4‘ .1
I

. it .. k .
1w...» V ...!...~....w... c
u... ._ u cw. .

 

 

a; . .
n finCfl Faun!“
“ti. r I
11. .c

Jul-fluhbws‘vvubwwmru

‘55..

 

.. .
ooly'
1. ‘.

 

J 1 f. .2...
.I: .

 

at.
I .f..c.~.

 

1.0 .I' .. i

. 1.575%.
EA...‘ . Jig-El
‘ ..:.,“.e#..,.m. ...n...u|.¢..,

.‘A .
I 6 1“, #0.
\ é «mule ,
. . ‘ , , , ‘ 517a.

rlslwilt'

$5ka

 

4.m....12....., . u.mn..,.w...n._

 

 

        

'vlalrll.

 

THESIS

Illllllilllllllllllllllllllllllllllll

3 1293 01688 7428

This is to certify that the

thesis entitled

FOCUS ON FORM: THE ROLE OF NEGATIVE FEEDBACK
IN GERMAN RELATIVE PRONOUN SELECTION

presented by

DANIEL PARK

has been accepted towards fulfillment
of the requirements for

M.A. , German
degree in

 

 

  

Major professor

Date 7-21-98

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

LIBRARY

MIchIgan State
Unlveralty

 

 

 

PLACE IN RETURN Box
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

MTE DUE I MTE DUE DATE DUE

 

 

 

EOCT 2 3 20+r _'
L.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 animus-p.14

FOCUS ON FORM: THE ROLE OF NEGATIVE FEEDBACK IN GERMAN
RELATIVE PRONOUN SELECTION

By
Daniel Park

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS
Department of Linguistics and Germanic, Slavic, Asian and African
languages

1998

ABSTRACT

FOCUS ON FORM: THE ROLE OF NEGATIVE FEEDBACK IN GERMAN
RELATIVE PRONOUN SELECTION

By
Daniel Park

This project contrasts several kinds of explicit negative
feedback within a CALL context. After completing a pre-test to
determine initial ability, subjects were divided into four groups,
three of which received treatment on the computer. The software
administered different kinds of negative feedback to different
groups. It was hypothesized that the group which received the
metalinguistic feedback would perform significantly better on the
post-tests than the group which received right/wrong feedback, and
that the group that was required to reflect on its incorrect responses
would perform better still. Test results showed that, although all
groups demonstrated a trend toward improvement, no group
performed significantly better than the others.

I also hypothesized that the two groups who received
‘ metalinguistic feedback would demonstrate positive attitudes toward
the software, in terms of both perceived benefit and enjoyment, as
measured by surveys. These survey results did demonstrate general

attitude improvement.

ACKNOWLEDGEMENTS

I wish to thank Lei Chen, Jyotirmoy Dey, and Connie Page for
their help in reviewing the collected data and providing statistical

advice.

i would also like to thank the members of my advisory
committee, who remained accessible and helpful throughout the

process of producing this study:

Patricia R. Paulsell (chair)
Thomas Lovik

Alison Mackey

Patrick McConeghy

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LIST OF TAEILS _ - - _- -.....- ---.------.-- - - -- -- vii
LIST OF FIGURES. ------ - .... - - - ------.ix
INTRODUCTION -1
CHAPTER 1
RESEARCH QUESTIONS, LIMITATIONS, AND TERMS ..................................... 4
Generalized Question -- - - - -4
Specific Questions 4
Limitations 5
Terms - 5
CHAPTER 2
HYPOTHESES AND THEORETICAL & EMPIRICAL SUPPORT ........................ 7
Hypotheses--- -_ - 7
Support ------ -7
Part I Input Enhancement--- - .- 8
“Fine-Grained” Analysis: Alertness
Orientation, and Detection ..... - - - . .---11
The Merger of Meaning and Structure.
Focus on Form - - - - -- 16
The Role of Negative Feedback. .................................................. 19
Reflection vs. Impulsiveness in L2 learners ........................ 22
Part II: Enjoyment and Perceived Benefit of CALL ....................... 23
CHAPTER 3
METHODS AND PROCEDURES - - _27
Institutional and Classroom Settings .................................................... 27
Subjects and Researchers-- ..... -- 27
Procedures for the Protection of Human Subjects .......................... 29
Method- -- -- - ........ - - - - - -29
Pre-test- -- - - - - ----.29
Treatment ----------- -- - 3O
Post—tests -- - ----- -- -- - - 38

