visuo-spatial scratchpad, a visuo-spatial writing topic should be more difficult

than a similar non visuo-spatial writing topic.

Some clues about the involvement of visuo-spatial processes in language tasks

come from in the hemispheric asymmetry literature. Ornstein, Herron, Johnstone and

Swencionis (1979) compared the EEG measure of brain activity of participants while they

read either technical passages or folktales. The left hemisphere acted the same with the

technical material and the folktales. However, the right hemisphere activated while the

subjects read the folk stories, but it did not while they read the technical material. Ornstein

et al., explain that technical material is almost exclusively imageless, while stories generate

lots of images, and that "the sense of a story emerges through style, images, and feelings"

(Ornstein, 1998, p. 73). Thus, reading may rely on the visuo-spatial processes associated








with the right hemisphere as well as the left hemisphere, depending on what is being read.

In another investigation of visuo-spatial information within language, Atwood

(1971) carried out an experiment in which participants heard either highly imageable

phrases such as "Nudist devouring bird", or highly abstract phrases such as "The intellect

of Einstein was a miracle." Each phrase was followed by a simple classification task

presentated either audtitorily or visually. Atwood reported a tendency for the imageable

phrases to be disrupted much more by processing the visually presented digits, while the

abstract sentences were disrupted more by auditory processing.

The current experiment tests Kellogg's (1996) claims that visuo-spatial resources

are involved in planning, particularly when writers plan by visualizing ideas. This

experiment tests this claim by comparing writing on two different topics. Both topics are

descriptive in nature, but one requires visualization and spatial memory while the other

does not. Quality and fluency are expected to be better on the non visuo-spatial topic

under dual-task conditions, but no differences are expected under single-task conditions.

Method

Participants

Fifty-eight general psychology students, from the University of Florida participated

in this experiment for which they received credit towards course requirement. To assess

differences in quality or fluency associated only with topic, the first twenty of these

students wrote two essays, one on each topic under single-task conditions. The remaining

thirty-eight students participated in all phases of the experiment except single-task writing

conditions. All reported that they were comfortable interacting with computers with a

mouse and were proficient typists.








Apparatus and Materials

Participants were each seated at an IBM-compatible 586 computer running

Microsoft Windows 95 with 15 inch monitors. Stimuli in all phases of the experiment

were created in Microsoft Visual Basic version 4. Two descriptive essay topics were

employed by this study. One essay required participants to describe the spatial layout of

the house they grew up in, while the other topic required them to describe the perfect job.

Procedure

There were 4 phases to this experiment; A typing test, single-task performance of a

visuo-spatial task, and two dual-task conditions. All phases of this experiment required

participants to interact with a computer.

In order to assess typing proficiency, participants were given 2 minutes to

duplicate a passage of text as quickly as the could. The visuo-spatial monitoring task from

experiment 2 was used here for 5 minutes. This task was designed to load the visuo-spatial

scratchpad but not the phonological loop. The computer displayed equal-sized arrows that

pointed up, down, left, or right. The participants' task was to maintain in memory a

representation of an x, y coordinate system and an imaginary placeholder at the 0,0

coordinate. They were instructed to move the imaginary placeholder one standard, but

arbitrary, unit of distance in the direction of the arrow presented. They were further

instructed to click the mouse button any time the placeholder moved away from either axis

and then returned to that axis. A new arrow was presented every 5 seconds and, on the

average, a response was required twice per minute. Crossings of the x- and y-axis

occurred equally often.








During each of two 10-minute dual-task phases of the experiment, participants

simultaneously performed the visuo-spatial task and wrote an essay on a given topic. The

essay topics "The perfect job" and a "The house you grew up in" were counterbalanced to

assess practice effects associated with order.

Results and Discussion

Essay quality was again assessed on a 6-point scale using the essay-sort method

(inter-rater reliability = 84%) (Madigan, Johnson, & Linton, 1994). Essay quality was also

assessed by Flesch-Kincaid grade level calculations which are based on average number of

words per sentence and the average number or syllables per word. Document fluency was

measured by the number of words generated per minute, and a signal detection paradigm

(Green & Swets, 1966) was employed in the visuo-spatial secondary task. Additional

calculations of the percentage of sentences that contained spatial information were

calculated for the essays written on the visuo-spatial topic under both single and dual-task

conditions.

As in experiments 2a and 2b, dual-task conditions here were first partitioned into

2.5 minute epochs for which individual d' scores were calculated. Again, patterns of d'

scores indicated that secondary task performance was not automated over time.

As illustrated in Table 2, secondary task d' scores were not significantly different

between the two writing topic conditions under either single or dual-task conditions, and

unlike experiments 2a and 2b, no decrement in d' scores was found when comparing

single to dual-task conditions (see figure 7). Failure to replicate the d' decrements

observed in experiments 2a and 2b may indicate differences in the research participant

pool from Summer to Spring semesters.











Table 2

Means and standard deviations for d' scores for the single-task and dual-task with visuo-
spatial and non-visuo-spatial writing topics.

Measure Single-Task Visuo-spatial Non-visuo-spatial
Topic Topic
Dual-task Dual-task


d'
Mean 4.36 3.70 3.73
SD 4.19 4.11 3.98


As expected, no quality, grade level, or fluency differences were found between

topics under single-task conditions (see Table 3) suggesting that there were no pre-

existing differences associated with the two topics.

Table 3

Means and standard deviations for Flesch-Kincaid grade level assessment and words per
minute for each writing topic under single and dual-task conditions.

Measure Visuo-spatial Non- Visuo- Visuo-spatial Non- Visuo-
Essay Single- spatial Essay Essay Dual-task spatial Essay
Task Single-Task Dual-task
T i........................ ..........................!.k......... W........................... ................................................................. .a ...a...s.........................
Grade Level
Mean 9.48 9.32 4.73 6.99
SD 2.06 2.11 1.68 1.9
Quality
Mean 3.25 3.22 2.65 2.78
SD .96 .92 .87 .88
WPM
Mean 18.8 19.7 15.1 16.5
SD .43 .45 .36 .47


As in experiment 2A and 2B, quality, fluency, and grade level measures were

disrupted by the addition of the secondary task (F (1,56) = 13.46, p < .001), (E (1,56)








10.23, p < .001) and (F (1,56) =63.19, p<.001) respectively (see Figure 7). The pattern of

results for quality and fluency under dual-task conditions indicated differences between

topics in the direction hypothesized, but those trends did not reach significance. Flesch-

Kincaid grade level calculations did support the hypothesis. The non visuo-spatial topic

grade levels (avg. = 6.99) were significantly better than grade levels for the visuo-spatial

topic (avg. = 4.73) under dual-task conditions as evidenced by a significant interaction

between topic and task condition (F (1, 56) = 13.40, p = .00 1).






house
fluency

quality

grade level

-50 -30 -10 10 30 50
Percent Decrement


Figure 7. Dual-task decrement in secondary task d', fluency, quality ratings, and Flesch-
Kincaid grade level assessments.


Although the expected differences between quality and fluency of the two topics

under dual-task conditions were not observed, the grade level assessment suggests greater

interference between the secondary task and writing on the visuo-spatial topic. Trends in

the quality and fluency ratings were consistent with the hypotheses, but differences did not

reach significance. One explanation for higher than expected visuo-spatial essay quality

was that writers might have avoided visuo-spatial planning in the presence of a visuo-









spatial secondary task. To test this possibility, the percentage of sentences that contained

some visual or spatial description was calculated. Sentences such as, "if you walk into the

master bedroom, on the right, after the closet, is a sink, then a bathroom" were considered

spatial sentences whereas sentences such as "My house was full of nice furniture that I

liked to play around" were considered non-spatial. The visuo-spatial essay written in the

presence of the secondary task contained significantly fewer spatial description sentences

than essays written on the same topic in isolation (56% Vs 70% respectively) (t (56) =1.9,

p= .03). More research is needed to control for the possibility that a phonological secondary

task might have the same effect, but this result suggests that the presence of the secondary

task changed the way the writers addressed the topic.














CHAPTER 5
GENERAL DISCUSSION

Experiment 1 illustrated the varying capacity in the phonological loop which

clearly presents explanatory difficulties for classes of models, including Baddeley and

Hitch's (1974) working memory, that postulate structures such as a phonological loop or

a visuo-spatial scratchpad to operate as binary, all-or-none mechanisms. Instead, our data

are compatible with an important, but seemingly subtle, variation that conceptualizes these

mechanisms as pools of resources that can operate in a continuous or graded fashion. The

distinction is important for writing theorists and researchers because it may help them to

make closer contact with colleagues who are exploring other aspects of human cognitive

processing, including speech production, language comprehension, or problem solving.

Questions of the processing nature of the visuo-spatial scratchpad remain

unresolved, but the methodology described in Experiment 1 lends itself well to this

question. Under dual-task conditions, if two visuo-spatial secondary tasks of varying

difficulty interfere equally with a primary task, then the all-or-none processing assumption

is supported. On the other hand, if the two visuo-spatial secondary tasks vary in the

relative degree to which they disrupt primary task performance, then graded processing is

supported.

Postulating that the visuo-spatial scratchpad and phonological loop comprise pools

or resources necessitates new guidelines for selecting secondary tasks. Five guidelines are

especially important for comparisons of visuo-spatial and verbal resource involvement: 1.








Secondary tasks should be equally demanding of attentional resources; 2. The

loading tasks should not change in the degree to which they draw on controlled processes

throughout the experiment; 3. The tasks should have similar structural composition,

requiring the same mode of input and output; 4. Response selection and execution should

occur equally often in the tasks; and 5. Finally, secondary tasks should require different

strategies for their successful performance.

Experiments 2a and 2b followed the secondary task guidelines described above.

Single-task d' scores were not significantly different between visuo-spatial and verbal

loading tasks, therefore suggesting that these tasks are equally demanding. To assure that

secondary tasks did not become automatic, 20-minute dual-task conditions were broken

down into 4 five-minute epochs for which d' scores were calculated. Patterns of d' scores

were inspected qualitatively on an individual basis with the intention of discarding

participants from the study if their performance changed over the course of the dual-task

conditions. No participants were excluded from this study based on this information.

Stimulus presentation modality and rate and response rate were identical for the visuo-

spatial and the verbal secondary tasks. Secondary tasks were constructed so that

performance differences would emerge if participants used the same strategies for each

task.

Experiment 2A reported a significant decrement in essay quality and fluency

associated with loading the phonological loop. These decrements in quality are consistent

with disruptions in comprehension in the presence of irrelevant speech (Martin, Wogalter,

& Forlanon, 1988), and lower holistic quality of essays and diminished syntactic

complexity as measured by the proportion of subordinate clauses occurring before the








main verb (Madigan & Linton, 1996). On the other hand, these findings seem to conflict

with Ransdell, Levy & Kellogg (in press) where an irrelevant speech manipulation

reportedly had no effect on essay quality. Inconsistencies here may point to differences

between irrelevant speech manipulations and the letter matching task employed in

Experiment 2A. In particular, response selection and execution stages required by the

letter matching task are likely to place greater demands on working memory, thus having a

greater impact on primary task performance.

Fluency decrements reported in Experiment 2A support Ransdell, Levy, &

Kellogg's (1997) finding that irrelevant speech reliably reduced fluency as measured by

words per minute. Madigan and Linton (1996) however, found no effect of irrelevant

speech on fluency as measured by word production times and pauses within selected

clauses. This raises questions concerning the reliability of different fluency measures.

Experiment 2B tested the hypothesis that a visuo-spatial secondary task

equal in difficulty to the phonological secondary task would be easier to time-share with

writing. This hypothesis is based on Kellogg's (1996) claims that writing depends more on

the phonological loop than on the visuo-spatial scratchpad. Results indicated that much

like Experiment 2, quality and fluency deteriorated significantly with the addition of the

visuo-spatial secondary task. Despite equal degradation in writing quality and fluency,

results suggested greater difficulty in timesharing writing with a phonological task because

the phonological secondary task performance fell much more drastically than the visuo-

spatial secondary task performance.

Experiments 2a and 2b therefore indicate that writing draws more on verbal

resources than visuo-spatial resources, which is consistent with EEG studies of writing








(Ornstein, Herron, Johnstone and Swencionis (1979). These results support a

multicomponent resource model of writing that specifies a division of visuo-spatial and

verbal resources instead of one undifferentiated pool of resources (Kahneman, 1973).

Undifferentiated models predict that equally demanding secondary tasks would be equally

interfering with a primary task. Based on single-task performance measures, our visuo-

spatial and phonological tasks were statistically equivalent, nevertheless, when the loading

tasks were performed at the same time that participants engaged in text production, the

phonological task clearly interfered more than the visuo-spatial task.

Support for a role of visuo-spatial working memory in writing is suggested by the

high interference between writing and a visuo-spatial task. However, as pointed out

earlier, interference could also be due to the response selection and execution aspects of

the visuo-spatial task. How can response selection and execution be teased out of a visuo-

spatial task without loosing some performance measure for that task? One possibility

involves postponing response selection and execution stages until after the dual-task

conditions are completed. For example, visuo-spatial stimuli may be presented as a

secondary task, and memory for the presented information can be tested after completion

of the dual-task condition, thus avoiding response selection and execution stages during

the dual-task condition. The disadvantage to this postponement ofjudgements is that this

type of visuo-spatial task shifts the emphasis from processing to storage, and this

distinction is not yet fully understood. Klapp and Netick (1988) report interference

between two processing tasks and two memory storage tasks, but not between a

processing and storage task suggesting that processing and storage resources are divided.

Wickens (1984) does not recognize storage and processing as an important distinction









though. Other methods of loading working memory in specific ways and then examining

their effects on the output of one or more systems or basic processes should be explored

further. One such method is embodied in the approach taken by Experiment 3.

Support for multicomponent resource models represents new trend in writing

research. Until recently, process-oriented writing research has been theoretically rooted in

accounts of attention based on undifferentiated capacity such as the model advanced by

Kahneman (1973). Writing research progressed under single capacity models as evidenced

by the influence of the Hayes & Flower (1982) model of writing and successful

development of computer-aided writing tools (Kellogg, 1994; Lea, Levy & Marek, 1995;

Lea, Rosen, Levy, Marek & Ransdell, 1995). Recent considerations of the role of limited

cognitive capacity in writing necessitate the more specific models of cognitive limitations

such as Baddeley's working memory model and multiple resource models.

Experiment 3 took a new approach to test specific hypotheses about the role of

visuo-spatial processes in writing. Specifically, this experiment tested Kellogg's (1996)

claims by manipulating writing topic so that one of two similar essays required more

visualization or spatial processing than the other. In a dual-task paradigm designed to load

the visuo-spatial scratchpad, the more visuo-spatial of the two topics should be more

difficult to timeshare.

Although significant differences were not found between fluency and quality, the

pattern of results matched our expectations. Significant differences in Flesch-Kincaid

grade level assessments supported the hypothesis indicating that more visuo-spatial

resources were required by the visuo-spatial writing topic. These findings are consistent

with Ornsetien, Herron, Johnstone, and Swencionis' (1979) findings that the right








hemisphere EEG is more active in reading passages containing imagery than technical

passages, and Atwood's (1971) finding that imageable phrases interfered more with a

visual than an auditory secondary task.

Visuo-spatial essay quality was higher than expected, raised concerns that

participants may have preserved essay quality by avoiding visuo-spatial planning by

focusing on non-spatial aspects of the house they grew up in. Closer examination of the

visuo-spatial essays revealed a significantly higher percentage of sentences devoted to

spatial description under single-task conditions. Further testing is needed to clarify

whether this effect is due to the visuo-spatial nature of the secondary task.

We expect this line of research to extend to the subprocess level of writing,

enabling investigations of the demands that formulating, executing and monitoring place

on the visuo-spatial scratchpad and phonological loop. These future investigations will

employ methods based on a resource capacity assumption of phonological loop and visuo-

spatial scratchpad processing, and will include the methods for selecting secondary tasks

that we addressed earlier.

Current research in our laboratory ( Levy, White, Lea, & Ransdell, 1998) extends

the methodologies introduced here to evaluate other claims from current working

memory-based models of text production. For example, we have recently devised a way to

overcome the inherent confounding in Experiments 2 and 3 between the specific stimuli

shown and the participants' response decision rules by holding the stimuli constant. Thus,

single alphabetic characters might be shown at the same rate as the stimuli presented as

before, but periodically they would simultaneously change color, case, font, physical

location, and value. Different groups of participants asked to respond whenever two








stimuli in succession were the same color or location (to engage the visuo-spatial

scratchpad), or whenever the adjacent stimuli both contained a long "e" phoneme (to

engage the phonological loop, or whenever they formed a 2-letter word (to focus on the

central executive). The physical stimuli would remain the same as would the physical

response. Any difficulties between the single tasks or between their effects on the writing

process would then simply be the result of the instructions designed to engage working

memory components differentially.

Clearly much remains to be discovered concerning visuo-spatial working memory

in general and specifically in the role of visuo-spatial processes in writing. It is likely that

developments in working memory and multiple resource theories will continue to influence

the way visuo-spatial and verbal processes are investigated.






APPENDIX A
Experiment 1 stimuli

1 wdjkm
2 ztbpk
3 hqscv
4 kpjfxn
5 fwykpv
6 rwfgkr
7 wt fj np k
8 qtpkhds





9 zcbmkyr
10 drfvyj np
11 k 1 y n ft r g
12 xtvhkwdg
13 wdvyhjlmq
14 rpyj nbx c w
15 zrxfhknpy
16 pkgvw
17 sryvx
18 wgthk





19 pyjnm
20 w d c v g
21 zrvhy
22 tgj vwd
23 wfthvk
24 wdgyjk
25 pfbszt
26 yj v zrn
27 kdrthq
28 htfvswg





29 srcgbkp
30 q z g 1 p y m
31 kgbtynr
32 ztyvbkp
33 lhkvrdg
34 qdz btny
35 frwxscbh
36 yhj klb fw
37 dztcybkn
38 dgj lkhfs





39 qtrwyphk
40 zdxfcgvhb
41 zmygcerxw
42 dj fkgmxnc
43 pkntbrcms
44 qzwdxgchb

45 g d fvy rnkm















REFERENCES


Atkinson, R.C., & Shiffrin, R.M. (1968). Human memory: A proposed system
and its control processes. In K.W. Spence & J.T. Spence (Eds.), The psychology of
learning and motivation: II. New York: Academic Press.

Atwood, G.E. (1971). An experimental study of visual imagination and memory.
Cognitive Psychology, 2, 290-299.

Baddeley, A.D. (1966). Short-term memory for word sequences as a function of
acoustic, semantic and formal similarity. Quarterly Journal of Experimental Psychology,
18, 362-366.

Baddeley, A.D. (1992). Is working memory working? The fifteenth Bartlett
lecture. The Quarterly Journal of Experimental Psychology, 44a (1), 1-31.

Baddeley, A.D. (1993). Working memory or working attention? In A. Baddeley &
L. Weiskrantz (Eds.) Attention: Selection awareness and control. (pp.152-170) New
York: Oxford.

Baddeley, A.D., & Hitch, G.J. (1974). Working memory. In G. Bower (Ed.), The
psychology of learning and motivation, Vol. III, pp.47-90, New York: Academic Press.

Baddeley, A.D., & Lewis, V. (1981). Inner active processes in reading: The inner
voice, the inner ear, and the inner eye. In A.M. Lesgold & C.A. Perfetti (Eds.), Interactive
processes in reading. Hillsdale, NJ: Lawrence Erlbaum Associates.

Baddeley, A.D., Lewis, V., Eldridge, M., & Thomson, N. (1984a). Attention and
retrieval from long-term memory. Journal of Experimental Psychology: General, 1113,
518-540.

Baddeley, A.D., Lewis, V., & Vallar, G. (1 984b). Exploring the articulatory loop.
The Quarterly Journal of Experimental Psychology, 36A, 233-252.

Baddeley, A.D., Papagno, C., & Vallar, G. (1988). When long-term learning
depends on short-term storage. Journal of Memory and Language, 27, 586-595.

Baddeley, A.D., Thomson, N., & Buchannon, M. (1975). Word length and the
structure of short-term memory. Journal of Verbal Learning and Verbal Behavior, 14,
575-589.








Bertelson, P. (1967). The psychological refractory period of choice reaction times
with regular and irregular ISI's. Acta Psychologica, 27, 45-56.

Bock, J.K. (1982). Toward a cognitive psychology of syntax: Information
processing contributions to sentence formulation. Psychological Review, 89, 1-47.

Boles, D.B., & Wickens, C.D. (1987). Display formatting in information
integration and nonintegration tasks. Human Factors, 29(4), 395-406.

Bourdin, B., & Fayol, M. (1994). Is written language production more difficult
than oral language production? A working memory approach. International Journal of
Psychology 29, 591-620.

Broadbent, D. (1958). Perception and communication. Oxford: Permagon.

Broadbent, D. (1982). Task communication and the selective intake of
information. Acta Psychologica, 50, 253-290.

Brooks, L.R. (1968). Spatial and verbal components of the act of recall, Canadian
Journal of Psychology, 22, 349-368.

Cherry, C. (1953). Some experiments on the recognition of speech with one and
two ears. Journal of the Acoustical Society of America, 23, 915-919.

Conrad, R., & Hull, A.J. (1964). Information, acoustic confusion and memory
span. British Journal of Psychology, 55, 429-432.

Craik, F., & Lockhart, F.S. (1972). Levels of processing: A framework for
memory research. Journal of Verbal Learning and Verbal Behavior. 11, 671-684.

Craik, F.I., & Watkins, M.J. (1973). The role of rehearsal in short-term memory.
Journal of Verbal Learning and Verbal Behavior, 12, 599-607.

Daneman, M., & Carpenter, P. (1980). Individual differences in working memory
and reading. Journal of Verbal Lerning & Verbal Behavior, 19, 450-466.

Daneman, M., & Greene, 1. (1986). Individual differences in comprehending and
producing words in context. Journal of Memory and Language, 25, 1-18.

De Renzi, E., & Nichelli, P. (1975). Verbal and nonverbal short-term memory
impairment following hemispheric damage. Cortex, 11, 341-353.

Dinet, J., & Passerault, J. J. (1998, July). Working memory and text production:
The role of the visuospatial sketchpad in writing argumentative and descriptive texts.
Paper presented at the Writing98 Conference, Poitiers, FR.









Ericsson, K.A., & Kintsch, W. (1995). Long-term working memory. Psychological
Review, 102, 211-245.

Fagot, C., & Pashler, H. (1992). Making two responses to a single object:
Exploring the central attentional bottleneck. Journal of Experimental Psychology: Human
Perception and Performance, 18, 1058-1079.

Fracker, M.L., & Wickens, C.D. (1989). Resources, confusions, and compatibility
in dual-axis tracking: Displays, controls, and dynamics. Journal of Experimental
Psychology: Human Perception and Performance, 15(1), 80-96.

Frick, R.W. (1988). Issues of representation and limited capacity in the visuo-
spatial scratchpad. British Journal of Psychology., 79, 289-308.

Friedman, A., & Polson, M. C. (1981). Hemispheres as independent resource
systems: Limited-capacity processing and cerebral specialization. Journal of Experimental
Psychology: Human Perception and Performance, 7(5) 1031-1058.

Galin, D., & Ornstein, R. (1972). Lateral specialization of cognitive mode: An
EEG study. Psychophysiology. 9, 412-418.

Gathercole, S., & Baddeley, A.D. (1993). Working memory and language. Hove,
UK: Lawrence Erlbaum Associates.

Green, D., & Swets, J. (1966) Signal detection theory and psychophysics. New
York: Wiley.

Hatano, G., & Osawa, K. (1983). Digit memory of grand experts in abacus-
derived mental calculation. Cognition, 15, 95-110.

Hayes, J. (1996). A new framework for understanding cognition and affect in
writing. In C.M. Levy & S. Ransdell (Eds.) The science of writing. (pp. 1-27) Mahwah,
NJ: Erlbaum.

Henry, F.M., & Rogers, D.E. (1960). Increased response latency for complicated
movements and a "memory drum" theory of neuromotor reaction. Research Quarterly,
31, 448-458.

Herdman, C.M., & Friedman, A. (1985). Multiple resources in divided attention: A
cross-modal test of the independence of hemispheric resources. Journal of Experimental
Psychology: Human Perception and Performance, 11 (1) 40-49.

Hirst, W., & Kalmar, D. (1987). Characterizing attentional resources. Journal of
Experimental Psychology: General, 116, 1, 68-81.








Hue, C., & Ericson, J.R. (1988). Short-term memory for Chinese characters and
radicals. Memory and Cognition, 16, 196-205.

Jefferey, G.C., & Underwood, G. (1995). The role of working memory in the
development of a writing skill: Learning to co-ordinate ideas within written text.
(Unpublished manuscript).

Jones, D.M., Miles, C., & Page, J. (1990). Disruption of proofreading by irrelevant
speech: Effects of attention, arousal or memory? Applied Cognitive Psychology, 4, 89-
108.

Kahneman, D. (1973) Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.

Kahneman, D., Triesman, A., & Gibbs, B.J_ (1992). The reviewing of object files:
Object specific integration of information. Cognitive Psychology, 24, 175-219.

Kantowitz, B.H. (1974). Double stimulation. In B.H. Kantowitz (Ed.), Human
information processing. Hillsdale, NJ.: Earlbaum.

Kellogg, R.T. (1996). A model of working memory in writing. In C.M. Levy & S.
Ransdell (Eds.) The science of writing. (pp. 57-71) Mahwah, NJ: Erlbaum.

Kinsbourne, M., & Hicks, R.(1978). Functional cerebral space. In J. Requin (Ed.),
Attention and performance VII. Hillsdale, NJ: Erlbaum.

Klapp, S.T., & Netick, A. (1988). Multiple resources for processing and storage in
short-term working memory, Human Factors 30(5), 617-632.

Knowles, W.B. (1963). Operator loading tasks. Human Factors, 5, 151-161.

Laberg, D. (1973). Attention and the measurement of perceptual learning. Memory
& Cognition, 1, 268-276.

Lea, J., Levy, C.M., & Marek-Lovejoy, P. (1995). The topic sentence display tool:
An evaluation of writing performance and process. Presented at the XIII Annual European
Conference on Writing and Computers, London.

Lea, J., Rosen, L., Levy, C.M., Marek-Lovejoy, P., & Ransdell, S.E. (1995). The
intelligent listening word processor: Prospects for the future of writing. Presented at the
XIII Annual European Conference on Writing and Computers, London.

Levy, C.M., & Marek, P. (1998). The role of working memory in writing
processes and performance. In G.C. Jeffery & M. Torrance (Eds.), High and low level








writing processes: Managing cognitive demands. Amsterdam: Amsterdam University
Press.

Levy, C. M., White, K., & Lea, J. (1998, July). The role of the visuo-spatial
scratchpad in writing: Testing Kellogg's multicomponent model. To be presented at the
Writing 98 Conference, Poitiers, FR.

Madigan, R.J., Johnson, S.E., & Linton, P.W. (1994 August). Working memory
capacity and the writing process. Paper presented at the American Psychological Society,
Washington, D.C.

Madigan, R.J., & Linton, P.W. (1996 October). Working memory and the writing
process. (unpublished manuscript).

Manin, K., & Johnson-Laird, P.N. (1982). The mental representation of spatial
descriptions. Memory & Cognition, 10, 181-187.

Martin, R.C., Shelton, J.R., & Yaffee, L.S. (1994). Language processing and
working memory: Neuropsychological evidence for separate phonological and semantic
capabilities. Journal of Memory and Language, 33, 83-111.

Martin, R.C., Wogalter, M.S., & Forlano, J.G. (1988). Reading comprehension in
the presence of unattended speech and music. Journal of Memory & Language, 27, 382-
398.

Marek, P., & Levy, C.M. (1996). The role of working memory in writing
processes and performance: II. Evaluation of formulating, monitoring and execution
processes. Paper presented at the European Conference on Writing and Computers,
Barcelona.

McFarland, K., & Ashton, R. (1978). The influence of concurrent task difficulty on
manual performance. Neurophysiologica, 16, 735-741.

Moray, N. (1959). Attention in dichotic listening. Quarterly Journal of
Experimental Psychology, 11, 56-60.

Moray, N. (1982). Subjective mental load. Human Factors, 23, 25-40.

Moray, N., Johannsen, G., Pew, R.W., Rasmussen, J., Sanders, A.F., & Wickens,
C.D. (1979). Report of the experimental psychology group. In N. Moray (Ed.), Mental
workload: Its theory and measurement. New York: Plenum.

Moscovitsch, M., & Klein, D. (1980). Material -specific perception for visual
words and faces. Journal of Experimental Psychology: Human Perception and
Performance, 6, 590-603.








Norman, D., & Bobrow, D. (1975). On data limited and resource limited
processing. Journal of Cognitive Psychology, 7, 44-60.

Navon, D. (1984). Resources-A theoretical soup stone? Psychological Review, 91
(2), 216-234.

Navon, D., & Gopher, D. (1979). On the economy of the human-information
processing system. Psychological Review, 86(3), 214-255.

Ornstein, R. (1997). The right mind: Making sense of the hemispheres. New York:
Harcourt Brace & Company.

Ornstein, R., Herron, J., Johnstone, J., & Swencionis, C. (1979). Differential right
hemisphere involvement in two reading tasks. Psychophysiology, 16(4), 398-401.

Paivio, A. (1986). Mental representations: A dual-coding approach. New York:
Oxford University Press.

Pashler, H. (1984). Processing stages in overlapping tasks: Evidence for a central
bottleneck. Journal of Experimental Psychology: Human Perception and Performance, 10,
358-377.

Pashler, H., & Carrier, M. (1995). Attentional limits in memory retrieval. Journal
of Experimental Psychology: Learning Memory and Cognition, 21(5), 1339-1348.

Pashler, H., & Johnston, J. C. (1989). Interference between temporally
overlapping tasks: Chronometric evidence for ventral postponement with or without
response grouping. Quarterly Journal of Experimental Psychology, 41 A, 19-45.

Ransdell, S., & Levy, C.M. (1996). Working memory constraints on writing
quality and fluency. In C.M. Levy & S. Ransdell (Eds.) The science of writing. (pp. 93-
105) Mahwah, NJ: Erlbaum.

Ransdell, S., & Levy, C.M. (1998). Writing, reading and speaking memory spans
and the importance of resourse flexibility. In G.C. Jeffery & M. Torrance (Eds.), High and
low level writing processes: managing cognitive demands. Amsterdam: Amsterdam
University Press.

Ransdell, S., Levy, C.M., & Kellogg, R.T. (in press). Concurrent loads on
working memory during text production. Memory.

Rolfe, J.M. (1971). The secondary task as a measure of mental load. In W.T.
Singleton, J.G. Fox, & D. Whitfield (Eds.) Measurement of man at work. London: Taylor
and Francis.









Salame', P., & Baddeley, A.D. (1982). Disruption of short-term memory by
unattended speech: Implications for the structure of working memory. Journal of Verbal
Learning and Verbal Behavior, 21, 150-164.

Shallice, T., & Warrington, E.K. (1970). Independent functioning of verbal
memory stores: A neuropsychological study. Quarterly Journal of Experimental
Psychology, 22, 261-273.

Shepard, R.N. (1978). The mental image. American Psychologist, 33, 125-137.

Sternberg, S. (1969). The discovery of processing stages: Extensions of Donders'
method. Acta Psychologica, 30, 276-315.

Thomson, J.A. (1983). Is continuous visual monitoring really necessary in visually
guided locomotion? Journal of Experimental Psychology: Human Perception and
Performance. 9, 427-443.

Triesman, A.M., & Davies, A. (1973). Divided attention to ear and eye. In S.
Komblum (Ed.). Attention and performance (Vol 4). New York: Academic Press.

Tulving, E. (1966). Subjective organization and effects of repetition in multi-trial
free-recall learning. Journal of Verbal Learning and Verbal Behavior, 6, 193-197.

Tzeng, O.J. (1973). Positive recency effects in delayed free recall. Journal of
Verbal Learning and Verbal Behavior, 12, 436-439.

Vallar, G., & Baddeley, A.D. (1984). Fractionation of working memory.
Neuropsychological evidence for a phonological short-term store. Journal of Verbal
Learning and Verbal Behavior, 23, 151-161.

Waters, G.S., Rochon, E., & Caplan, D. (1992). The role of high-level speech
planning in rehearsal: Evidence from patients with apraxia of speech. Journal of Memory
and Language, 31, 54-73.

Welford, A.T. (1967). Single channel operations in the brain. Acta Psycholog ca.
27,5-22.

Wickens, C.D. (1976). The effects of divided attention on information processing
in manual tracking. Journal of Experimental Psychology: Human Perception and
Performance, 2(1), 1-13.

Wickens, C.D. (1980). Processing resourc demands of failure detection in dynamic
systems. Journal of Experimental Psychology: Human Perception and Performance, 6(3),
564-577.





84

Wickens, C.D. (1984). Processing resources in attention. In R. Parasuraman (Ed.)
Varieties of attention. New York: Academic Press.

Yee, P.L., Hunt, E., & Pellegringo, J.W. (1991). Coordinating cognitive
information: Task effects and individual differencs in integrating information from several
sources. Cognitive Psychology, 23, 615-680.














BIOGRAPHICAL SKETCH

Joseph D. Lea III was born November 28, 1970, in Tuscon, Arizona, to Joseph D.

Lea Jr. and Billie Ann Lea. Joseph (Josh) majored in psychology at the University of

North Carolina at Wilmington and graduated cum laude with honors in psychology in

May, 1993. Joseph (Joe) entered the Cognitive and Sensory Processes doctoral program

at the University of Florida in August, 1993. After graduation Joseph will begin work at

IBM in San Jose, California as a human factors engineer in software development.








I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Docor osoh,

C" Xchael*Le Chairan
Professor of Psychology

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequa e, in scope and
quality, as a dissertation for the degree of Dotrg hilopy., in

Barton Weitz,
J.C. Penny Eminent Scholar

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree ofDoctor-g sophy

Shari Ellis,
Assistant Professor of Psychology

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in sope and
quality, as a dissertation for the degree of Doctor Phpopy.


Professor of Psychology

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Doctw of Philoso hy.

Stepheb-Blessing,
Assistant Professor of Psychology

This dissertation was submitted to the Graduate Faculty of the Department of
Psychology in the College of Liberal Arts and Sciences and to the Graduate School
and was accepted as partial fulfillment of the requirements for the degree of Doctor
of Philosophy.

August, 1998
Dean, Graduate School




Full Text
VERBAL AND VISUO-SPAXIAL PROCESSING DEMANDS IN WRITING
By
JOSEPH D. LEA
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1998

ACKNOWLEDGEMENTS
I would like to thank my chair, Dr. Michael Levy, for all of the time and effort he
has given to this project and for all of the support he has given me over the years. I would
also like to thank my other committee members, Dr. Ira Fischler, Dr. Stephen Blessing,
Dr. Bart Weitz, and Dr. Shari Ellis for their time and suggestions which have greatly
improved this project. In addition, I would like to thank Dr. Lise Abrams for attending
the defense, and for her many helpful comments and suggestions. I would also like to
take this opportunity to thank my classmates Pam Marek, Brian Howland, Nancy
Lincoln, Michele Simmons, Ryan West, Katherine White, Glenn Gunzelman, and Dean
Sabatinelli for their camaraderie. Special thanks to IBM’s Santa Teresa Laboratory for
the incentive to finish this project on schedule. Finally, I would like to thank my wife,
Elena Schuhmann, my parents, and my grandfather for all of their support and
encouragement.

TABLE OF CONTENTS
Page
AKNO WLEDGEMENT S ii
ABSTRACT v
CHAPTERS
1 INTRODUCTION 1
Resources and. Structures 4
Capacity Theories 5
Structural Theories 8
Multiple Resource Theories 12
Working Memory 15
Working Memory in Writing 20
Adapted Working Memory Model 29
Summary 32
2 EXPERIMENT 1 33
Method 35
Results and Discussion 36
3 EXPERIMENT 2 39
Experiment 2A 39
Method 44
Results and Discussion 45
EXPERIMENT 2B 48
Method 50
Results and Discussion 52
Combined Analysis of Experiments 2A and 2B 54
4 EXPERIMENT 3 58
Method 59
Results and Discussion 61
5 GENERAL DISCUSSION 65

APPENDIX 72
REFERENCES 77
BIOGRAPHICAL
SKETCH 85
iv

Abstract of Dissertation Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
VERBAL AND VISUO-SPATIAL PROCESSING DEMANDS IN WRITING
By
Joseph D. Lea III
August, 1998
Chaiiperson: C. Michael Levy
Major Department: Psychology
Writing research has recently begun to consider the role that limitations in
cognitive capacity play in the writing process, thus obligating writing researchers to
examine existing theories of cognitive limitations. This dissertation reviews four major
classes of cognitive processing models that might serve as theoretical bases for extension
to written language production. Four experiments assess the merits of a
reconceptualization of Kellogg's (1996) model of working memory and writing by
emphasizing the role of resource pools rather than binary functioning cognitive
structures. Empirical findings support the recent trend toward multicomponent classes of
working memory or resource models, particularly those that distinguish between visuo-
spatial and verbal/phonological processing. Results also indicate that there is a significant
visuo-spatial component in the planning stages of writing.