 

DependentVariables---- -- -- -..38

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Experimental Design - - -- - - - - 39
Data Collection Procedures ...... - - -- ----39
Timing - - - -- ..... - 39
CHAPTER 4
RESULTS--- - ..... - - 41
Post-Treatment Test Results -- ------------ - - - - _----41
Chi-square Analyses. -- -- ---------- - - - - 42
ANOVAAnaIyses.- - - --------- -- ---..-------- -- ------ 44
Survey Results- ................. . ..... -- ........ . ........... -- 47
Subjects’ReflectionI-H - - -- - -- -- -- - -- -51
Summary. --- - - - - 52
CHAPTER 5
DISCUSSION- -- - - - - - -- 54
Section 1: Discussion of Pre-test/Post—test Results ......................... 55
Section 11: Recommendations for Future Research. ........................ 59
Section III: Discussion of Survey Data. ................................................ 62
Attitude Improvement. .................................................................. 63
Perceived Benefit..-" - ...... - - .......... - - - 64
Section IV: Discussion of Reflection Data. ........................................... 64
CONCLUSIONS- -- -- - - - - --66
APPENDIXA: PRE-TFS'T--- ..... . - ------.69
APPENDDI B: IMMEDIATE POST-TEST ............................................................ 7 1
APPENDIX C: DELAYED POST-TEST ................................................................... 73
APPENDIX D: PRE-TFST SURVEY FORM ........................................................... 7S
APPENDD< E: POST-TEST SURVEY FORM ......................................................... 76
APPENDIX F: CONSENT FORMW - ...... - .----..77
APPENDIX C: CHI-SQUARE TEST RESULTS ..................................................... 78

APPENDD( H: MET ALINGUISTIC FEEDBACK

 

FIRST TREATMENT SESSION .................................................... 79
APPENDIX 1: MET ALINGUISTIC FEEDBACK

SECOND TREATMENT SESSION - - -- . - - 87
LISI‘OF REFERENCES. - - ------------------------ - - - - ....... -95

 

vi

LIST OF TABLES

Table 1 - Means (Standard Deviations) from Brandl’s

 

 

 

follow-up interviews- -- - -- - -- - - - .--.-----2--6
Table 2 - Class percentages in each group ..................................................... 28
Table 3 - Rates of attrition for each group. - 28
Table 4 - Instructors and groups--. -_. ----.- ----.-.-. --- ----29
Table 5 - Declensional forms of relative pronouns .................................... 32
Table 6 - Algebraic design of the treatments ............................................... 39
Table 7 - Mean pre-test scores of all groups ................................................ 39
Table 8 - Developers vs. non-developers for chi-square

test: production, pre-test to immediate post-test ................... 44
Table 9 - Kinds of ANOVA procedures run .................................................... 45

Table 10 - Score improvement in the recognition
section, pretest to immediate post-test. ................................ 46

Table 1 1 - Score improvement in the production
section, pre—test to immediate post-test. ................................ 46

Table 12 - Score improvement in the recognition
section, pre-test to delayed post-test ...................................... 46

Table 13 - Score improvement in the production
section, pre-test to delayed post-test. ..................................... 47

Table 14 - Post-treatment attitude improvement
among groups -- -- ---------- - ------ 49

vii

Table 15 - Subjects’ self—assessed improvement in
relative pronoun selection .................................... - - -- -50

Table 16 - Summary of the results in regard

 

to the three hypotheses ................................................................. 53

Table 17 - Percentages of correctly answered items
in the metalinguistic group ............................................................ 58

Table 18 - Percentages of correctly answered items
in the rigthrong group ................................................................ 58

Table 19 - Percentages of correctly answered items
in the reflection group ................ - -- - -.--58
Table 20 — Research questions and responses .............................................. 66
Table 21 - Chi-square figures, pre-test to immediate post-test .......... 78

viii

LIST OF FIGURES

Figure 1 — Long’s breakdown of negative evidence ...................................... 6
Figure 2 - The components of attention. ......................................................... 14
Figure 3 - Treatment questions. ......................................................................... 34
Figure 4 - Post-treatment test questions ....................................................... 34