CHAPTER 1
INTRODUCTION
Writing research has recently begun to consider the role that limitations in
cognitive capacity play in the writing process. A common assumption is that writers have a
limited cognitive capacity, the size and/or efficiency of which varies between individuals.
Research on cognitive capacity and writing is relatively new compared to studies of
capacity limitations in reading, and to spoken language production and comprehension
(see Gathercole and Baddeley, 1993 for an overview), obligating writing researchers to
develop and refine appropriate models necessary for theoretical development and testing.
Fortunately, the nature of human processing and storage limitations has long interested
psychologists working within an information processing framework, and accounts of
other language production tasks have led to the development of constructs such as
working memory, resource capacity, and mental workload.
In part, the contribution of this dissertation lies in the extension of debates in the
working memory and attention literature to the nearly untouched domain of writing. As
both a thinking and language production task, writing research may challenge these
existing models. Four approaches to cognitive limitations are introduced below, followed
by a critical analysis of information processing theory development, and rationale for the
formation of a hybrid model that the employs elements of these models with the greatest
explanatory power for written language production tasks.
1

2
• Structural theories were popularized in the 1950s with experimental investigations of
the dichotic listening task which revealed that attention was severely limited (Cherry,
1953; Broadbent, 1958; Moray, 1959). Several classes of theoretical models were
generated to localize the point in information processing where the bottleneck occurs.
Early-selection theories claimed that this bottleneck is in the early processes such as
perception, and late-selection theories pointed to stages of decision making and
response selection as the limiting stages of information processing.
• Early human factors research in the measurement of mental workload led to the
conceptualization of the human operator possessing a pool of limited-capacity
resources (Knowles, 1963). This resource capacity approach offers the simplest
framework for interpreting the demands placed on the writer by postulating an
undifferentiated pool of mental resources (Kahneman, 1973). Performance on any
given task declines only when the aggregate of task demands exceed the capacity of
resources.
• An alternate conception of this resource economy makes a distinction in the type of
resources available at any given moment. This “multiple resource” approach posits that
there may be various types of resources and that different tasks may require different
types of resources in various compositions. The most widely supported division of
resources is a neurologically plausible division between visuo-spatial and verbal
resources (Wickens, 1984; Friedman & Poison, 1981). Friedman and Poison (1981)
V
proposed a framework with the intention of extending multiple resource models to our
understanding of cerebral specialization phenomena. They suggest that the physically
distinct cerebral hemispheres can be seen as independent resource systems, and that

3
the left and right hemispheres together form a system of two mutually inaccessible and
finite pools of resources.
® Baddeley & Hitch’s (1974) model of working memory is a multi-component system
for both the storage and the processing of verbal and visuo-spatial information. It has
three primary components: a limited-capacity central executive and two peripheral
slave systems. One of these slave systems, the phonological loop, handles verbal and
auditory information while the other, the visuo-spatial scratchpad, is specialized for
processing visual and spatial information.
Most approaches to cognitive processing limitations recognize a division in
processing for verbal and visuo-spatial tasks, owing to the vast literature on hemispheric
differences. Since the doctrine of cerebral localization in the late 1700’s, particular
cognitive functions have been assigned to specific regions of the brain. Generally, visuo-
spatial processing has been localized to the right hemisphere while the left hemisphere is
believed to play a stronger role in verbal tasks. Popular multiple resource models
(Wickens, 1984; Friedman & Poison, 1981) and Baddeley and Hitch’s (1974) working
memory model build on this neurological separation of visuo-spatial and verbal
processing.
As a writer plans ideas, translates ideas into prose, types or handwrites sentences,
and monitors all of these activities, many demands are placed on temporary storage and
processing capacity. This dissertation seeks to provide an account of the division of labor
between visuo-spatial and verbal processing demands and capacity limitations as they
influence the writing process. Special treatment will be given to the wholly untouched
question of how visuo-spatial resources constrain writing. The balance of this chapter

4
critically reviews the literature regarding the models introduced above, focusing on their
assumptions about the nature of human cognitive limitations. The review concludes with a
summary and critique of existing literature, followed by a proposal for an adapted theory
of working memory which is tested in Experiment 1. Experiments 2a and 2b compare the
involvement of phonological and visuo-spatial processing in essay writing, and Experiment
3 investigates the role that writing topic plays in the resource demands of writing.
Resources and Structures
Two hypothetical constructs have greatly influenced theory development
addressing cognitive limitations. First, we will consider the notion of limited processing
resources that are demanded by non-automatic tasks. According to resource theories,
performance declines only when the aggregate of task demands exceed the available
resources. For example, when someone uses a word processor to translate an idea into
text, we might expect that simultaneously generating new ideas and editing grammatical
and spelling mistakes would cause a disruption in the rate at which text is generated.
Juggling that writing task with talking on the telephone or listening to the radio would
further drive down performance. A second construct proposed to explain processing
efficiency is referred to as structure. According to a structural view (Pashler &Johnston,
1989), when people attempt to simultaneously engage in competing mental operations,
incompatible stages or processes are initiated, which results in interference between tasks.
In other words, two tasks compete when they each simultaneously require a single
common process. For example, we would expect interference between two tasks such as
conversing on the telephone while simultaneously listening to a radio interview, more so
than the dual-task combination than carrying on a conversation while riding a bike,

5
presumably because similar processes are demanded by producing and comprehending
audible speech.
Two classes of theoretical models of attention have emerged from consideration of
these hypothetical variables. These are capacity or resource theories, described by
Kahneman (1973), and structural theories which began with Broadbent (1958) and
recently have been elaborated by Pashler and his associates (Pashler & Johnston, 1989;
Pashler & Carrier, 1995). After considering structural and capacity theories in more
detail, we will examine multiple resource models and Baddeley’s working memory model
as each offer a rapprochement of structural and capacity theories. Further, Experiment 1
will reveal empirical findings suggesting that writing research may be best guided by a
model that combines the strengths of both working memory and multiple resource
theories.
Capacity Theories
How is it that we may become less conscious-rather than become entirely unaware
- of some things when we become more conscious of others? Capacity theories seek to
explain this common experience by suggesting that some resource pool is approaching its
limitation. Thus, when a primary task demands more of these resources, fewer are
available for a concurrent secondary task. Whereas structural theories assume that
structures are dedicated to one task at a time, the resource view assumes that capacity can
be allocated between separate activities in graded quantity.
Nearly all of the original paradigms investigating the nature of resources (or mental
capacity or effort) involve presenting two tasks to an individual to process simultaneously,
or so-called dual-task conditions. Resource theories in large part were created to explain

6
instances where two processes were simultaneously activated. Perhaps the earliest task
(although not a traditional dual-task) was the Stroop task where reading processes were
observed to compete with color naming processes for a common pathway. Although
theorizing about pathways has evolved into structural theories of attention, the basic
question about mechanisms of interference is the same. In the 1950s British psychologists
Broadbent, Moray, and others began working with the dichotic listening task, to test
various structural models of the selection of attention; for example, late versus early
selection. The probe technique emerged shortly thereafter, where participants respond to a
discrete stimulus while performing some primary task. A participant’s baseline reaction
time was subtracted from his or her reaction time to a tone presented during performance
of a primary task in order to compute an interference reaction time, which was assumed to
be an index of mental effort.
Kahneman (1973) brought unity to the concept of capacity as an intervening
variable in dual-task performance. His ideas and synthesis of the research at that time
facilitated the evolution of the resource metaphor from an intuition to a quantitative theory
with testable predictions. Since the early 1970s, capacity theories have continued to
influence basic and applied research. Capacity theories are especially important in applied
research, where workload measurement is investigated. Capacity theories have been
applied to aircraft cockpit design (Rolfe, 1971), the failure detection of autopilot systems
(Wickens & Kessel, 1980), target tracking systems used by military pilots (Fracker &
Wickens, 1989), and in the design of visual displays requiring the integration of
information from several sources (Yee, Hunt, & Pellegrino, 1991; Boles & Wickens,
1987). The notion of resources has also played a role in the more basic pursuits of

7
learning, memory, and writing research. Researchers studying automaticity have also
benefited from the resource conceptualization of attention. As a task becomes more
automatic, fewer resources are needed for its successful completion (Laberge, 1973).
Despite the explanatory power over dual-task performance and success in applied
settings, capacity theories have received some criticism. Navon (1984) argued that most
effects interpreted in terms of resources can also be accommodated well by theories that
do not assume any limit on resources and by theories in which the resources construct
does not exist at all. Navon describes a diverse range of phenomenon that are typically
discussed in terms of resources, and takes issue with how necessary resource terminology
is for dealing with these phenomena. Manipulations of motivation level through payoff
schemes may be interpreted in the resource framework as regulating the amount of
resources allocated to the task. Navon argues that resources are not necessary to explain
the effect of motivation though, and he suggests that processing may be modulated
without any change in the supply of resources. In manipulations of task difficulty, the
more difficult a task, the more resources are said to be needed by that task. Navon
suggests that other factors such as poor luminance contrast might impair the quality of
sensory data available to a process independently of resources. Manipulations of task
complexity are typically explained by the notion that more complex tasks require more
resources. Navon argues that independent of resources, more complex tasks may require
more steps or mental operations, thus presenting more opportunity for errors. Dual-task
deficits are typically thought to degrade the performance of the primary task by curtailing
the amount of resources in its service. Navon points to other possible sources of
interference such as structural interference, some form of cross talk similar to that in

8
parallel phone lines, or interfering patterns of activation in which inhibition necessary for
one task disrupts activation on concurrent task.
In defense of resource theories, some researchers (Herdman & Friedman, 1985;
Wickens, 1976) claim that distinguishing between resource and structure theories may be
accomplished by manipulation of task priority in dual-task conditions. If performance can
be traded off from primary to secondary tasks and vice versa, then their interpretation is
that the two tasks interfere because the are consuming common resources. If the two tasks
compete for a common structure then the expectation is that performance will not trade
off between the two tasks in a gradual manner. Navon (1984) counters by arguing that
even single structural mechanisms could be timeshared between tasks by some schedule of
queuing or alternation.
Despite continuing debate, the resources construct continues to influence
information processing theory development and human factors concerns, and is at the very
least a useful metaphor. It seems likely that resource theories will continue to flourish as
long as continue to be useful in applied psychology.
Structural Theories
Early studies of the limitations of attention often involved the dichotic presentation
of verbal material. Subjects typically wore stereo headphones, and different messages were
simultaneously presented to each ear. Participants attended to one of the two messages
and repeated it aloud immediately after hearing it (shadowing). This technique ensured
that the participants focused on one of the two messages. Experimental investigations of
the dichotic listening task revealed that attention was severely limited (Cherry, 1953;
Broadbent, 1958; Moray, 1959). Several classes of theoretical models were generated to

9
localize the point in information processing where the bottleneck occurred. Early-selection
theories claimed that the bottleneck was in the early processes such as perception, and
late-selection theories pointed to stages of decision making and response selection as the
limiting stages of information processing.
Another paradigm was also often used in the investigation of the limitations of
attention. The psychological refractory period (PRP) paradigm (Bertelson, 1967; Welford,
1967), also known as the overlapping tasks paradigm (Pashler, 1984), involves the
presentation of two stimuli (S1 and S2) in rapid succession. Participants make a response
to each stimulus (R1 and R2) as quickly as possible. Using this paradigm, many
researchers (Kantowitz, 1974; Pashler, 1984; Pashler & Johnston, 1989; Fagot & Pashler,
1992) concluded, as did the late-selection theorists, that limitations in processing were
localized in the response initiation phase.
A subsequent extension of this structural model by the late-selection theorists
postulated that there is not a single stage or mental operation that acts as the source of
interference. Instead, a limited-capacity central processor, when engaged by one task, is
unavailable to a second task requiring the same operation. Thus, the performance of the
second task will necessarily deteriorate. By suggesting that there are a number of
operations that require the exclusive attention of the limited-capacity central processor in
order to proceed, this view permits more than one bottleneck in the information
processing system. Restrictions in the ability to carry on multiple mental computations
may occur at a variety of stages, including perceptual identification, decision and response
selection, response initiation and execution (Pashler & Johnston, 1989), and motor
mechanisms (Kahneman, 1973).

10
Broadbent (1982) criticized studies of continuous performance that rely on
accuracy measures because they cannot discriminate between the predictions of capacity
theory regarding simultaneous mental processing of both tasks and the predictions of
structural models regarding strategy switching. Pashler and Johnston (1989) incorporated
Broadbent’s suggestions when they investigated dual-task interference with the PRP. They
argued that this paradigm potentially provides much more detailed information about the
time course of dual-task interference than is obtained in continuous dual-task studies.
Pashler and Johnston divided models of dual-task interference into two categories:
1. capacity sharing models discussed earlier, and 2. “postponement models” which are
structural in nature. In postponement models, some bottleneck stage or process cannot
operate simultaneously for each of two overlapping tasks. As a result, processing of this
stage in the second task is literally postponed. This produces the relatively straightforward
hypothesis that as stimulus onset asynchrony (SOA) between SI and S2 is reduced, there
should come a point at which any further reduction in the SOA produces a corresponding
increase in the duration of R2. However, even when this prediction is confirmed, it does
not seem to contradict predictions from the capacity theory.
Kahneman (1973) addressed the notion that structural theories and a limited
capacity theory may account for the same findings. He observed that the assumptions of
structural theories are more restricting than the limited capacity models, and that
predictions, especially for the PRP, have typically failed to be confirmed. Pashler and
Johnston concede that the PRP paradigm has been studied somewhat less of late perhaps
reflecting Kahneman’s early concerns.

11
A major finding that has frequently been observed with the overlapping tasks
paradigm is a slowing of the Rlrelative to single-task performance (Kahneman, 1973).
Similar findings in discrete movement research are thought to indicate the increased effort
needed to compile a more complex movement (Henry & Rogers, 1960). R2 slowing is
readily accounted for with the capacity models because both tasks are assumed to be
performed with depleted allocations of capacity. Postponement models, in contrast, do not
predict this slowing directly. It has been suggested that R1 slowing may result from a
grouping strategy in which the subject essentially treats SI and S2 as a compound stimulus
and selects a corresponding compound response (Pashler & Johnston, 1989). Other
strategies have been proposed as well, but in any case, R1 slowing by itself is not thought
to be especially diagnostic of the underlying causes of dual-task interference.
Because both capacity and structural models can provide a plausible account of the
basic results, more analytic tests that make distinctive, non-obvious predictions are
necessary. Pashler (1984) used a chronometric approach to develop a method for testing
postponement models inspired by additive factors logic (Sternberg, 1969). By
manipulating task factors that increased or decreased the duration of selected stages of
processing and looking for effects that were additive (or underadditive), Pashler was able
to test hypotheses about the locus of the singe-channel bottleneck. Pashler reported a
preliminary study using this methodology that found underadditive effects that were
consistent with the predictions from a postponement model, and could not be explained
with capacity theory. This study supported the late-selection class of theories, suggesting
that a bottleneck exists either in the decision or response selection stages or in both stages.

12
Postponement models predict that effects at stages prior to the bottleneck should
become progressively smaller at shorter SOAs. As SOA shortens, the wait for the
processor is thought to become progressively longer, and the proportion of trials on which
delay in early stimulus processing of S2 will have any effect on RT2 progressively
decreases. Although Pashler (1984) only used one SOA and could not test this prediction,
Pashler and Johnston’s findings (1989) lend stronger support to the postponement model
with a bottleneck at the stages of decision making and/or response selection. In particular,
results indicating underadditivity present a problem for capacity theory as capacity models
predict that dual-task slowing of RT2 should interact overadditively with factors that
increase the difficulty of the second task.
There seems to be a general acknowledgement in the dual-task literature that
structural theories have very little empirical support outside of isolated laboratory tasks.
Structural theories have played a large role in studies of cognitive limitations, particularly
in their focus on mental stages. It is interesting in a historical sense that structural
assumptions have been carried over into the Baddeley and Hitch (1974) model of working
memory that will be discussed in detail later. Baddeley and Hitch, after all, were trained in
Broadbent’s lab where much of the early structural theorizing and testing took place.
Multiple Resource Theories
A search for a compromise to the structure/capacity debate has led directly to
multiple resource models (Navon & Gopher, 1979) where resources are declared to reside
within structures or pools. Until multiple resource models emerged, interference was
thought to depend exclusively on either the extent that two tasks draw on a common
resource pool or structural limitations where some stage of processing cannot proceed in

13
parallel with another. Neither of these accounts explains all of the phenomena related to
interference though. The central capacity notion cannot explain why some secondary
tasks interfere more with one primary task than another equally difficult primary task. For
example, vocal responses interfere more than spatial responses with recall of a sentence,
but less than spatial responses with recall of a line diagram (Brooks, 1968). Pure structural
models are also inadequate because processes that share similar stages or mechanisms may
interfere with each other, but they seldom block each other completely (Navon & Gopher,
1979). For example, when Triesman and Davies (1973) presented participants with stimuli
simultaneously to the same modality, performance was impaired, but not entirely impeded.
Multiple resource theory posits that there may be various types of resources and
different tasks may require different types of resources in various compositions resources
(Norman & Bobrow, 1975; Navon & Gopher, 1979, Wickens, 1984). On the one hand
this approach is structural in the sense that it identifies the source of interference as
overlapping mechanisms. On the other hand, multiple resource theory is a capacity
approach, because it does not assume that a mechanism can be accessed and used by only
one process at a time. Rather, mechanisms have a capacity that can be shared by several
processes until those processes demand more resources than the capacity allows. Thus,
multiple resource theory is a marriage of capacity and structural models where structure is
important, but each structure is limited by its own capacity.
Wickens (1984) proposed a multiple resource model that accounts for a great deal
of the empirical findings from dual-task studies. This model assumes that the extent to
which two tasks share common resource pools determines how much they will interfere in
a dual-task paradigm. In particular, the resource architecture of Wickens’s model contains

14
three dimensions derived from traditional psychological dichotomies: (a) Based on the
general finding that it is easier to time-share an auditory and visual task than two auditory
or two visual tasks, Wickens proposed a separation of resource pools specialized for
dealing with different processing modalities, (b) The processing codes dimension
distinguishes information that is spatial or analog from that which is verbal or linguistic.
(c) Wickens proposed a separation of processing stages based on the finding that tasks
demanding either response processes or cognitive/perceptual processes will interfere with
each other to a greater extent than will a perceptual and a response task.
Friedman and Poison (1981) proposed a framework with the intention of extending
the resource metaphor to our understanding of cerebral specialization phenomena.
Essentially a special case of a multiple resources model of limited-capacity, they suggest
that the physically distinct cerebral hemispheres can be seen as independent resource
systems. Specifically, the left and right hemispheres together form a system of two
mutually inaccessible and finite pools of resources. This conceptualization differs from
other multiple resource models in that it is based, at least grossly, on the physical structure
of the brain. In some respects this anatomical basis goes beyond the intent of earlier
multiple resource theories, while at the same time taking a step back to the most clearly
supported resource differentiation of all, the distinction of processing which is coded
verbally from visuo-spatially coded processing. Ultimately however, their theory, like
other multiple resource models, seeks to provide insight into the mechanisms that might be
responsible for those patterns of task interference not easily explained by models
postulating a single pool of undifferentiated resources.

15
Although the existence of multiple resource pools appeals to many researchers
(Wickens, 1984; Friedman & Poison, 1981; Norman & Bobrow, 1975), others are
skeptical (Navon, 1984). Navon and Gopher (1979) noted that multiple resource theory
might leave researchers disconcerted by the prospect of devaluation of the precious time-
honored concept of attention and that the proliferation of resources might seem strange
or threatening. Those who favored the more conservative central capacity theory argued
that multiple resource theory is not logically falsifiable and that it is impossible to know
how many resource pools there are. Therefore, new resource pools could be postulated to
account for any pattern of results. Multiple resource theory, as a class of models, may not
need to be falsifiable to be worthwhile. Rather than struggling to find a critical test for the
entire approach, researchers might do better by focusing on generating and testing specific
multiple resource models that delineate the composition of resources.
Working Memory
Like the multiple resource approach, Baddeley and Hitch’s (1974) model of
working memory is another hybrid of structural and capacity models. Instead of resources
within structures, this model implies both resource and structural components. The
model’s hybrid nature often goes overlooked, perhaps because working memory has been
primarily connected with higher level language tasks such as speech production,
vocabulary acquisition, and speech comprehension rather than the typical lower level
cognitive studies of processing limits where the capacity/structure debate flourished.
Working memory has recently become a popular construct in writing research because
speech production, vocabulary acquisition and speech comprehension presumably share
some characteristics with written language production.

16
The Baddeley and Hitch (1974) model evolved when neurological and
experimental data did not fit assumptions concerning the functioning of the unitary short¬
term store in the modal model (Atkinson & Shiffren, 1968). In the modal model, the short¬
term store was positioned as the route information must take to gain access into a long¬
term store, but neurological evidence indicated that poor performance on an auditory
memory span task was not linked to a long-term learning deficit (Shallice & Warrington,
1970). Nor did an impairment in the short-term store interfere with comprehension and
production of speech (Vallar & Baddeley, 1984). Second, the short-term store was
considered to play a major role in retrieval from the long-term store. However, a digit
span task performed concurrently with the retrieval phase of a free recall task, did not
depress accuracy of retrieval (Baddeley, et al., 1984a). Neither did rehearsal of a 6-digit
number influence recall of paired associates (Baddeley, Thomson, & Buchanan, 1975).
Third, a key postulate of the modal model was that the probability of information being
transferred to long-term store was heightened by increased rehearsal. Yet, the time spent
rehearsing target items interspersed at varying intervals in a longer list of words was
unrelated to recall (Craik & Watkins, 1973). Nor did frequent rehearsal prior to the start
of a free recall experiment improve recall of the rehearsed items (Tulving, 1966). Finally,
another tenet of the modal model was the recency effect in free recall tasks was derived
from information that remained in the short-term store. This was inconsistent with findings
that the recency effect remained even when participants counted backwards by three’s for
20 seconds after each word (Tzeng, 1973).
Baddeley’s model is a multi-component system for both the storage and the
processing of verbal and visuo-spatial tasks. It has three primary components: a limited-

17
capacity central executive and two peripheral slave systems. One of these slave systems
handles verbal and auditory information (the phonological loop). The other is specialized
for visual and spatial information (the visuo-spatial scratchpad) These slave systems do
not represent pools of limited resources that can act in parallel like the central executive.
Instead, they operate serially, in the same sense that a computer’s printer may receive
multiple print jobs, but can only perform one print job at a time.
The phonological loop is assumed to be most important in the production,
comprehension, and development of language. Architecturally, it has two subcomponents:
(a) a phonological store that is believed to hold phonological information for
approximately 2 seconds and (b) an articulatory mechanism that is implicated in the
transfer of written verbal material to the phonological loop. Because the working memory
model emerged from research on verbal tasks, the phonological loop has received much
more study than either the central executive or the visuo-spatial scratchpad. There are five
empirical sources of support for the phonological loop. First, evidence for the articulatory
control process comes from the word length effect: memory span is smaller for long words
than that for short words (Baddeley, Thomson, & Buchanan, 1975). This is assumed to
occur because rehearsal takes longer for longer words than for short words, allowing
more decay to occur before the next rehearsal cycle. It is typically found that people
remember as many words as they can read in 2 seconds (Baddeley, Thomson, &
Buchanan, 1975). Second, performance is disrupted when participants vocalize some
predefined pattern of speech (e.g., tee tah, tee tah ) while they simultaneously perform
some primary task. This articulatory suppression is claimed to engage the phonological
store and block its ability to participate in accomplishing the primaiy task. Disruptions

18
have been observed in a variety of tasks, including vocabulary acquisition and serial
learning (Baddeley, et al., 1984b). Third, performance degradation occurs when
participants attempt to perform a primary task while irrelevant speech is presented that
they are instructed to ignore. This procedure has been shown to disrupt serial recall of
visually presented lists (Salame’ & Baddeley, 1982). Fourth, acoustic or phonological
similarity effects are common where the dissimilar sounding items are recalled better than
similar sounding items is interpreted as evidence that the phonological store is speech-
based (Conrad & Hull, 1964; Baddeley, 1966). Words or nonwords that sound alike
interfere with each other in memory span tests more than semantically-related words that
differ in their acoustic properties. Fifth, neuropsychological patients frequently have
specific phonological loop deficits, but suffer no general cognitive impairment (Baddeley,
Papagno, & Vallar, 1988).
The visuo-spatial scratchpad has been less researched than the phonological loop,
but the notion that spatial and verbal processes may each draw on functionally separate
resources is well supported (Wickens, 1984). Where phonological similarity effects have
been observed in tasks believed to require the phonological loop, visual similarity effects
have been observed in visuo-spatial tasks. It is common for stimuli that look alike to
interfere with each other. Hue and Ericsson (1988) observed this phenomenon with
Chinese characters in a study using participants who presumably had no experience with
the Chinese language. Frick (1988) elaborated on the visual similarity effect, arguing on
the basis of the frequency of errors like “P” being mistaken for “R” that images in visuo-
spatial working memory are unparsed. Further evidence for the existence of a separate
visuo-spatial and verbal storage systems comes from neuropsychological studies where

19
patients with right posterior lesions can be markedly impaired on tests of memory span for
movements to different spatial locations, despite having normal auditory-verbal memory
spans (De Renzi & Nichelli, 1975).
A separate visuo-spatial processing system is also claimed to be involved in
planning and executing spatial tasks. For example, Japanese abacus experts can perform
complex calculations without the aid of the abacus, and appear to do so by simulating the
apparatus using visuo-spatial working memory (Hatano & Osawa, 1983). Researchers
have also implicated the visuo-spatial scratchpad in keeping track of changes in the visual
perceptual world overtime (Kahneman, Triesman, & Gibbs, 1992), maintaining
orientation in space and directing spatial movement (Thomson, 1983), and comprehending
certain types of verbal information (Manin & Johnson-Laird, 1982).
The central executive component of Baddeley’s model is described as a limited-
capacity mechanism responsible for coordinating tasks and managing the two slave
systems. It operates in a fundamentally different way than the two slave systems, because
it may be actively involved in meeting the demands of two or more concurrent tasks.
Accordingly, performance declines only when the collective task demands exceed the
capacity of resources.
Baddeley acknowledges (1992) that embarrassingly little direct research has
focused on the central executive and that it is somewhat of a catch-all component
responsible for a wide range of attentional duties. Nevertheless, one source of evidence
derives from the study of Alzheimer’s disease, a disorder typically associated with deficits
in tasks believed to depend primarily on central executive functioning. In one study
(Baddeley, 1993), Alzheimer’s patients were trained to do a visual task and a verbal task.

20
They then performed both tasks together in a dual-task paradigm. As the disease
progressed, single-task performance remained unchanged, but dual-task performance
dropped markedly. This effect was interpreted as support for the modularity of the central
executive, reasoning that time-sharing ability is accomplished entirely by the central
executive.
The Baddeley model has competitors other than single and multiple resource
models (Kellogg, 1996). Martin, Shelton, and Yaffee (1994) postulate that verbal working
memory may be further partitioned into subsystems for phonological and semantic
representations. Hirst and Kalmar (1987) indicate that working memory might also be
characterized to explain interference as a function of the semantic similarity of processing.
Klapp and Netick (1988) suggest that working memory resources may be further divided
between processing and storage. Ericsson and Kintsch (1995) challenge the utility of
postulating resource limits and raise the point that experts in a domain circumvent short¬
term capacity limitations through the structures of long-term working memory.
Working Memory in Writing
Kellogg (1996) proposed a model of writing that details how specific writing
processes rely on the main components of Baddeley’s working memory model. In this
proposal, Kellogg distinguishes among three language production processes: formulating,
executing, and monitoring. Each of these involves two basic level processes. Formulation
consists of planning ideas and translating them into sentences that may later be
handwritten or typed. Execution includes programming (controlling motor movements)
and executing (typing, handwriting, or dictation). Monitoring includes reading and
editing. The model is very clear that writers do not necessarily progress serially from

21
formulating text to executing it, and then to monitoring what they have written. Instead,
the model supports simultaneous activation of each of these processes, provided that the
demands placed on the central executive do not exceed its capacity limitations. Thus, the
model anticipates that the typing of a word or phrase may take place simultaneously with
the formulation of new material or reading previously written material. This is possible
only when execution is well-practiced and can proceed virtually automatically, so that
central executive resources are not needed. In contrast, the basic processes of formulation
and monitoring are much less likely to ever become automated.
Kellogg considers the demands that the basic processes of writing make on
working memory. He hypothesizes that formulation places the heaviest burden on working
memory, particularly on the central executive, but also on both slave components.
Planning demands the resources of the visuo-spatial scratchpad particularly when writers
plan by visualizing ideas, organizational schemes, supporting graphics, appearances of the
orthography and layout. Creating ideas (Shepard, 1978) and recalling them from long¬
term memory (Paivio, 1986) can invoke visual imagery. Gathercole and Baddeley (1993)
claim that planning also requires central executive resources in all its facets such as
generating ideas, trying out various organization schemes, or debating the appropriate
tone for a given audience. When formulating, writers covertly talk to themselves in the
form of inner speech or pre-text as they are generating sentences (Baddeley & Lewis,
1981), thus activating the phonological loop. In particular, phonological representations of
the words and sentences are stored in the phonological loop. To connect the phonological
loop with formulation tasks Kellogg notes cases of apraxic and dyspraxic patients who
make phonemic and other linguistic mistakes in speech and who also are impaired in

22
memory span tests and fail to show phonological similarity and word length effects
(Waters, Rochon, & Caplan, 1992). Translating, unlike conversational speech production
(Bock, 1982), demands central executive resources when the writer struggles to find just
the right words and sentence structures. Long pauses and high degrees of expended
cognitive effort suggest the involvement of the central executive in these cases (Kellogg,
1996) as well as the positive correlation of memory span with the ability to select lexical
items for use in a sentence (Daneman & Greene, 1986). Kellogg adds that the
involvement of the central executive in translation may vary with the demands of the task
at hand. A writer composing polished final draft prose on the first attempt may require
central executive resources to a greater extent than a writer composing a more
conversational email to a friend.
Typing and handwriting (execution) are believed to make minimal demands on the
central executive when highly practiced. In young children however, Bourdin & Fayol
(1994) found that handwriting demands more capacity than speaking, consistent with the
notion that novel activities of all kinds require the central executive to control the schemas
used in motor output (Kellogg, 1996).
Gathercole and Baddeley (1993) claim that reading, one of the basic processes of
monitoring, requires the resources of both the phonological loop and the central executive.
Kellogg suggests that the most significant demand of monitoring stems from editing, not
reading, because editing takes so many forms, ranging from the detection of a motor
programming effort to a revision in the organization of ideas in a text (Kellogg, 1996).

23
Relevant Working Memory and Writing Research
Owing in part to its newness, little research has been reported that explicitly tests
the implications of Kellogg’s model. One prediction of the Kellogg model suggests that
the phonological loop plays a strong role in monitoring. Jones, Miles, and Page (1990)
introduced an irrelevant speech manipulation to explore phonological loop involvement in
proofreading. Although proofreading is clearly not the same as editing one’s own writing,
it does involve some aspects of the monitoring process as described by Kellogg (1996).
Jones, et al. compared participants’ ability to detect two types of errors, contextual
(grammatical and word choice) and non-contextual (spelling and typographical errors),
while they heard either meaningful or reversed speech. The irrelevant speech consisted of
a taped lecture either played forward (meaningful speech) or backward (reversed speech).
The detection rate for non-contextual errors was disrupted in the meaningful speech
condition, but this effect did not extend to contextual errors. No effects were observed for
the reversed irrelevant speech manipulation, nor were there differences between reversed
irrelevant speech and silence in a second experiment, lending support to the idea that a
semantic component is important in the irrelevant speech effect. These results are
consistent with Kellogg’s (1996) predictions that the phonological loop is involved in the
detection of spelling and typographical errors, but inconsistent with the prediction that
irrelevant speech would hinder the detection of grammatical and word choice errors.
Investigations of irrelevant speech effects on reading comprehension also failed to
support the role of the phonological loop as a temporary store for sentence
representations (Martin, Wogalter, & Forlanon, 1988). In the initial investigation,
comprehension scores declined in two conditions involving irrelevant speech compared to

24
a silent control. The irrelevant speech consisted of either words in a meaningful order or
the same words randomly scrambled. Comprehension in the two speech conditions was
not reliably different. In a second experiment, Martin et al. (1988) manipulated
instrumentation (present or not) and lyrics (sung, spoken, or none) within subjects.
Findings indicated that the presence of verbal material adversely affected comprehension,
while instrumentation did not.
Madigan and Linton (1996) asked participants to compose two essays on a word
processor, one while hearing irrelevant speech and the other in silence. Prior to the writing
session, participants completed a reading span task designed by Daneman and Carpenter
(1980). This reading span test is used to assess capacity by incorporating a processing and
a storage component. Participants read aloud an increasingly larger set of sentences. After
each set, they recalled the last word of each sentence. Madigan and Linton based the
working memory span on the number of last words correctly recalled. Those who scored
in the top third of the sample were classified as high capacity, while those in the bottom
third were classified as low capacity.
Irrelevant speech reliably reduced the holistic quality of the essays, also causing a
decline in syntactic complexity (the proportion of subordinate clauses that occur before
the main verb) among high span writers. Writing fluency, as measured by word production
times and pauses within selected clauses, were unaffected by the irrelevant speech
manipulation. The investigators argue that the reduction in holistic quality is inconsistent
with assumptions of the Kellogg model related to the relatively minor influence of the
phonological loop when compared to the central executive, on writing quality. Kellogg

25
(1996), however, disagrees, contending that “the small effect size is consistent with the
hypothesis that irrelevant speech disrupts only translating and reading” (p 69).
Ransdell, Levy and Kellogg (1977) used irrelevant speech and memory load
manipulations comparing performance of writers with high and low writing span in a series
of experiments. They hypothesized that irrelevant speech, linked only to translating and
reading, would have a relatively small effect on fluency and quality compared to a memory
load relying on central executive resources. They also expected that the number of pauses
would decline when the central executive rather than the phonological loop was affected
by the manipulation.
In each of three experiments, participants composed two 10-minute essays. In the
first experiment, one essay was written while hearing irrelevant speech, the other in
silence. In the second experiment, they wrote with or without a six digit concurrent load.
In the final investigation, participants wrote one essay in the presence of unattended
irrelevant speech, and one essay while they were instructed to attended to the irrelevant
speech, periodically responding “yes” or “no” to questions related to either phonological,
spatial or semantic characteristics of the message. Because the semantic task was
positioned as more heavily dependent on central executive capacity, it was expected that
decrements in performance would be greatest in this condition.
Results of the first experiment demonstrated that irrelevant speech reliably reduced
fluency, and an increase in the number of pauses longer than five seconds, but did not
influence the percentage of pauses at clause boundaries. Essay quality was similar in the
irrelevant speech and silent conditions, an unexpected finding. The researchers suggested
two possible explanations for the immunity of essay quality to the speech manipulation.

26
First, they considered the possibility that the phonological loop is activated only for
complex sentences. Second, they raised the possibility that the articulatory loop was more
intimately involved in the writing task than the phonological loop, and that irrelevant
speech derived its influence via corrupting material in the store.
In the second experiment, memory load also resulted in a reliable decrement in
fluency and in sentence length. Only the fluency of low span writers was adversely affected
by the concurrent task. Concurrent load also altered the temporal patterns of composing.
When remembering and recalling digits, participants paused longer and more often. High
span writers also paused less frequently at clause boundaries. Furthermore, unlike
exposure to irrelevant speech, the concurrent task also reliably reduced essay quality.
In the third experiment, attending and responding to irrelevant speech led to a
reliable decrement in writing fluency, sentence length and pause frequency compared to
baseline, but pause length and location were similar in both conditions. Additionally,
overall writing quality did not decline in the attended speech condition, although scores of
high span writers with “good sentence” instructions were adversely affected in the
organization and development subgroup. The type of question answered in the attended
speech condition (phonological, spatial, or semantic) did not affect any dependent
measure. Of the three secondary tasks, accuracy was higher in the semantic condition than
in the spatial or phonological conditions which were equivalent.
The most consistent result throughout the three experiments was the expected
decrement in fluency as a result of the experimental manipulation. Neither unattended nor
attended speech resulted in a quality decrement. In contrast, the six-digit concurrent load
did adversely affect quality relative to a control condition. The six-digit load was also the

27
only task that lengthened pause duration and reduced the proportion of pauses at clause
boundaries. Here, there is a trade-off between quality and fluency that involves an increase
in the number of pauses rather than in their duration.
To test Kellogg’s (1996) predictions of the phonological loop involvement in
translating ideas into sentences (formulation) and reading text (monitoring), Marek and
Levy (1996) designed a series of computer-based tasks to simulate these primary writing
processes. These writing subtasks included creating sentences from groups of words
(formulation), copying an existing text (execution) and proofreading (monitoring). Writers
first performed each of the three tasks in silence, then repeated each task while listening to
computer-generated speech. Irrelevant speech during the formulation task reliably reduced
both the number of target words used in the sentences and sentence quality, a measure
based on a combination of meaningfulness and technical quality. Irrelevant speech did not
influence performance measures related to execution or monitoring as predicted by the
model but did lend support to the role of the phonological loop in formulation.
One possible explanation for the null result found for monitoring relates to the
proportion of editing versus reading in the tasks. In the proofreading task, twenty errors
were embedded into approximately 150 words of text. Participants were asked not only to
detect the errors, but to identify them as well, within a 5-minute time period. Beyond the
obvious possibility that the results reflect a difference between proofreading someone
else’s work and reviewing one’s own writing, the task was not designed to enable
recording of the amount of time spent reading versus the amount of time spent editing.
Nor did the researchers record or control how much of the selection was actually read.

28
The detection rate for errors ranged from 5% to 56%, but whether this primarily illustrates
differing ability to identify the mistakes or differences in reading fluency remain unknown.
Another prediction of the Kellogg model suggests that loading the central
executive should detract from writing fluency and quality the most, because the central
executive plays a part in all three writing processes. Jeffery and Underwood (1995)
conducted an experiment in which participants created sentences from information
presented, with either a 0, 3, or 6 digit preload. If this construction task appropriately
reflects formulation, the Kellogg model predicts that loading the central executive should
disrupt its performance. Rather than an overall disruption, Jefferey and Underwood found
a fluency-quality tradeoff. Although the preload actually reduced sentence initiation time,
quality of the sentences, as measured by the level of idea coordination, also dropped in the
preload conditions.
In a second experiment, participants copied rather than constructed sentences,
again with or without a preload. In this context, sentence initiation time under a 6-digit
preload was higher than in the non-preload condition. Because semantic processing is
required only in the construction task, Jefferey and Underwood concluded that the
decrease in sentence initiation time in their first experiment was attributable to a lack of
available resources for semantic processing in the preload conditions. More generally,
differences between preload and control conditions suggest that semantic processing
during writing involves the central executive.
Jefferey and Underwood positioned the primary task in these experiments as
constructing rather than formulating sentences. However, other writing researchers
(Kellogg, 1996; Ransdell & Levy, 1996) have suggested that constructing sentences

29
represents one aspect of formulation in writing. Debate surrounding inconsistencies at this
level underscores the difficulty involved in studying the specific components of a complex
task such as writing. It raises questions concerning the relationship of the part to the
whole, and calls attention to the problem of developing a satisfactory operational
definition for a specific component that has been separated from its task context. Although
investigations of writing at the subcomponent level are important, great difficulty
surrounds this practice and some aspects of writing may not be profitably reduced to this
level. Thus, this dissertation investigates essay writing as a complete task rather than
breaking it down into formulation, execution, and monitoring.
Adapted Working Memory Model
The balance of this chapter closely examines the assumptions of Baddeley’s
working memory model and proposes an adaptation that brings working memory into
closer alignment with the attention literature while simultaneously preserving its
orientation toward complex language tasks. These adaptations to the working memory
model have subtle implications on the dual-task methods employed by this dissertation and
other studies of this nature. Before addressing these methodological implications,
however, a closer analysis of working memory is needed.
There are several interesting assumptions that are commonly made about the
processing capabilities of the two slave systems. First, in comparisons of visuo-spatial and
phonological loads, the structural assumption implies that the relative difficulty of the
secondary tasks is irrelevant because any secondary task, regardless of difficulty, will
engage the visuo-spatial scratchpad or phonological loop entirely, or not at all. Therefore,
a phenomenologically easy visuo-spatial task, such as identifying in which ear a sound is

30
played through headphones, should load the visuo-spatial scratchpad as much as a more
difficult spatial task, such as mentally rotating a three dimensional object. If we assume,
instead, that the slave systems represent pools of resources, then the difficulty of the
secondary tasks becomes a variable of interest. This variable should be experimentally
controlled because of the common finding that increasing the difficulty of a secondary task
almost always results in a performance degradation in the primary task (Wickens, 1984;
Navon & Gopher, 1979).
If we assume the slave systems are resource pools, then when we compare the
relative effects of visuo-spatial and phonological load tasks on a primary task, it becomes
important to equalize the overall difficulty of the loading tasks. Consider the implications
of mistakenly assuming the slave systems are structures. If we disregard the difficulty of
loading tasks, a disproportionately more difficult phonological task would disrupt the
primary task of writing more than the visuo-spatial secondary task, possibly leading to the
conclusion that the visuo-spatial scratchpad plays a smaller role in writing than it actually
does. Thus, when comparing the contributions made by the two slave systems to a primary
task under dual-task conditions, it may be more risky to assume that these components are
structures rather than resource pools.
Another implication of Baddeley’s theoretical assumptions arises when comparing
effects across studies. Under the assumption of structural slave systems, differences in
secondary task structure (including characteristics such as the frequency of response
selection and execution as well as input and output modalities) will not affect the extent to
which that task engages its targeted component. Resource theorists argue that just as task
difficulty is likely to influence primary and secondary task interference, the frequency of

31
response selection and the input and response modality are likely to affect the degree to
which they interfere with a primary task.
Overall, a structural assumption allows for much easier secondary task selection,
but the results may be inaccurate if the visuo-spatial scratchpad and phonological loop are
in fact pools of resources rather than structures. The more rigorous secondary task
selection methods associated with a resource assumption are acceptable regardless of
whether the visuo-spatial scratchpad and phonological loop act as structures or resources.
In light of these considerations, a new model of working memory was proposed
that posits two mutually inaccessible pools of resources for verbal and visuo-spatial
processing and storage such as Friedman and Poison’s (1981) multiple resource model.
The current model differs from Friedman and Poison (1981) in its inclusion of Baddeley’s
notion of a limited capacity central executive system that acts for a wide range of
attentional duties including coordinating information between the visuo-spatial and verbal
systems.
Experiment 1 seeks empirical support for this adapted working memory model, by
testing the hypothesis that the phonological loop may be partially disrupted within subjects
rather than disrupted entirely or not at all.
Summary
As a writer plans ideas, translates ideas into prose, types or handwrites sentences,
and monitors all of these activities, many demands are placed on the writer’s temporary
storage and processing capacity. Investigations of writing at this level are fairly new, thus
obligating writing researchers to turn to existing theories of cognitive limitations for a
framework in which these demands may be assessed.

32
Theory development surrounding processing limitations has typically employed
two hypothetical constructs, resources and structures. Early accounts of processing
demands were explained by either resource (Kahneman, 1973) or structural (Broadbent,
1958) models, but more recently elaborate models have evolved that embody both
resource and structural aspects. The two most prevalent of such models are multiple
resource models (Norman & Bobrow, 1975; Wickens, 1984; Friedman & Poison, 1981)
and Baddeley’s working memory model (Baddeley & Hitch, 1974). Both propose
divisions in visuo-spatial and verbal processing, generally in alignment with the
hemispheric asymmetry literature.
Kellogg (1996) proposed a model of writing that details how specific writing
processes rely on the main components of Baddeley’s working memory model. Before
testing the specific claims of Kellogg’s model, a closer examination of working memory as
a model of cognitive limitations is necessary. In particular, the assumptions of the
structural nature of the phonological loop and visuo-spatial scratchpad seem outdated.
This dissertation proposes an adaptation to the working memory model whereby
the phonological loop and visuo-spatial scratchpad operate as pools of mutually
inaccessible resources for verbal and visuo-spatial processing and storage similar to
Friedman and Poison’s multiple resource model (1981). This hybrid model of working
memory and multiple resources is tested in Experiment 1. Experiments 2a and 2b consider
the division of labor between visuo-spatial and verbal processing and storage components,
and Experiment 3 addresses the role of writing topic in the resource demands of writing.

CHAPTER 2
EXPERIMENT 1
If the phonological loop and visuo-spatial scratchpad of Baddeley’s working
memory model are engaged in a binary, all-or-none fashion, it follows that in a dual-task
paradigm, secondary tasks of varying difficulty will have the same loading effect. This
assumption is comparatively easy to test. For example, in a dual-task experiment, if two
phonological or visuo-spatial secondary tasks of varying difficulty interfere equally with a
primary task, then the all-or-none processing assumption is supported. Conversely, if the
two phonological or visuo-spatial secondary tasks vary in the relative degree to which they
disrupt primary task performance, then graded processing is supported.
Baddeley’s working memory model predicts no differences in performance related
to type of irrelevant speech, because speech sounds, regardless of meaning, are believed to
have obligatory access to the phonological store. Those sounds are believed to engage the
phonological loop entirely due to its structural nature. Inconsistent with these predictions,
Jones, Miles, and Page (1990) found that meaningful speech interfered more with the
detection of spelling and typographical errors in a proofreading task than the same speech
played backward. However, in a second experiment, Jones et al. found no difference
between the effect of reversed speech and a silent control on error detection rates,
suggesting that reversed speech may not engage the phonological loop at all. It could be
argued, however, that there may be so little phonemic similarity between forward and
backward speech that it may not be reasonable to claim that speech played backward is
33

34
really “speech.” Further, Levy and Marek (1998) report similar patterns of result in a
writing study when computer-generated speech was “natural” and when the individual
words in each sentence were randomly rearranged, but no changes made to their phonemic
attributes.
A recent study using magnetic resonance imaging revealed that irrelevant speech
played backward is processed in different areas of the brain than is forward speech, raising
questions whether the phonological loop acts alone in processing speech. Baddeley’s
working memory model again predicts no differences in performance related to type of
irrelevant speech, because speech sounds, regardless of meaning, are believed to have
obligatory access to the phonological store, and are believed to engage the phonological
loop entirely due to its structural nature.
Experiment 1 compares two types of irrelevant speech to test the hypothesis that
the phonological loop can be differentially loaded in a dual-task paradigm. One recording
was made of a passage read by the experimenter, and the other featured the same passage
generated using a text-to-speech computer program. The Baddeley model predicts that
accuracy on a primary memory task would be best in the absence of irrelevant speech.
When either the human-generated or the computer-generated speech is delivered while
participants perform another (primary) task, performance on the latter should decline, but
the performance decrements should be equal for the two types of irrelevant speech. This
would be so even if the materials delivered in a computer-synthesized speech generated a
lesser processing load (because it lacked nuances, had reduced inflections, and so forth)
than the speech produced by a human voice. In contrast, our modified model, which

35
anticipates such secondary tasks producing differential processing loads, would predict
differential outcomes on the primary task.
Method
Participants
Twenty-six undergraduate students from the University of Florida participated in
this experiment in partial fulfillment of a course requirement. One participant was excluded
from the data because he or she stopped recording the recalled consonants. Participants,
were neither vision nor hearing impaired.
Materials
Sets of five to nine randomly generated consonants were printed on transparencies
and presentated on an overhead projector. Consonant sets contained no duplicate letters,
and no two sets were identical. Participants recorded their responses on printed answer
sheets during the free recall phase of the experiment.
Two tapes of irrelevant speech were recorded. One featured a children’s fable read
by the experimenter. The other featured the same story generated via computer running a
text-to-speech program. The content and acoustic energy level of the two taped
presentations was the same.
Procedure
The to-be-remembered consonant sets were shown to the participants for 10
seconds each. Participants then had 20-seconds to recall the consonants. This sequence
was repeated using 15 different sets of consonants for the baseline (no irrelevant speech)
and each of the experimental conditions. During the experimental conditions, participants

36
heard one of two types of irrelevant speech during the presentation portion of each trial.
On alternate trials, the irrelevant speech was human-generated or generated via computer.
Results and Discussion
The patterns in the data are quite compelling (see Figure 1). For nearly all
participants, the human-generated speech disrupted letter recall more than computer¬
generated speech. A within-subjects analysis of variance showed that both human-
generated irrelevant speech and computer-generated irrelevant speech interfered with
participants recall of consonants (F (1,23) = 27.65, p<001 and F (1,23) = 15.72, p<001
respectively). Recall accuracy was lower in the presence of human-generated speech than
the computer-generated speech (F (1,23) = 33.85, p<001). This pattern of results matches
their subjective reports that the computer-generated speech was easier to disregard. If the
phonological loop is engaged in an all-or-none fashion, it follows that the irrelevant
computer-generated speech should have the same effect as irrelevant human-generated
speech. These results are more consistent with the notion that the phonological loop is
comprised of a pool of resources that can be engaged in a graded fashion. Further testing
is needed to determine if the visuo-spatial scratchpad also operates as a resource pool
rather than a structure.

37
Figure 1. Retention of consonant strings as a function of memory set size and irrelevant
speech condition.
Regardless of whether we assume that the visuo-spatial scratchpad operates as a
pool of resources like the phonological loop, because the phonological loop operates as a
resource pool, it is necessary to adopt the following criteria for selecting secondary tasks
when making comparisons between these two working memory components. First,
secondary tasks should be equally demanding of attentional resources. Equivalency may be
accomplished by comparing secondary task performance in isolation, or by placing each
secondary task in a probe-tone paradigm in which subjects perform the load task as a
primary task and are interrupted occasionally by a tone to which they respond as quickly
as possible. Reaction times in probe tone paradigms are typically taken as an index of the
mental effort demanded by the primary task. Second, the loading tasks should not change
in the degree to which they draw on controlled processes throughout the experiment. If a
load can become automated over the course of the dual-task conditions, then by the most
common definitions of automaticity, it no longer draws on attentional resources, thus

38
allowing participants to work on the primary task as if they were working under single¬
task conditions. Third, the tasks should have similar structural composition, requiring the
same mode of input and output. Fourth, response selection and execution should occur
equally often in the tasks. Finally, secondary tasks should require different strategies for
their successful performance. For example, a visuo-spatial task should not be easily
performed by re-coding visuo-spatial stimuli into phonological stimuli. Thus, secondary
tasks should be constructed so that performance differences will emerge if participants use
the same strategies for each task. The secondary task selection process for Experiment 2B
serves to illustrate these selection criterion.
Once appropriate secondary tasks are selected, the rationale for dual-task studies is
relatively straightforward. By selectively engaging one component of working memory in
a secondary task, the primary task may then only draw on the remaining components.
Thus, using a dual-task paradigm, experiment 2A investigates the contribution of verbal
resources to writing in a dual-task paradigm, while experiment 2B investigates the
contribution of visuo-spatial resources to writing. Given the adaptations we have made to
the working memory model, comparisons of the relative involvement of verbal and visuo-
spatial processing can be made across these two experiments by adhering to the secondary
task selection criteria described above.

CHAPTER 3
EXPERIMENT 2
Experiment 2A
Most resource models (Wickens, 1984; Friedman & Poison, 1981) and working
memory theories (Baddeley & Hitch, 1974) acknowledge a division in processing for
verbal and visuo-spatial tasks, in alignment with the literature on hemispheric differences.
Generally, visuo-spatial processing is localized to the right hemisphere while verbal
processing is identified with the left hemisphere.
Brooks (1968) demonstrated a clear dissociation of verbal and visuo-spatial
processing in an experiment were participants were presented with a block capital letter
with the bottom left hand comer marked with a star. Participants were asked to look away
from the letter and, holding it in memory, to go around the letter clockwise from the star
responding “yes” if the comer in question pointed outward or “no” if the comer pointed
inward. Hence for the block letter F, the correct responses would be “yes, yes, yes, yes, no
, no , yes, yes, no.” Brooks also developed a verbal task where participants were
presented with a sentence like, “A bird in hand is not in the bush”. Participants were asked
to hold this sentence in memory, and then successively categorize each word as either a
noun (in which case participants would respond “yes”), or a verb (“no”). Hence, for that
particular sentence the correct sequence would be “no, yes, no, yes, no, no, no, no, yes”.
Brooks used two methods of responding, either spoken or manual (this involved pointing
to a series of yeses or nos scattered irregularly down the response sheet). He observed a
39

40
clear interaction between type of memory task and mode of response, with the visual letter
task being performed more accurately when the response was spoken than when it
involved pointing, while the reverse was the case for the sentence task. This finding
demonstrates the interference of tasks that share similar processes, and points to separate
systems for verbal and visuo-spatial memory.
Subsequent selective interference effects have been observed: (a) Concurrent
visuo-spatial pursuit tracking disrupted performance on Brook’s (1968) spatial task, but
not on a corresponding abstract task (Baddeley, 1975); (b) A verbal task interfered with
right hand dowel balancing (Kinsbourne & Hicks, 1978); and (c) a spatial task interfered
with left hand dowel balancing (McFarland and Ashton, 1978). Also, in the perceptual
domain, Moscovitch and Klein (1980) observed that recognition performance was more
impaired when two spatial targets were presented simultaneously (a face and a random
polygon), than when a spatial and a verbal target were presented simultaneously.
EEG research supports the intuition that writing, like other verbal tasks, is
supported by the left hemisphere. Gallin & Omstein (1972) set out to select areas of the
brain that the split-brain and brain lesion research had shown were involved in different
tasks, and to record the brain’s electrical activity while a person was doing these tasks. In
general, they found that ordinary people, doing ordinary activities, turned on and off the
two sides of their brain appropriately. Specifically, Gallin & Omstein (1972) reported that
the right hemisphere showed more alpha activity than the left while writing a letter and the
left hemisphere showed more beta activity. While the person was arranging blocks, the left
hemisphere showed more alpha than the right side, and the right hemisphere showed more
beta waves. So when people write, they turn on the left hemisphere and turn off the right

41
side of the brain. While arranging blocks in space, they turn on the right side and turn off
the left hemisphere.
Baddeley (1993) investigated the working memory demands of chess by selectively
disrupting components of working memory in a dual-task paradigm. Two groups of
expertise were compared: weak club players formed the basic group and an advanced
group was comprised of expert chess players with Elo gradings ranging from 130 to 240.
Participants were required to memorize and recall chess positions as a primary task, while
performing one of three secondary tasks or a control condition where no concurrent
activity was performed. To disrupt the phonological loop, participants were given a
secondary task that required them to utter the word “the” at the rate of one per second
(articulatory suppression). The visuo-spatial scratchpad was disrupted by requiring
participants to tap the keypad of a computer at a one-second rate following a fixed order.
Random letter generation, again at a one-second rate, was required to disrupt the central
executive component of working memory. Overall, performance in the two groups
differed greatly, but the pattern of interference was equivalent, with articulatory
suppression had little effect, while both finger tapping, and random letter generation
produced greater disruption. These findings suggest that verbal processes play little or no
role in memory for chess positions compared to the role of visuo-spatial or central
executive processes.
Several criticism can be made of Baddeley’s dual-task methods in the study of
chess. Because participants performance was not measured on any of the secondary tasks
there is no evidence that participants performed these tasks as they were instructed.
Participants may simply have slowed or even stopped performing the secondary task as

42
primary task demands increased. Additionally, it is possible that participant’s performance
of these tasks became automatic, thus requiring little or no resources. These secondary
tasks implicitly assume all-or-none processing of the phonological loop, visuo-spatial
scratchpad and the central executive because no attempts were made to ensure that these
tasks are equally demanding.
The purpose of the present experiment was similar to Baddeley’s (1993)
investigation of the working memory requirements of chess, but instead of chess, this
study assess the contributions of the phonological loop to essay writing in a dual-task
paradigm. Writing researchers have addressed this issue by employing irrelevant speech
manipulations while measuring essay quality and fluency (Ransdell & Levy, 1998;
Ransdell, Levy, & Kellogg, 1997), proofreading ability (Jones, Miles, Page, 1990),
temporary storage of sentence representations (Martin, Wogalter, & Forlanon, 1988), and
creating sentences from groups of words, proofreading, and copying text (Marek & Levy,
1996).
Because the long-term research goal was to compare phonological loop
(experiment 2A) and visuo-spatial scratchpad (experiment 2B) contributions to writing, a
new secondary task was needed which, unlike irrelevant speech, required measurable
responses from the participant. A signal detection paradigm (Green & Swets, 1966)
paradigm was implemented because it would allow for easy comparison of performance to
a similar visuo-spatial secondary task. Signal detection is a well-established method used
by engineers and by psychophysicists for estimating an observer’s discriminative capacity
or sensitivity to a signal independent of the observer’s response bias. The measure of
sensitivity is a pure number represented by the symbol, d’ which is based on the

43
participants’ number of hits and false alarms. The greater the value of d’, the better the
individual’s performance.
Additional efforts were necessary to assure that this secondary task did not
become automatic. Dual-task conditions were designed so the data could be parted into
five-minute epochs for which individual d’ scores could be calculated. Thus, patterns of
d’ scores could be inspected qualitatively on an individual basis and participants who
showed dramatic changes in secondary task performance over time could be eliminated
from the study. No participants were excluded from this study based on this information.
Figure 2 illustrates the average secondary task performance and the percentage of writing
as a function of these five-minute epochs.
Time (5-minute epochs)
Figure 2. Average percentage written and dual task performance as a function of time.

44
Method
Participants
Forty-one University of Florida general psychology students participated in this
study. All participants reported to be proficient touch typists (avg. typing speed was 32
wpm) and had experience interacting with a computer using a mouse.
Materials and Apparatus
Participants wrote essays on 2 open-ended topics, “The perfect job” and “The
greatest high” using a Windows-based word processor developed especially for this
research effort. The program supported basic functions of text entry, cutting and pasting,
and mouse control at the point of insertion. The program also enabled the presentation of
the secondary task stimuli (directional symbols) in a large (120 point) sans serif font
displayed to the left of the text composition window. The characters subtended
approximately 3 degrees of visual angle and were readily identifiable from the writers’
peripheral vision.
Procedure
Participants first completed a 2-minute typing test in which they were asked to
transcribe as quickly as possible text that appeared on their screens. Next, they spent 20
minutes writing an essay on one of the two topics. Participants were given verbal notice 5
minutes before their time expired.
Participants next performed a task designed to load the phonological loop under
single-task conditions for 5 minutes. This task was designed to load the phonological loop,
but not the visuo-spatial scratchpad. During this task, a single letter or a single digit was
shown for 5 seconds. At the end of this period, the visible stimulus was erased and

45
replaced by another single character. The characters typically alternated between letters
and digits. The participant’s task was to indicate - by pressing the mouse button -
whenever the category (letter or digit) remained the same from one presentation to the
next (i.e. two consecutive letters or numbers). A response was required, on the average,
twice per minute on a variable interval schedule. Consecutive letters and consecutive digits
occurred equally often. If participants responded correctly a green circle appeared instead
of the next character, but if they missed an opportunity to respond correctly a red circle
appeared instead of the next character. False alarms did not disrupt the sequence of
character presentation, and no feedback was given to indicate a false alarm had been
made.
In the final, dual-task phase of the experiment, participants wrote for 20 minutes
on whichever topic they had not already addressed earlier under single-task conditions. At
the same time, they simultaneously performed the phonological loading task that they had
just finished. Instructions for the loading task, stimuli presentation rate, and method of
responding were identical to the single-task conditions. Participants were instructed to
perform both tasks to the best of their ability. To maintain interest in the task, the research
promised that the person who performed the best overall would win $100.
Results and Discussion
Document quality was assessed on a 6-point scale using the essay-sort method
(Madigan, Johnson, & Linton, 1994) where two independent raters sorted essays first into
3 categories of quality, and then rated essays as high or low quality within category (inter¬
rater reliability = 88%). Document fluency was measured by the number of words

46
generated per minute. A signal detection paradigm (Green & Swets, 1966) was employed
in the phonological secondary task.
As illustrated in Figure 3, performance in the present experiment deteriorated
significantly from single to dual-task conditions on all measures. Words produced per
minute dropped 21% (t (42) = 7,19, p< 001) and the quality of the essays declined 12.3%
(t (42) = 2.71, p=.01) with the addition of the phonological loop secondary task. Most
compelling was a 60% decrement in d’ associated with the addition of the secondary task
(t(42) = 8.91, p <001).
Quality
Fluency (WPM)
2nd task (d')
-80 -60 -40 -20 0 20 40 60 80
Percent Change
Figure 3. Decrement in quality, fluency, and secondary task d’ from single to dual-task
phonological load conditions.
Kellogg’s (1996) predictions for the involvement of the phonological loop in
writing were that this component would play a strong role in monitoring, particularly in
the detection of spelling and typographical errors. Also, the phonological loop should

have a minor influence on writing quality relative to the central executive, but more
influence on quality than the visuo-spatial scratchpad.
47
Decrements in quality associated with the phonological secondary task seem to
conflict with Levy, Ransdell & Kellogg (1997) who failed to find an effect of irrelevant
speech on essay quality. Inconsistencies here can be explained by the greater demands of
the letter matching task due to response selection and execution stages that were required.
It is important to keep in mind that the primary purpose of this experiment is the eventual
comparison to the effects of a visuo-spatial loading condition, and therefore the response
selection and execution stages are necessary, and will algebraically cancel each other out
when that comparison is made.
Researchers using an irrelevant speech manipulation have observed disruptions in
the detection rate for non-contextual errors (Jones, Miles, and Page, 1990), diminished
comprehension (Martin, Wogalter, & Forlanon, 1988), lower holistic quality of essays and
diminished syntactic complexity as measured by the proportion of subordinate clauses that
occur before the main verb (Madigan & Linton, 1996).
The present experiment supports Ransdell, Levy & Kellogg’s (in press) finding
that irrelevant speech reliably reduced fluency as measured by words per minute. Madigan
and Linton (1996), however, found no effect of irrelevant speech on fluency as measured
by word production times and pauses within selected clauses. This calls into question the
relationship between different measures of writing fluency.
Further questions involve how a visuo-spatial secondary task will effect essay
quality, fluency, and d’ scores compared to the phonological secondary task observed in

48
this experiment 2B describes the selection and effects of a visuo-spatial secondary task,
followed by the combined analysis of experiments 2a and 2b.
Experiment 2B
Because of the impressive amount of research on the effects of irrelevant speech
on the variety of tasks discussed earlier, the focus of considerable recent work in writing
has also focused on variables that might influence the engagement of the phonological
loop. The role of visuo-spatial processing in writing has received very little attention
compared to that of the phonological loop (Dinet & Passerault, 1998; Levy, White, &
Lea, 1998).
Kellogg (1996) claimed that visuo-spatial resources are involved in planning,
particularly when writers plan by visualizing ideas, organizational schemes, supporting
graphics, appearances of the orthography and layout. In one of the only studies
investigating the role of the visuo-spatial scratchpad in writing, Ransdell, Levy, & Kellogg
(1997) had participants write one essay in the presence of unattended irrelevant speech,
and one essay while they were instructed to attend to the irrelevant speech periodically
responding “yes” or “no” to questions related to either phonological, spatial or semantic
characteristics of the message. The type of question answered in the attended speech
condition did not affect any dependent measure. One interpretation of this lack of findings
is that the combination of making judgements requiring a response on irrelevant speech
may have simultaneously loaded the phonological loop (with irrelevant speech), the central
executive (by requiring response selection and execution) and then either the phonological
loop (phonological judgements) or the central executive (semantic judgements) again or
the visuo-spatial scratchpad (with spatial judgements).

49
In the present experiment, we use the methodology underlying Experiment 2 A to
focus on how processing in the visuo-spatial scratchpad influences writing. The
hypothesis was that a visuo-spatial task, like the phonological task in the previous
experiment, would cause a significant decrement in writing quality and fluency.
Additionally, when comparing the dual-task performance of a visuo-spatial task and
writing to dual-task performance of a phonological task and writing (Experiment 2A),
dual-task performance would be better in the visuo-spatial condition.
In order for performance here to be compared to Experiment 2A, a secondary task was
needed to engage the visuo-spatial scratchpad in a signal detection paradigm where no
significant differences exist between the visuo-spatial and phonological single task d’
scores. Additionally, this visuo-spatial secondary task would also need to be designed so
that d’ scores could be calculated for each 5-minute epoch, to assess whether that task
became automated. Figure 4 illustrates the average secondary task performance and the
average percentage of writing as a function of these five-minute epochs.

Figure 4. Average percentage written and dual-task performance as a function of time.
Stimulus presentation modality and rate and the rate of responding needed to be
equivalent for the visuo-spatial and the phonological secondary tasks. Additionally, the
visuo-spatial secondary task had to be constructed so that performance differences would
emerge if participants encoded the stimuli verbally as they did in the phonological loop
secondary task.
Method
Participants
Forty-four University of Florida general psychology students participated in this
study. All participants reported to be proficient typists (avg. typing speed was 34 wpm)
and had experience interacting with a computer using a mouse.

51
Apparatus and Materials
Participants wrote essays on 2 open-ended topics, “The perfect job” and “The
greatest high”. They used a Windows-based word processor written especially for this
research effort. The program supported basic functions of text entry, cutting and pasting,
and mouse control at the point of insertion. The program also enabled the presentation of
the secondary task stimuli (directional symbols) in a large (120 point) sans serif font
displayed to the left of the text composition window. The characters subtended
approximately 3 degrees of visual angle and were readily identifiable from the writers’
peripheral vision.
Procedure
Participants first completed a 2-minute typing test in which they were asked to
transcribe as quickly as possible text that appeared on their screens. Participants then
wrote an essay for 20 minutes on one of the two topics. Half wrote on one topic, while the
other half wrote on the other. Participants were given notice 5 minutes before their time
expired.
Participants next performed a visuo-spatial task for 5 minutes. This task was
designed to load the visuo-spatial scratchpad but not the phonological loop. Rather than
letters and digits, the computer displayed equal-sized arrows that pointed up, down, left,
or right. The participants’ task was to maintain in memory a representation of an x, y
coordinate system and an imaginary placeholder at the 0,0 coordinate. They were
instructed to move the imaginary placeholder one standard, but arbitrary, unit of distance
in the direction of the arrow presented. They were further instructed to click the mouse
button any time the placeholder moved away from either axis and then returned to either

52
axis. A new arrow was presented every 5 seconds and, on the average, a response was
required twice per minute. Crossings of the x- and y-axis occurred equally often. If
participants responded correctly a green circle appeared instead of the next arrow, but if
they missed an opportunity to respond correctly a red circle appeared instead of the next
arrow. False alarms did not disrupt the sequence of arrow presentation, and no feedback
was given to indicate a false alarm had been made.
In the final, dual-task phase of the experiment, participants wrote for 20 minutes
on whichever topic they had not already addressed earlier as the primary task. At the same
time, they simultaneously performed the secondary loading task that they had just finished.
Instructions for the loading task, stimuli presentation rate, and method of responding were
identical to the single-task conditions. Participants were instructed to perform both tasks
to the best of their ability. Again, to maintain interest in the task, the researcher promised
that the person who performed the best overall would win $100.
Results and Discussion
Like experiment 2A, document quality was assessed on a 6-point scale using the
essay-sort method (Madigan, Johnson, & Linton, 1994) where two independent raters
sorted essays first into 3 categories of quality, and then rated essays as high or low quality
within category (inter-rater reliability = 84%). Document fluency was measured by the
number of words generated per minute. A signal detection paradigm (Green & Swets,
1966) was employed in the phonological secondary task.
The secondary task met all of the criteria established for secondary task selection,
most important of which was that there was no significant difference between visuo-spatial
and phonological d’ scores under single task conditions (3.50 and 4.09 respectively). If the

53
visuo-spatial task had been re-coded into a verbal task then a phonological representation
of the presented arrows (up, up, right, down, up, left, etc) would need to be maintained in
working memory. If this had been the case then we would have expected performance to
be much worse compared to the phonological secondary task because participants would
need to hold more information (up to 10 chunks) in memory than they would on the verbal
secondary task where only the last character needed to be remembered. If they instead
adopted a visuo-spatial coding strategy and stored the location of a mental placeholder as
they were instructed, then only one chunk needed to be maintained in memory and single¬
task performance would be equivalent to the phonological secondary task. No significant
differences were found between the phonological d’ (4.09) and the visuo-spatial d’ (3.5)
under single task conditions.
Results indicated that much like experiment 2, performance in the present
experiment deteriorated significantly from single to dual-task conditions on all measures
(see Figure 5). Words produced per minute dropped 13% (t (39) = 3.17, g<01) and the
quality of the essays declined 13.6% (t (39) = 2.95, g< 01) with the addition of the
phonological loop secondary task. A 44% decrement was observed in d’ comparing single
to dual-task conditions (t(39) = 6.01, p <.001).

54
Quality
Fluency (WPM)
2nd task (d')
-80 -60 -40 -20 0 20 40 60 80
Percent Change
Figure 5. Decrement in quality, fluency, and secondary task d’ from single to dual-task
visuo-spatial load conditions.
Combined Analysis of Experiments 2 A and 2B
As Figure 6. shows, selectively engaging either component of working memory
reliably deteriorated fluency, quality and secondary task performance. Table 3 lists means
and standard deviations for fluency, quality and secondary task d’ scores under single and
dual-task conditions. Writing fluency slowed significantly overall, F(l, 81) - 49.06, p
<.001, and as Kellogg’s model anticipates, there was a significant interaction between the
component of working memory that was loaded and the testing conditions (either single or
dual-task), indicating that the phonological secondary task disrupted fluency more than the
visuo-spatial secondary task, F(l,81) = 3.99, p = .04. Interestingly, the d’ measured during
the phonological secondary task declined much more than during the visuo-spatial
scratchpad, as reflected in the interaction between loading task and the testing conditions,
F(l, 81) = 5.79, p < .01. The decrement in d’ associated with the dual-task for the visuo-

55
spatial group was 44%. For the phonological group, the cP fell 60%. Although quality was
not affected more by one working memory load than another, both fluency and d’ scores
deteriorated more in the pairing of the phonological task with writing than in the pairing of
the visuo-spatial task and writing. Despite the equal degradation in writing quality, these
results suggest greater difficulty in timesharing writing with a phonological task than a
with visuo-spatial task which is consistent with Kellogg’s (1996) predictions.
Table 3.
Means and standard errors for d’. writing speed, and writing quality scores for the single
and dual visuo-spatial and phonological tasks.
Measure
Visuo-spatial
Single Task
Visuo-spatial
Dual-task
Phonological
Single Task
Phonological
Dual-task
d’
Mean
3.50
1.96
4.09
1.64
SD
1.72
0.85
1.73
0.67
WPM
Mean
14.7
12.8
16.2
12.8
SD
0.79
0.70
0.62
0.57
Quality
Mean
3.16
2.73
3.18
2.79
SD
1.04
.95
0.92
0.82

56
Percent Change
Figure 6. Decrement in quality, fluency, and secondary task d’ from single to dual-task
conditions separated by resource loading condition.
These findings support the recent trend toward multi-component models of
working memory, and in particular those that distinguish between visuo-spatial and verbal
resources. It is clear that these results do not support the undifferentiated class of models
that dominated early writing process research. Undifferentiated models would predict that
equally demanding secondary tasks would be equally interfering with the writing primary
task. Our phonological secondary task clearly interfered more than the visuo-spatial
secondary task despite no single-task performance differences (in fact, trends in the single¬
task data indicated that the visuo-spatial task was slightly more difficult than the
phonological task although differences were not significant).
In the future this line of research may be extended to the subprocess level of
writing where we can investigate the demands that formulating, executing and monitoring
place on the visuo-spatial scratchpad and phonological loop. These future investigations

57
will employ methods based on a resource capacity assumption of phonological loop and
visuo-spatial scratchpad processing, and will include the methods for selecting secondary
tasks that we specified earlier.

CHAPTER 4
EXPERIMENT 3
Experiments 2a and 2b lend support to the notion that visuo-spatial processes are
active in writing, because of the size of decrements in writing quality and fluency observed
in the presence of the visuo-spatial secondary task. However, questions about the specific
role of visuo-spatial processes in writing remain unanswered. Kellogg (1996) suggests that
planning may draw on visuo-spatial processes when ideas, organizational schemes, or
appearances of the orthography and layout are visualized. The current experiment tests
Kellogg’s claims by manipulating the writing topic so that one of two similar essays
requires more visuo-spatial processing than the other. Under dual-task conditions designed
to load the visuo-spatial scratchpad, a visuo-spatial writing topic should be more difficult
than a similar non visuo-spatial writing topic.
Some clues about the involvement of visuo-spatial processes in language tasks
come from in the hemispheric asymmetry literature. Omstein, Herron, Johnstone and
Swencionis (1979) compared the EEG measure of brain activity of participants while they
read either technical passages or folktales. The left hemisphere acted the same with the
technical material and the folktales. However, the right hemisphere activated while the
subjects read the folk stories, but it did not while they read the technical material. Ornstein
et al., explain that technical material is almost exclusively imageless, while stories generate
lots of images, and that “the sense of a story emerges through style, images, and feelings”
(Omstein, 1998, p. 73). Thus, reading may rely on the visuo-spatial processes associated
58

59
with the right hemisphere as well as the left hemisphere, depending on what is being read.
In another investigation of visuo-spatial information within language, Atwood
(1971) carried out an experiment in which participants heard either highly imageable
phrases such as “Nudist devouring bird”, or highly abstract phrases such as “The intellect
of Einstein was a miracle.” Each phrase was followed by a simple classification task
presentated either audtitorily or visually. Atwood reported a tendency for the imageable
phrases to be disrupted much more by processing the visually presented digits, while the
abstract sentences were disrupted more by auditory processing.
The current experiment tests Kellogg’s (1996) claims that visuo-spatial resources
are involved in planning, particularly when writers plan by visualizing ideas. This
experiment tests this claim by comparing writing on two different topics. Both topics are
descriptive in nature, but one requires visualization and spatial memory while the other
does not. Quality and fluency are expected to be better on the non visuo-spatial topic
under dual-task conditions, but no differences are expected under single-task conditions.
Method
Participants
Fifty-eight general psychology students, from the University of Florida participated
in this experiment for which they received credit towards course requirement. To assess
differences in quality or fluency associated only with topic, the first twenty of these
students wrote two essays, one on each topic under single-task conditions. The remaining
thirty-eight students participated in all phases of the experiment except single-task writing
conditions. All reported that they were comfortable interacting with computers with a
mouse and were proficient typists.

60
Apparatus and Materials
Participants were each seated at an IBM-compatible 586 computer running
Microsoft Windows 95 with 15 inch monitors. Stimuli in all phases of the experiment
were created in Microsoft Visual Basic version 4. Two descriptive essay topics were
employed by this study. One essay required participants to describe the spatial layout of
the house they grew up in, while the other topic required them to describe the perfect job.
Procedure
There were 4 phases to this experiment; A typing test, single-task performance of a
visuo-spatial task, and two dual-task conditions. All phases of this experiment required
participants to interact with a computer.
In order to assess typing proficiency, participants were given 2 minutes to
duplicate a passage of text as quickly as the could. The visuo-spatial monitoring task from
experiment 2 was used here for 5 minutes. This task was designed to load the visuo-spatial
scratchpad but not the phonological loop. The computer displayed equal-sized arrows that
pointed up, down, left, or right. The participants' task was to maintain in memory a
representation of an x, y coordinate system and an imaginary placeholder at the 0,0
coordinate. They were instructed to move the imaginary placeholder one standard, but
arbitrary, unit of distance in the direction of the arrow presented. They were further
instructed to click the mouse button any time the placeholder moved away from either axis
and then returned to that axis. A new arrow was presented every 5 seconds and, on the
average, a response was required twice per minute. Crossings of the x- and y-axis
occurred equally often.