Figure 5 - Improvement in test scores between
groups and between tests. .............................................................. 47

Figure 6 - Post-treatment attitude and perceived
benefit on an ordinal scale of 1 to S ............................................ S 1

Figure 7 - Clause type percentage of the total correct
responses on the immediate and delayed post-tests ......... S9

INTRODUCTION

In recent years, much of the research in second language
acquisition (SLA) has dealt with the process of making input more
comprehensible to L2 learners. This process, known as “input
enhancement” (Sharwood Smith, 1991), attempts to focus learners’
attention on form, which entails a concentration on grammatical
structure, but within a meaning-based context. As a means of
achieving this goal, SLA researchers (Fotos, 1993; long, 1996; P.
Robinson, 1995; Schmidt, 1993; Tomlin & Villa, 1994) maintain that
“attention” is a key factor in the detection of a given structure, where
it is then available for further processing. In other words, when
learning an L2, the learner must be able to cognitively process
linguistic information for L2 acquisition to take place. long (1996)
states that attention “is widely claimed to be both necessary and
sufficient for extracting items from a stimulus array and storing
them in long-term memory” (p. 426). The current study attempted
to test these theories by comparing three groups: in two of the
groups, the study will endeavor to stimulate subjects’ attention, by
administering explicit negative feedback which constrains learners to
focus on form.

Also, by measuring student attitudes toward the computer-

assisted language learning (CALL) program which administered the

2

treatment, this study will attempt to determine the effectiveness of
such a program in motivating the student to use CALL software. In
addition, a CALL situation is well-suited to testing the effectiveness
of explicit negative feedback, because it creates an environment in
which feedback to students is consistent‘ , immediate, and
repeatable. Students feel less self-conscious because correction
comes from a machine rather than by a human in a classroom
situation (in front of their peers) (Pennington, 1996). In gathering
more data about this relationship, we as researchers may better
determine how L2 instructors can best select the tools to make input
more salient to their students.

Since the tool of computer technology was first made available
to language learners in the late 1960’s, the field of CALL has
undergone several developments as new approaches and
methodologies were introduced (Warschauer, 1997). Warschauer
outlines a brief history of CALL, the first phase of which was
behavioristic (stimulus-response) CALL, popular in the 1960’s and
70’s. Most programs introduced in this phase were of the “drill-and-
practice” kind (or “drill and kill,” as it was known near the end of its
popularity). At the dawn of the 1980’s, communicative CALL was
introduced (Warschauer, 1997), in which programs made an attempt
to teach grammar implicitly, and concentrate on meaning rather than
on form. With the introduction of communicative CALL, it seems that

‘ Due to the fact that students could have various motives for their incorrect responses. it is
possible that the feedback in this study was not entirely consistent.

 

3

the detractors of the drill-type programs which used explicit
negative feedback may have acted too hastily in dismissing them. As
discussed in detail in chapter 2, tasks which focus on form can be a
valid means of promoting acquisition.

CALL will never constitute a teaching methodology in and of
itself (Garrett, 1991; Warschauer, 1997); trends in CALL generally
mirror trends in SLA, or as Warschauer (1997) states, “The
effectiveness of CALL cannot reside in the medium itself but only in
how it is put to use” (p. 17). With new research into the fields of
attention and noticing, CALL specialists are seeing a resurgence of
the kind of software that had its heyday twenty-five years ago, but
with a new twist. Rather than concentrating on grammatical forms
(isolated linguistic structures), programs with a new focus on form
(structure, but within a communicative context) have taken
advantage of the knowledge gained by SLA researchers in recent
years. This study utilized form-focused CALL activities in order to
determine if a focus on form can truly make relative pronouns more

salient to learners.

Chapter 1

RESEARCH QUESTIONS, LIMITATIONS, AND TERMS
The main focus of this study is to examine the influence of
various kinds of feedback on the correct selection of German relative
pronouns. In conducting empirical research in the psychometric
tradition in order to contrast three kinds of explicit negative
feedback in a CALL environment, this study is concerned with the
following generalized question and specific questions:
WW
0 Does the type of negative feedback given influence subjects’
improvement in the accurate selection of German relative
pronouns?
W
1. To what extent does the reception of metalinguistic feedback
influence subject performance?
2. How does allowing the subjects extra time to reflect upon their
incorrect responses affect overall performance?
3. Is there a relationship between the kind of feedback-received
and the subjects’ perceived benefit from the CALL exercise?
4. Is there a relationship between the kind of feedback received

and the subjects’ overall enjoyment of the CALL exercise?