61
During each of two 10-minute dual-task phases of the experiment, participants
simultaneously performed the visuo-spatial task and wrote an essay on a given topic. The
essay topics "The perfect job" and a "The house you grew up in" were counterbalanced to
assess practice effects associated with order.
Results and Discussion
Essay quality was again assessed on a 6-point scale using the essay-sort method
(inter-rater reliability = 84%) (Madigan, Johnson, & Linton, 1994). Essay quality was also
assessed by Flesch-Kincaid grade level calculations which are based on average number of
words per sentence and the average number or syllables per word. Document fluency was
measured by the number of words generated per minute, and a signal detection paradigm
(Green & Swets, 1966) was employed in the visuo-spatial secondary task. Additional
calculations of the percentage of sentences that contained spatial information were
calculated for the essays written on the visuo-spatial topic under both single and dual-task
conditions.
As in experiments 2a and 2b, dual-task conditions here were first partitioned into
2.5 minute epochs for which individual d’ scores were calculated. Again, patterns of d’
scores indicated that secondary task performance was not automated over time.
As illustrated in Table 2, secondary task d’ scores were not significantly different
between the two writing topic conditions under either single or dual-task conditions, and
unlike experiments 2a and 2b, no decrement in d’ scores was found when comparing
single to dual-task conditions (see figure 7). Failure to replicate the d’ decrements
observed in experiments 2a and 2b may indicate differences in the research participant
pool from Summer to Spring semesters.

62
Table 2
Means and standard deviations for d’ scores for the single-task and dual-task with visuo-
spatial and non-visuo-spatial writing topics.
Measure
Single-Task
Visuo-spatial
Topic
Dual-task
Non-visuo-spatial
Topic
Dual-task
d’
Mean
4.36
3.70
3.73
SD
4.19
4.11
3.98
As expected, no quality, grade level, or fluency differences were found between
topics under single-task conditions (see Table 3) suggesting that there were no pre¬
existing differences associated with the two topics.
Table 3
Means and standard deviations for Flesch-Kincaid grade level assessment and words per
minute for each writing topic under single and dual-task conditions.
Measure
Visuo-spatial
Essay Single-
Task
Non- Visuo-
spatial Essay
Single-Task
Visuo-spatial
Essay Dual-task
Non- Visuo-
spatial Essay
Dual-task
Grade Level
Mean
9.48
9.32
4.73
6.99
SD
2.06
2.11
1.68
1.9
Quality
Mean
3.25
3.22
2.65
2.78
SD
.96
.92
.87
.88
WPM
Mean
18.8
19.7
15.1
16.5
SD
.43
.45
.36
.47
As in experiment 2A and 2B, quality, fluency, and grade level measures were
disrupted by the addition of the secondary task (F (1,56) = 13.46, p < .001), (F (1,56) =

63
10.23, p < .001) and (F (1,56) =63.19, p<001) respectively (see Figure 7). The pattern of
results for quality and fluency under dual-task conditions indicated differences between
topics in the direction hypothesized, but those trends did not reach significance. Flesch-
Kincaid grade level calculations did support the hypothesis. The non visuo-spatial topic
grade levels (avg. = 6.99) were significantly better than grade levels for the visuo-spatial
topic (avg. = 4.73) under dual-task conditions as evidenced by a significant interaction
between topic and task condition (F (1, 56) = 13.40, p = .001).
-50 -30 -10 10 30
Percent Decrement
50
Figure 7. Dual-task decrement in secondary task d’, fluency, quality ratings, and Flesch-
Kincaid grade level assessments.
Although the expected differences between quality and fluency of the two topics
under dual-task conditions were not observed, the grade level assessment suggests greater
interference between the secondary task and writing on the visuo-spatial topic. Trends in
the quality and fluency ratings were consistent with the hypotheses, but differences did not
reach significance. One explanation for higher than expected visuo-spatial essay quality
was that writers might have avoided visuo-spatial planning in the presence of a visuo-

64
spatial secondary task. To test this possibility, the percentage of sentences that contained
some visual or spatial description was calculated. Sentences such as, "if you walk into the
master bedroom, on the right, after the closet, is a sink, then a bathroom" were considered
spatial sentences whereas sentences such as "My house was full of nice furniture that I
liked to play around" were considered non-spatial. The visuo-spatial essay written in the
presence of the secondary task contained significantly fewer spatial description sentences
than essays written on the same topic in isolation (56% Vs 70% respectively) (t (56) =1.9,
P= .03). More research is needed to control for the possibility that a phonological secondary
task might have the same effect, but this result suggests that the presence of the secondary
task changed the way the writers addressed the topic.

CHAPTER 5
GENERAL DISCUSSION
Experiment 1 illustrated the varying capacity in the phonological loop which
clearly presents explanatory difficulties for classes of models, including Baddeley and
Hitch’s (1974) working memory, that postulate structures such as a phonological loop or
a visuo-spatial scratchpad to operate as binary, all-or-none mechanisms. Instead, our data
are compatible with an important, but seemingly subtle, variation that conceptualizes these
mechanisms as pools of resources that can operate in a continuous or graded fashion. The
distinction is important for writing theorists and researchers because it may help them to
make closer contact with colleagues who are exploring other aspects of human cognitive
processing, including speech production, language comprehension, or problem solving.
Questions of the processing nature of the visuo-spatial scratchpad remain
unresolved, but the methodology described in Experiment 1 lends itself well to this
question. Under dual-task conditions, if two visuo-spatial secondary tasks of varying
difficulty interfere equally with a primary task, then the all-or-none processing assumption
is supported. On the other hand, if the two visuo-spatial secondary tasks vary in the
relative degree to which they disrupt primary task performance, then graded processing is
supported.
Postulating that the visuo-spatial scratchpad and phonological loop comprise pools
or resources necessitates new guidelines for selecting secondary tasks. Five guidelines are
especially important for comparisons of visuo-spatial and verbal resource involvement: 1.
65

66
Secondary tasks should be equally demanding of attentional resources; 2. The
loading tasks should not change in the degree to which they draw on controlled processes
throughout the experiment; 3. The tasks should have similar structural composition,
requiring the same mode of input and output; 4. Response selection and execution should
occur equally often in the tasks; and 5. Finally, secondary tasks should require different
strategies for their successful performance.
Experiments 2a and 2b followed the secondary task guidelines described above.
Single-task d’ scores were not significantly different between visuo-spatial and verbal
loading tasks, therefore suggesting that these tasks are equally demanding. To assure that
secondary tasks did not become automatic, 20-minute dual-task conditions were broken
down into 4 five-minute epochs for which d’ scores were calculated. Patterns of d’ scores
were inspected qualitatively on an individual basis with the intention of discarding
participants from the study if their performance changed over the course of the dual-task
conditions. No participants were excluded from this study based on this information.
Stimulus presentation modality and rate and response rate were identical for the visuo-
spatial and the verbal secondary tasks. Secondary tasks were constructed so that
performance differences would emerge if participants used the same strategies for each
task.
Experiment 2A reported a significant decrement in essay quality and fluency
associated with loading the phonological loop. These decrements in quality are consistent
with disruptions in comprehension in the presence of irrelevant speech (Martin, Wogalter,
& Forlanon, 1988), and lower holistic quality of essays and diminished syntactic
complexity as measured by the proportion of subordinate clauses occurring before the

67
main verb (Madigan & Linton, 1996). On the other hand, these findings seem to conflict
with Ransdell, Levy & Kellogg (in press) where an irrelevant speech manipulation
reportedly had no effect on essay quality. Inconsistencies here may point to differences
between irrelevant speech manipulations and the letter matching task employed in
Experiment 2A. In particular, response selection and execution stages required by the
letter matching task are likely to place greater demands on working memory, thus having a
greater impact on primary task performance.
Fluency decrements reported in Experiment 2A support Ransdell, Levy, &
Kellogg’s (1997) finding that irrelevant speech reliably reduced fluency as measured by
words per minute. Madigan and Linton (1996) however, found no effect of irrelevant
speech on fluency as measured by word production times and pauses within selected
clauses. This raises questions concerning the reliability of different fluency measures.
Experiment 2B tested the hypothesis that a visuo-spatial secondary task
equal in difficulty to the phonological secondary task would be easier to time-share with
writing. This hypothesis is based on Kellogg’s (1996) claims that writing depends more on
the phonological loop than on the visuo-spatial scratchpad. Results indicated that much
like Experiment 2, quality and fluency deteriorated significantly with the addition of the
visuo-spatial secondary task. Despite equal degradation in writing quality and fluency,
results suggested greater difficulty in timesharing writing with a phonological task because
the phonological secondary task performance fell much more drastically than the visuo-
spatial secondary task performance.
Experiments 2a and 2b therefore indicate that writing draws more on verbal
resources than visuo-spatial resources, which is consistent with EEG studies of writing

68
(Omstein, Herron, Johnstone and Swencionis (1979). These results support a
multicomponent resource model of writing that specifies a division of visuo-spatial and
verbal resources instead of one undifferentiated pool of resources (Kahneman, 1973).
Undifferentiated models predict that equally demanding secondary tasks would be equally
interfering with a primary task. Based on single-task performance measures, our visuo-
spatial and phonological tasks were statistically equivalent, nevertheless, when the loading
tasks were performed at the same time that participants engaged in text production, the
phonological task clearly interfered more than the visuo-spatial task.
Support for a role of visuo-spatial working memory in writing is suggested by the
high interference between writing and a visuo-spatial task. However, as pointed out
earlier, interference could also be due to the response selection and execution aspects of
the visuo-spatial task. How can response selection and execution be teased out of a visuo-
spatial task without loosing some performance measure for that task? One possibility
involves postponing response selection and execution stages until after the dual-task
conditions are completed. For example, visuo-spatial stimuli may be presented as a
secondary task, and memory for the presented information can be tested after completion
of the dual-task condition, thus avoiding response selection and execution stages during
the dual-task condition. The disadvantage to this postponement of judgements is that this
type of visuo-spatial task shifts the emphasis from processing to storage, and this
distinction is not yet fully understood. Klapp and Netick (1988) report interference
between two processing tasks and two memory storage tasks, but not between a
processing and storage task suggesting that processing and storage resources are divided.
Wickens (1984) does not recognize storage and processing as an important distinction

69
though. Other methods of loading working memory in specific ways and then examining
their effects on the output of one or more systems or basic processes should be explored
further. One such method is embodied in the approach taken by Experiment 3.
Support for multicomponent resource models represents new trend in writing
research. Until recently, process-oriented writing research has been theoretically rooted in
accounts of attention based on undifferentiated capacity such as the model advanced by
Kahneman (1973). Writing research progressed under single capacity models as evidenced
by the influence of the Hayes & Flower (1982) model of writing and successful
development of computer-aided writing tools (Kellogg, 1994; Lea, Levy & Marek, 1995;
Lea, Rosen, Levy, Marek & Ransdell, 1995). Recent considerations of the role of limited
cognitive capacity in writing necessitate the more specific models of cognitive limitations
such as Baddeley’s working memory model and multiple resource models.
Experiment 3 took a new approach to test specific hypotheses about the role of
visuo-spatial processes in writing. Specifically, this experiment tested Kellogg’s (1996)
claims by manipulating writing topic so that one of two similar essays required more
visualization or spatial processing than the other. In a dual-task paradigm designed to load
the visuo-spatial scratchpad, the more visuo-spatial of the two topics should be more
difficult to timeshare.
Although significant differences were not found between fluency and quality, the
pattern of results matched our expectations. Significant differences in Flesch-Kincaid
grade level assessments supported the hypothesis indicating that more visuo-spatial
resources were required by the visuo-spatial writing topic. These findings are consistent
with Omsetien, Herron, Johnstone, and Swencionis’ (1979) findings that the right

70
hemisphere EEG is more active in reading passages containing imagery than technical
passages, and Atwood’s (1971) finding that imageable phrases interfered more with a
visual than an auditory secondary task.
Visuo-spatial essay quality was higher than expected, raised concerns that
participants may have preserved essay quality by avoiding visuo-spatial planning by
focusing on non-spatial aspects of the house they grew up in. Closer examination of the
visuo-spatial essays revealed a significantly higher percentage of sentences devoted to
spatial description under single-task conditions. Further testing is needed to clarify
whether this effect is due to the visuo-spatial nature of the secondary task.
We expect this line of research to extend to the subprocess level of writing,
enabling investigations of the demands that formulating, executing and monitoring place
on the visuo-spatial scratchpad and phonological loop. These future investigations will
employ methods based on a resource capacity assumption of phonological loop and visuo-
spatial scratchpad processing, and will include the methods for selecting secondary tasks
that we addressed earlier.
Current research in our laboratory ( Levy, White, Lea, & Ransdell, 1998) extends
the methodologies introduced here to evaluate other claims from current working
memory-based models of text production. For example, we have recently devised a way to
overcome the inherent confounding in Experiments 2 and 3 between the specific stimuli
shown and the participants’ response decision rules by holding the stimuli constant. Thus,
single alphabetic characters might be shown at the same rate as the stimuli presented as
before, but periodically they would simultaneously change color, case, font, physical
location, and value. Different groups of participants asked to respond whenever two

71
stimuli in succession were the same color or location (to engage the visuo-spatial
scratchpad), or whenever the adjacent stimuli both contained a long “e” phoneme (to
engage the phonological loop, or whenever they formed a 2-letter word (to focus on the
central executive). The physical stimuli would remain the same as would the physical
response. Any difficulties between the single tasks or between their effects on the writing
process would then simply be the result of the instructions designed to engage working
memory components differentially.
Clearly much remains to be discovered concerning visuo-spatial working memory
in general and specifically in the role of visuo-spatial processes in writing. It is likely that
developments in working memory and multiple resource theories will continue to influence
the way visuo-spatial and verbal processes are investigated.

APPENDIX A
Experiment 1 stimuli
1 w d j k m
2 z t b p k
3 h q s c v
4 k p j f x n
5 f w y k p v
6 r w f g k r
7 w t f j n p k
8 qtpkhds
72

9
10
11
12
13
14
15
16
17
18
73
z c b m k y r
d r f v y j n p
k 1 y n f t r g
xtvhkwdg
wdvyhj lmq
rpyjnbxcw
zrxfhknpy
p k g v w
s r y v x
w g t h k

19
20
21
22
23
24
25
26
27
28
74
P y J n m
w d c v g
z r v h y
t g j v w d
w f t h y k
wdgyjk
p f b s z t
y j v z r n
k d r t h q
h t f v s w g

29
30
31
32
33
34
35
36
37
38
75
s r c g b k p
q z g 1 p y m
k g b t y n r
z t y v b k p
1 h k v r d g
q d z b t n y
frwxscbh
y h j k 1 b f w
dztcybkn
d g j 1 k h f s

39 qtrwyphk
40 zdxfcgvhb
41 zmygcerxw
42 dj fkgmxnc
43 pkntbrcms
44 qzwdxgchb
45 gdfvyrnkm

REFERENCES
Atkinson, R.C., & Shiffrin, R.M. (1968). Human memory: A proposed system
and its control processes. In K.W. Spence & J.T. Spence (Eds.), The psychology of
learning and motivation: II. New York: Academic Press.
Atwood, G.E. (1971). An experimental study of visual imagination and memory.
Cognitive Psychology, 2. 290-299.
Baddeley, A.D. (1966). Short-term memory for word sequences as a function of
acoustic, semantic and formal similarity. Quarterly Journal of Experimental Psychology.
ÃœL 362-366.
Baddeley, A.D. (1992). Is working memory working? The fifteenth Bartlett
lecture. The Quarterly Journal of Experimental Psychology, 44a (IT 1-31.
Baddeley, A.D. (1993). Working memory or working attention? In A. Baddeley &
L. Weiskrantz (Eds.) Attention: Selection awareness and control, (pp. 152-170) New
York: Oxford.
Baddeley, A.D., & Hitch, G.J. (1974). Working memory. In G. Bower (Ed.), The
psychology of learning and motivation. Vol. Ill, pp.47-90, New York: Academic Press.
Baddeley, A.D., & Lewis, V. (1981). Inner active processes in reading: The inner
voice, the imier ear, and the inner eye. In A.M. Lesgold & C.A. Perfetti (Eds.), Interactive
processes in reading. Hillsdale, NJ: Lawrence Erlbaum Associates.
Baddeley, A.D., Lewis, V., Eldridge, M., & Thomson, N. (1984a). Attention and
retrieval from long-term memory. Journal of Experimental Psychology: General, 1113,
518-540.
Baddeley, A.D., Lewis, V., & Vallar, G. (1984b). Exploring the articulatory loop.
The Quarterly Journal of Experimental Psychology, 36A, 233-252.
Baddeley, A.D., Papagno, C., & Vallar, G. (1988). When long-term learning
depends on short-term storage. Journal of Memory and Language. 27. 586-595.
Baddeley, A.D., Thomson, N., & Buchannon, M. (1975). Word length and the
structure of short-term memory. Journal of Verbal Learning and Verbal Behavior. 14.
575-589.
77

78
Bertelson, P. (1967). The psychological refractory period of choice reaction times
with regular and irregular ISPs. Acta Psychologica. 27. 45-56.
Bock, J.K. (1982). Toward a cognitive psychology of syntax: Information
processing contributions to sentence formulation. Psychological Review. 89. 1-47.
Boles, D.B., & Wickens, C.D. (1987). Display formatting in information
integration and nonintegration tasks. Human Factors. 29(4). 395-406.
Bourdin, B., & Fayol, M. (1994). Is written language production more difficult
than oral language production? A working memory approach. International Journal of
Psychology, 29. 591-620.
Broadbent, D. (1958). Perception and communication. Oxford: Permagon.
Broadbent, D. (1982). Task communication and the selective intake of
information. Acta Psychologica, 50. 253-290.
Brooks, L.R. (1968). Spatial and verbal components of the act of recall. Canadian
Journal of Psychology, 22, 349-368.
Cherry, C. (1953). Some experiments on the recognition of speech with one and
two ears. Journal of the Acoustical Society of America. 23. 915-919.
Conrad, R., & Hull, A.J. (1964). Information, acoustic confusion and memory
span. British Journal of Psychology, 55. 429-432.
Craik, F.I., & Lockhart, F.S. (1972). Levels of processing: A framework for
memory research. Journal of Verbal Learning and Verbal Behavior. 11. 671-684.
Craik, F.I., & Watkins, M.J. (1973). The role of rehearsal in short-term memory.
Journal of Verbal Learning and Verbal Behavior. 12. 599-607.
Daneman, M., & Carpenter, P. (1980). Individual differences in working memory
and reading. Journal of Verbal Leming & Verbal Behavior. 19. 450-466.
Daneman, M., & Greene, I. (1986). Individual differences in comprehending and
producing words in context. Journal of Memory and Language. 25. 1-18.
De Renzi, E., & Nichelli, P. (1975). Verbal and nonverbal short-term memory
impairment following hemispheric damage. Cortex. 11. 341-353.
Dinet, J., & Passerault, J. J. (1998, July). Working memory and text production:
The role of the visuospatial sketchpad in writing argumentative and descriptive texts.
Paper presented at the Writing98 Conference, Poitiers, FR.

79
Ericsson, K.A., & Kintsch, W. (1995). Long-term working memory. Psychological
Review. 102. 211-245.
Fagot, C., & Pashler, H. (1992). Making two responses to a single object:
Exploring the central attentional bottleneck. Journal of Experimental Psychology; Human
Perception and Performance. 18, 1058-1079.
Fracker, M.L., & Wickens, C.D. (1989). Resources, contusions, and compatibility
in dual-axis tracking: Displays, controls, and dynamics. Journal of Experimental
Psychology: Human Perception and Performance. 15(T1 80-96.
Frick, R.W. (1988). Issues of representation and limited capacity in the visuo-
spatial scratchpad. British Journal of Psychology. 79. 289-308.
Friedman, A., & Poison, M. C. (1981). Hemispheres as independent resource
systems: Limited-capacity processing and cerebral specialization. Journal of Experimental
Psychology: Human Perception and Performance. 7(5) 1031-1058.
Galin, D., & Omstein, R. (1972). Lateral specialization of cognitive mode: An
EEG study. Psychophysiology. 9. 412-418.
Gathercole, S., & Baddeley, A.D. (1993). Working memory and language. Hove,
UK: Lawrence Erlbaum Associates.
Green, D., & Swets, J. (1966) Signal detection theory and psychophysics. New
York: Wiley.
Hatano, G., & Osawa, K. (1983). Digit memory of grand experts in abacus-
derived mental calculation. Cognition. 15. 95-110.
Hayes, J. (1996). A new framework for understanding cognition and affect in
writing. In C.M. Levy & S. Ransdell (Eds.) The science of writing, (pp. 1-27) Mahwah,
NJ: Erlbaum.
Henry, F.M., & Rogers, D.E. (1960). Increased response latency for complicated
movements and a “memory drum” theory of neuromotor reaction. Research Quarterly,
31, 448-458.
Herdman, C.M., & Friedman, A. (1985). Multiple resources in divided attention: A
cross-modal test of the independence of hemispheric resources. Journal of Experimental
Psychology: Human Perception and Performance. 11(1) 40-49.
Hirst, W., & Kalmar, D. (1987). Characterizing attentional resources. Journal of
Experimental Psychology: General. 116. 1, 68-81.

80
Hue, C., & Ericson, J.R. (1988). Short-term memory for Chinese characters and
radicals. Memory and Cognition. 16. 196-205.
JefFerey, G.C., & Underwood, G. (1995). The role of working memory in the
development of a writing skill: Learning to co-ordinate ideas within written text.
(Unpublished manuscript).
Jones, D.M., Miles, C., & Page, J. (1990). Disruption of proofreading by irrelevant
speech: Effects of attention, arousal or memory? Applied Cognitive Psychology. 4. 89-
108.
Kahneman, D. (1973) Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.
Kahneman, D., Triesman, A., & Gibbs, B.J. (1992). The reviewing of object files:
Object specific integration of information. Cognitive Psychology, 24. 175-219.
Kantowitz, B.H. (1974). Double stimulation. In B.H. Kantowitz (Ed.), Human
information processing. Hillsdale, NJ.: Earlbaum.
Kellogg, R.T. (1996). A model of working memory in writing. In C.M. Levy & S.
Ransdell (Eds.) The science of writing, (pp. 57-71) Mahwah, NJ: Erlbaum.
Kinsboume, M., & Hicks, R.(1978). Functional cerebral space. In J. Requin (Ed.),
Attention and performance VII. Hillsdale, NJ: Erlbaum.
Klapp, S.T., & Netick, A. (1988). Multiple resources for processing and storage in
short-term working memory, Human Factors. 3015). 617-632.
Knowles, W.B. (1963). Operator loading tasks. Human Factors. 5. 151-161.
Laberg, D. (1973). Attention and the measurement of perceptual learning. Memory
& Cognition. 1. 268-276.
Lea, J., Levy, C.M., & Marek-Lovejoy, P. (1995). The topic sentence display tool:
An evaluation of writing performance and process. Presented at the XIII Annual European
Conference on Writing and Computers, London.
Lea, J., Rosen, L., Levy, C.M., Marek-Lovejoy, P., & Ransdell, S.E. (1995). The
intelligent listening word processor: Prospects for the future of writing. Presented at the
XIII Annual European Conference on Writing and Computers, London.
Levy, C.M., & Marek, P. (1998). The role of working memory in writing
processes and performance. In G.C. Jeffery & M. Torrance (Eds.), High and low level

81
writing processes: Managing cognitive demands. Amsterdam: Amsterdam University
Press.
Levy, C. M., White, K., & Lea, J. (1998, July). The role of the visuo-spatial
scratchpad in writing: Testing Kellogg’s multicomponent model. To be presented at the
Writing 98 Conference, Poitiers, FR.
Madigan, R.J., Johnson, S.E., & Linton, P.W. (1994 August). Working memory
capacity and the writing process. Paper presented at the American Psychological Society,
Washington, D.C.
Madigan, R.J., & Linton, P.W. (1996 October). Working memory and the writing
process, (unpublished manuscript).
Manin, K., & Johnson-Laird, P.N. (1982). The mental representation of spatial
descriptions. Memory & Cognition. 10. 181-187.
Martin, R.C., Shelton, J.R., & Yaffee, L.S. (1994). Language processing and
working memory: Neuropsychological evidence for separate phonological and semantic
capabilities. Journal of Memory and Language. 33. 83-11 1.
Martin, R.C., Wogalter, M.S., & Forlano, J.G. (1988). Reading comprehension in
the presence of unattended speech and music. Journal of Memory & Language. 27. 382-
398.
Marek, P., & Levy, C.M. (1996). The role of working memory in writing
processes and performance: II. Evaluation of formulating, monitoring and execution
processes. Paper presented at the European Conference on Writing and Computers,
Barcelona.
McFarland, K., & Ashton, R. (1978). The influence of concurrent task difficulty on
manual performance. Neurophvsiologica, 16. 735-741.
Moray, N. (1959). Attention in dichotic listening. Quarterly Journal of
Experimental Psychology, 1L 56-60.
Moray, N. (1982). Subjective mental load. Human Factors. 23, 25-40.
Moray, N., Johannsen, G., Pew, R.W., Rasmussen, J., Sanders, A.F., & Wickens,
C.D. (1979). Report of the experimental psychology group. In N. Moray (Ed.), Mental
workload: Its theory and measurement. New York: Plenum.
Moscovitsch, M., & Klein, D. (1980). Material -specific perception for visual
words and faces. Journal of Experimental Psychology: Human Perception and
Performance, 6. 590-603.

82
Norman, D., & Bobrow, D. (1975). On data limited and resource limited
processing. Journal of Cognitive Psychology. 7. 44-60.
Navon, D. (1984). Resources-A theoretical soup stone? Psychological Review. 91
(2), 216-234.
Navon, D., & Gopher, D. (1979). On the economy of the human-information
processing system. Psychological Review. 86(3). 214-255.
Omstein, R. (1997). The right mind: Making sense of the hemispheres. New York:
Harcourt Brace & Company.
Omstein, R., Herron, J., Johnstone, J., & Swencionis, C. (1979). Differential right
hemisphere involvement in two reading tasks. Psychophysiology, 16(4). 398-401.
Paivio, A. (1986). Mental representations: A dual-coding approach. New York:
Oxford University Press.
Pashler, H. (1984). Processing stages in overlapping tasks: Evidence for a central
bottleneck. Journal of Experimental Psychology: Human Perception and Performance. 10.
358-377.
Pashler, H., & Carrier, M. (1995). Attentional limits in memory retrieval. Journal
of Experimental Psychology: Learning Memory and Cognition. 21(51. 1339-1348.
Pashler, H, & Johnston, J. C. (1989). Interference between temporally
overlapping tasks: Chronometric evidence for ventral postponement with or without
response grouping. Quarterly Journal of Experimental Psychology. 41 A. 19-45.
Ransdell, S., & Levy, C.M. (1996). Working memory constraints on writing
quality and fluency. In C.M. Levy & S. Ransdell (Eds.) The science of writing, (pp. 93-
105) Mahwah, NJ: Erlbaum.
Ransdell, S., & Levy, C.M. (1998). Writing, reading and speaking memory spans
and the importance of resourse flexibility. In G.C. Jeffery & M. Torrance (Eds.), High and
low level writing processes: managing cognitive demands. Amsterdam: Amsterdam
University Press.
Ransdell, S., Levy, C.M., & Kellogg, R.T. (in press). Concurrent loads on
working memory during text production. Memory.
Rolfe, J.M. (1971). The secondary task as a measure of mental load. In W.T.
Singleton, J.G. Fox, & D. Whitfield (Eds.) Measurement of man at work. London: Taylor
and Francis.

83
Salame', P., & Baddeley, A.D. (1982). Disruption of short-term memory by
unattended speech: Implications for the structure of working memory. Journal of Verbal
Learning and Verbal Behavior. 21. 150-164.
Shallice, T., & Warrington, E.K. (1970). Independent functioning of verbal
memory stores: A neuropsychological study. Quarterly Journal of Experimental
Psychology, 22. 261-273.
Shepard, R.N. (1978). The mental image. American Psychologist, 33. 125-137,
Sternberg, S. (1969). The discovery of processing stages: Extensions of Donders’
method. Acta Psvchologica. 30. 276-315.
Thomson, J.A. (1983). Is continuous visual monitoring really necessary in visually
guided locomotion? Journal of Experimental Psychology: Human Perception and
Performance. 9. 427-443.
Triesman, A.M., & Davies, A. (1973). Divided attention to ear and eye. In S.
Komblum (Ed.). Attention and performance (Vol 4). New York: Academic Press.
Tulving, E. (1966). Subjective organization and effects of repetition in multi-trial
free-recall learning. Journal of Verbal Learning and Verbal Behavior. 6. 193-197.
Tzeng, O.J. (1973). Positive recency effects in delayed free recall. Journal of
Verbal Learning and Verbal Behavior. 12. 436-439.
Vallar, G., & Baddeley, A.D. (1984). Fractionation of working memory.
Neuropsychological evidence for a phonological short-term store. Journal of Verbal
Learning and Verbal Behavior. 23. 151-161.
Waters, G.S., Rochon, E., & Caplan, D. (1992). The role of high-level speech
planning in rehearsal: Evidence from patients with apraxia of speech. Journal of Memory
and Language. 31. 54-73.
Welford, A.T. (1967). Single channel operations in the brain. Acta Psvchologica.
27, 5-22.
Wickens, C.D. (1976). The effects of divided attention on information processing
in manual tracking. Journal of Experimental Psychology: Human Perception and
Performance. 2(1). 1-13.
Wickens, C.D. (1980). Processing resourc demands of failure detection in dynamic
systems. Journal of Experimental Psychology: Human Perception and Performance. 6(3),
564-577.

84
Wickens, C.D. (1984). Processing resources in attention. In R. Parasuraman (Ed.)
Varieties of attention. New York: Academic Press.
Yee, P.L., Hunt, E., & Pellegringo, J.W. (1991). Coordinating cognitive
information: Task effects and individual differencs in integrating information from several
sources. Cognitive Psychology, 23. 615-680.

BIOGRAPHICAL SKETCH
Joseph D. Lea III was born November 28, 1970, in Tuscon, Arizona, to Joseph D.
Lea Jr. and Billie Ann Lea. Joseph (Josh) majored in psychology at the University of
North Carolina at Wilmington and graduated cum laude with honors in psychology in
May, 1993. Joseph (Joe) entered the Cognitive and Sensory Processes doctoral program
at the University of Florida in August, 1993. After graduation Joseph will begin work at
IBM in San Jose, California as a human factors engineer in software development.
85

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree ofDoctorofPhü,osoph\
C. Michael Levy; Chairarían
Professor of Psychology
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Do^rofyjPhilqpopJiy^
Barton Weitz,
J.C. Penny Eminent Scholar
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Doctor^f Plpf^so^h^^
Shari Ellis,
Assistant Professor of Psychology
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in spope and
quality, as a dissertation for the degree of Doctorar 1
’ Fischler,
Professor of Psychology
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Doctor of Philosophy.
Stephen Blessing, 0
Assistant Professor of Psychology
This dissertation was submitted to the Graduate Faculty of the Department of
Psychology in the College of Liberal Arts and Sciences and to the Graduate School
and was accepted as partial fulfillment of the requirements for the degree of Doctor
of Philosophy.
August, 1998
Dean, Graduate School




PAGE 1

9(5%$/ $1' 9,68263$;,$/ 352&(66,1* '(0$1'6 ,1 :5,7,1* %\ -26(3+ /($ $ ',66(57$7,21 35(6(17(' 72 7+( *5$'8$7( 6&+22/ 2) 7+( 81,9(56,7< 2) )/25,'$ ,1 3$57,$/ )8/),//0(17 2) 7+( 5(48,5(0(176 )25 7+( '(*5(( 2) '2&725 2) 3+,/2623+< 81,9(56,7< 2) )/25,'$

PAGE 2

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fV 6DQWD 7HUHVD /DERUDWRU\ IRU WKH LQFHQWLYH WR ILQLVK WKLV SURMHFW RQ VFKHGXOH )LQDOO\ ZRXOG OLNH WR WKDQN P\ ZLIH (OHQD 6FKXKPDQQ P\ SDUHQWV DQG P\ JUDQGIDWKHU IRU DOO RI WKHLU VXSSRUW DQG HQFRXUDJHPHQW

PAGE 3

7$%/( 2) &217(176 3DJH $.12 :/('*(0(17 6 LL $%675$&7 Y &+$37(56 ,1752'8&7,21 5HVRXUFHV DQG 6WUXFWXUHV &DSDFLW\ 7KHRULHV 6WUXFWXUDO 7KHRULHV 0XOWLSOH 5HVRXUFH 7KHRULHV :RUNLQJ 0HPRU\ :RUNLQJ 0HPRU\ LQ :ULWLQJ $GDSWHG :RUNLQJ 0HPRU\ 0RGHO 6XPPDU\ (;3(5,0(17 0HWKRG 5HVXOWV DQG 'LVFXVVLRQ (;3(5,0(17 ([SHULPHQW $ 0HWKRG 5HVXOWV DQG 'LVFXVVLRQ (;3(5,0(17 % 0HWKRG 5HVXOWV DQG 'LVFXVVLRQ &RPELQHG $QDO\VLV RI ([SHULPHQWV $ DQG % (;3(5,0(17 0HWKRG 5HVXOWV DQG 'LVFXVVLRQ *(1(5$/ ',6&866,21

PAGE 4

$33(1',; 5()(5(1&(6 %,2*5$3+,&$/ 6.(7&+ LY

PAGE 5

$EVWUDFW RI 'LVVHUWDWLRQ 3UHVHQWHG WR WKH *UDGXDWH 6FKRRO RI WKH 8QLYHUVLW\ RI )ORULGD LQ 3DUWLDO )XOILOOPHQW RI WKH 5HTXLUHPHQWV IRU WKH 'HJUHH RI 'RFWRU RI 3KLORVRSK\ 9(5%$/ $1' 9,68263$7,$/ 352&(66,1* '(0$1'6 ,1 :5,7,1* %\ -RVHSK /HD ,,, $XJXVW &KDLLSHUVRQ & 0LFKDHO /HY\ 0DMRU 'HSDUWPHQW 3V\FKRORJ\ :ULWLQJ UHVHDUFK KDV UHFHQWO\ EHJXQ WR FRQVLGHU WKH UROH WKDW OLPLWDWLRQV LQ FRJQLWLYH FDSDFLW\ SOD\ LQ WKH ZULWLQJ SURFHVV WKXV REOLJDWLQJ ZULWLQJ UHVHDUFKHUV WR H[DPLQH H[LVWLQJ WKHRULHV RI FRJQLWLYH OLPLWDWLRQV 7KLV GLVVHUWDWLRQ UHYLHZV IRXU PDMRU FODVVHV RI FRJQLWLYH SURFHVVLQJ PRGHOV WKDW PLJKW VHUYH DV WKHRUHWLFDO EDVHV IRU H[WHQVLRQ WR ZULWWHQ ODQJXDJH SURGXFWLRQ )RXU H[SHULPHQWV DVVHVV WKH PHULWV RI D UHFRQFHSWXDOL]DWLRQ RI .HOORJJnV f PRGHO RI ZRUNLQJ PHPRU\ DQG ZULWLQJ E\ HPSKDVL]LQJ WKH UROH RI UHVRXUFH SRROV UDWKHU WKDQ ELQDU\ IXQFWLRQLQJ FRJQLWLYH VWUXFWXUHV (PSLULFDO ILQGLQJV VXSSRUW WKH UHFHQW WUHQG WRZDUG PXOWLFRPSRQHQW FODVVHV RI ZRUNLQJ PHPRU\ RU UHVRXUFH PRGHOV SDUWLFXODUO\ WKRVH WKDW GLVWLQJXLVK EHWZHHQ YLVXR VSDWLDO DQG YHUEDOSKRQRORJLFDO SURFHVVLQJ 5HVXOWV DOVR LQGLFDWH WKDW WKHUH LV D VLJQLILFDQW YLVXRVSDWLDO FRPSRQHQW LQ WKH SODQQLQJ VWDJHV RI ZULWLQJ

PAGE 6

&+$37(5 ,1752'8&7,21 :ULWLQJ UHVHDUFK KDV UHFHQWO\ EHJXQ WR FRQVLGHU WKH UROH WKDW OLPLWDWLRQV LQ FRJQLWLYH FDSDFLW\ SOD\ LQ WKH ZULWLQJ SURFHVV $ FRPPRQ DVVXPSWLRQ LV WKDW ZULWHUV KDYH D OLPLWHG FRJQLWLYH FDSDFLW\ WKH VL]H DQGRU HIILFLHQF\ RI ZKLFK YDULHV EHWZHHQ LQGLYLGXDOV 5HVHDUFK RQ FRJQLWLYH FDSDFLW\ DQG ZULWLQJ LV UHODWLYHO\ QHZ FRPSDUHG WR VWXGLHV RI FDSDFLW\ OLPLWDWLRQV LQ UHDGLQJ DQG WR VSRNHQ ODQJXDJH SURGXFWLRQ DQG FRPSUHKHQVLRQ VHH *DWKHUFROH DQG %DGGHOH\ IRU DQ RYHUYLHZf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

PAGE 7

f 6WUXFWXUDO WKHRULHV ZHUH SRSXODUL]HG LQ WKH V ZLWK H[SHULPHQWDO LQYHVWLJDWLRQV RI WKH GLFKRWLF OLVWHQLQJ WDVN ZKLFK UHYHDOHG WKDW DWWHQWLRQ ZDV VHYHUHO\ OLPLWHG &KHUU\ %URDGEHQW 0RUD\ f 6HYHUDO FODVVHV RI WKHRUHWLFDO PRGHOV ZHUH JHQHUDWHG WR ORFDOL]H WKH SRLQW LQ LQIRUPDWLRQ SURFHVVLQJ ZKHUH WKH ERWWOHQHFN RFFXUV (DUO\VHOHFWLRQ WKHRULHV FODLPHG WKDW WKLV ERWWOHQHFN LV LQ WKH HDUO\ SURFHVVHV VXFK DV SHUFHSWLRQ DQG ODWHVHOHFWLRQ WKHRULHV SRLQWHG WR VWDJHV RI GHFLVLRQ PDNLQJ DQG UHVSRQVH VHOHFWLRQ DV WKH OLPLWLQJ VWDJHV RI LQIRUPDWLRQ SURFHVVLQJ f (DUO\ KXPDQ IDFWRUV UHVHDUFK LQ WKH PHDVXUHPHQW RI PHQWDO ZRUNORDG OHG WR WKH FRQFHSWXDOL]DWLRQ RI WKH KXPDQ RSHUDWRU SRVVHVVLQJ D SRRO RI OLPLWHGFDSDFLW\ UHVRXUFHV .QRZOHV f 7KLV UHVRXUFH FDSDFLW\ DSSURDFK RIIHUV WKH VLPSOHVW IUDPHZRUN IRU LQWHUSUHWLQJ WKH GHPDQGV SODFHG RQ WKH ZULWHU E\ SRVWXODWLQJ DQ XQGLIIHUHQWLDWHG SRRO RI PHQWDO UHVRXUFHV .DKQHPDQ f 3HUIRUPDQFH RQ DQ\ JLYHQ WDVN GHFOLQHV RQO\ ZKHQ WKH DJJUHJDWH RI WDVN GHPDQGV H[FHHG WKH FDSDFLW\ RI UHVRXUFHV f $Q DOWHUQDWH FRQFHSWLRQ RI WKLV UHVRXUFH HFRQRP\ PDNHV D GLVWLQFWLRQ LQ WKH W\SH RI UHVRXUFHV DYDLODEOH DW DQ\ JLYHQ PRPHQW 7KLV fPXOWLSOH UHVRXUFHf DSSURDFK SRVLWV WKDW WKHUH PD\ EH YDULRXV W\SHV RI UHVRXUFHV DQG WKDW GLIIHUHQW WDVNV PD\ UHTXLUH GLIIHUHQW W\SHV RI UHVRXUFHV LQ YDULRXV FRPSRVLWLRQV 7KH PRVW ZLGHO\ VXSSRUWHG GLYLVLRQ RI UHVRXUFHV LV D QHXURORJLFDOO\ SODXVLEOH GLYLVLRQ EHWZHHQ YLVXRVSDWLDO DQG YHUEDO UHVRXUFHV :LFNHQV )ULHGPDQ t 3RLVRQ f )ULHGPDQ DQG 3RLVRQ f 9 SURSRVHG D IUDPHZRUN ZLWK WKH LQWHQWLRQ RI H[WHQGLQJ PXOWLSOH UHVRXUFH PRGHOV WR RXU XQGHUVWDQGLQJ RI FHUHEUDO VSHFLDOL]DWLRQ SKHQRPHQD 7KH\ VXJJHVW WKDW WKH SK\VLFDOO\ GLVWLQFW FHUHEUDO KHPLVSKHUHV FDQ EH VHHQ DV LQGHSHQGHQW UHVRXUFH V\VWHPV DQG WKDW

PAGE 8

WKH OHIW DQG ULJKW KHPLVSKHUHV WRJHWKHU IRUP D V\VWHP RI WZR PXWXDOO\ LQDFFHVVLEOH DQG ILQLWH SRROV RI UHVRXUFHV p %DGGHOH\ t +LWFKfV f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fV SDUWLFXODU FRJQLWLYH IXQFWLRQV KDYH EHHQ DVVLJQHG WR VSHFLILF UHJLRQV RI WKH EUDLQ *HQHUDOO\ YLVXR VSDWLDO SURFHVVLQJ KDV EHHQ ORFDOL]HG WR WKH ULJKW KHPLVSKHUH ZKLOH WKH OHIW KHPLVSKHUH LV EHOLHYHG WR SOD\ D VWURQJHU UROH LQ YHUEDO WDVNV 3RSXODU PXOWLSOH UHVRXUFH PRGHOV :LFNHQV )ULHGPDQ t 3RLVRQ f DQG %DGGHOH\ DQG +LWFKfV f ZRUNLQJ PHPRU\ PRGHO EXLOG RQ WKLV QHXURORJLFDO VHSDUDWLRQ RI YLVXRVSDWLDO DQG YHUEDO SURFHVVLQJ $V D ZULWHU SODQV LGHDV WUDQVODWHV LGHDV LQWR SURVH W\SHV RU KDQGZULWHV VHQWHQFHV DQG PRQLWRUV DOO RI WKHVH DFWLYLWLHV PDQ\ GHPDQGV DUH SODFHG RQ WHPSRUDU\ VWRUDJH DQG SURFHVVLQJ FDSDFLW\ 7KLV GLVVHUWDWLRQ VHHNV WR SURYLGH DQ DFFRXQW RI WKH GLYLVLRQ RI ODERU EHWZHHQ YLVXRVSDWLDO DQG YHUEDO SURFHVVLQJ GHPDQGV DQG FDSDFLW\ OLPLWDWLRQV DV WKH\ LQIOXHQFH WKH ZULWLQJ SURFHVV 6SHFLDO WUHDWPHQW ZLOO EH JLYHQ WR WKH ZKROO\ XQWRXFKHG TXHVWLRQ RI KRZ YLVXRVSDWLDO UHVRXUFHV FRQVWUDLQ ZULWLQJ 7KH EDODQFH RI WKLV FKDSWHU

PAGE 9

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t-RKQVWRQ f ZKHQ SHRSOH DWWHPSW WR VLPXOWDQHRXVO\ HQJDJH LQ FRPSHWLQJ PHQWDO RSHUDWLRQV LQFRPSDWLEOH VWDJHV RU SURFHVVHV DUH LQLWLDWHG ZKLFK UHVXOWV LQ LQWHUIHUHQFH EHWZHHQ WDVNV ,Q RWKHU ZRUGV WZR WDVNV FRPSHWH ZKHQ WKH\ HDFK VLPXOWDQHRXVO\ UHTXLUH D VLQJOH FRPPRQ SURFHVV )RU H[DPSOH ZH ZRXOG H[SHFW LQWHUIHUHQFH EHWZHHQ WZR WDVNV VXFK DV FRQYHUVLQJ RQ WKH WHOHSKRQH ZKLOH VLPXOWDQHRXVO\ OLVWHQLQJ WR D UDGLR LQWHUYLHZ PRUH VR WKDQ WKH GXDOWDVN FRPELQDWLRQ WKDQ FDUU\LQJ RQ D FRQYHUVDWLRQ ZKLOH ULGLQJ D ELNH

PAGE 10

SUHVXPDEO\ EHFDXVH VLPLODU SURFHVVHV DUH GHPDQGHG E\ SURGXFLQJ DQG FRPSUHKHQGLQJ DXGLEOH VSHHFK 7ZR FODVVHV RI WKHRUHWLFDO PRGHOV RI DWWHQWLRQ KDYH HPHUJHG IURP FRQVLGHUDWLRQ RI WKHVH K\SRWKHWLFDO YDULDEOHV 7KHVH DUH FDSDFLW\ RU UHVRXUFH WKHRULHV GHVFULEHG E\ .DKQHPDQ f DQG VWUXFWXUDO WKHRULHV ZKLFK EHJDQ ZLWK %URDGEHQW f DQG UHFHQWO\ KDYH EHHQ HODERUDWHG E\ 3DVKOHU DQG KLV DVVRFLDWHV 3DVKOHU t -RKQVWRQ 3DVKOHU t &DUULHU f $IWHU FRQVLGHULQJ VWUXFWXUDO DQG FDSDFLW\ WKHRULHV LQ PRUH GHWDLO ZH ZLOO H[DPLQH PXOWLSOH UHVRXUFH PRGHOV DQG %DGGHOH\f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f LQYROYH SUHVHQWLQJ WZR WDVNV WR DQ LQGLYLGXDO WR SURFHVV VLPXOWDQHRXVO\ RU VRFDOOHG GXDOWDVN FRQGLWLRQV 5HVRXUFH WKHRULHV LQ ODUJH SDUW ZHUH FUHDWHG WR H[SODLQ

PAGE 11

LQVWDQFHV ZKHUH WZR SURFHVVHV ZHUH VLPXOWDQHRXVO\ DFWLYDWHG 3HUKDSV WKH HDUOLHVW WDVN DOWKRXJK QRW D WUDGLWLRQDO GXDOWDVNf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fV EDVHOLQH UHDFWLRQ WLPH ZDV VXEWUDFWHG IURP KLV RU KHU UHDFWLRQ WLPH WR D WRQH SUHVHQWHG GXULQJ SHUIRUPDQFH RI D SULPDU\ WDVN LQ RUGHU WR FRPSXWH DQ LQWHUIHUHQFH UHDFWLRQ WLPH ZKLFK ZDV DVVXPHG WR EH DQ LQGH[ RI PHQWDO HIIRUW .DKQHPDQ f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f WKH IDLOXUH GHWHFWLRQ RI DXWRSLORW V\VWHPV :LFNHQV t .HVVHO f WDUJHW WUDFNLQJ V\VWHPV XVHG E\ PLOLWDU\ SLORWV )UDFNHU t :LFNHQV f DQG LQ WKH GHVLJQ RI YLVXDO GLVSOD\V UHTXLULQJ WKH LQWHJUDWLRQ RI LQIRUPDWLRQ IURP VHYHUDO VRXUFHV
PAGE 12

OHDUQLQJ PHPRU\ DQG ZULWLQJ UHVHDUFK 5HVHDUFKHUV VWXG\LQJ DXWRPDWLFLW\ KDYH DOVR EHQHILWHG IURP WKH UHVRXUFH FRQFHSWXDOL]DWLRQ RI DWWHQWLRQ $V D WDVN EHFRPHV PRUH DXWRPDWLF IHZHU UHVRXUFHV DUH QHHGHG IRU LWV VXFFHVVIXO FRPSOHWLRQ /DEHUJH f 'HVSLWH WKH H[SODQDWRU\ SRZHU RYHU GXDOWDVN SHUIRUPDQFH DQG VXFFHVV LQ DSSOLHG VHWWLQJV FDSDFLW\ WKHRULHV KDYH UHFHLYHG VRPH FULWLFLVP 1DYRQ f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

PAGE 13

SDUDOOHO SKRQH OLQHV RU LQWHUIHULQJ SDWWHUQV RI DFWLYDWLRQ LQ ZKLFK LQKLELWLRQ QHFHVVDU\ IRU RQH WDVN GLVUXSWV DFWLYDWLRQ RQ FRQFXUUHQW WDVN ,Q GHIHQVH RI UHVRXUFH WKHRULHV VRPH UHVHDUFKHUV +HUGPDQ t )ULHGPDQ :LFNHQV f FODLP WKDW GLVWLQJXLVKLQJ EHWZHHQ UHVRXUFH DQG VWUXFWXUH WKHRULHV PD\ EH DFFRPSOLVKHG E\ PDQLSXODWLRQ RI WDVN SULRULW\ LQ GXDOWDVN FRQGLWLRQV ,I SHUIRUPDQFH FDQ EH WUDGHG RII IURP SULPDU\ WR VHFRQGDU\ WDVNV DQG YLFH YHUVD WKHQ WKHLU LQWHUSUHWDWLRQ LV WKDW WKH WZR WDVNV LQWHUIHUH EHFDXVH WKH DUH FRQVXPLQJ FRPPRQ UHVRXUFHV ,I WKH WZR WDVNV FRPSHWH IRU D FRPPRQ VWUXFWXUH WKHQ WKH H[SHFWDWLRQ LV WKDW SHUIRUPDQFH ZLOO QRW WUDGH RII EHWZHHQ WKH WZR WDVNV LQ D JUDGXDO PDQQHU 1DYRQ f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f 7KLV WHFKQLTXH HQVXUHG WKDW WKH SDUWLFLSDQWV IRFXVHG RQ RQH RI WKH WZR PHVVDJHV ([SHULPHQWDO LQYHVWLJDWLRQV RI WKH GLFKRWLF OLVWHQLQJ WDVN UHYHDOHG WKDW DWWHQWLRQ ZDV VHYHUHO\ OLPLWHG &KHUU\ %URDGEHQW 0RUD\ f 6HYHUDO FODVVHV RI WKHRUHWLFDO PRGHOV ZHUH JHQHUDWHG WR

PAGE 14

ORFDOL]H WKH SRLQW LQ LQIRUPDWLRQ SURFHVVLQJ ZKHUH WKH ERWWOHQHFN RFFXUUHG (DUO\VHOHFWLRQ WKHRULHV FODLPHG WKDW WKH ERWWOHQHFN ZDV LQ WKH HDUO\ SURFHVVHV VXFK DV SHUFHSWLRQ DQG ODWHVHOHFWLRQ WKHRULHV SRLQWHG WR VWDJHV RI GHFLVLRQ PDNLQJ DQG UHVSRQVH VHOHFWLRQ DV WKH OLPLWLQJ VWDJHV RI LQIRUPDWLRQ SURFHVVLQJ $QRWKHU SDUDGLJP ZDV DOVR RIWHQ XVHG LQ WKH LQYHVWLJDWLRQ RI WKH OLPLWDWLRQV RI DWWHQWLRQ 7KH SV\FKRORJLFDO UHIUDFWRU\ SHULRG 353f SDUDGLJP %HUWHOVRQ :HOIRUG f DOVR NQRZQ DV WKH RYHUODSSLQJ WDVNV SDUDGLJP 3DVKOHU f LQYROYHV WKH SUHVHQWDWLRQ RI WZR VWLPXOL 6 DQG 6f LQ UDSLG VXFFHVVLRQ 3DUWLFLSDQWV PDNH D UHVSRQVH WR HDFK VWLPXOXV 5 DQG 5f DV TXLFNO\ DV SRVVLEOH 8VLQJ WKLV SDUDGLJP PDQ\ UHVHDUFKHUV .DQWRZLW] 3DVKOHU 3DVKOHU t -RKQVWRQ )DJRW t 3DVKOHU f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t -RKQVWRQ f DQG PRWRU PHFKDQLVPV .DKQHPDQ f

PAGE 15

%URDGEHQW f FULWLFL]HG VWXGLHV RI FRQWLQXRXV SHUIRUPDQFH WKDW UHO\ RQ DFFXUDF\ PHDVXUHV EHFDXVH WKH\ FDQQRW GLVFULPLQDWH EHWZHHQ WKH SUHGLFWLRQV RI FDSDFLW\ WKHRU\ UHJDUGLQJ VLPXOWDQHRXV PHQWDO SURFHVVLQJ RI ERWK WDVNV DQG WKH SUHGLFWLRQV RI VWUXFWXUDO PRGHOV UHJDUGLQJ VWUDWHJ\ VZLWFKLQJ 3DVKOHU DQG -RKQVWRQ f LQFRUSRUDWHG %URDGEHQWfV VXJJHVWLRQV ZKHQ WKH\ LQYHVWLJDWHG GXDOWDVN LQWHUIHUHQFH ZLWK WKH 353 7KH\ DUJXHG WKDW WKLV SDUDGLJP SRWHQWLDOO\ SURYLGHV PXFK PRUH GHWDLOHG LQIRUPDWLRQ DERXW WKH WLPH FRXUVH RI GXDOWDVN LQWHUIHUHQFH WKDQ LV REWDLQHG LQ FRQWLQXRXV GXDOWDVN VWXGLHV 3DVKOHU DQG -RKQVWRQ GLYLGHG PRGHOV RI GXDOWDVN LQWHUIHUHQFH LQWR WZR FDWHJRULHV FDSDFLW\ VKDULQJ PRGHOV GLVFXVVHG HDUOLHU DQG fSRVWSRQHPHQW PRGHOVf ZKLFK DUH VWUXFWXUDO LQ QDWXUH ,Q SRVWSRQHPHQW PRGHOV VRPH ERWWOHQHFN VWDJH RU SURFHVV FDQQRW RSHUDWH VLPXOWDQHRXVO\ IRU HDFK RI WZR RYHUODSSLQJ WDVNV $V D UHVXOW SURFHVVLQJ RI WKLV VWDJH LQ WKH VHFRQG WDVN LV OLWHUDOO\ SRVWSRQHG 7KLV SURGXFHV WKH UHODWLYHO\ VWUDLJKWIRUZDUG K\SRWKHVLV WKDW DV VWLPXOXV RQVHW DV\QFKURQ\ 62$f EHWZHHQ 6, DQG 6 LV UHGXFHG WKHUH VKRXOG FRPH D SRLQW DW ZKLFK DQ\ IXUWKHU UHGXFWLRQ LQ WKH 62$ SURGXFHV D FRUUHVSRQGLQJ LQFUHDVH LQ WKH GXUDWLRQ RI 5 +RZHYHU HYHQ ZKHQ WKLV SUHGLFWLRQ LV FRQILUPHG LW GRHV QRW VHHP WR FRQWUDGLFW SUHGLFWLRQV IURP WKH FDSDFLW\ WKHRU\ .DKQHPDQ f DGGUHVVHG WKH QRWLRQ WKDW VWUXFWXUDO WKHRULHV DQG D OLPLWHG FDSDFLW\ WKHRU\ PD\ DFFRXQW IRU WKH VDPH ILQGLQJV +H REVHUYHG WKDW WKH DVVXPSWLRQV RI VWUXFWXUDO WKHRULHV DUH PRUH UHVWULFWLQJ WKDQ WKH OLPLWHG FDSDFLW\ PRGHOV DQG WKDW SUHGLFWLRQV HVSHFLDOO\ IRU WKH 353 KDYH W\SLFDOO\ IDLOHG WR EH FRQILUPHG 3DVKOHU DQG -RKQVWRQ FRQFHGH WKDW WKH 353 SDUDGLJP KDV EHHQ VWXGLHG VRPHZKDW OHVV RI ODWH SHUKDSV UHIOHFWLQJ .DKQHPDQfV HDUO\ FRQFHUQV

PAGE 16

$ PDMRU ILQGLQJ WKDW KDV IUHTXHQWO\ EHHQ REVHUYHG ZLWK WKH RYHUODSSLQJ WDVNV SDUDGLJP LV D VORZLQJ RI WKH 5OUHODWLYH WR VLQJOHWDVN SHUIRUPDQFH .DKQHPDQ f 6LPLODU ILQGLQJV LQ GLVFUHWH PRYHPHQW UHVHDUFK DUH WKRXJKW WR LQGLFDWH WKH LQFUHDVHG HIIRUW QHHGHG WR FRPSLOH D PRUH FRPSOH[ PRYHPHQW +HQU\ t 5RJHUV f 5 VORZLQJ LV UHDGLO\ DFFRXQWHG IRU ZLWK WKH FDSDFLW\ PRGHOV EHFDXVH ERWK WDVNV DUH DVVXPHG WR EH SHUIRUPHG ZLWK GHSOHWHG DOORFDWLRQV RI FDSDFLW\ 3RVWSRQHPHQW PRGHOV LQ FRQWUDVW GR QRW SUHGLFW WKLV VORZLQJ GLUHFWO\ ,W KDV EHHQ VXJJHVWHG WKDW 5 VORZLQJ PD\ UHVXOW IURP D JURXSLQJ VWUDWHJ\ LQ ZKLFK WKH VXEMHFW HVVHQWLDOO\ WUHDWV 6, DQG 6 DV D FRPSRXQG VWLPXOXV DQG VHOHFWV D FRUUHVSRQGLQJ FRPSRXQG UHVSRQVH 3DVKOHU t -RKQVWRQ f 2WKHU VWUDWHJLHV KDYH EHHQ SURSRVHG DV ZHOO EXW LQ DQ\ FDVH 5 VORZLQJ E\ LWVHOI LV QRW WKRXJKW WR EH HVSHFLDOO\ GLDJQRVWLF RI WKH XQGHUO\LQJ FDXVHV RI GXDOWDVN LQWHUIHUHQFH %HFDXVH ERWK FDSDFLW\ DQG VWUXFWXUDO PRGHOV FDQ SURYLGH D SODXVLEOH DFFRXQW RI WKH EDVLF UHVXOWV PRUH DQDO\WLF WHVWV WKDW PDNH GLVWLQFWLYH QRQREYLRXV SUHGLFWLRQV DUH QHFHVVDU\ 3DVKOHU f XVHG D FKURQRPHWULF DSSURDFK WR GHYHORS D PHWKRG IRU WHVWLQJ SRVWSRQHPHQW PRGHOV LQVSLUHG E\ DGGLWLYH IDFWRUV ORJLF 6WHUQEHUJ f %\ PDQLSXODWLQJ WDVN IDFWRUV WKDW LQFUHDVHG RU GHFUHDVHG WKH GXUDWLRQ RI VHOHFWHG VWDJHV RI SURFHVVLQJ DQG ORRNLQJ IRU HIIHFWV WKDW ZHUH DGGLWLYH RU XQGHUDGGLWLYHf 3DVKOHU ZDV DEOH WR WHVW K\SRWKHVHV DERXW WKH ORFXV RI WKH VLQJHFKDQQHO ERWWOHQHFN 3DVKOHU UHSRUWHG D SUHOLPLQDU\ VWXG\ XVLQJ WKLV PHWKRGRORJ\ WKDW IRXQG XQGHUDGGLWLYH HIIHFWV WKDW ZHUH FRQVLVWHQW ZLWK WKH SUHGLFWLRQV IURP D SRVWSRQHPHQW PRGHO DQG FRXOG QRW EH H[SODLQHG ZLWK FDSDFLW\ WKHRU\ 7KLV VWXG\ VXSSRUWHG WKH ODWHVHOHFWLRQ FODVV RI WKHRULHV VXJJHVWLQJ WKDW D ERWWOHQHFN H[LVWV HLWKHU LQ WKH GHFLVLRQ RU UHVSRQVH VHOHFWLRQ VWDJHV RU LQ ERWK VWDJHV

PAGE 17

3RVWSRQHPHQW PRGHOV SUHGLFW WKDW HIIHFWV DW VWDJHV SULRU WR WKH ERWWOHQHFN VKRXOG EHFRPH SURJUHVVLYHO\ VPDOOHU DW VKRUWHU 62$V $V 62$ VKRUWHQV WKH ZDLW IRU WKH SURFHVVRU LV WKRXJKW WR EHFRPH SURJUHVVLYHO\ ORQJHU DQG WKH SURSRUWLRQ RI WULDOV RQ ZKLFK GHOD\ LQ HDUO\ VWLPXOXV SURFHVVLQJ RI 6 ZLOO KDYH DQ\ HIIHFW RQ 57 SURJUHVVLYHO\ GHFUHDVHV $OWKRXJK 3DVKOHU f RQO\ XVHG RQH 62$ DQG FRXOG QRW WHVW WKLV SUHGLFWLRQ 3DVKOHU DQG -RKQVWRQfV ILQGLQJV f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f PRGHO RI ZRUNLQJ PHPRU\ WKDW ZLOO EH GLVFXVVHG LQ GHWDLO ODWHU %DGGHOH\ DQG +LWFK DIWHU DOO ZHUH WUDLQHG LQ %URDGEHQWfV ODE ZKHUH PXFK RI WKH HDUO\ VWUXFWXUDO WKHRUL]LQJ DQG WHVWLQJ WRRN SODFH 0XOWLSOH 5HVRXUFH 7KHRULHV $ VHDUFK IRU D FRPSURPLVH WR WKH VWUXFWXUHFDSDFLW\ GHEDWH KDV OHG GLUHFWO\ WR PXOWLSOH UHVRXUFH PRGHOV 1DYRQ t *RSKHU f ZKHUH UHVRXUFHV DUH GHFODUHG WR UHVLGH ZLWKLQ VWUXFWXUHV RU SRROV 8QWLO PXOWLSOH UHVRXUFH PRGHOV HPHUJHG LQWHUIHUHQFH ZDV WKRXJKW WR GHSHQG H[FOXVLYHO\ RQ HLWKHU WKH H[WHQW WKDW WZR WDVNV GUDZ RQ D FRPPRQ UHVRXUFH SRRO RU VWUXFWXUDO OLPLWDWLRQV ZKHUH VRPH VWDJH RI SURFHVVLQJ FDQQRW SURFHHG LQ

PAGE 18

SDUDOOHO ZLWK DQRWKHU 1HLWKHU RI WKHVH DFFRXQWV H[SODLQV DOO RI WKH SKHQRPHQD UHODWHG WR LQWHUIHUHQFH WKRXJK 7KH FHQWUDO FDSDFLW\ QRWLRQ FDQQRW H[SODLQ ZK\ VRPH VHFRQGDU\ WDVNV LQWHUIHUH PRUH ZLWK RQH SULPDU\ WDVN WKDQ DQRWKHU HTXDOO\ GLIILFXOW SULPDU\ WDVN )RU H[DPSOH YRFDO UHVSRQVHV LQWHUIHUH PRUH WKDQ VSDWLDO UHVSRQVHV ZLWK UHFDOO RI D VHQWHQFH EXW OHVV WKDQ VSDWLDO UHVSRQVHV ZLWK UHFDOO RI D OLQH GLDJUDP %URRNV f 3XUH VWUXFWXUDO PRGHOV DUH DOVR LQDGHTXDWH EHFDXVH SURFHVVHV WKDW VKDUH VLPLODU VWDJHV RU PHFKDQLVPV PD\ LQWHUIHUH ZLWK HDFK RWKHU EXW WKH\ VHOGRP EORFN HDFK RWKHU FRPSOHWHO\ 1DYRQ t *RSKHU f )RU H[DPSOH ZKHQ 7ULHVPDQ DQG 'DYLHV f SUHVHQWHG SDUWLFLSDQWV ZLWK VWLPXOL VLPXOWDQHRXVO\ WR WKH VDPH PRGDOLW\ SHUIRUPDQFH ZDV LPSDLUHG EXW QRW HQWLUHO\ LPSHGHG 0XOWLSOH UHVRXUFH WKHRU\ SRVLWV WKDW WKHUH PD\ EH YDULRXV W\SHV RI UHVRXUFHV DQG GLIIHUHQW WDVNV PD\ UHTXLUH GLIIHUHQW W\SHV RI UHVRXUFHV LQ YDULRXV FRPSRVLWLRQV UHVRXUFHV 1RUPDQ t %REURZ 1DYRQ t *RSKHU :LFNHQV f 2Q WKH RQH KDQG WKLV DSSURDFK LV VWUXFWXUDO LQ WKH VHQVH WKDW LW LGHQWLILHV WKH VRXUFH RI LQWHUIHUHQFH DV RYHUODSSLQJ PHFKDQLVPV 2Q WKH RWKHU KDQG PXOWLSOH UHVRXUFH WKHRU\ LV D FDSDFLW\ DSSURDFK EHFDXVH LW GRHV QRW DVVXPH WKDW D PHFKDQLVP FDQ EH DFFHVVHG DQG XVHG E\ RQO\ RQH SURFHVV DW D WLPH 5DWKHU PHFKDQLVPV KDYH D FDSDFLW\ WKDW FDQ EH VKDUHG E\ VHYHUDO SURFHVVHV XQWLO WKRVH SURFHVVHV GHPDQG PRUH UHVRXUFHV WKDQ WKH FDSDFLW\ DOORZV 7KXV PXOWLSOH UHVRXUFH WKHRU\ LV D PDUULDJH RI FDSDFLW\ DQG VWUXFWXUDO PRGHOV ZKHUH VWUXFWXUH LV LPSRUWDQW EXW HDFK VWUXFWXUH LV OLPLWHG E\ LWV RZQ FDSDFLW\ :LFNHQV f SURSRVHG D PXOWLSOH UHVRXUFH PRGHO WKDW DFFRXQWV IRU D JUHDW GHDO RI WKH HPSLULFDO ILQGLQJV IURP GXDOWDVN VWXGLHV 7KLV PRGHO DVVXPHV WKDW WKH H[WHQW WR ZKLFK WZR WDVNV VKDUH FRPPRQ UHVRXUFH SRROV GHWHUPLQHV KRZ PXFK WKH\ ZLOO LQWHUIHUH LQ D GXDOWDVN SDUDGLJP ,Q SDUWLFXODU WKH UHVRXUFH DUFKLWHFWXUH RI :LFNHQVfV PRGHO FRQWDLQV

PAGE 19

WKUHH GLPHQVLRQV GHULYHG IURP WUDGLWLRQDO SV\FKRORJLFDO GLFKRWRPLHV Df %DVHG RQ WKH JHQHUDO ILQGLQJ WKDW LW LV HDVLHU WR WLPHVKDUH DQ DXGLWRU\ DQG YLVXDO WDVN WKDQ WZR DXGLWRU\ RU WZR YLVXDO WDVNV :LFNHQV SURSRVHG D VHSDUDWLRQ RI UHVRXUFH SRROV VSHFLDOL]HG IRU GHDOLQJ ZLWK GLIIHUHQW SURFHVVLQJ PRGDOLWLHV Ef 7KH SURFHVVLQJ FRGHV GLPHQVLRQ GLVWLQJXLVKHV LQIRUPDWLRQ WKDW LV VSDWLDO RU DQDORJ IURP WKDW ZKLFK LV YHUEDO RU OLQJXLVWLF Ff :LFNHQV SURSRVHG D VHSDUDWLRQ RI SURFHVVLQJ VWDJHV EDVHG RQ WKH ILQGLQJ WKDW WDVNV GHPDQGLQJ HLWKHU UHVSRQVH SURFHVVHV RU FRJQLWLYHSHUFHSWXDO SURFHVVHV ZLOO LQWHUIHUH ZLWK HDFK RWKHU WR D JUHDWHU H[WHQW WKDQ ZLOO D SHUFHSWXDO DQG D UHVSRQVH WDVN )ULHGPDQ DQG 3RLVRQ f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

PAGE 20

$OWKRXJK WKH H[LVWHQFH RI PXOWLSOH UHVRXUFH SRROV DSSHDOV WR PDQ\ UHVHDUFKHUV :LFNHQV )ULHGPDQ t 3RLVRQ 1RUPDQ t %REURZ f RWKHUV DUH VNHSWLFDO 1DYRQ f 1DYRQ DQG *RSKHU f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fV f PRGHO RI ZRUNLQJ PHPRU\ LV DQRWKHU K\EULG RI VWUXFWXUDO DQG FDSDFLW\ PRGHOV ,QVWHDG RI UHVRXUFHV ZLWKLQ VWUXFWXUHV WKLV PRGHO LPSOLHV ERWK UHVRXUFH DQG VWUXFWXUDO FRPSRQHQWV 7KH PRGHOfV K\EULG QDWXUH RIWHQ JRHV RYHUORRNHG SHUKDSV EHFDXVH ZRUNLQJ PHPRU\ KDV EHHQ SULPDULO\ FRQQHFWHG ZLWK KLJKHU OHYHO ODQJXDJH WDVNV VXFK DV VSHHFK SURGXFWLRQ YRFDEXODU\ DFTXLVLWLRQ DQG VSHHFK FRPSUHKHQVLRQ UDWKHU WKDQ WKH W\SLFDO ORZHU OHYHO FRJQLWLYH VWXGLHV RI SURFHVVLQJ OLPLWV ZKHUH WKH FDSDFLW\VWUXFWXUH GHEDWH IORXULVKHG :RUNLQJ PHPRU\ KDV UHFHQWO\ EHFRPH D SRSXODU FRQVWUXFW LQ ZULWLQJ UHVHDUFK EHFDXVH VSHHFK SURGXFWLRQ YRFDEXODU\ DFTXLVLWLRQ DQG VSHHFK FRPSUHKHQVLRQ SUHVXPDEO\ VKDUH VRPH FKDUDFWHULVWLFV ZLWK ZULWWHQ ODQJXDJH SURGXFWLRQ

PAGE 21

7KH %DGGHOH\ DQG +LWFK f PRGHO HYROYHG ZKHQ QHXURORJLFDO DQG H[SHULPHQWDO GDWD GLG QRW ILW DVVXPSWLRQV FRQFHUQLQJ WKH IXQFWLRQLQJ RI WKH XQLWDU\ VKRUWn WHUP VWRUH LQ WKH PRGDO PRGHO $WNLQVRQ t 6KLIIUHQ f ,Q WKH PRGDO PRGHO WKH VKRUWn WHUP VWRUH ZDV SRVLWLRQHG DV WKH URXWH LQIRUPDWLRQ PXVW WDNH WR JDLQ DFFHVV LQWR D ORQJn WHUP VWRUH EXW QHXURORJLFDO HYLGHQFH LQGLFDWHG WKDW SRRU SHUIRUPDQFH RQ DQ DXGLWRU\ PHPRU\ VSDQ WDVN ZDV QRW OLQNHG WR D ORQJWHUP OHDUQLQJ GHILFLW 6KDOOLFH t :DUULQJWRQ f 1RU GLG DQ LPSDLUPHQW LQ WKH VKRUWWHUP VWRUH LQWHUIHUH ZLWK FRPSUHKHQVLRQ DQG SURGXFWLRQ RI VSHHFK 9DOODU t %DGGHOH\ f 6HFRQG WKH VKRUWWHUP VWRUH ZDV FRQVLGHUHG WR SOD\ D PDMRU UROH LQ UHWULHYDO IURP WKH ORQJWHUP VWRUH +RZHYHU D GLJLW VSDQ WDVN SHUIRUPHG FRQFXUUHQWO\ ZLWK WKH UHWULHYDO SKDVH RI D IUHH UHFDOO WDVN GLG QRW GHSUHVV DFFXUDF\ RI UHWULHYDO %DGGHOH\ HW DO Df 1HLWKHU GLG UHKHDUVDO RI D GLJLW QXPEHU LQIOXHQFH UHFDOO RI SDLUHG DVVRFLDWHV %DGGHOH\ 7KRPVRQ t %XFKDQDQ f 7KLUG D NH\ SRVWXODWH RI WKH PRGDO PRGHO ZDV WKDW WKH SUREDELOLW\ RI LQIRUPDWLRQ EHLQJ WUDQVIHUUHG WR ORQJWHUP VWRUH ZDV KHLJKWHQHG E\ LQFUHDVHG UHKHDUVDO
PAGE 22

FDSDFLW\ FHQWUDO H[HFXWLYH DQG WZR SHULSKHUDO VODYH V\VWHPV 2QH RI WKHVH VODYH V\VWHPV KDQGOHV YHUEDO DQG DXGLWRU\ LQIRUPDWLRQ WKH SKRQRORJLFDO ORRSf 7KH RWKHU LV VSHFLDOL]HG IRU YLVXDO DQG VSDWLDO LQIRUPDWLRQ WKH YLVXRVSDWLDO VFUDWFKSDGf 7KHVH VODYH V\VWHPV GR QRW UHSUHVHQW SRROV RI OLPLWHG UHVRXUFHV WKDW FDQ DFW LQ SDUDOOHO OLNH WKH FHQWUDO H[HFXWLYH ,QVWHDG WKH\ RSHUDWH VHULDOO\ LQ WKH VDPH VHQVH WKDW D FRPSXWHUfV SULQWHU PD\ UHFHLYH PXOWLSOH SULQW MREV EXW FDQ RQO\ SHUIRUP RQH SULQW MRE DW D WLPH 7KH SKRQRORJLFDO ORRS LV DVVXPHG WR EH PRVW LPSRUWDQW LQ WKH SURGXFWLRQ FRPSUHKHQVLRQ DQG GHYHORSPHQW RI ODQJXDJH $UFKLWHFWXUDOO\ LW KDV WZR VXEFRPSRQHQWV Df D SKRQRORJLFDO VWRUH WKDW LV EHOLHYHG WR KROG SKRQRORJLFDO LQIRUPDWLRQ IRU DSSUR[LPDWHO\ VHFRQGV DQG Ef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t %XFKDQDQ f 7KLV LV DVVXPHG WR RFFXU EHFDXVH UHKHDUVDO WDNHV ORQJHU IRU ORQJHU ZRUGV WKDQ IRU VKRUW ZRUGV DOORZLQJ PRUH GHFD\ WR RFFXU EHIRUH WKH QH[W UHKHDUVDO F\FOH ,W LV W\SLFDOO\ IRXQG WKDW SHRSOH UHPHPEHU DV PDQ\ ZRUGV DV WKH\ FDQ UHDG LQ VHFRQGV %DGGHOH\ 7KRPVRQ t %XFKDQDQ f 6HFRQG SHUIRUPDQFH LV GLVUXSWHG ZKHQ SDUWLFLSDQWV YRFDOL]H VRPH SUHGHILQHG SDWWHUQ RI VSHHFK HJ WHH WDK WHH WDK f ZKLOH WKH\ VLPXOWDQHRXVO\ SHUIRUP VRPH SULPDU\ WDVN 7KLV DUWLFXODWRU\ VXSSUHVVLRQ LV FODLPHG WR HQJDJH WKH SKRQRORJLFDO VWRUH DQG EORFN LWV DELOLW\ WR SDUWLFLSDWH LQ DFFRPSOLVKLQJ WKH SULPDL\ WDVN 'LVUXSWLRQV