Limitaticns

1. Subjects in the study were undergraduate students enrolled in
beginning-level German classes at a large public university. The
findings of this study can only be generalized to those institutions
which matriculate students of similar academic backgrounds.

2. Due to the difficulties involved with scheduling groups in the lab,
intact classes were used in this study. The use of intact groups was
the only means of assuring that subjects would use the software
specific to their group. Since relative clauses were taught to students
during the same semester that this study was conducted, it is
possible that individual instructors contributed to some variation,
though all instructors work from identical syllabi, readings, and
quizzes.

Terms

Card - A HyperCard term referring to an individual screen within the
program.

Courseware - Software used for a pedagogical purpose.

Feedback - This study utilizes Long’s (1997, p. 10) definition of
negative feedback, yet concentrating solely on overt error correction:
“Reactive implicit or explicit information derivable from events or
an interlocutor’s verbal or non-verbal behavior which indicates that
a learner’s speech or writing (...) was ungrammatical, opaque, or

inappropriate” [emphasis added]. See Figure 1 (Long, 1997).

 

Negative Evidence

/\

 

 

preemptive reactive
explicit implicit
grammar my
ov ert error (communication recasts

correction breakdown) /\

simple complex

 

 

 

Figure 1. Long’s breakdown of negative evidence [emphasis added].

Hypertext - a non-linear approach to information in which text is
linked to other text in a non-linear way (Bradin, 1996).
Metalinguistic Feedback - feedback on an incorrect response which
provides the learner with grammatical information as to why his/ her
response was wrong.

Stack - A term referring to a collection of cards which form a
HyperCard document.

Task - This study uses Gass’ (1997) definition of task: “...a piece of
work that must be completed.” Although the term is generally
applied to meaning-focused activities, Gass also notes that a focus on

form is “...an essential part of the learning process” (p. 153).

Chapter 2
HYPOTHESES AND THEORETICAL & EMPIRICAL SUPPORT

W
1 . Students who utilize a computerized program focusing on relative
pronouns, and featuring metalinguistic feedback, will score
significantly higher on a post-test which tests them on these forms
than students who only receive right/wrong feedback and are not
instructed to reflect on their responses.
2 . Students who only receive indications of correct/ incorrect
responses from the computer and who are required to reflect on
their incorrect responses, so that they may determine for themselves
why the responses are incorrect, and then receive metalinguistic
feedback, will score higher than subjects who only receive
metalinguistic feedback.
3 . Subjects in the two groups who receive metalinguistic feedback
will report both greater enjoyment and greater perceived benefit of
their CALL experience on a standardized post—test survey than those
students in the right/wrong group.
Snnmm

In order to justify the previously mentioned hypotheses, I will
cite recent theoretical work and empirical studies, which aided me in

designing the overall project and pointed to the possible outcomes of

7

8

my own research. I have divided this literature into two parts. The
first provides support for formulating hypotheses 1 and 2, since the
trivo differ only in respect to the amount of reflection required of the
participant. This part examines one aspect of the role of input
enhancement, addresses the notions of noticing and attention, as well
as the use of negative evidence as enhanced input to facilitate
acquisition. In addition, a small section which examines research on
learner impulsiveness supports only hypothesis 2. The second part
of this review, which supports hypothesis 3, addresses the role of
CALL in SLA research, particularly in regard to student attitudes
toward using it. Although CALL is clearly not the main focus of this
study, I wished to assess subjects’ enjoyment and perceived benefit
from the method of treatment; recent studies support my hypothesis
that both would be generally positive.
W

A number of theories exist on the process of making input
more salient for the L2 learner, one aspect of which is known as
“input enhancement” (Sharwood Smith, 1991). This process has also
been called “consciousness raising” (Sharwood Smith, 1991; Schmidt,
1993), but both researchers dismiss the term as vague and confusing,
and Schmidt states that “the terms ‘conscious’ and ‘unconscious’ have
accumulated so many conflicting meanings that they are unusable

and need to be replaced with specified and testable constructs” (p.