PAGE 23

KDYH EHHQ REVHUYHG LQ D YDULHW\ RI WDVNV LQFOXGLQJ YRFDEXODU\ DFTXLVLWLRQ DQG VHULDO OHDUQLQJ %DGGHOH\ HW DO Ef 7KLUG SHUIRUPDQFH GHJUDGDWLRQ RFFXUV ZKHQ SDUWLFLSDQWV DWWHPSW WR SHUIRUP D SULPDU\ WDVN ZKLOH LUUHOHYDQW VSHHFK LV SUHVHQWHG WKDW WKH\ DUH LQVWUXFWHG WR LJQRUH 7KLV SURFHGXUH KDV EHHQ VKRZQ WR GLVUXSW VHULDO UHFDOO RI YLVXDOO\ SUHVHQWHG OLVWV 6DODPHf t %DGGHOH\ f )RXUWK DFRXVWLF RU SKRQRORJLFDO VLPLODULW\ HIIHFWV DUH FRPPRQ ZKHUH WKH GLVVLPLODU VRXQGLQJ LWHPV DUH UHFDOOHG EHWWHU WKDQ VLPLODU VRXQGLQJ LWHPV LV LQWHUSUHWHG DV HYLGHQFH WKDW WKH SKRQRORJLFDO VWRUH LV VSHHFK EDVHG &RQUDG t +XOO %DGGHOH\ f :RUGV RU QRQZRUGV WKDW VRXQG DOLNH LQWHUIHUH ZLWK HDFK RWKHU LQ PHPRU\ VSDQ WHVWV PRUH WKDQ VHPDQWLFDOO\UHODWHG ZRUGV WKDW GLIIHU LQ WKHLU DFRXVWLF SURSHUWLHV )LIWK QHXURSV\FKRORJLFDO SDWLHQWV IUHTXHQWO\ KDYH VSHFLILF SKRQRORJLFDO ORRS GHILFLWV EXW VXIIHU QR JHQHUDO FRJQLWLYH LPSDLUPHQW %DGGHOH\ 3DSDJQR t 9DOODU f 7KH YLVXRVSDWLDO VFUDWFKSDG KDV EHHQ OHVV UHVHDUFKHG WKDQ WKH SKRQRORJLFDO ORRS EXW WKH QRWLRQ WKDW VSDWLDO DQG YHUEDO SURFHVVHV PD\ HDFK GUDZ RQ IXQFWLRQDOO\ VHSDUDWH UHVRXUFHV LV ZHOO VXSSRUWHG :LFNHQV f :KHUH SKRQRORJLFDO VLPLODULW\ HIIHFWV KDYH EHHQ REVHUYHG LQ WDVNV EHOLHYHG WR UHTXLUH WKH SKRQRORJLFDO ORRS YLVXDO VLPLODULW\ HIIHFWV KDYH EHHQ REVHUYHG LQ YLVXRVSDWLDO WDVNV ,W LV FRPPRQ IRU VWLPXOL WKDW ORRN DOLNH WR LQWHUIHUH ZLWK HDFK RWKHU +XH DQG (ULFVVRQ f REVHUYHG WKLV SKHQRPHQRQ ZLWK &KLQHVH FKDUDFWHUV LQ D VWXG\ XVLQJ SDUWLFLSDQWV ZKR SUHVXPDEO\ KDG QR H[SHULHQFH ZLWK WKH &KLQHVH ODQJXDJH )ULFN f HODERUDWHG RQ WKH YLVXDO VLPLODULW\ HIIHFW DUJXLQJ RQ WKH EDVLV RI WKH IUHTXHQF\ RI HUURUV OLNH f3f EHLQJ PLVWDNHQ IRU f5f WKDW LPDJHV LQ YLVXR VSDWLDO ZRUNLQJ PHPRU\ DUH XQSDUVHG )XUWKHU HYLGHQFH IRU WKH H[LVWHQFH RI D VHSDUDWH YLVXRVSDWLDO DQG YHUEDO VWRUDJH V\VWHPV FRPHV IURP QHXURSV\FKRORJLFDO VWXGLHV ZKHUH

PAGE 24

SDWLHQWV ZLWK ULJKW SRVWHULRU OHVLRQV FDQ EH PDUNHGO\ LPSDLUHG RQ WHVWV RI PHPRU\ VSDQ IRU PRYHPHQWV WR GLIIHUHQW VSDWLDO ORFDWLRQV GHVSLWH KDYLQJ QRUPDO DXGLWRU\YHUEDO PHPRU\ VSDQV 'H 5HQ]L t 1LFKHOOL f $ VHSDUDWH YLVXRVSDWLDO SURFHVVLQJ V\VWHP LV DOVR FODLPHG WR EH LQYROYHG LQ SODQQLQJ DQG H[HFXWLQJ VSDWLDO WDVNV )RU H[DPSOH -DSDQHVH DEDFXV H[SHUWV FDQ SHUIRUP FRPSOH[ FDOFXODWLRQV ZLWKRXW WKH DLG RI WKH DEDFXV DQG DSSHDU WR GR VR E\ VLPXODWLQJ WKH DSSDUDWXV XVLQJ YLVXRVSDWLDO ZRUNLQJ PHPRU\ +DWDQR t 2VDZD f 5HVHDUFKHUV KDYH DOVR LPSOLFDWHG WKH YLVXRVSDWLDO VFUDWFKSDG LQ NHHSLQJ WUDFN RI FKDQJHV LQ WKH YLVXDO SHUFHSWXDO ZRUOG RYHUWLPH .DKQHPDQ 7ULHVPDQ t *LEEV f PDLQWDLQLQJ RULHQWDWLRQ LQ VSDFH DQG GLUHFWLQJ VSDWLDO PRYHPHQW 7KRPVRQ f DQG FRPSUHKHQGLQJ FHUWDLQ W\SHV RI YHUEDO LQIRUPDWLRQ 0DQLQ t -RKQVRQ/DLUG f 7KH FHQWUDO H[HFXWLYH FRPSRQHQW RI %DGGHOH\fV PRGHO LV GHVFULEHG DV D OLPLWHG FDSDFLW\ PHFKDQLVP UHVSRQVLEOH IRU FRRUGLQDWLQJ WDVNV DQG PDQDJLQJ WKH WZR VODYH V\VWHPV ,W RSHUDWHV LQ D IXQGDPHQWDOO\ GLIIHUHQW ZD\ WKDQ WKH WZR VODYH V\VWHPV EHFDXVH LW PD\ EH DFWLYHO\ LQYROYHG LQ PHHWLQJ WKH GHPDQGV RI WZR RU PRUH FRQFXUUHQW WDVNV $FFRUGLQJO\ SHUIRUPDQFH GHFOLQHV RQO\ ZKHQ WKH FROOHFWLYH WDVN GHPDQGV H[FHHG WKH FDSDFLW\ RI UHVRXUFHV %DGGHOH\ DFNQRZOHGJHV f WKDW HPEDUUDVVLQJO\ OLWWOH GLUHFW UHVHDUFK KDV IRFXVHG RQ WKH FHQWUDO H[HFXWLYH DQG WKDW LW LV VRPHZKDW RI D FDWFKDOO FRPSRQHQW UHVSRQVLEOH IRU D ZLGH UDQJH RI DWWHQWLRQDO GXWLHV 1HYHUWKHOHVV RQH VRXUFH RI HYLGHQFH GHULYHV IURP WKH VWXG\ RI $O]KHLPHUfV GLVHDVH D GLVRUGHU W\SLFDOO\ DVVRFLDWHG ZLWK GHILFLWV LQ WDVNV EHOLHYHG WR GHSHQG SULPDULO\ RQ FHQWUDO H[HFXWLYH IXQFWLRQLQJ ,Q RQH VWXG\ %DGGHOH\ f $O]KHLPHUfV SDWLHQWV ZHUH WUDLQHG WR GR D YLVXDO WDVN DQG D YHUEDO WDVN

PAGE 25

7KH\ WKHQ SHUIRUPHG ERWK WDVNV WRJHWKHU LQ D GXDOWDVN SDUDGLJP $V WKH GLVHDVH SURJUHVVHG VLQJOHWDVN SHUIRUPDQFH UHPDLQHG XQFKDQJHG EXW GXDOWDVN SHUIRUPDQFH GURSSHG PDUNHGO\ 7KLV HIIHFW ZDV LQWHUSUHWHG DV VXSSRUW IRU WKH PRGXODULW\ RI WKH FHQWUDO H[HFXWLYH UHDVRQLQJ WKDW WLPHVKDULQJ DELOLW\ LV DFFRPSOLVKHG HQWLUHO\ E\ WKH FHQWUDO H[HFXWLYH 7KH %DGGHOH\ PRGHO KDV FRPSHWLWRUV RWKHU WKDQ VLQJOH DQG PXOWLSOH UHVRXUFH PRGHOV .HOORJJ f 0DUWLQ 6KHOWRQ DQG
PAGE 26

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f DQG UHFDOOLQJ WKHP IURP ORQJn WHUP PHPRU\ 3DLYLR f FDQ LQYRNH YLVXDO LPDJHU\ *DWKHUFROH DQG %DGGHOH\ f FODLP WKDW SODQQLQJ DOVR UHTXLUHV FHQWUDO H[HFXWLYH UHVRXUFHV LQ DOO LWV IDFHWV VXFK DV JHQHUDWLQJ LGHDV WU\LQJ RXW YDULRXV RUJDQL]DWLRQ VFKHPHV RU GHEDWLQJ WKH DSSURSULDWH WRQH IRU D JLYHQ DXGLHQFH :KHQ IRUPXODWLQJ ZULWHUV FRYHUWO\ WDON WR WKHPVHOYHV LQ WKH IRUP RI LQQHU VSHHFK RU SUHWH[W DV WKH\ DUH JHQHUDWLQJ VHQWHQFHV %DGGHOH\ t /HZLV f WKXV DFWLYDWLQJ WKH SKRQRORJLFDO ORRS ,Q SDUWLFXODU SKRQRORJLFDO UHSUHVHQWDWLRQV RI WKH ZRUGV DQG VHQWHQFHV DUH VWRUHG LQ WKH SKRQRORJLFDO ORRS 7R FRQQHFW WKH SKRQRORJLFDO ORRS ZLWK IRUPXODWLRQ WDVNV .HOORJJ QRWHV FDVHV RI DSUD[LF DQG G\VSUD[LF SDWLHQWV ZKR PDNH SKRQHPLF DQG RWKHU OLQJXLVWLF PLVWDNHV LQ VSHHFK DQG ZKR DOVR DUH LPSDLUHG LQ

PAGE 27

PHPRU\ VSDQ WHVWV DQG IDLO WR VKRZ SKRQRORJLFDO VLPLODULW\ DQG ZRUG OHQJWK HIIHFWV :DWHUV 5RFKRQ t &DSODQ f 7UDQVODWLQJ XQOLNH FRQYHUVDWLRQDO VSHHFK SURGXFWLRQ %RFN f GHPDQGV FHQWUDO H[HFXWLYH UHVRXUFHV ZKHQ WKH ZULWHU VWUXJJOHV WR ILQG MXVW WKH ULJKW ZRUGV DQG VHQWHQFH VWUXFWXUHV /RQJ SDXVHV DQG KLJK GHJUHHV RI H[SHQGHG FRJQLWLYH HIIRUW VXJJHVW WKH LQYROYHPHQW RI WKH FHQWUDO H[HFXWLYH LQ WKHVH FDVHV .HOORJJ f DV ZHOO DV WKH SRVLWLYH FRUUHODWLRQ RI PHPRU\ VSDQ ZLWK WKH DELOLW\ WR VHOHFW OH[LFDO LWHPV IRU XVH LQ D VHQWHQFH 'DQHPDQ t *UHHQH f .HOORJJ DGGV WKDW WKH LQYROYHPHQW RI WKH FHQWUDO H[HFXWLYH LQ WUDQVODWLRQ PD\ YDU\ ZLWK WKH GHPDQGV RI WKH WDVN DW KDQG $ ZULWHU FRPSRVLQJ SROLVKHG ILQDO GUDIW SURVH RQ WKH ILUVW DWWHPSW PD\ UHTXLUH FHQWUDO H[HFXWLYH UHVRXUFHV WR D JUHDWHU H[WHQW WKDQ D ZULWHU FRPSRVLQJ D PRUH FRQYHUVDWLRQDO HPDLO WR D IULHQG 7\SLQJ DQG KDQGZULWLQJ H[HFXWLRQf DUH EHOLHYHG WR PDNH PLQLPDO GHPDQGV RQ WKH FHQWUDO H[HFXWLYH ZKHQ KLJKO\ SUDFWLFHG ,Q \RXQJ FKLOGUHQ KRZHYHU %RXUGLQ t )D\RO f IRXQG WKDW KDQGZULWLQJ GHPDQGV PRUH FDSDFLW\ WKDQ VSHDNLQJ FRQVLVWHQW ZLWK WKH QRWLRQ WKDW QRYHO DFWLYLWLHV RI DOO NLQGV UHTXLUH WKH FHQWUDO H[HFXWLYH WR FRQWURO WKH VFKHPDV XVHG LQ PRWRU RXWSXW .HOORJJ f *DWKHUFROH DQG %DGGHOH\ f FODLP WKDW UHDGLQJ RQH RI WKH EDVLF SURFHVVHV RI PRQLWRULQJ UHTXLUHV WKH UHVRXUFHV RI ERWK WKH SKRQRORJLFDO ORRS DQG WKH FHQWUDO H[HFXWLYH .HOORJJ VXJJHVWV WKDW WKH PRVW VLJQLILFDQW GHPDQG RI PRQLWRULQJ VWHPV IURP HGLWLQJ QRW UHDGLQJ EHFDXVH HGLWLQJ WDNHV VR PDQ\ IRUPV UDQJLQJ IURP WKH GHWHFWLRQ RI D PRWRU SURJUDPPLQJ HIIRUW WR D UHYLVLRQ LQ WKH RUJDQL]DWLRQ RI LGHDV LQ D WH[W .HOORJJ f

PAGE 28

5HOHYDQW :RUNLQJ 0HPRU\ DQG :ULWLQJ 5HVHDUFK 2ZLQJ LQ SDUW WR LWV QHZQHVV OLWWOH UHVHDUFK KDV EHHQ UHSRUWHG WKDW H[SOLFLWO\ WHVWV WKH LPSOLFDWLRQV RI .HOORJJfV PRGHO 2QH SUHGLFWLRQ RI WKH .HOORJJ PRGHO VXJJHVWV WKDW WKH SKRQRORJLFDO ORRS SOD\V D VWURQJ UROH LQ PRQLWRULQJ -RQHV 0LOHV DQG 3DJH f LQWURGXFHG DQ LUUHOHYDQW VSHHFK PDQLSXODWLRQ WR H[SORUH SKRQRORJLFDO ORRS LQYROYHPHQW LQ SURRIUHDGLQJ $OWKRXJK SURRIUHDGLQJ LV FOHDUO\ QRW WKH VDPH DV HGLWLQJ RQHfV RZQ ZULWLQJ LW GRHV LQYROYH VRPH DVSHFWV RI WKH PRQLWRULQJ SURFHVV DV GHVFULEHG E\ .HOORJJ f -RQHV HW DO FRPSDUHG SDUWLFLSDQWVf DELOLW\ WR GHWHFW WZR W\SHV RI HUURUV FRQWH[WXDO JUDPPDWLFDO DQG ZRUG FKRLFHf DQG QRQFRQWH[WXDO VSHOOLQJ DQG W\SRJUDSKLFDO HUURUVf ZKLOH WKH\ KHDUG HLWKHU PHDQLQJIXO RU UHYHUVHG VSHHFK 7KH LUUHOHYDQW VSHHFK FRQVLVWHG RI D WDSHG OHFWXUH HLWKHU SOD\HG IRUZDUG PHDQLQJIXO VSHHFKf RU EDFNZDUG UHYHUVHG VSHHFKf 7KH GHWHFWLRQ UDWH IRU QRQFRQWH[WXDO HUURUV ZDV GLVUXSWHG LQ WKH PHDQLQJIXO VSHHFK FRQGLWLRQ EXW WKLV HIIHFW GLG QRW H[WHQG WR FRQWH[WXDO HUURUV 1R HIIHFWV ZHUH REVHUYHG IRU WKH UHYHUVHG LUUHOHYDQW VSHHFK PDQLSXODWLRQ QRU ZHUH WKHUH GLIIHUHQFHV EHWZHHQ UHYHUVHG LUUHOHYDQW VSHHFK DQG VLOHQFH LQ D VHFRQG H[SHULPHQW OHQGLQJ VXSSRUW WR WKH LGHD WKDW D VHPDQWLF FRPSRQHQW LV LPSRUWDQW LQ WKH LUUHOHYDQW VSHHFK HIIHFW 7KHVH UHVXOWV DUH FRQVLVWHQW ZLWK .HOORJJfV f SUHGLFWLRQV WKDW WKH SKRQRORJLFDO ORRS LV LQYROYHG LQ WKH GHWHFWLRQ RI VSHOOLQJ DQG W\SRJUDSKLFDO HUURUV EXW LQFRQVLVWHQW ZLWK WKH SUHGLFWLRQ WKDW LUUHOHYDQW VSHHFK ZRXOG KLQGHU WKH GHWHFWLRQ RI JUDPPDWLFDO DQG ZRUG FKRLFH HUURUV ,QYHVWLJDWLRQV RI LUUHOHYDQW VSHHFK HIIHFWV RQ UHDGLQJ FRPSUHKHQVLRQ DOVR IDLOHG WR VXSSRUW WKH UROH RI WKH SKRQRORJLFDO ORRS DV D WHPSRUDU\ VWRUH IRU VHQWHQFH UHSUHVHQWDWLRQV 0DUWLQ :RJDOWHU t )RUODQRQ f ,Q WKH LQLWLDO LQYHVWLJDWLRQ FRPSUHKHQVLRQ VFRUHV GHFOLQHG LQ WZR FRQGLWLRQV LQYROYLQJ LUUHOHYDQW VSHHFK FRPSDUHG WR

PAGE 29

D VLOHQW FRQWURO 7KH LUUHOHYDQW VSHHFK FRQVLVWHG RI HLWKHU ZRUGV LQ D PHDQLQJIXO RUGHU RU WKH VDPH ZRUGV UDQGRPO\ VFUDPEOHG &RPSUHKHQVLRQ LQ WKH WZR VSHHFK FRQGLWLRQV ZDV QRW UHOLDEO\ GLIIHUHQW ,Q D VHFRQG H[SHULPHQW 0DUWLQ HW DO f PDQLSXODWHG LQVWUXPHQWDWLRQ SUHVHQW RU QRWf DQG O\ULFV VXQJ VSRNHQ RU QRQHf ZLWKLQ VXEMHFWV )LQGLQJV LQGLFDWHG WKDW WKH SUHVHQFH RI YHUEDO PDWHULDO DGYHUVHO\ DIIHFWHG FRPSUHKHQVLRQ ZKLOH LQVWUXPHQWDWLRQ GLG QRW 0DGLJDQ DQG /LQWRQ f DVNHG SDUWLFLSDQWV WR FRPSRVH WZR HVVD\V RQ D ZRUG SURFHVVRU RQH ZKLOH KHDULQJ LUUHOHYDQW VSHHFK DQG WKH RWKHU LQ VLOHQFH 3ULRU WR WKH ZULWLQJ VHVVLRQ SDUWLFLSDQWV FRPSOHWHG D UHDGLQJ VSDQ WDVN GHVLJQHG E\ 'DQHPDQ DQG &DUSHQWHU f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f DPRQJ KLJK VSDQ ZULWHUV :ULWLQJ IOXHQF\ DV PHDVXUHG E\ ZRUG SURGXFWLRQ WLPHV DQG SDXVHV ZLWKLQ VHOHFWHG FODXVHV ZHUH XQDIIHFWHG E\ WKH LUUHOHYDQW VSHHFK PDQLSXODWLRQ 7KH LQYHVWLJDWRUV DUJXH WKDW WKH UHGXFWLRQ LQ KROLVWLF TXDOLW\ LV LQFRQVLVWHQW ZLWK DVVXPSWLRQV RI WKH .HOORJJ PRGHO UHODWHG WR WKH UHODWLYHO\ PLQRU LQIOXHQFH RI WKH SKRQRORJLFDO ORRS ZKHQ FRPSDUHG WR WKH FHQWUDO H[HFXWLYH RQ ZULWLQJ TXDOLW\ .HOORJJ

PAGE 30

f KRZHYHU GLVDJUHHV FRQWHQGLQJ WKDW fWKH VPDOO HIIHFW VL]H LV FRQVLVWHQW ZLWK WKH K\SRWKHVLV WKDW LUUHOHYDQW VSHHFK GLVUXSWV RQO\ WUDQVODWLQJ DQG UHDGLQJf S f 5DQVGHOO /HY\ DQG .HOORJJ f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f\HVf RU fQRf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

PAGE 31

)LUVW WKH\ FRQVLGHUHG WKH SRVVLELOLW\ WKDW WKH SKRQRORJLFDO ORRS LV DFWLYDWHG RQO\ IRU FRPSOH[ VHQWHQFHV 6HFRQG WKH\ UDLVHG WKH SRVVLELOLW\ WKDW WKH DUWLFXODWRU\ ORRS ZDV PRUH LQWLPDWHO\ LQYROYHG LQ WKH ZULWLQJ WDVN WKDQ WKH SKRQRORJLFDO ORRS DQG WKDW LUUHOHYDQW VSHHFK GHULYHG LWV LQIOXHQFH YLD FRUUXSWLQJ PDWHULDO LQ WKH VWRUH ,Q WKH VHFRQG H[SHULPHQW PHPRU\ ORDG DOVR UHVXOWHG LQ D UHOLDEOH GHFUHPHQW LQ IOXHQF\ DQG LQ VHQWHQFH OHQJWK 2QO\ WKH IOXHQF\ RI ORZ VSDQ ZULWHUV ZDV DGYHUVHO\ DIIHFWHG E\ WKH FRQFXUUHQW WDVN &RQFXUUHQW ORDG DOVR DOWHUHG WKH WHPSRUDO SDWWHUQV RI FRPSRVLQJ :KHQ UHPHPEHULQJ DQG UHFDOOLQJ GLJLWV SDUWLFLSDQWV SDXVHG ORQJHU DQG PRUH RIWHQ +LJK VSDQ ZULWHUV DOVR SDXVHG OHVV IUHTXHQWO\ DW FODXVH ERXQGDULHV )XUWKHUPRUH XQOLNH H[SRVXUH WR LUUHOHYDQW VSHHFK WKH FRQFXUUHQW WDVN DOVR UHOLDEO\ UHGXFHG HVVD\ TXDOLW\ ,Q WKH WKLUG H[SHULPHQW DWWHQGLQJ DQG UHVSRQGLQJ WR LUUHOHYDQW VSHHFK OHG WR D UHOLDEOH GHFUHPHQW LQ ZULWLQJ IOXHQF\ VHQWHQFH OHQJWK DQG SDXVH IUHTXHQF\ FRPSDUHG WR EDVHOLQH EXW SDXVH OHQJWK DQG ORFDWLRQ ZHUH VLPLODU LQ ERWK FRQGLWLRQV $GGLWLRQDOO\ RYHUDOO ZULWLQJ TXDOLW\ GLG QRW GHFOLQH LQ WKH DWWHQGHG VSHHFK FRQGLWLRQ DOWKRXJK VFRUHV RI KLJK VSDQ ZULWHUV ZLWK fJRRG VHQWHQFHf LQVWUXFWLRQV ZHUH DGYHUVHO\ DIIHFWHG LQ WKH RUJDQL]DWLRQ DQG GHYHORSPHQW VXEJURXS 7KH W\SH RI TXHVWLRQ DQVZHUHG LQ WKH DWWHQGHG VSHHFK FRQGLWLRQ SKRQRORJLFDO VSDWLDO RU VHPDQWLFf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

PAGE 32

RQO\ WDVN WKDW OHQJWKHQHG SDXVH GXUDWLRQ DQG UHGXFHG WKH SURSRUWLRQ RI SDXVHV DW FODXVH ERXQGDULHV +HUH WKHUH LV D WUDGHRII EHWZHHQ TXDOLW\ DQG IOXHQF\ WKDW LQYROYHV DQ LQFUHDVH LQ WKH QXPEHU RI SDXVHV UDWKHU WKDQ LQ WKHLU GXUDWLRQ 7R WHVW .HOORJJfV f SUHGLFWLRQV RI WKH SKRQRORJLFDO ORRS LQYROYHPHQW LQ WUDQVODWLQJ LGHDV LQWR VHQWHQFHV IRUPXODWLRQf DQG UHDGLQJ WH[W PRQLWRULQJf 0DUHN DQG /HY\ f GHVLJQHG D VHULHV RI FRPSXWHUEDVHG WDVNV WR VLPXODWH WKHVH SULPDU\ ZULWLQJ SURFHVVHV 7KHVH ZULWLQJ VXEWDVNV LQFOXGHG FUHDWLQJ VHQWHQFHV IURP JURXSV RI ZRUGV IRUPXODWLRQf FRS\LQJ DQ H[LVWLQJ WH[W H[HFXWLRQf DQG SURRIUHDGLQJ PRQLWRULQJf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fV ZRUN DQG UHYLHZLQJ RQHfV RZQ ZULWLQJ WKH WDVN ZDV QRW GHVLJQHG WR HQDEOH UHFRUGLQJ RI WKH DPRXQW RI WLPH VSHQW UHDGLQJ YHUVXV WKH DPRXQW RI WLPH VSHQW HGLWLQJ 1RU GLG WKH UHVHDUFKHUV UHFRUG RU FRQWURO KRZ PXFK RI WKH VHOHFWLRQ ZDV DFWXDOO\ UHDG

PAGE 33

7KH GHWHFWLRQ UDWH IRU HUURUV UDQJHG IURP b WR b EXW ZKHWKHU WKLV SULPDULO\ LOOXVWUDWHV GLIIHULQJ DELOLW\ WR LGHQWLI\ WKH PLVWDNHV RU GLIIHUHQFHV LQ UHDGLQJ IOXHQF\ UHPDLQ XQNQRZQ $QRWKHU SUHGLFWLRQ RI WKH .HOORJJ PRGHO VXJJHVWV WKDW ORDGLQJ WKH FHQWUDO H[HFXWLYH VKRXOG GHWUDFW IURP ZULWLQJ IOXHQF\ DQG TXDOLW\ WKH PRVW EHFDXVH WKH FHQWUDO H[HFXWLYH SOD\V D SDUW LQ DOO WKUHH ZULWLQJ SURFHVVHV -HIIHU\ DQG 8QGHUZRRG f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t /HY\ f KDYH VXJJHVWHG WKDW FRQVWUXFWLQJ VHQWHQFHV

PAGE 34

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f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

PAGE 35

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t *RSKHU f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fV WKHRUHWLFDO DVVXPSWLRQV DULVHV ZKHQ FRPSDULQJ HIIHFWV DFURVV VWXGLHV 8QGHU WKH DVVXPSWLRQ RI VWUXFWXUDO VODYH V\VWHPV GLIIHUHQFHV LQ VHFRQGDU\ WDVN VWUXFWXUH LQFOXGLQJ FKDUDFWHULVWLFV VXFK DV WKH IUHTXHQF\ RI UHVSRQVH VHOHFWLRQ DQG H[HFXWLRQ DV ZHOO DV LQSXW DQG RXWSXW PRGDOLWLHVf ZLOO QRW DIIHFW WKH H[WHQW WR ZKLFK WKDW WDVN HQJDJHV LWV WDUJHWHG FRPSRQHQW 5HVRXUFH WKHRULVWV DUJXH WKDW MXVW DV WDVN GLIILFXOW\ LV OLNHO\ WR LQIOXHQFH SULPDU\ DQG VHFRQGDU\ WDVN LQWHUIHUHQFH WKH IUHTXHQF\ RI

PAGE 36

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fV f PXOWLSOH UHVRXUFH PRGHO 7KH FXUUHQW PRGHO GLIIHUV IURP )ULHGPDQ DQG 3RLVRQ f LQ LWV LQFOXVLRQ RI %DGGHOH\f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fV WHPSRUDU\ VWRUDJH DQG SURFHVVLQJ FDSDFLW\ ,QYHVWLJDWLRQV RI ZULWLQJ DW WKLV OHYHO DUH IDLUO\ QHZ WKXV REOLJDWLQJ ZULWLQJ UHVHDUFKHUV WR WXUQ WR H[LVWLQJ WKHRULHV RI FRJQLWLYH OLPLWDWLRQV IRU D IUDPHZRUN LQ ZKLFK WKHVH GHPDQGV PD\ EH DVVHVVHG

PAGE 37

7KHRU\ GHYHORSPHQW VXUURXQGLQJ SURFHVVLQJ OLPLWDWLRQV KDV W\SLFDOO\ HPSOR\HG WZR K\SRWKHWLFDO FRQVWUXFWV UHVRXUFHV DQG VWUXFWXUHV (DUO\ DFFRXQWV RI SURFHVVLQJ GHPDQGV ZHUH H[SODLQHG E\ HLWKHU UHVRXUFH .DKQHPDQ f RU VWUXFWXUDO %URDGEHQW f PRGHOV EXW PRUH UHFHQWO\ HODERUDWH PRGHOV KDYH HYROYHG WKDW HPERG\ ERWK UHVRXUFH DQG VWUXFWXUDO DVSHFWV 7KH WZR PRVW SUHYDOHQW RI VXFK PRGHOV DUH PXOWLSOH UHVRXUFH PRGHOV 1RUPDQ t %REURZ :LFNHQV )ULHGPDQ t 3RLVRQ f DQG %DGGHOH\fV ZRUNLQJ PHPRU\ PRGHO %DGGHOH\ t +LWFK f %RWK SURSRVH GLYLVLRQV LQ YLVXRVSDWLDO DQG YHUEDO SURFHVVLQJ JHQHUDOO\ LQ DOLJQPHQW ZLWK WKH KHPLVSKHULF DV\PPHWU\ OLWHUDWXUH .HOORJJ f SURSRVHG D PRGHO RI ZULWLQJ WKDW GHWDLOV KRZ VSHFLILF ZULWLQJ SURFHVVHV UHO\ RQ WKH PDLQ FRPSRQHQWV RI %DGGHOH\fV ZRUNLQJ PHPRU\ PRGHO %HIRUH WHVWLQJ WKH VSHFLILF FODLPV RI .HOORJJfV PRGHO D FORVHU H[DPLQDWLRQ RI ZRUNLQJ PHPRU\ DV D PRGHO RI FRJQLWLYH OLPLWDWLRQV LV QHFHVVDU\ ,Q SDUWLFXODU WKH DVVXPSWLRQV RI WKH VWUXFWXUDO QDWXUH RI WKH SKRQRORJLFDO ORRS DQG YLVXRVSDWLDO VFUDWFKSDG VHHP RXWGDWHG 7KLV GLVVHUWDWLRQ SURSRVHV DQ DGDSWDWLRQ WR WKH ZRUNLQJ PHPRU\ PRGHO ZKHUHE\ WKH SKRQRORJLFDO ORRS DQG YLVXRVSDWLDO VFUDWFKSDG RSHUDWH DV SRROV RI PXWXDOO\ LQDFFHVVLEOH UHVRXUFHV IRU YHUEDO DQG YLVXRVSDWLDO SURFHVVLQJ DQG VWRUDJH VLPLODU WR )ULHGPDQ DQG 3RLVRQfV PXOWLSOH UHVRXUFH PRGHO f 7KLV K\EULG PRGHO RI ZRUNLQJ PHPRU\ DQG PXOWLSOH UHVRXUFHV LV WHVWHG LQ ([SHULPHQW ([SHULPHQWV D DQG E FRQVLGHU WKH GLYLVLRQ RI ODERU EHWZHHQ YLVXRVSDWLDO DQG YHUEDO SURFHVVLQJ DQG VWRUDJH FRPSRQHQWV DQG ([SHULPHQW DGGUHVVHV WKH UROH RI ZULWLQJ WRSLF LQ WKH UHVRXUFH GHPDQGV RI ZULWLQJ

PAGE 38

&+$37(5 (;3(5,0(17 ,I WKH SKRQRORJLFDO ORRS DQG YLVXRVSDWLDO VFUDWFKSDG RI %DGGHOH\f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fV ZRUNLQJ PHPRU\ PRGHO SUHGLFWV QR GLIIHUHQFHV LQ SHUIRUPDQFH UHODWHG WR W\SH RI LUUHOHYDQW VSHHFK EHFDXVH VSHHFK VRXQGV UHJDUGOHVV RI PHDQLQJ DUH EHOLHYHG WR KDYH REOLJDWRU\ DFFHVV WR WKH SKRQRORJLFDO VWRUH 7KRVH VRXQGV DUH EHOLHYHG WR HQJDJH WKH SKRQRORJLFDO ORRS HQWLUHO\ GXH WR LWV VWUXFWXUDO QDWXUH ,QFRQVLVWHQW ZLWK WKHVH SUHGLFWLRQV -RQHV 0LOHV DQG 3DJH f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

PAGE 39

UHDOO\ fVSHHFKf )XUWKHU /HY\ DQG 0DUHN f UHSRUW VLPLODU SDWWHUQV RI UHVXOW LQ D ZULWLQJ VWXG\ ZKHQ FRPSXWHUJHQHUDWHG VSHHFK ZDV fQDWXUDOf DQG ZKHQ WKH LQGLYLGXDO ZRUGV LQ HDFK VHQWHQFH ZHUH UDQGRPO\ UHDUUDQJHG EXW QR FKDQJHV PDGH WR WKHLU SKRQHPLF DWWULEXWHV $ UHFHQW VWXG\ XVLQJ PDJQHWLF UHVRQDQFH LPDJLQJ UHYHDOHG WKDW LUUHOHYDQW VSHHFK SOD\HG EDFNZDUG LV SURFHVVHG LQ GLIIHUHQW DUHDV RI WKH EUDLQ WKDQ LV IRUZDUG VSHHFK UDLVLQJ TXHVWLRQV ZKHWKHU WKH SKRQRORJLFDO ORRS DFWV DORQH LQ SURFHVVLQJ VSHHFK %DGGHOH\f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f WDVN SHUIRUPDQFH RQ WKH ODWWHU VKRXOG GHFOLQH EXW WKH SHUIRUPDQFH GHFUHPHQWV VKRXOG EH HTXDO IRU WKH WZR W\SHV RI LUUHOHYDQW VSHHFK 7KLV ZRXOG EH VR HYHQ LI WKH PDWHULDOV GHOLYHUHG LQ D FRPSXWHUV\QWKHVL]HG VSHHFK JHQHUDWHG D OHVVHU SURFHVVLQJ ORDG EHFDXVH LW ODFNHG QXDQFHV KDG UHGXFHG LQIOHFWLRQV DQG VR IRUWKf WKDQ WKH VSHHFK SURGXFHG E\ D KXPDQ YRLFH ,Q FRQWUDVW RXU PRGLILHG PRGHO ZKLFK

PAGE 40

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fV IDEOH UHDG E\ WKH H[SHULPHQWHU 7KH RWKHU IHDWXUHG WKH VDPH VWRU\ JHQHUDWHG YLD FRPSXWHU UXQQLQJ D WH[WWRVSHHFK SURJUDP 7KH FRQWHQW DQG DFRXVWLF HQHUJ\ OHYHO RI WKH WZR WDSHG SUHVHQWDWLRQV ZDV WKH VDPH 3URFHGXUH 7KH WREHUHPHPEHUHG FRQVRQDQW VHWV ZHUH VKRZQ WR WKH SDUWLFLSDQWV IRU VHFRQGV HDFK 3DUWLFLSDQWV WKHQ KDG VHFRQGV WR UHFDOO WKH FRQVRQDQWV 7KLV VHTXHQFH ZDV UHSHDWHG XVLQJ GLIIHUHQW VHWV RI FRQVRQDQWV IRU WKH EDVHOLQH QR LUUHOHYDQW VSHHFKf DQG HDFK RI WKH H[SHULPHQWDO FRQGLWLRQV 'XULQJ WKH H[SHULPHQWDO FRQGLWLRQV SDUWLFLSDQWV

PAGE 41

KHDUG RQH RI WZR W\SHV RI LUUHOHYDQW VSHHFK GXULQJ WKH SUHVHQWDWLRQ SRUWLRQ RI HDFK WULDO 2Q DOWHUQDWH WULDOV WKH LUUHOHYDQW VSHHFK ZDV KXPDQJHQHUDWHG RU JHQHUDWHG YLD FRPSXWHU 5HVXOWV DQG 'LVFXVVLRQ 7KH SDWWHUQV LQ WKH GDWD DUH TXLWH FRPSHOOLQJ VHH )LJXUH f )RU QHDUO\ DOO SDUWLFLSDQWV WKH KXPDQJHQHUDWHG VSHHFK GLVUXSWHG OHWWHU UHFDOO PRUH WKDQ FRPSXWHUn JHQHUDWHG VSHHFK $ ZLWKLQVXEMHFWV DQDO\VLV RI YDULDQFH VKRZHG WKDW ERWK KXPDQ JHQHUDWHG LUUHOHYDQW VSHHFK DQG FRPSXWHUJHQHUDWHG LUUHOHYDQW VSHHFK LQWHUIHUHG ZLWK SDUWLFLSDQWV UHFDOO RI FRQVRQDQWV ) f S DQG ) f S UHVSHFWLYHO\f 5HFDOO DFFXUDF\ ZDV ORZHU LQ WKH SUHVHQFH RI KXPDQJHQHUDWHG VSHHFK WKDQ WKH FRPSXWHUJHQHUDWHG VSHHFK ) f Sf 7KLV SDWWHUQ RI UHVXOWV PDWFKHV WKHLU VXEMHFWLYH UHSRUWV WKDW WKH FRPSXWHUJHQHUDWHG VSHHFK ZDV HDVLHU WR GLVUHJDUG ,I WKH SKRQRORJLFDO ORRS LV HQJDJHG LQ DQ DOORUQRQH IDVKLRQ LW IROORZV WKDW WKH LUUHOHYDQW FRPSXWHUJHQHUDWHG VSHHFK VKRXOG KDYH WKH VDPH HIIHFW DV LUUHOHYDQW KXPDQJHQHUDWHG VSHHFK 7KHVH UHVXOWV DUH PRUH FRQVLVWHQW ZLWK WKH QRWLRQ WKDW WKH SKRQRORJLFDO ORRS LV FRPSULVHG RI D SRRO RI UHVRXUFHV WKDW FDQ EH HQJDJHG LQ D JUDGHG IDVKLRQ )XUWKHU WHVWLQJ LV QHHGHG WR GHWHUPLQH LI WKH YLVXRVSDWLDO VFUDWFKSDG DOVR RSHUDWHV DV D UHVRXUFH SRRO UDWKHU WKDQ D VWUXFWXUH

PAGE 42

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

PAGE 43

DOORZLQJ SDUWLFLSDQWV WR ZRUN RQ WKH SULPDU\ WDVN DV LI WKH\ ZHUH ZRUNLQJ XQGHU VLQJOHn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

PAGE 44

&+$37(5 (;3(5,0(17 ([SHULPHQW $ 0RVW UHVRXUFH PRGHOV :LFNHQV )ULHGPDQ t 3RLVRQ f DQG ZRUNLQJ PHPRU\ WKHRULHV %DGGHOH\ t +LWFK f DFNQRZOHGJH D GLYLVLRQ LQ SURFHVVLQJ IRU YHUEDO DQG YLVXRVSDWLDO WDVNV LQ DOLJQPHQW ZLWK WKH OLWHUDWXUH RQ KHPLVSKHULF GLIIHUHQFHV *HQHUDOO\ YLVXRVSDWLDO SURFHVVLQJ LV ORFDOL]HG WR WKH ULJKW KHPLVSKHUH ZKLOH YHUEDO SURFHVVLQJ LV LGHQWLILHG ZLWK WKH OHIW KHPLVSKHUH %URRNV f GHPRQVWUDWHG D FOHDU GLVVRFLDWLRQ RI YHUEDO DQG YLVXRVSDWLDO SURFHVVLQJ LQ DQ H[SHULPHQW ZHUH SDUWLFLSDQWV ZHUH SUHVHQWHG ZLWK D EORFN FDSLWDO OHWWHU ZLWK WKH ERWWRP OHIW KDQG FRPHU PDUNHG ZLWK D VWDU 3DUWLFLSDQWV ZHUH DVNHG WR ORRN DZD\ IURP WKH OHWWHU DQG KROGLQJ LW LQ PHPRU\ WR JR DURXQG WKH OHWWHU FORFNZLVH IURP WKH VWDU UHVSRQGLQJ f\HVf LI WKH FRPHU LQ TXHVWLRQ SRLQWHG RXWZDUG RU fQRf LI WKH FRPHU SRLQWHG LQZDUG +HQFH IRU WKH EORFN OHWWHU ) WKH FRUUHFW UHVSRQVHV ZRXOG EH f\HV \HV \HV \HV QR QR \HV \HV QRf %URRNV DOVR GHYHORSHG D YHUEDO WDVN ZKHUH SDUWLFLSDQWV ZHUH SUHVHQWHG ZLWK D VHQWHQFH OLNH f$ ELUG LQ KDQG LV QRW LQ WKH EXVKf 3DUWLFLSDQWV ZHUH DVNHG WR KROG WKLV VHQWHQFH LQ PHPRU\ DQG WKHQ VXFFHVVLYHO\ FDWHJRUL]H HDFK ZRUG DV HLWKHU D QRXQ LQ ZKLFK FDVH SDUWLFLSDQWV ZRXOG UHVSRQG f\HVff RU D YHUE fQRff +HQFH IRU WKDW SDUWLFXODU VHQWHQFH WKH FRUUHFW VHTXHQFH ZRXOG EH fQR \HV QR \HV QR QR QR QR \HVf %URRNV XVHG WZR PHWKRGV RI UHVSRQGLQJ HLWKHU VSRNHQ RU PDQXDO WKLV LQYROYHG SRLQWLQJ WR D VHULHV RI \HVHV RU QRV VFDWWHUHG LUUHJXODUO\ GRZQ WKH UHVSRQVH VKHHWf +H REVHUYHG D

PAGE 45

FOHDU LQWHUDFWLRQ EHWZHHQ W\SH RI PHPRU\ WDVN DQG PRGH RI UHVSRQVH ZLWK WKH YLVXDO OHWWHU WDVN EHLQJ SHUIRUPHG PRUH DFFXUDWHO\ ZKHQ WKH UHVSRQVH ZDV VSRNHQ WKDQ ZKHQ LW LQYROYHG SRLQWLQJ ZKLOH WKH UHYHUVH ZDV WKH FDVH IRU WKH VHQWHQFH WDVN 7KLV ILQGLQJ GHPRQVWUDWHV WKH LQWHUIHUHQFH RI WDVNV WKDW VKDUH VLPLODU SURFHVVHV DQG SRLQWV WR VHSDUDWH V\VWHPV IRU YHUEDO DQG YLVXRVSDWLDO PHPRU\ 6XEVHTXHQW VHOHFWLYH LQWHUIHUHQFH HIIHFWV KDYH EHHQ REVHUYHG Df &RQFXUUHQW YLVXRVSDWLDO SXUVXLW WUDFNLQJ GLVUXSWHG SHUIRUPDQFH RQ %URRNfV f VSDWLDO WDVN EXW QRW RQ D FRUUHVSRQGLQJ DEVWUDFW WDVN %DGGHOH\ f Ef $ YHUEDO WDVN LQWHUIHUHG ZLWK ULJKW KDQG GRZHO EDODQFLQJ .LQVERXUQH t +LFNV f DQG Ff D VSDWLDO WDVN LQWHUIHUHG ZLWK OHIW KDQG GRZHO EDODQFLQJ 0F)DUODQG DQG $VKWRQ f $OVR LQ WKH SHUFHSWXDO GRPDLQ 0RVFRYLWFK DQG .OHLQ f REVHUYHG WKDW UHFRJQLWLRQ SHUIRUPDQFH ZDV PRUH LPSDLUHG ZKHQ WZR VSDWLDO WDUJHWV ZHUH SUHVHQWHG VLPXOWDQHRXVO\ D IDFH DQG D UDQGRP SRO\JRQf WKDQ ZKHQ D VSDWLDO DQG D YHUEDO WDUJHW ZHUH SUHVHQWHG VLPXOWDQHRXVO\ ((* UHVHDUFK VXSSRUWV WKH LQWXLWLRQ WKDW ZULWLQJ OLNH RWKHU YHUEDO WDVNV LV VXSSRUWHG E\ WKH OHIW KHPLVSKHUH *DOOLQ t 2PVWHLQ f VHW RXW WR VHOHFW DUHDV RI WKH EUDLQ WKDW WKH VSOLWEUDLQ DQG EUDLQ OHVLRQ UHVHDUFK KDG VKRZQ ZHUH LQYROYHG LQ GLIIHUHQW WDVNV DQG WR UHFRUG WKH EUDLQfV HOHFWULFDO DFWLYLW\ ZKLOH D SHUVRQ ZDV GRLQJ WKHVH WDVNV ,Q JHQHUDO WKH\ IRXQG WKDW RUGLQDU\ SHRSOH GRLQJ RUGLQDU\ DFWLYLWLHV WXUQHG RQ DQG RII WKH WZR VLGHV RI WKHLU EUDLQ DSSURSULDWHO\ 6SHFLILFDOO\ *DOOLQ t 2PVWHLQ f UHSRUWHG WKDW WKH ULJKW KHPLVSKHUH VKRZHG PRUH DOSKD DFWLYLW\ WKDQ WKH OHIW ZKLOH ZULWLQJ D OHWWHU DQG WKH OHIW KHPLVSKHUH VKRZHG PRUH EHWD DFWLYLW\ :KLOH WKH SHUVRQ ZDV DUUDQJLQJ EORFNV WKH OHIW KHPLVSKHUH VKRZHG PRUH DOSKD WKDQ WKH ULJKW VLGH DQG WKH ULJKW KHPLVSKHUH VKRZHG PRUH EHWD ZDYHV 6R ZKHQ SHRSOH ZULWH WKH\ WXUQ RQ WKH OHIW KHPLVSKHUH DQG WXUQ RII WKH ULJKW

PAGE 46

VLGH RI WKH EUDLQ :KLOH DUUDQJLQJ EORFNV LQ VSDFH WKH\ WXUQ RQ WKH ULJKW VLGH DQG WXUQ RII WKH OHIW KHPLVSKHUH %DGGHOH\ f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fWKHf DW WKH UDWH RI RQH SHU VHFRQG DUWLFXODWRU\ VXSSUHVVLRQf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fV GXDOWDVN PHWKRGV LQ WKH VWXG\ RI FKHVV %HFDXVH SDUWLFLSDQWV SHUIRUPDQFH ZDV QRW PHDVXUHG RQ DQ\ RI WKH VHFRQGDU\ WDVNV WKHUH LV QR HYLGHQFH WKDW SDUWLFLSDQWV SHUIRUPHG WKHVH WDVNV DV WKH\ ZHUH LQVWUXFWHG 3DUWLFLSDQWV PD\ VLPSO\ KDYH VORZHG RU HYHQ VWRSSHG SHUIRUPLQJ WKH VHFRQGDU\ WDVN DV

PAGE 47

SULPDU\ WDVN GHPDQGV LQFUHDVHG $GGLWLRQDOO\ LW LV SRVVLEOH WKDW SDUWLFLSDQWfV SHUIRUPDQFH RI WKHVH WDVNV EHFDPH DXWRPDWLF WKXV UHTXLULQJ OLWWOH RU QR UHVRXUFHV 7KHVH VHFRQGDU\ WDVNV LPSOLFLWO\ DVVXPH DOORUQRQH SURFHVVLQJ RI WKH SKRQRORJLFDO ORRS YLVXRVSDWLDO VFUDWFKSDG DQG WKH FHQWUDO H[HFXWLYH EHFDXVH QR DWWHPSWV ZHUH PDGH WR HQVXUH WKDW WKHVH WDVNV DUH HTXDOO\ GHPDQGLQJ 7KH SXUSRVH RI WKH SUHVHQW H[SHULPHQW ZDV VLPLODU WR %DGGHOH\fV f LQYHVWLJDWLRQ RI WKH ZRUNLQJ PHPRU\ UHTXLUHPHQWV RI FKHVV EXW LQVWHDG RI FKHVV WKLV VWXG\ DVVHVV WKH FRQWULEXWLRQV RI WKH SKRQRORJLFDO ORRS WR HVVD\ ZULWLQJ LQ D GXDOWDVN SDUDGLJP :ULWLQJ UHVHDUFKHUV KDYH DGGUHVVHG WKLV LVVXH E\ HPSOR\LQJ LUUHOHYDQW VSHHFK PDQLSXODWLRQV ZKLOH PHDVXULQJ HVVD\ TXDOLW\ DQG IOXHQF\ 5DQVGHOO t /HY\ 5DQVGHOO /HY\ t .HOORJJ f SURRIUHDGLQJ DELOLW\ -RQHV 0LOHV 3DJH f WHPSRUDU\ VWRUDJH RI VHQWHQFH UHSUHVHQWDWLRQV 0DUWLQ :RJDOWHU t )RUODQRQ f DQG FUHDWLQJ VHQWHQFHV IURP JURXSV RI ZRUGV SURRIUHDGLQJ DQG FRS\LQJ WH[W 0DUHN t /HY\ f %HFDXVH WKH ORQJWHUP UHVHDUFK JRDO ZDV WR FRPSDUH SKRQRORJLFDO ORRS H[SHULPHQW $f DQG YLVXRVSDWLDO VFUDWFKSDG H[SHULPHQW %f FRQWULEXWLRQV WR ZULWLQJ D QHZ VHFRQGDU\ WDVN ZDV QHHGHG ZKLFK XQOLNH LUUHOHYDQW VSHHFK UHTXLUHG PHDVXUDEOH UHVSRQVHV IURP WKH SDUWLFLSDQW $ VLJQDO GHWHFWLRQ SDUDGLJP *UHHQ t 6ZHWV f SDUDGLJP ZDV LPSOHPHQWHG EHFDXVH LW ZRXOG DOORZ IRU HDV\ FRPSDULVRQ RI SHUIRUPDQFH WR D VLPLODU YLVXRVSDWLDO VHFRQGDU\ WDVN 6LJQDO GHWHFWLRQ LV D ZHOOHVWDEOLVKHG PHWKRG XVHG E\ HQJLQHHUV DQG E\ SV\FKRSK\VLFLVWV IRU HVWLPDWLQJ DQ REVHUYHUfV GLVFULPLQDWLYH FDSDFLW\ RU VHQVLWLYLW\ WR D VLJQDO LQGHSHQGHQW RI WKH REVHUYHUfV UHVSRQVH ELDV 7KH PHDVXUH RI VHQVLWLYLW\ LV D SXUH QXPEHU UHSUHVHQWHG E\ WKH V\PERO Gf ZKLFK LV EDVHG RQ WKH

PAGE 48

SDUWLFLSDQWVf QXPEHU RI KLWV DQG IDOVH DODUPV 7KH JUHDWHU WKH YDOXH RI Gf WKH EHWWHU WKH LQGLYLGXDOfV SHUIRUPDQFH $GGLWLRQDO HIIRUWV ZHUH QHFHVVDU\ WR DVVXUH WKDW WKLV VHFRQGDU\ WDVN GLG QRW EHFRPH DXWRPDWLF 'XDOWDVN FRQGLWLRQV ZHUH GHVLJQHG VR WKH GDWD FRXOG EH SDUWHG LQWR ILYHPLQXWH HSRFKV IRU ZKLFK LQGLYLGXDO Gf VFRUHV FRXOG EH FDOFXODWHG 7KXV SDWWHUQV RI Gf VFRUHV FRXOG EH LQVSHFWHG TXDOLWDWLYHO\ RQ DQ LQGLYLGXDO EDVLV DQG SDUWLFLSDQWV ZKR VKRZHG GUDPDWLF FKDQJHV LQ VHFRQGDU\ WDVN SHUIRUPDQFH RYHU WLPH FRXOG EH HOLPLQDWHG IURP WKH VWXG\ 1R SDUWLFLSDQWV ZHUH H[FOXGHG IURP WKLV VWXG\ EDVHG RQ WKLV LQIRUPDWLRQ )LJXUH LOOXVWUDWHV WKH DYHUDJH VHFRQGDU\ WDVN SHUIRUPDQFH DQG WKH SHUFHQWDJH RI ZULWLQJ DV D IXQFWLRQ RI WKHVH ILYHPLQXWH HSRFKV L V e fVV ‘D 2L / 0P00 7LPH PLQXWH HSRFKVf )LJXUH $YHUDJH SHUFHQWDJH ZULWWHQ DQG GXDO WDVN SHUIRUPDQFH DV D IXQFWLRQ RI WLPH

PAGE 49

0HWKRG 3DUWLFLSDQWV )RUW\RQH 8QLYHUVLW\ RI )ORULGD JHQHUDO SV\FKRORJ\ VWXGHQWV SDUWLFLSDWHG LQ WKLV VWXG\ $OO SDUWLFLSDQWV UHSRUWHG WR EH SURILFLHQW WRXFK W\SLVWV DYJ W\SLQJ VSHHG ZDV ZSPf DQG KDG H[SHULHQFH LQWHUDFWLQJ ZLWK D FRPSXWHU XVLQJ D PRXVH 0DWHULDOV DQG $SSDUDWXV 3DUWLFLSDQWV ZURWH HVVD\V RQ RSHQHQGHG WRSLFV f7KH SHUIHFW MREf DQG f7KH JUHDWHVW KLJKf XVLQJ D :LQGRZVEDVHG ZRUG SURFHVVRU GHYHORSHG HVSHFLDOO\ IRU WKLV UHVHDUFK HIIRUW 7KH SURJUDP VXSSRUWHG EDVLF IXQFWLRQV RI WH[W HQWU\ FXWWLQJ DQG SDVWLQJ DQG PRXVH FRQWURO DW WKH SRLQW RI LQVHUWLRQ 7KH SURJUDP DOVR HQDEOHG WKH SUHVHQWDWLRQ RI WKH VHFRQGDU\ WDVN VWLPXOL GLUHFWLRQDO V\PEROVf LQ D ODUJH SRLQWf VDQV VHULI IRQW GLVSOD\HG WR WKH OHIW RI WKH WH[W FRPSRVLWLRQ ZLQGRZ 7KH FKDUDFWHUV VXEWHQGHG DSSUR[LPDWHO\ GHJUHHV RI YLVXDO DQJOH DQG ZHUH UHDGLO\ LGHQWLILDEOH IURP WKH ZULWHUVf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

PAGE 50

UHSODFHG E\ DQRWKHU VLQJOH FKDUDFWHU 7KH FKDUDFWHUV W\SLFDOO\ DOWHUQDWHG EHWZHHQ OHWWHUV DQG GLJLWV 7KH SDUWLFLSDQWfV WDVN ZDV WR LQGLFDWH E\ SUHVVLQJ WKH PRXVH EXWWRQ ZKHQHYHU WKH FDWHJRU\ OHWWHU RU GLJLWf UHPDLQHG WKH VDPH IURP RQH SUHVHQWDWLRQ WR WKH QH[W LH WZR FRQVHFXWLYH OHWWHUV RU QXPEHUVf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t /LQWRQ f ZKHUH WZR LQGHSHQGHQW UDWHUV VRUWHG HVVD\V ILUVW LQWR FDWHJRULHV RI TXDOLW\ DQG WKHQ UDWHG HVVD\V DV KLJK RU ORZ TXDOLW\ ZLWKLQ FDWHJRU\ LQWHUn UDWHU UHOLDELOLW\ bf 'RFXPHQW IOXHQF\ ZDV PHDVXUHG E\ WKH QXPEHU RI ZRUGV

PAGE 51

JHQHUDWHG SHU PLQXWH $ VLJQDO GHWHFWLRQ SDUDGLJP *UHHQ t 6ZHWV f ZDV HPSOR\HG LQ WKH SKRQRORJLFDO VHFRQGDU\ WDVN $V LOOXVWUDWHG LQ )LJXUH SHUIRUPDQFH LQ WKH SUHVHQW H[SHULPHQW GHWHULRUDWHG VLJQLILFDQWO\ IURP VLQJOH WR GXDOWDVN FRQGLWLRQV RQ DOO PHDVXUHV :RUGV SURGXFHG SHU PLQXWH GURSSHG b W f J f DQG WKH TXDOLW\ RI WKH HVVD\V GHFOLQHG b W f S f ZLWK WKH DGGLWLRQ RI WKH SKRQRORJLFDO ORRS VHFRQGDU\ WDVN 0RVW FRPSHOOLQJ ZDV D b GHFUHPHQW LQ Gf DVVRFLDWHG ZLWK WKH DGGLWLRQ RI WKH VHFRQGDU\ WDVN Wf S f 4XDOLW\ )OXHQF\ :30f QG WDVN Gnf 3HUFHQW &KDQJH )LJXUH 'HFUHPHQW LQ TXDOLW\ IOXHQF\ DQG VHFRQGDU\ WDVN Gf IURP VLQJOH WR GXDOWDVN SKRQRORJLFDO ORDG FRQGLWLRQV .HOORJJfV f SUHGLFWLRQV IRU WKH LQYROYHPHQW RI WKH SKRQRORJLFDO ORRS LQ ZULWLQJ ZHUH WKDW WKLV FRPSRQHQW ZRXOG SOD\ D VWURQJ UROH LQ PRQLWRULQJ SDUWLFXODUO\ LQ WKH GHWHFWLRQ RI VSHOOLQJ DQG W\SRJUDSKLFDO HUURUV $OVR WKH SKRQRORJLFDO ORRS VKRXOG

PAGE 52

KDYH D PLQRU LQIOXHQFH RQ ZULWLQJ TXDOLW\ UHODWLYH WR WKH FHQWUDO H[HFXWLYH EXW PRUH LQIOXHQFH RQ TXDOLW\ WKDQ WKH YLVXRVSDWLDO VFUDWFKSDG 'HFUHPHQWV LQ TXDOLW\ DVVRFLDWHG ZLWK WKH SKRQRORJLFDO VHFRQGDU\ WDVN VHHP WR FRQIOLFW ZLWK /HY\ 5DQVGHOO t .HOORJJ f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f GLPLQLVKHG FRPSUHKHQVLRQ 0DUWLQ :RJDOWHU t )RUODQRQ f ORZHU KROLVWLF TXDOLW\ RI HVVD\V DQG GLPLQLVKHG V\QWDFWLF FRPSOH[LW\ DV PHDVXUHG E\ WKH SURSRUWLRQ RI VXERUGLQDWH FODXVHV WKDW RFFXU EHIRUH WKH PDLQ YHUE 0DGLJDQ t /LQWRQ f 7KH SUHVHQW H[SHULPHQW VXSSRUWV 5DQVGHOO /HY\ t .HOORJJfV LQ SUHVVf ILQGLQJ WKDW LUUHOHYDQW VSHHFK UHOLDEO\ UHGXFHG IOXHQF\ DV PHDVXUHG E\ ZRUGV SHU PLQXWH 0DGLJDQ DQG /LQWRQ f KRZHYHU IRXQG QR HIIHFW RI LUUHOHYDQW VSHHFK RQ IOXHQF\ DV PHDVXUHG E\ ZRUG SURGXFWLRQ WLPHV DQG SDXVHV ZLWKLQ VHOHFWHG FODXVHV 7KLV FDOOV LQWR TXHVWLRQ WKH UHODWLRQVKLS EHWZHHQ GLIIHUHQW PHDVXUHV RI ZULWLQJ IOXHQF\ )XUWKHU TXHVWLRQV LQYROYH KRZ D YLVXRVSDWLDO VHFRQGDU\ WDVN ZLOO HIIHFW HVVD\ TXDOLW\ IOXHQF\ DQG Gf VFRUHV FRPSDUHG WR WKH SKRQRORJLFDO VHFRQGDU\ WDVN REVHUYHG LQ

PAGE 53

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t 3DVVHUDXOW /HY\ :KLWH t /HD f .HOORJJ f FODLPHG WKDW YLVXRVSDWLDO UHVRXUFHV DUH LQYROYHG LQ SODQQLQJ SDUWLFXODUO\ ZKHQ ZULWHUV SODQ E\ YLVXDOL]LQJ LGHDV RUJDQL]DWLRQDO VFKHPHV VXSSRUWLQJ JUDSKLFV DSSHDUDQFHV RI WKH RUWKRJUDSK\ DQG OD\RXW ,Q RQH RI WKH RQO\ VWXGLHV LQYHVWLJDWLQJ WKH UROH RI WKH YLVXRVSDWLDO VFUDWFKSDG LQ ZULWLQJ 5DQVGHOO /HY\ t .HOORJJ f KDG SDUWLFLSDQWV ZULWH RQH HVVD\ LQ WKH SUHVHQFH RI XQDWWHQGHG LUUHOHYDQW VSHHFK DQG RQH HVVD\ ZKLOH WKH\ ZHUH LQVWUXFWHG WR DWWHQG WR WKH LUUHOHYDQW VSHHFK SHULRGLFDOO\ UHVSRQGLQJ f\HVf RU fQRf WR TXHVWLRQV UHODWHG WR HLWKHU SKRQRORJLFDO VSDWLDO RU VHPDQWLF FKDUDFWHULVWLFV RI WKH PHVVDJH 7KH W\SH RI TXHVWLRQ DQVZHUHG LQ WKH DWWHQGHG VSHHFK FRQGLWLRQ GLG QRW DIIHFW DQ\ GHSHQGHQW PHDVXUH 2QH LQWHUSUHWDWLRQ RI WKLV ODFN RI ILQGLQJV LV WKDW WKH FRPELQDWLRQ RI PDNLQJ MXGJHPHQWV UHTXLULQJ D UHVSRQVH RQ LUUHOHYDQW VSHHFK PD\ KDYH VLPXOWDQHRXVO\ ORDGHG WKH SKRQRORJLFDO ORRS ZLWK LUUHOHYDQW VSHHFKf WKH FHQWUDO H[HFXWLYH E\ UHTXLULQJ UHVSRQVH VHOHFWLRQ DQG H[HFXWLRQf DQG WKHQ HLWKHU WKH SKRQRORJLFDO ORRS SKRQRORJLFDO MXGJHPHQWVf RU WKH FHQWUDO H[HFXWLYH VHPDQWLF MXGJHPHQWVf DJDLQ RU WKH YLVXRVSDWLDO VFUDWFKSDG ZLWK VSDWLDO MXGJHPHQWVf

PAGE 54

,Q WKH SUHVHQW H[SHULPHQW ZH XVH WKH PHWKRGRORJ\ XQGHUO\LQJ ([SHULPHQW $ WR IRFXV RQ KRZ SURFHVVLQJ LQ WKH YLVXRVSDWLDO VFUDWFKSDG LQIOXHQFHV ZULWLQJ 7KH K\SRWKHVLV ZDV WKDW D YLVXRVSDWLDO WDVN OLNH WKH SKRQRORJLFDO WDVN LQ WKH SUHYLRXV H[SHULPHQW ZRXOG FDXVH D VLJQLILFDQW GHFUHPHQW LQ ZULWLQJ TXDOLW\ DQG IOXHQF\ $GGLWLRQDOO\ ZKHQ FRPSDULQJ WKH GXDOWDVN SHUIRUPDQFH RI D YLVXRVSDWLDO WDVN DQG ZULWLQJ WR GXDOWDVN SHUIRUPDQFH RI D SKRQRORJLFDO WDVN DQG ZULWLQJ ([SHULPHQW $f GXDOWDVN SHUIRUPDQFH ZRXOG EH EHWWHU LQ WKH YLVXRVSDWLDO FRQGLWLRQ ,Q RUGHU IRU SHUIRUPDQFH KHUH WR EH FRPSDUHG WR ([SHULPHQW $ D VHFRQGDU\ WDVN ZDV QHHGHG WR HQJDJH WKH YLVXRVSDWLDO VFUDWFKSDG LQ D VLJQDO GHWHFWLRQ SDUDGLJP ZKHUH QR VLJQLILFDQW GLIIHUHQFHV H[LVW EHWZHHQ WKH YLVXRVSDWLDO DQG SKRQRORJLFDO VLQJOH WDVN Gf VFRUHV $GGLWLRQDOO\ WKLV YLVXRVSDWLDO VHFRQGDU\ WDVN ZRXOG DOVR QHHG WR EH GHVLJQHG VR WKDW Gf VFRUHV FRXOG EH FDOFXODWHG IRU HDFK PLQXWH HSRFK WR DVVHVV ZKHWKHU WKDW WDVN EHFDPH DXWRPDWHG )LJXUH LOOXVWUDWHV WKH DYHUDJH VHFRQGDU\ WDVN SHUIRUPDQFH DQG WKH DYHUDJH SHUFHQWDJH RI ZULWLQJ DV D IXQFWLRQ RI WKHVH ILYHPLQXWH HSRFKV

PAGE 55

7QH PLQXWH HSRFKVf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f DQG KDG H[SHULHQFH LQWHUDFWLQJ ZLWK D FRPSXWHU XVLQJ D PRXVH

PAGE 56

$SSDUDWXV DQG 0DWHULDOV 3DUWLFLSDQWV ZURWH HVVD\V RQ RSHQHQGHG WRSLFV f7KH SHUIHFW MREf DQG f7KH JUHDWHVW KLJKf 7KH\ XVHG D :LQGRZVEDVHG ZRUG SURFHVVRU ZULWWHQ HVSHFLDOO\ IRU WKLV UHVHDUFK HIIRUW 7KH SURJUDP VXSSRUWHG EDVLF IXQFWLRQV RI WH[W HQWU\ FXWWLQJ DQG SDVWLQJ DQG PRXVH FRQWURO DW WKH SRLQW RI LQVHUWLRQ 7KH SURJUDP DOVR HQDEOHG WKH SUHVHQWDWLRQ RI WKH VHFRQGDU\ WDVN VWLPXOL GLUHFWLRQDO V\PEROVf LQ D ODUJH SRLQWf VDQV VHULI IRQW GLVSOD\HG WR WKH OHIW RI WKH WH[W FRPSRVLWLRQ ZLQGRZ 7KH FKDUDFWHUV VXEWHQGHG DSSUR[LPDWHO\ GHJUHHV RI YLVXDO DQJOH DQG ZHUH UHDGLO\ LGHQWLILDEOH IURP WKH ZULWHUVf SHULSKHUDO YLVLRQ 3URFHGXUH 3DUWLFLSDQWV ILUVW FRPSOHWHG D PLQXWH W\SLQJ WHVW LQ ZKLFK WKH\ ZHUH DVNHG WR WUDQVFULEH DV TXLFNO\ DV SRVVLEOH WH[W WKDW DSSHDUHG RQ WKHLU VFUHHQV 3DUWLFLSDQWV WKHQ ZURWH DQ HVVD\ IRU PLQXWHV RQ RQH RI WKH WZR WRSLFV +DOI ZURWH RQ RQH WRSLF ZKLOH WKH RWKHU KDOI ZURWH RQ WKH RWKHU 3DUWLFLSDQWV ZHUH JLYHQ QRWLFH PLQXWHV EHIRUH WKHLU WLPH H[SLUHG 3DUWLFLSDQWV QH[W SHUIRUPHG D YLVXRVSDWLDO WDVN IRU PLQXWHV 7KLV WDVN ZDV GHVLJQHG WR ORDG WKH YLVXRVSDWLDO VFUDWFKSDG EXW QRW WKH SKRQRORJLFDO ORRS 5DWKHU WKDQ OHWWHUV DQG GLJLWV WKH FRPSXWHU GLVSOD\HG HTXDOVL]HG DUURZV WKDW SRLQWHG XS GRZQ OHIW RU ULJKW 7KH SDUWLFLSDQWVf WDVN ZDV WR PDLQWDLQ LQ PHPRU\ D UHSUHVHQWDWLRQ RI DQ [ \ FRRUGLQDWH V\VWHP DQG DQ LPDJLQDU\ SODFHKROGHU DW WKH FRRUGLQDWH 7KH\ ZHUH LQVWUXFWHG WR PRYH WKH LPDJLQDU\ SODFHKROGHU RQH VWDQGDUG EXW DUELWUDU\ XQLW RI GLVWDQFH LQ WKH GLUHFWLRQ RI WKH DUURZ SUHVHQWHG 7KH\ ZHUH IXUWKHU LQVWUXFWHG WR FOLFN WKH PRXVH EXWWRQ DQ\ WLPH WKH SODFHKROGHU PRYHG DZD\ IURP HLWKHU D[LV DQG WKHQ UHWXUQHG WR HLWKHU

PAGE 57

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t /LQWRQ f ZKHUH WZR LQGHSHQGHQW UDWHUV VRUWHG HVVD\V ILUVW LQWR FDWHJRULHV RI TXDOLW\ DQG WKHQ UDWHG HVVD\V DV KLJK RU ORZ TXDOLW\ ZLWKLQ FDWHJRU\ LQWHUUDWHU UHOLDELOLW\ bf 'RFXPHQW IOXHQF\ ZDV PHDVXUHG E\ WKH QXPEHU RI ZRUGV JHQHUDWHG SHU PLQXWH $ VLJQDO GHWHFWLRQ SDUDGLJP *UHHQ t 6ZHWV f ZDV HPSOR\HG LQ WKH SKRQRORJLFDO VHFRQGDU\ WDVN 7KH VHFRQGDU\ WDVN PHW DOO RI WKH FULWHULD HVWDEOLVKHG IRU VHFRQGDU\ WDVN VHOHFWLRQ PRVW LPSRUWDQW RI ZKLFK ZDV WKDW WKHUH ZDV QR VLJQLILFDQW GLIIHUHQFH EHWZHHQ YLVXRVSDWLDO DQG SKRQRORJLFDO Gf VFRUHV XQGHU VLQJOH WDVN FRQGLWLRQV DQG UHVSHFWLYHO\f ,I WKH

PAGE 58

YLVXRVSDWLDO WDVN KDG EHHQ UHFRGHG LQWR D YHUEDO WDVN WKHQ D SKRQRORJLFDO UHSUHVHQWDWLRQ RI WKH SUHVHQWHG DUURZV XS XS ULJKW GRZQ XS OHIW HWFf ZRXOG QHHG WR EH PDLQWDLQHG LQ ZRUNLQJ PHPRU\ ,I WKLV KDG EHHQ WKH FDVH WKHQ ZH ZRXOG KDYH H[SHFWHG SHUIRUPDQFH WR EH PXFK ZRUVH FRPSDUHG WR WKH SKRQRORJLFDO VHFRQGDU\ WDVN EHFDXVH SDUWLFLSDQWV ZRXOG QHHG WR KROG PRUH LQIRUPDWLRQ XS WR FKXQNVf LQ PHPRU\ WKDQ WKH\ ZRXOG RQ WKH YHUEDO VHFRQGDU\ WDVN ZKHUH RQO\ WKH ODVW FKDUDFWHU QHHGHG WR EH UHPHPEHUHG ,I WKH\ LQVWHDG DGRSWHG D YLVXRVSDWLDO FRGLQJ VWUDWHJ\ DQG VWRUHG WKH ORFDWLRQ RI D PHQWDO SODFHKROGHU DV WKH\ ZHUH LQVWUXFWHG WKHQ RQO\ RQH FKXQN QHHGHG WR EH PDLQWDLQHG LQ PHPRU\ DQG VLQJOHn WDVN SHUIRUPDQFH ZRXOG EH HTXLYDOHQW WR WKH SKRQRORJLFDO VHFRQGDU\ WDVN 1R VLJQLILFDQW GLIIHUHQFHV ZHUH IRXQG EHWZHHQ WKH SKRQRORJLFDO Gf f DQG WKH YLVXRVSDWLDO Gf f XQGHU VLQJOH WDVN FRQGLWLRQV 5HVXOWV LQGLFDWHG WKDW PXFK OLNH H[SHULPHQW SHUIRUPDQFH LQ WKH SUHVHQW H[SHULPHQW GHWHULRUDWHG VLJQLILFDQWO\ IURP VLQJOH WR GXDOWDVN FRQGLWLRQV RQ DOO PHDVXUHV VHH )LJXUH f :RUGV SURGXFHG SHU PLQXWH GURSSHG b W f Jf DQG WKH TXDOLW\ RI WKH HVVD\V GHFOLQHG b W f J f ZLWK WKH DGGLWLRQ RI WKH SKRQRORJLFDO ORRS VHFRQGDU\ WDVN $ b GHFUHPHQW ZDV REVHUYHG LQ Gf FRPSDULQJ VLQJOH WR GXDOWDVN FRQGLWLRQV Wf S f

PAGE 59

4XDOLW\ )OXHQF\ :30f QG WDVN Gnf 3HUFHQW &KDQJH )LJXUH 'HFUHPHQW LQ TXDOLW\ IOXHQF\ DQG VHFRQGDU\ WDVN Gf IURP VLQJOH WR GXDOWDVN YLVXRVSDWLDO ORDG FRQGLWLRQV &RPELQHG $QDO\VLV RI ([SHULPHQWV $ DQG % $V )LJXUH VKRZV VHOHFWLYHO\ HQJDJLQJ HLWKHU FRPSRQHQW RI ZRUNLQJ PHPRU\ UHOLDEO\ GHWHULRUDWHG IOXHQF\ TXDOLW\ DQG VHFRQGDU\ WDVN SHUIRUPDQFH 7DEOH OLVWV PHDQV DQG VWDQGDUG GHYLDWLRQV IRU IOXHQF\ TXDOLW\ DQG VHFRQGDU\ WDVN Gf VFRUHV XQGHU VLQJOH DQG GXDOWDVN FRQGLWLRQV :ULWLQJ IOXHQF\ VORZHG VLJQLILFDQWO\ RYHUDOO )O f S DQG DV .HOORJJfV PRGHO DQWLFLSDWHV WKHUH ZDV D VLJQLILFDQW LQWHUDFWLRQ EHWZHHQ WKH FRPSRQHQW RI ZRUNLQJ PHPRU\ WKDW ZDV ORDGHG DQG WKH WHVWLQJ FRQGLWLRQV HLWKHU VLQJOH RU GXDOWDVNf LQGLFDWLQJ WKDW WKH SKRQRORJLFDO VHFRQGDU\ WDVN GLVUXSWHG IOXHQF\ PRUH WKDQ WKH YLVXRVSDWLDO VHFRQGDU\ WDVN )Of S ,QWHUHVWLQJO\ WKH Gf PHDVXUHG GXULQJ WKH SKRQRORJLFDO VHFRQGDU\ WDVN GHFOLQHG PXFK PRUH WKDQ GXULQJ WKH YLVXRVSDWLDO VFUDWFKSDG DV UHIOHFWHG LQ WKH LQWHUDFWLRQ EHWZHHQ ORDGLQJ WDVN DQG WKH WHVWLQJ FRQGLWLRQV )O f S 7KH GHFUHPHQW LQ Gf DVVRFLDWHG ZLWK WKH GXDOWDVN IRU WKH YLVXR

PAGE 60

VSDWLDO JURXS ZDV b )RU WKH SKRQRORJLFDO JURXS WKH F3 IHOO b $OWKRXJK TXDOLW\ ZDV QRW DIIHFWHG PRUH E\ RQH ZRUNLQJ PHPRU\ ORDG WKDQ DQRWKHU ERWK IOXHQF\ DQG Gf VFRUHV GHWHULRUDWHG PRUH LQ WKH SDLULQJ RI WKH SKRQRORJLFDO WDVN ZLWK ZULWLQJ WKDQ LQ WKH SDLULQJ RI WKH YLVXRVSDWLDO WDVN DQG ZULWLQJ 'HVSLWH WKH HTXDO GHJUDGDWLRQ LQ ZULWLQJ TXDOLW\ WKHVH UHVXOWV VXJJHVW JUHDWHU GLIILFXOW\ LQ WLPHVKDULQJ ZULWLQJ ZLWK D SKRQRORJLFDO WDVN WKDQ D ZLWK YLVXRVSDWLDO WDVN ZKLFK LV FRQVLVWHQW ZLWK .HOORJJfV f SUHGLFWLRQV 7DEOH 0HDQV DQG VWDQGDUG HUURUV IRU Gf ZULWLQJ VSHHG DQG ZULWLQJ TXDOLW\ VFRUHV IRU WKH VLQJOH DQG GXDO YLVXRVSDWLDO DQG SKRQRORJLFDO WDVNV 0HDVXUH 9LVXRVSDWLDO 6LQJOH 7DVN 9LVXRVSDWLDO 'XDOWDVN 3KRQRORJLFDO 6LQJOH 7DVN 3KRQRORJLFDO 'XDOWDVN Gf 0HDQ 6' :30 0HDQ 6' 4XDOLW\ 0HDQ 6'