208V.

Although Schmidt makes clear that intention (defined by
Schmidt as the conscious desire to acquire a certain structure) is not
always necessary in second language acquisition, he contends that
attention is necessary. Attention, as defined by Schmidt, is the state
of awareness most conducive to noticing, the process by which a
linguistic structure is detected and then available for further
processing. Attention is regarded by many (Fotos, 1993; P. Robinson,
1995; Schmidt, 1993) as the key element involved in encoding
linguistic information into long-term memory. Elaborating on this,
Schmidt (p. 209) states that, “the subjective experience of ‘noticing’ is
the necessary and sufficient condition for the conversion of input to
intake.”

However, Robinson points out the difficulty in the usage of the
term “attention.” The concept, according to Robinson (1995), has
three uses:

1. “It can be used to describe the processes involved in ‘selecting’

the information to be processed and stored in memory” (p.

287). In other words, attentional resources home in on specific

information for the learner to detect.

2. “It can be used to describe our ‘capacity’ for processing

information” (p. 287). Limited capacity systems represent the

 

’ Despite this warning, the term “conscious” in its various forms

( “consciousness,” “unconscious,” “subconscious,” etc.) appears in the body of
Schmidt’s article no fewer than 40 times, which may point to its pervasiveness
in past research.

1O

belief that only a finite amount of information could be

attended to at one time. This meant that students, with their

limited selective attention, could not focus on form and
meaning at the same time. Robinson (1995) and Tomlin & Villa

(1994) note, however, that more recent theories put an

emphasis on the subjects’ control of attentional resources

rather than on their total attentional capacity. In other words,
although attention may have certain limits, “these limits can be

distributed among several tasks” (Tomlin & Villa, 1994, p. 188).

3. “Finally, it can be used to describe the mental ‘effort’ involved

in processing information” (p. 287). Tomlin & Villa (1994)

state that controlled processes require attention, and this may

inhibit the completion of other attention-demanding tasks.

However, they acknowledge the possibility of processing two

such tasks simultaneously, provided the tasks are somehow

compatible.

Tomlin & Villa (1994) added a fourth component to these uses,
which they call the “course-grained” uses of attention, namely that
attention is a question of the control of information and action. This
examines “the process by which information is regulated as to
whether or not it receives attentional resources” (p. 189). They point
out that these four uses may at times contradict one another, and

that a more precise analysis, or “fine—grained analysis” of attention is

necessary.

11

“Fine—Grained” Analysis: Alertness, Orientation, and
Detection

Tomlin & Villa’s (1994) analysis of attention differs slightly
from Schmidt’s (1993), because they disagree with Schmidt’s claim
that awareness is a prerequisite of noticing. Schmidt’s formulation of
the Noticing Hypothesis was partially based on his diary study of his
own attempts to acquire Portuguese. Tomlin & Villa (1994) call his
study (as well as all diary studies) into question, due to the temporal
irregularity of such studies. They state that “diary studies
encompass spans of time as large as several weeks, but the cognitive
processing of L2 input takes place in relatively brief spans of time,
seconds or even parts of seconds” (p. 185), and cite this as a major
limitation in discovering how attention or noticing function during
the normal time course for a L2 learner. They suggest that conscious
awareness (noticing) may not be as critical for acquisition as other
processes, namely detection and orientation, as these are processes of
attention that can be separated from awareness.

According to Tomlin and Villa (1994), there are three separate
but related components of attention: alertness, orientation, and
detection. I will discuss these each in detail, as well as relate the role
of awareness in relation to these components of attention.

Alertness is regarded as an overall readiness to receive
incoming stimuli. Tomlin & Villa (1994) cite two important aspects

of alertness. First, the degree of alertness stipulates the speed at

12

which information is selected for processing. A warning, perhaps an
auditory signal that the stimulus is coming, can increase learner
alertness. Second, as a disadvantage of high alertness, rapid selection
of information can lead to breakdowns in accuracy. In SLA, too much
eagerness on the part of the learner can bring about such an effect.
In this study, I had the expectation that sitting at the computer in
anticipation of the treatment would bring about the required
alertness in the subjects. The reflection group helped to determine if
over-alertness was a problem, which is discussed in detail in this
chapter’s section on learner impulsiveness.

While alertness indicates a general readiness to receive data,
orientation is the process by which attentional resources can be
directed at a specific kind of stimulus. Tomlin & Villa (1994) state
that the main idea of orientation is the notion that “the specific
aligning of attention (‘orienting’) on a stimulus has facilitative or
inhibitory consequences for further processing depending on
whether information occurs as expected or not expected” (p. 191).

By orienting subjects with explicit metalinguistic feedback toward
the specific target structure of relative pronouns, it was anticipated
that this would aid detection, and that subjects’ ability to accurately
select relative pronouns would significantly improve.

This orienting of attentional resources facilitates the third
aspect of attention, detection. Detection is the process in which

cognitive registration of sensory information occurs. Detected

13

information is the stimulus that has been selected for further
processing, and thus absorbs a lot of attentional resources. In so
doing, it causes significant interference with the processing of other
input. In SLA, detected information is encoded into memory and
available for further cognitive processing (testing and hypothesis
formation). In this study, once detection had occurred, I anticipated
that subjects would process this information to the point that they
could appropriately produce relative pronouns.

It is important to note that detection is most similar to what
Schmidt (1993) labels “noticing.” The principal difference between
these two terms is illustrated by the views which researchers have
on the role of awareness in SLA. Awareness is generally viewed
(both in SLA and cognitive science) as “a particular state of mind in
which an individual has undergone a specific subjective experience
of some cognitive content or external stimulus” (Tomlin & Villa,
1994, p. 193). They cite three prerequisites that learners must meet
if they are to be labeled as aware (p. 193):

1. They must show a “behavioral or cognitive change” that can be
attributed to the experience.

2. They must explicitly state that they were aware of the
experience.

3. They must be able to describe the subjective experience.
Schmidt (1993) claims that acquisition is impossible without

awareness. According the Tomlin & Villa (1994), however, none of

14

the three components of attention require awareness. In addition,
they state that detection requires neither alertness nor orientation,
though these processes certainly facilitate detection. They assert
that information can be cognitively detected, although the learner is
not aware of this occurrence. Therefore, awareness requires
attention, but not vice versa. Schmidt’s (1993) view of noticing is
regarded by Tomlin & Villa (1994) as detection within selective
attention. They state that awareness “plays a potential support role
for detection, helping to set up the circumstances for detection, but it
does not directly lead to detection itself” (p. 199). Figure 2 is the
model of attention set forth by Tomlin & Villa, in examining the

relationships of the various components of attention.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Awareness
, Alertness Orientation 7
Detection F Hypothesis Formation
I
I Input Utterance I

 

Figure 2. The components of attention (Tomlin 8: Villa, 1994, p. 197).

Figure 2 shows that awareness may augment alertness and

15

orientation. Alertness facilitates orientation, and both alertness and
orientation facilitate detection, although these factors do not directly
lead to detection. Once the input has been detected, it is available for
testing and hypothesis formation.

An empirical study by Fotos (1993), in an effort to see how
task type affects acquisition, illustrated how input enhancement can
increase learner awareness, thus facilitating acquisition. Fotos
assigned 160 Japanese university ESL learners (most of whom were
male) to three groups: a grammar task group, which participated in
three grammar input enhancement (dubbed “consciousness raising”
by Fotos) tasks, a grammar lesson group, which received formal
instruction on the same target structures (indirect object placement,
adverb placement, and relative clause usage), and a control group,
which performed communicative tasks with no grammatical content.

After each treatment, which was administered in three cycles
of three weeks each (one week for each structure), all groups
participated in noticing exercises. Subjects were read a story with
the target structure embedded within; they then answered questions
about the story. Participants then received the text to go over their
answers, and were to underline any “special use of English.”

During analysis of data, Fotos determined the content-based
noticings (proverbs, unusual phrases, etc.) to be fairly equivalent
between groups. The grammar lesson group did notice more

examples of indirect object placement than the other groups, but in

16

general, the grammar-based noticings of the two experimental
groups were roughly equal. The grammar lesson and grammar task
groups noticed significantly more target structures than did the
control group, which achieved almost no grammar-based noticings.
Generally, at least fifty percent of all subjects in the experimental
groups made correct noticings, and many noticed multiple structures.

In addition, the experimental groups participated in a pre- and
post-testing regime to determine gains in proficiency. Again, there
were significant gains made within groups, yet no significant
difference between the two groups when it came to general
improvement. Fotos (1993) goes on to report the delayed effects of
the experiment for both experimental groups, stating that, “...once
their consciousness had been raised, the learners continued to be
aware of the structures, noticing them significantly one and two
weeks later in communicative input” (p. 397). Despite the limitation
of intact classes in the study, the treatments raised the alertness of
the participants and oriented them to the specific grammatical
structures to be tested, thus aiding their detection.
The Merger of Meaning and Structure: Focus on Form

In Long’s (1996) research on the environmental factors of L2
acquisition, he determined that comprehensible input alone is not
sufficient for acquisition to occur. He advocated tasks that made
grammatical forms more accessible to learners, by placing them in a

communicative, meaning-based context. Long states that

17

“environmental contributions to acquisition are mediated by
selective attention and the learner’s developing L2 processing
capacity, and that these resources are brought together most
usefully, though not exclusively, during negotiation for meaning" (p.
414).

This negotiation for meaning, during which a given target
structure could be made more salient, was labeled by Long (1996) as
focus on form (p. 429). Form-focused task activities often prove
useful in giving students context-rich input. In addition to being
more interesting than the rote memorization of grammar rules
(which Long calls focus on forms), form-focused activities can also
prove at least as effective as grammar-oriented learning in the
acquisition of grammatical structure, as the following study
illustrates.

In an SLA study which, like the present one, targeted the
acquisition of relative clauses (specifically object of a preposition
clauses), Doughty (1991) sought to compare the effects of form-
focused versus grammar-focused instruction on relativization in
English. She split 20 (10 male, 10 female) intermediate-level ESL
students into three groups: a group which was given meaning-
oriented treatment (or MOG), a group which received grammatical
rule-oriented treatment (or ROG), and a control group (or COG).

After a pre-test to determine pre-treatment proficiency, all

groups completed one computerized reading comprehension lesson

18

every day for 10 days. These texts contained five or six sentences,
all of which contained an object of a preposition type relative clause
[e.g. “I know the people you talked with” (Doughty, 1991, p. 436)].
During treatment, the subjects were assigned to read a given text,
during which they were guided through four steps. First, they had to
skim the text for overall content. This allowed subjects to attain a
general familiarity with the material before proceeding further.

The second portion, called Reading for Understanding, was the
only section that differed between groups. Sentences of the text
appeared on the top half of the screen, with an “instructional
window” on the bottom half. In their instructional window, the MOG
group received semantic and lexical rephrasings, as well as overall
clarification of sentence meaning. Also, the head nouns and relative
pronouns appeared in capitalized form. The ROG group viewed a
program called “Animated Grammar” in their instructional window,
in which relative clauses were broken down into two sentences with
the aid of animation. This was supplemented by explicit statement
of grammatical rules. The COG group was provided only with the
sentences of the original text, with no instructional aid whatsoever.

Following this section, all groups completed a scanning exercise,
during which they were to answer questions about the text’s content.
Finally, participants had to write a summary of the text in their L1.
After their treatments, the subjects took a post-test to examine their

proficiency gains. Although all groups improved significantly, the

19

MOG and ROG achieved massive gains over the COG, and were even
able to generalize their knowledge to relative clause forms other
than object of a preposition.

Although the limited number of subjects and no means of
assessing long-term gains were hindrances in Doughty’s study, her
results support form-focused instruction. The MOG, which was
treated with a focus on form, improved as much as the ROG. The
MOG had the additional advantage of being the only group that
demonstrated a firm understanding of text content, even though
relative pronouns in English generally contribute little to the
meaning of a given sentence.

This is not necessarily the case in German, however. Although
German relative pronouns are also non-salient forms which carry
little semantic meaning by themselves, they stipulate both gender
and case within the clause. An error in pronoun selection could give
the sentence a drastically different meaning. It is important to make
these differences clear to the learner, thereby providing them with
grammatical instruction within a meaningful context.

The Role of Negative Feedback

Upon examination of the theories of attention and noticing in
L2 acquisition, one fundamental question arises: through which
methods can input be enhanced and thus made more salient to the
learner? Shaiwood Smith (1991), among many others, has noted

that this process may be problematic, as externally created salience

20

may fail to have the desired effect, for a number of reasons. First,
the learner may not be developmentally ready to acquire the target
form. Also, as attention is an internal process, other factors may
have hold of the learners’ attentional resources, preventing detection.

As opposed to positive evidence, negative evidence (the kind of
feedback used for all groups in the present study) is defined as
“information about what is not possible in the grammar” (Sharwood
Smith, 1991). However, when learners receive information that
clashes with their own internal grammars, a condition occurs that
Sharwood Smith (1991) refers to as DAZED (detection of anomaly
with zero development). In other words, a learner may know that a
given utterance is incorrect, but unless she/ he is given further
feedback, she/ he may remain unaware of the source of the error.
Although the internalization of negative evidence may increase the
learner’s error detection capabilities, it does not necessarily lead to
correct production (Sharwood Smith 1991 ). This clearly establishes a
need for metalinguistic feedback, in order to clarify why a certain
response is ungrammatical, and thus pointing learners in the
direction of what is grammatical instead of what is not.

This is what Carroll and Swain (1993) discovered in an ESL
study which, like the present one, investigated the effects of various
kinds of feedback. They divided 100 adult native speakers of
Spanish equally into five groups. When subjects made an error

during treatment, group A received explicit rule statements

21

pertaining to the error. Participants in Group B were told when they
were wrong, with no further elaboration (referred to as explicit
utterance rejection). Group C subjects were given the correct form
when they committed an error. Group D subjects, upon giving an
incorrect response, were asked if they were sure. The control group
received no feedback.

Although all treatment groups demonstrated significant
improvement in comparison to the control group, Group A
participants significantly outperformed all other groups on both an
immediate and delayed post-test (with the exception of Group C on
the immediate post-test). Although the study had the serious
limitations of only one treatment session and no control for time on
task, it lends credence to the notion that metalinguistic feedback has
a positive effect on the acquisition of target structures.

In a similar study on input enhancement, Nagata and Swisher
(1995) examined methods of making Japanese particles and
passivization more salient to American university students. During
this project, they investigated the effectiveness of computerized
metalinguistic feedback vs. a more traditional method of feedback
which only told the student if a word was unexpected or missing.
Although the study had the limitation of using one class consisting of
38 subjects, their results revealed that the metalinguistic feedback
was much more effective after testing all subjects on the treated

grammatical structures. The metalinguistic software, which used a

22

complicated parsing system (akin to artificial intelligence) to identify
errors and tailor the feedback to the individual subject, also supports
the value of input enhancement in L2 instruction.

Reflection vs. lmpulsiveness in L2 Learners

As mentioned in Tomlin & Villa’s (1994) article on attention,
when learners’ alertness is heightened, the speed of their responses
is much quicker. However, this has the unfortunate side effect of an
increased number of errors. As a means of promoting more
reflective behavior, Meredith (1978) conducted a quasi-experimental
classroom study that tested impulsiveness in subjects, with the
anticipation that students will catch their own errors when
compelled to reflect before answering a question.

He then examined student scores on a language laboratory
program (on cassette tape) when some students were forced to wait
before responding to a question. Impulsive students’ scores rose
when they had to “stop and reflect” before entering their responses.
Meredith (1978) states that “...allowing such subjects to respond as
quickly as they wish serves only to encourage premature and
inaccurate responding” (p. 326). The present study takes a slight
variation on this theme. Rather than having students wait before
answering a question, they must reflect on and write about their
previously-entered incorrect responses, with the anticipation that
impulsive students will catch some their own errors before seeing

the metalinguistic feedback. This period of reflection is the only

23

differentiating aspect between the reflection and metalinguistic
groups.
W

The third hypothesis posed in this study examines the role of
awareness in the process of using the treatment software. That is,
how does the subjective experience of working with the courseware
contribute to the enjoyment and perceived benefit of the treatment?
Also, how does the perceived improvement compare with actual
proficiency gains? Early CALL research focused predominantly on
whether such a treatment method was more or less effective than
normal classroom instruction (Chapelle 8: Jamieson, 1989). For the
purposes of this study, however, I chose to examine the subjective
experience of the learner; several empirical studies support the
notion that the utilization