PAGE 61

4XDOLW\ )OXHQF\ :30f QG WDVN Gnf 3HUFHQW &KDQJH )LJXUH 'HFUHPHQW LQ TXDOLW\ IOXHQF\ DQG VHFRQGDU\ WDVN Gf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n WDVN GDWD LQGLFDWHG WKDW WKH YLVXRVSDWLDO WDVN ZDV VOLJKWO\ PRUH GLIILFXOW WKDQ WKH SKRQRORJLFDO WDVN DOWKRXJK GLIIHUHQFHV ZHUH QRW VLJQLILFDQWf ,Q WKH IXWXUH WKLV OLQH RI UHVHDUFK PD\ EH H[WHQGHG WR WKH VXESURFHVV OHYHO RI ZULWLQJ ZKHUH ZH FDQ LQYHVWLJDWH WKH GHPDQGV WKDW IRUPXODWLQJ H[HFXWLQJ DQG PRQLWRULQJ SODFH RQ WKH YLVXRVSDWLDO VFUDWFKSDG DQG SKRQRORJLFDO ORRS 7KHVH IXWXUH LQYHVWLJDWLRQV

PAGE 62

ZLOO HPSOR\ PHWKRGV EDVHG RQ D UHVRXUFH FDSDFLW\ DVVXPSWLRQ RI SKRQRORJLFDO ORRS DQG YLVXRVSDWLDO VFUDWFKSDG SURFHVVLQJ DQG ZLOO LQFOXGH WKH PHWKRGV IRU VHOHFWLQJ VHFRQGDU\ WDVNV WKDW ZH VSHFLILHG HDUOLHU

PAGE 63

&+$37(5 (;3(5,0(17 ([SHULPHQWV D DQG E OHQG VXSSRUW WR WKH QRWLRQ WKDW YLVXRVSDWLDO SURFHVVHV DUH DFWLYH LQ ZULWLQJ EHFDXVH RI WKH VL]H RI GHFUHPHQWV LQ ZULWLQJ TXDOLW\ DQG IOXHQF\ REVHUYHG LQ WKH SUHVHQFH RI WKH YLVXRVSDWLDO VHFRQGDU\ WDVN +RZHYHU TXHVWLRQV DERXW WKH VSHFLILF UROH RI YLVXRVSDWLDO SURFHVVHV LQ ZULWLQJ UHPDLQ XQDQVZHUHG .HOORJJ f VXJJHVWV WKDW SODQQLQJ PD\ GUDZ RQ YLVXRVSDWLDO SURFHVVHV ZKHQ LGHDV RUJDQL]DWLRQDO VFKHPHV RU DSSHDUDQFHV RI WKH RUWKRJUDSK\ DQG OD\RXW DUH YLVXDOL]HG 7KH FXUUHQW H[SHULPHQW WHVWV .HOORJJfV FODLPV E\ PDQLSXODWLQJ WKH ZULWLQJ WRSLF VR WKDW RQH RI WZR VLPLODU HVVD\V UHTXLUHV PRUH YLVXRVSDWLDO SURFHVVLQJ WKDQ WKH RWKHU 8QGHU GXDOWDVN FRQGLWLRQV GHVLJQHG WR ORDG WKH YLVXRVSDWLDO VFUDWFKSDG D YLVXRVSDWLDO ZULWLQJ WRSLF VKRXOG EH PRUH GLIILFXOW WKDQ D VLPLODU QRQ YLVXRVSDWLDO ZULWLQJ WRSLF 6RPH FOXHV DERXW WKH LQYROYHPHQW RI YLVXRVSDWLDO SURFHVVHV LQ ODQJXDJH WDVNV FRPH IURP LQ WKH KHPLVSKHULF DV\PPHWU\ OLWHUDWXUH 2PVWHLQ +HUURQ -RKQVWRQH DQG 6ZHQFLRQLV f FRPSDUHG WKH ((* PHDVXUH RI EUDLQ DFWLYLW\ RI SDUWLFLSDQWV ZKLOH WKH\ UHDG HLWKHU WHFKQLFDO SDVVDJHV RU IRONWDOHV 7KH OHIW KHPLVSKHUH DFWHG WKH VDPH ZLWK WKH WHFKQLFDO PDWHULDO DQG WKH IRONWDOHV +RZHYHU WKH ULJKW KHPLVSKHUH DFWLYDWHG ZKLOH WKH VXEMHFWV UHDG WKH IRON VWRULHV EXW LW GLG QRW ZKLOH WKH\ UHDG WKH WHFKQLFDO PDWHULDO 2UQVWHLQ HW DO H[SODLQ WKDW WHFKQLFDO PDWHULDO LV DOPRVW H[FOXVLYHO\ LPDJHOHVV ZKLOH VWRULHV JHQHUDWH ORWV RI LPDJHV DQG WKDW fWKH VHQVH RI D VWRU\ HPHUJHV WKURXJK VW\OH LPDJHV DQG IHHOLQJVf 2PVWHLQ S f 7KXV UHDGLQJ PD\ UHO\ RQ WKH YLVXRVSDWLDO SURFHVVHV DVVRFLDWHG

PAGE 64

ZLWK WKH ULJKW KHPLVSKHUH DV ZHOO DV WKH OHIW KHPLVSKHUH GHSHQGLQJ RQ ZKDW LV EHLQJ UHDG ,Q DQRWKHU LQYHVWLJDWLRQ RI YLVXRVSDWLDO LQIRUPDWLRQ ZLWKLQ ODQJXDJH $WZRRG f FDUULHG RXW DQ H[SHULPHQW LQ ZKLFK SDUWLFLSDQWV KHDUG HLWKHU KLJKO\ LPDJHDEOH SKUDVHV VXFK DV f1XGLVW GHYRXULQJ ELUGf RU KLJKO\ DEVWUDFW SKUDVHV VXFK DV f7KH LQWHOOHFW RI (LQVWHLQ ZDV D PLUDFOHf (DFK SKUDVH ZDV IROORZHG E\ D VLPSOH FODVVLILFDWLRQ WDVN SUHVHQWDWHG HLWKHU DXGWLWRULO\ RU YLVXDOO\ $WZRRG UHSRUWHG D WHQGHQF\ IRU WKH LPDJHDEOH SKUDVHV WR EH GLVUXSWHG PXFK PRUH E\ SURFHVVLQJ WKH YLVXDOO\ SUHVHQWHG GLJLWV ZKLOH WKH DEVWUDFW VHQWHQFHV ZHUH GLVUXSWHG PRUH E\ DXGLWRU\ SURFHVVLQJ 7KH FXUUHQW H[SHULPHQW WHVWV .HOORJJfV f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

PAGE 65

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n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

PAGE 66

'XULQJ HDFK RI WZR PLQXWH GXDOWDVN SKDVHV RI WKH H[SHULPHQW SDUWLFLSDQWV VLPXOWDQHRXVO\ SHUIRUPHG WKH YLVXRVSDWLDO WDVN DQG ZURWH DQ HVVD\ RQ D JLYHQ WRSLF 7KH HVVD\ WRSLFV 7KH SHUIHFW MRE DQG D 7KH KRXVH \RX JUHZ XS LQ ZHUH FRXQWHUEDODQFHG WR DVVHVV SUDFWLFH HIIHFWV DVVRFLDWHG ZLWK RUGHU 5HVXOWV DQG 'LVFXVVLRQ (VVD\ TXDOLW\ ZDV DJDLQ DVVHVVHG RQ D SRLQW VFDOH XVLQJ WKH HVVD\VRUW PHWKRG LQWHUUDWHU UHOLDELOLW\ bf 0DGLJDQ -RKQVRQ t /LQWRQ f (VVD\ TXDOLW\ ZDV DOVR DVVHVVHG E\ )OHVFK.LQFDLG JUDGH OHYHO FDOFXODWLRQV ZKLFK DUH EDVHG RQ DYHUDJH QXPEHU RI ZRUGV SHU VHQWHQFH DQG WKH DYHUDJH QXPEHU RU V\OODEOHV SHU ZRUG 'RFXPHQW IOXHQF\ ZDV PHDVXUHG E\ WKH QXPEHU RI ZRUGV JHQHUDWHG SHU PLQXWH DQG D VLJQDO GHWHFWLRQ SDUDGLJP *UHHQ t 6ZHWV f ZDV HPSOR\HG LQ WKH YLVXRVSDWLDO VHFRQGDU\ WDVN $GGLWLRQDO FDOFXODWLRQV RI WKH SHUFHQWDJH RI VHQWHQFHV WKDW FRQWDLQHG VSDWLDO LQIRUPDWLRQ ZHUH FDOFXODWHG IRU WKH HVVD\V ZULWWHQ RQ WKH YLVXRVSDWLDO WRSLF XQGHU ERWK VLQJOH DQG GXDOWDVN FRQGLWLRQV $V LQ H[SHULPHQWV D DQG E GXDOWDVN FRQGLWLRQV KHUH ZHUH ILUVW SDUWLWLRQHG LQWR PLQXWH HSRFKV IRU ZKLFK LQGLYLGXDO Gf VFRUHV ZHUH FDOFXODWHG $JDLQ SDWWHUQV RI Gf VFRUHV LQGLFDWHG WKDW VHFRQGDU\ WDVN SHUIRUPDQFH ZDV QRW DXWRPDWHG RYHU WLPH $V LOOXVWUDWHG LQ 7DEOH VHFRQGDU\ WDVN Gf VFRUHV ZHUH QRW VLJQLILFDQWO\ GLIIHUHQW EHWZHHQ WKH WZR ZULWLQJ WRSLF FRQGLWLRQV XQGHU HLWKHU VLQJOH RU GXDOWDVN FRQGLWLRQV DQG XQOLNH H[SHULPHQWV D DQG E QR GHFUHPHQW LQ Gf VFRUHV ZDV IRXQG ZKHQ FRPSDULQJ VLQJOH WR GXDOWDVN FRQGLWLRQV VHH ILJXUH f )DLOXUH WR UHSOLFDWH WKH Gf GHFUHPHQWV REVHUYHG LQ H[SHULPHQWV D DQG E PD\ LQGLFDWH GLIIHUHQFHV LQ WKH UHVHDUFK SDUWLFLSDQW SRRO IURP 6XPPHU WR 6SULQJ VHPHVWHUV

PAGE 67

7DEOH 0HDQV DQG VWDQGDUG GHYLDWLRQV IRU Gf VFRUHV IRU WKH VLQJOHWDVN DQG GXDOWDVN ZLWK YLVXR VSDWLDO DQG QRQYLVXRVSDWLDO ZULWLQJ WRSLFV 0HDVXUH 6LQJOH7DVN 9LVXRVSDWLDO 7RSLF 'XDOWDVN 1RQYLVXRVSDWLDO 7RSLF 'XDOWDVN Gf 0HDQ 6' $V H[SHFWHG QR TXDOLW\ JUDGH OHYHO RU IOXHQF\ GLIIHUHQFHV ZHUH IRXQG EHWZHHQ WRSLFV XQGHU VLQJOHWDVN FRQGLWLRQV VHH 7DEOH f VXJJHVWLQJ WKDW WKHUH ZHUH QR SUHn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f S f ) f

PAGE 68

S f DQG ) f Sf UHVSHFWLYHO\ VHH )LJXUH f 7KH SDWWHUQ RI UHVXOWV IRU TXDOLW\ DQG IOXHQF\ XQGHU GXDOWDVN FRQGLWLRQV LQGLFDWHG GLIIHUHQFHV EHWZHHQ WRSLFV LQ WKH GLUHFWLRQ K\SRWKHVL]HG EXW WKRVH WUHQGV GLG QRW UHDFK VLJQLILFDQFH )OHVFK .LQFDLG JUDGH OHYHO FDOFXODWLRQV GLG VXSSRUW WKH K\SRWKHVLV 7KH QRQ YLVXRVSDWLDO WRSLF JUDGH OHYHOV DYJ f ZHUH VLJQLILFDQWO\ EHWWHU WKDQ JUDGH OHYHOV IRU WKH YLVXRVSDWLDO WRSLF DYJ f XQGHU GXDOWDVN FRQGLWLRQV DV HYLGHQFHG E\ D VLJQLILFDQW LQWHUDFWLRQ EHWZHHQ WRSLF DQG WDVN FRQGLWLRQ ) f S f )LJXUH 'XDOWDVN GHFUHPHQW LQ VHFRQGDU\ WDVN Gf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

PAGE 69

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b 9V b UHVSHFWLYHO\f W f 3 f 0RUH UHVHDUFK LV QHHGHG WR FRQWURO IRU WKH SRVVLELOLW\ WKDW D SKRQRORJLFDO VHFRQGDU\ WDVN PLJKW KDYH WKH VDPH HIIHFW EXW WKLV UHVXOW VXJJHVWV WKDW WKH SUHVHQFH RI WKH VHFRQGDU\ WDVN FKDQJHG WKH ZD\ WKH ZULWHUV DGGUHVVHG WKH WRSLF

PAGE 70

&+$37(5 *(1(5$/ ',6&866,21 ([SHULPHQW LOOXVWUDWHG WKH YDU\LQJ FDSDFLW\ LQ WKH SKRQRORJLFDO ORRS ZKLFK FOHDUO\ SUHVHQWV H[SODQDWRU\ GLIILFXOWLHV IRU FODVVHV RI PRGHOV LQFOXGLQJ %DGGHOH\ DQG +LWFKfV f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

PAGE 71

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f VFRUHV ZHUH QRW VLJQLILFDQWO\ GLIIHUHQW EHWZHHQ YLVXRVSDWLDO DQG YHUEDO ORDGLQJ WDVNV WKHUHIRUH VXJJHVWLQJ WKDW WKHVH WDVNV DUH HTXDOO\ GHPDQGLQJ 7R DVVXUH WKDW VHFRQGDU\ WDVNV GLG QRW EHFRPH DXWRPDWLF PLQXWH GXDOWDVN FRQGLWLRQV ZHUH EURNHQ GRZQ LQWR ILYHPLQXWH HSRFKV IRU ZKLFK Gf VFRUHV ZHUH FDOFXODWHG 3DWWHUQV RI Gf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t )RUODQRQ f DQG ORZHU KROLVWLF TXDOLW\ RI HVVD\V DQG GLPLQLVKHG V\QWDFWLF FRPSOH[LW\ DV PHDVXUHG E\ WKH SURSRUWLRQ RI VXERUGLQDWH FODXVHV RFFXUULQJ EHIRUH WKH

PAGE 72

PDLQ YHUE 0DGLJDQ t /LQWRQ f 2Q WKH RWKHU KDQG WKHVH ILQGLQJV VHHP WR FRQIOLFW ZLWK 5DQVGHOO /HY\ t .HOORJJ LQ SUHVVf ZKHUH DQ LUUHOHYDQW VSHHFK PDQLSXODWLRQ UHSRUWHGO\ KDG QR HIIHFW RQ HVVD\ TXDOLW\ ,QFRQVLVWHQFLHV KHUH PD\ SRLQW WR GLIIHUHQFHV EHWZHHQ LUUHOHYDQW VSHHFK PDQLSXODWLRQV DQG WKH OHWWHU PDWFKLQJ WDVN HPSOR\HG LQ ([SHULPHQW $ ,Q SDUWLFXODU UHVSRQVH VHOHFWLRQ DQG H[HFXWLRQ VWDJHV UHTXLUHG E\ WKH OHWWHU PDWFKLQJ WDVN DUH OLNHO\ WR SODFH JUHDWHU GHPDQGV RQ ZRUNLQJ PHPRU\ WKXV KDYLQJ D JUHDWHU LPSDFW RQ SULPDU\ WDVN SHUIRUPDQFH )OXHQF\ GHFUHPHQWV UHSRUWHG LQ ([SHULPHQW $ VXSSRUW 5DQVGHOO /HY\ t .HOORJJfV f ILQGLQJ WKDW LUUHOHYDQW VSHHFK UHOLDEO\ UHGXFHG IOXHQF\ DV PHDVXUHG E\ ZRUGV SHU PLQXWH 0DGLJDQ DQG /LQWRQ f KRZHYHU IRXQG QR HIIHFW RI LUUHOHYDQW VSHHFK RQ IOXHQF\ DV PHDVXUHG E\ ZRUG SURGXFWLRQ WLPHV DQG SDXVHV ZLWKLQ VHOHFWHG FODXVHV 7KLV UDLVHV TXHVWLRQV FRQFHUQLQJ WKH UHOLDELOLW\ RI GLIIHUHQW IOXHQF\ PHDVXUHV ([SHULPHQW % WHVWHG WKH K\SRWKHVLV WKDW D YLVXRVSDWLDO VHFRQGDU\ WDVN HTXDO LQ GLIILFXOW\ WR WKH SKRQRORJLFDO VHFRQGDU\ WDVN ZRXOG EH HDVLHU WR WLPHVKDUH ZLWK ZULWLQJ 7KLV K\SRWKHVLV LV EDVHG RQ .HOORJJfV f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

PAGE 73

2PVWHLQ +HUURQ -RKQVWRQH DQG 6ZHQFLRQLV f 7KHVH UHVXOWV VXSSRUW D PXOWLFRPSRQHQW UHVRXUFH PRGHO RI ZULWLQJ WKDW VSHFLILHV D GLYLVLRQ RI YLVXRVSDWLDO DQG YHUEDO UHVRXUFHV LQVWHDG RI RQH XQGLIIHUHQWLDWHG SRRO RI UHVRXUFHV .DKQHPDQ f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f UHSRUW LQWHUIHUHQFH EHWZHHQ WZR SURFHVVLQJ WDVNV DQG WZR PHPRU\ VWRUDJH WDVNV EXW QRW EHWZHHQ D SURFHVVLQJ DQG VWRUDJH WDVN VXJJHVWLQJ WKDW SURFHVVLQJ DQG VWRUDJH UHVRXUFHV DUH GLYLGHG :LFNHQV f GRHV QRW UHFRJQL]H VWRUDJH DQG SURFHVVLQJ DV DQ LPSRUWDQW GLVWLQFWLRQ

PAGE 74

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f :ULWLQJ UHVHDUFK SURJUHVVHG XQGHU VLQJOH FDSDFLW\ PRGHOV DV HYLGHQFHG E\ WKH LQIOXHQFH RI WKH +D\HV t )ORZHU f PRGHO RI ZULWLQJ DQG VXFFHVVIXO GHYHORSPHQW RI FRPSXWHUDLGHG ZULWLQJ WRROV .HOORJJ /HD /HY\ t 0DUHN /HD 5RVHQ /HY\ 0DUHN t 5DQVGHOO f 5HFHQW FRQVLGHUDWLRQV RI WKH UROH RI OLPLWHG FRJQLWLYH FDSDFLW\ LQ ZULWLQJ QHFHVVLWDWH WKH PRUH VSHFLILF PRGHOV RI FRJQLWLYH OLPLWDWLRQV VXFK DV %DGGHOH\fV ZRUNLQJ PHPRU\ PRGHO DQG PXOWLSOH UHVRXUFH PRGHOV ([SHULPHQW WRRN D QHZ DSSURDFK WR WHVW VSHFLILF K\SRWKHVHV DERXW WKH UROH RI YLVXRVSDWLDO SURFHVVHV LQ ZULWLQJ 6SHFLILFDOO\ WKLV H[SHULPHQW WHVWHG .HOORJJfV f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f f ILQGLQJV WKDW WKH ULJKW

PAGE 75

KHPLVSKHUH ((* LV PRUH DFWLYH LQ UHDGLQJ SDVVDJHV FRQWDLQLQJ LPDJHU\ WKDQ WHFKQLFDO SDVVDJHV DQG $WZRRGfV f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t 5DQVGHOO f H[WHQGV WKH PHWKRGRORJLHV LQWURGXFHG KHUH WR HYDOXDWH RWKHU FODLPV IURP FXUUHQW ZRUNLQJ PHPRU\EDVHG PRGHOV RI WH[W SURGXFWLRQ )RU H[DPSOH ZH KDYH UHFHQWO\ GHYLVHG D ZD\ WR RYHUFRPH WKH LQKHUHQW FRQIRXQGLQJ LQ ([SHULPHQWV DQG EHWZHHQ WKH VSHFLILF VWLPXOL VKRZQ DQG WKH SDUWLFLSDQWVf UHVSRQVH GHFLVLRQ UXOHV E\ KROGLQJ WKH VWLPXOL FRQVWDQW 7KXV VLQJOH DOSKDEHWLF FKDUDFWHUV PLJKW EH VKRZQ DW WKH VDPH UDWH DV WKH VWLPXOL SUHVHQWHG DV EHIRUH EXW SHULRGLFDOO\ WKH\ ZRXOG VLPXOWDQHRXVO\ FKDQJH FRORU FDVH IRQW SK\VLFDO ORFDWLRQ DQG YDOXH 'LIIHUHQW JURXSV RI SDUWLFLSDQWV DVNHG WR UHVSRQG ZKHQHYHU WZR

PAGE 76

VWLPXOL LQ VXFFHVVLRQ ZHUH WKH VDPH FRORU RU ORFDWLRQ WR HQJDJH WKH YLVXRVSDWLDO VFUDWFKSDGf RU ZKHQHYHU WKH DGMDFHQW VWLPXOL ERWK FRQWDLQHG D ORQJ fHf SKRQHPH WR HQJDJH WKH SKRQRORJLFDO ORRS RU ZKHQHYHU WKH\ IRUPHG D OHWWHU ZRUG WR IRFXV RQ WKH FHQWUDO H[HFXWLYHf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

PAGE 77

$33(1',; $ ([SHULPHQW VWLPXOL Z G M N P ] W E S N K T V F Y N S M I [ Q I Z \ N S Y U Z I J N U Z W I M Q S N TWSNKGV

PAGE 78

] F E P N \ U G U I Y \ M Q S N \ Q I W U J [WYKNZGJ ZGY\KM OPT US\MQE[FZ ]U[IKNQS\ S N J Y Z V U \ Y [ Z J W K N

PAGE 79

3 \ Q P Z G F Y J ] U Y K \ W J M Y Z G Z I W K \ N ZGJ\MN S I E V ] W \ M Y ] U Q N G U W K T K W I Y V Z J

PAGE 80

V U F J E N S T ] J S \ P N J E W \ Q U ] W \ Y E N S K N Y U G J T G ] E W Q \ IUZ[VFEK \ K M N E I Z G]WF\ENQ G J M N K I V

PAGE 81

TWUZ\SKN ]G[IFJYKE ]P\JFHU[Z GM INJP[QF SNQWEUFPV T]ZG[JFKE JGIY\UQNP

PAGE 82

5()(5(1&(6 $WNLQVRQ 5& t 6KLIIULQ 50 f +XPDQ PHPRU\ $ SURSRVHG V\VWHP DQG LWV FRQWURO SURFHVVHV ,Q .: 6SHQFH t -7 6SHQFH (GVf 7KH SV\FKRORJ\ RI OHDUQLQJ DQG PRWLYDWLRQ ,, 1HZ
PAGE 83

%HUWHOVRQ 3 f 7KH SV\FKRORJLFDO UHIUDFWRU\ SHULRG RI FKRLFH UHDFWLRQ WLPHV ZLWK UHJXODU DQG LUUHJXODU ,63V $FWD 3V\FKRORJLFD %RFN -. f 7RZDUG D FRJQLWLYH SV\FKRORJ\ RI V\QWD[ ,QIRUPDWLRQ SURFHVVLQJ FRQWULEXWLRQV WR VHQWHQFH IRUPXODWLRQ 3V\FKRORJLFDO 5HYLHZ %ROHV '% t :LFNHQV &' f 'LVSOD\ IRUPDWWLQJ LQ LQIRUPDWLRQ LQWHJUDWLRQ DQG QRQLQWHJUDWLRQ WDVNV +XPDQ )DFWRUV f %RXUGLQ % t )D\RO 0 f ,V ZULWWHQ ODQJXDJH SURGXFWLRQ PRUH GLIILFXOW WKDQ RUDO ODQJXDJH SURGXFWLRQ" $ ZRUNLQJ PHPRU\ DSSURDFK ,QWHUQDWLRQDO -RXUQDO RI 3V\FKRORJ\ %URDGEHQW f 3HUFHSWLRQ DQG FRPPXQLFDWLRQ 2[IRUG 3HUPDJRQ %URDGEHQW f 7DVN FRPPXQLFDWLRQ DQG WKH VHOHFWLYH LQWDNH RI LQIRUPDWLRQ $FWD 3V\FKRORJLFD %URRNV /5 f 6SDWLDO DQG YHUEDO FRPSRQHQWV RI WKH DFW RI UHFDOO &DQDGLDQ -RXUQDO RI 3V\FKRORJ\ &KHUU\ & f 6RPH H[SHULPHQWV RQ WKH UHFRJQLWLRQ RI VSHHFK ZLWK RQH DQG WZR HDUV -RXUQDO RI WKH $FRXVWLFDO 6RFLHW\ RI $PHULFD &RQUDG 5 t +XOO $f ,QIRUPDWLRQ DFRXVWLF FRQIXVLRQ DQG PHPRU\ VSDQ %ULWLVK -RXUQDO RI 3V\FKRORJ\ &UDLN ), t /RFNKDUW )6 f /HYHOV RI SURFHVVLQJ $ IUDPHZRUN IRU PHPRU\ UHVHDUFK -RXUQDO RI 9HUEDO /HDUQLQJ DQG 9HUEDO %HKDYLRU &UDLN ), t :DWNLQV 0f 7KH UROH RI UHKHDUVDO LQ VKRUWWHUP PHPRU\ -RXUQDO RI 9HUEDO /HDUQLQJ DQG 9HUEDO %HKDYLRU 'DQHPDQ 0 t &DUSHQWHU 3 f ,QGLYLGXDO GLIIHUHQFHV LQ ZRUNLQJ PHPRU\ DQG UHDGLQJ -RXUQDO RI 9HUEDO /HPLQJ t 9HUEDO %HKDYLRU 'DQHPDQ 0 t *UHHQH f ,QGLYLGXDO GLIIHUHQFHV LQ FRPSUHKHQGLQJ DQG SURGXFLQJ ZRUGV LQ FRQWH[W -RXUQDO RI 0HPRU\ DQG /DQJXDJH 'H 5HQ]L ( t 1LFKHOOL 3 f 9HUEDO DQG QRQYHUEDO VKRUWWHUP PHPRU\ LPSDLUPHQW IROORZLQJ KHPLVSKHULF GDPDJH &RUWH[ 'LQHW t 3DVVHUDXOW -XO\f :RUNLQJ PHPRU\ DQG WH[W SURGXFWLRQ 7KH UROH RI WKH YLVXRVSDWLDO VNHWFKSDG LQ ZULWLQJ DUJXPHQWDWLYH DQG GHVFULSWLYH WH[WV 3DSHU SUHVHQWHG DW WKH :ULWLQJ &RQIHUHQFH 3RLWLHUV )5

PAGE 84

(ULFVVRQ .$ t .LQWVFK : f /RQJWHUP ZRUNLQJ PHPRU\ 3V\FKRORJLFDO 5HYLHZ )DJRW & t 3DVKOHU + f 0DNLQJ WZR UHVSRQVHV WR D VLQJOH REMHFW ([SORULQJ WKH FHQWUDO DWWHQWLRQDO ERWWOHQHFN -RXUQDO RI ([SHULPHQWDO 3V\FKRORJ\ +XPDQ 3HUFHSWLRQ DQG 3HUIRUPDQFH )UDFNHU 0/ t :LFNHQV &' f 5HVRXUFHV FRQWXVLRQV DQG FRPSDWLELOLW\ LQ GXDOD[LV WUDFNLQJ 'LVSOD\V FRQWUROV DQG G\QDPLFV -RXUQDO RI ([SHULPHQWDO 3V\FKRORJ\ +XPDQ 3HUFHSWLRQ DQG 3HUIRUPDQFH 7/ )ULFN 5: f ,VVXHV RI UHSUHVHQWDWLRQ DQG OLPLWHG FDSDFLW\ LQ WKH YLVXR VSDWLDO VFUDWFKSDG %ULWLVK -RXUQDO RI 3V\FKRORJ\ )ULHGPDQ $ t 3RLVRQ 0 & f +HPLVSKHUHV DV LQGHSHQGHQW UHVRXUFH V\VWHPV /LPLWHGFDSDFLW\ SURFHVVLQJ DQG FHUHEUDO VSHFLDOL]DWLRQ -RXUQDO RI ([SHULPHQWDO 3V\FKRORJ\ +XPDQ 3HUFHSWLRQ DQG 3HUIRUPDQFH f *DOLQ t 2PVWHLQ 5 f /DWHUDO VSHFLDOL]DWLRQ RI FRJQLWLYH PRGH $Q ((* VWXG\ 3V\FKRSK\VLRORJ\ *DWKHUFROH 6 t %DGGHOH\ $' f :RUNLQJ PHPRU\ DQG ODQJXDJH +RYH 8. /DZUHQFH (UOEDXP $VVRFLDWHV *UHHQ t 6ZHWV f 6LJQDO GHWHFWLRQ WKHRU\ DQG SV\FKRSK\VLFV 1HZ
PAGE 85

+XH & t (ULFVRQ -5 f 6KRUWWHUP PHPRU\ IRU &KLQHVH FKDUDFWHUV DQG UDGLFDOV 0HPRU\ DQG &RJQLWLRQ -HI)HUH\ *& t 8QGHUZRRG f 7KH UROH RI ZRUNLQJ PHPRU\ LQ WKH GHYHORSPHQW RI D ZULWLQJ VNLOO /HDUQLQJ WR FRRUGLQDWH LGHDV ZLWKLQ ZULWWHQ WH[W 8QSXEOLVKHG PDQXVFULSWf -RQHV '0 0LOHV & t 3DJH f 'LVUXSWLRQ RI SURRIUHDGLQJ E\ LUUHOHYDQW VSHHFK (IIHFWV RI DWWHQWLRQ DURXVDO RU PHPRU\" $SSOLHG &RJQLWLYH 3V\FKRORJ\ .DKQHPDQ f $WWHQWLRQ DQG HIIRUW (QJOHZRRG &OLIIV 13UHQWLFH+DOO .DKQHPDQ 7ULHVPDQ $ t *LEEV %f 7KH UHYLHZLQJ RI REMHFW ILOHV 2EMHFW VSHFLILF LQWHJUDWLRQ RI LQIRUPDWLRQ &RJQLWLYH 3V\FKRORJ\ .DQWRZLW] %+ f 'RXEOH VWLPXODWLRQ ,Q %+ .DQWRZLW] (Gf +XPDQ LQIRUPDWLRQ SURFHVVLQJ +LOOVGDOH 1(DUOEDXP .HOORJJ 57 f $ PRGHO RI ZRUNLQJ PHPRU\ LQ ZULWLQJ ,Q &0 /HY\ t 6 5DQVGHOO (GVf 7KH VFLHQFH RI ZULWLQJ SS f 0DKZDK 1(UOEDXP .LQVERXPH 0 t +LFNV 5f )XQFWLRQDO FHUHEUDO VSDFH ,Q 5HTXLQ (Gf $WWHQWLRQ DQG SHUIRUPDQFH 9,, +LOOVGDOH 1(UOEDXP .ODSS 67 t 1HWLFN $ f 0XOWLSOH UHVRXUFHV IRU SURFHVVLQJ DQG VWRUDJH LQ VKRUWWHUP ZRUNLQJ PHPRU\ +XPDQ )DFWRUV .QRZOHV :% f 2SHUDWRU ORDGLQJ WDVNV +XPDQ )DFWRUV /DEHUJ f $WWHQWLRQ DQG WKH PHDVXUHPHQW RI SHUFHSWXDO OHDUQLQJ 0HPRU\ t &RJQLWLRQ /HD /HY\ &0 t 0DUHN/RYHMR\ 3 f 7KH WRSLF VHQWHQFH GLVSOD\ WRRO $Q HYDOXDWLRQ RI ZULWLQJ SHUIRUPDQFH DQG SURFHVV 3UHVHQWHG DW WKH ;,,, $QQXDO (XURSHDQ &RQIHUHQFH RQ :ULWLQJ DQG &RPSXWHUV /RQGRQ /HD 5RVHQ / /HY\ &0 0DUHN/RYHMR\ 3 t 5DQVGHOO 6( f 7KH LQWHOOLJHQW OLVWHQLQJ ZRUG SURFHVVRU 3URVSHFWV IRU WKH IXWXUH RI ZULWLQJ 3UHVHQWHG DW WKH ;,,, $QQXDO (XURSHDQ &RQIHUHQFH RQ :ULWLQJ DQG &RPSXWHUV /RQGRQ /HY\ &0 t 0DUHN 3 f 7KH UROH RI ZRUNLQJ PHPRU\ LQ ZULWLQJ SURFHVVHV DQG SHUIRUPDQFH ,Q *& -HIIHU\ t 0 7RUUDQFH (GVf +LJK DQG ORZ OHYHO

PAGE 86

ZULWLQJ SURFHVVHV 0DQDJLQJ FRJQLWLYH GHPDQGV $PVWHUGDP $PVWHUGDP 8QLYHUVLW\ 3UHVV /HY\ & 0 :KLWH t /HD -XO\f 7KH UROH RI WKH YLVXRVSDWLDO VFUDWFKSDG LQ ZULWLQJ 7HVWLQJ .HOORJJfV PXOWLFRPSRQHQW PRGHO 7R EH SUHVHQWHG DW WKH :ULWLQJ &RQIHUHQFH 3RLWLHUV )5 0DGLJDQ 5-RKQVRQ 6( t /LQWRQ 3: $XJXVWf :RUNLQJ PHPRU\ FDSDFLW\ DQG WKH ZULWLQJ SURFHVV 3DSHU SUHVHQWHG DW WKH $PHULFDQ 3V\FKRORJLFDO 6RFLHW\ :DVKLQJWRQ '& 0DGLJDQ 5t /LQWRQ 3: 2FWREHUf :RUNLQJ PHPRU\ DQG WKH ZULWLQJ SURFHVV XQSXEOLVKHG PDQXVFULSWf 0DQLQ t -RKQVRQ/DLUG 31 f 7KH PHQWDO UHSUHVHQWDWLRQ RI VSDWLDO GHVFULSWLRQV 0HPRU\ t &RJQLWLRQ 0DUWLQ 5& 6KHOWRQ -5 t
PAGE 87

1RUPDQ t %REURZ f 2Q GDWD OLPLWHG DQG UHVRXUFH OLPLWHG SURFHVVLQJ -RXUQDO RI &RJQLWLYH 3V\FKRORJ\ 1DYRQ f 5HVRXUFHV$ WKHRUHWLFDO VRXS VWRQH" 3V\FKRORJLFDO 5HYLHZ f 1DYRQ t *RSKHU f 2Q WKH HFRQRP\ RI WKH KXPDQLQIRUPDWLRQ SURFHVVLQJ V\VWHP 3V\FKRORJLFDO 5HYLHZ f 2PVWHLQ 5 f 7KH ULJKW PLQG 0DNLQJ VHQVH RI WKH KHPLVSKHUHV 1HZ
PAGE 88

6DODPHn 3 t %DGGHOH\ $' f 'LVUXSWLRQ RI VKRUWWHUP PHPRU\ E\ XQDWWHQGHG VSHHFK ,PSOLFDWLRQV IRU WKH VWUXFWXUH RI ZRUNLQJ PHPRU\ -RXUQDO RI 9HUEDO /HDUQLQJ DQG 9HUEDO %HKDYLRU 6KDOOLFH 7 t :DUULQJWRQ (. f ,QGHSHQGHQW IXQFWLRQLQJ RI YHUEDO PHPRU\ VWRUHV $ QHXURSV\FKRORJLFDO VWXG\ 4XDUWHUO\ -RXUQDO RI ([SHULPHQWDO 3V\FKRORJ\ 6KHSDUG 51 f 7KH PHQWDO LPDJH $PHULFDQ 3V\FKRORJLVW 6WHUQEHUJ 6 f 7KH GLVFRYHU\ RI SURFHVVLQJ VWDJHV ([WHQVLRQV RI 'RQGHUVf PHWKRG $FWD 3VYFKRORJLFD 7KRPVRQ -$ f ,V FRQWLQXRXV YLVXDO PRQLWRULQJ UHDOO\ QHFHVVDU\ LQ YLVXDOO\ JXLGHG ORFRPRWLRQ" -RXUQDO RI ([SHULPHQWDO 3V\FKRORJ\ +XPDQ 3HUFHSWLRQ DQG 3HUIRUPDQFH 7ULHVPDQ $0 t 'DYLHV $ f 'LYLGHG DWWHQWLRQ WR HDU DQG H\H ,Q 6 .RPEOXP (Gf $WWHQWLRQ DQG SHUIRUPDQFH 9RO f 1HZ
PAGE 89

:LFNHQV &' f 3URFHVVLQJ UHVRXUFHV LQ DWWHQWLRQ ,Q 5 3DUDVXUDPDQ (Gf 9DULHWLHV RI DWWHQWLRQ 1HZ
PAGE 90

%,2*5$3+,&$/ 6.(7&+ -RVHSK /HD ,,, ZDV ERUQ 1RYHPEHU LQ 7XVFRQ $UL]RQD WR -RVHSK /HD -U DQG %LOOLH $QQ /HD -RVHSK -RVKf PDMRUHG LQ SV\FKRORJ\ DW WKH 8QLYHUVLW\ RI 1RUWK &DUROLQD DW :LOPLQJWRQ DQG JUDGXDWHG FXP ODXGH ZLWK KRQRUV LQ SV\FKRORJ\ LQ 0D\ -RVHSK -RHf HQWHUHG WKH &RJQLWLYH DQG 6HQVRU\ 3URFHVVHV GRFWRUDO SURJUDP DW WKH 8QLYHUVLW\ RI )ORULGD LQ $XJXVW $IWHU JUDGXDWLRQ -RVHSK ZLOO EHJLQ ZRUN DW ,%0 LQ 6DQ -RVH &DOLIRUQLD DV D KXPDQ IDFWRUV HQJLQHHU LQ VRIWZDUH GHYHORSPHQW

PAGE 91

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


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EGML2TJ6X_D3OLWC INGEST_TIME 2014-07-17T22:26:13Z PACKAGE AA00022877_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES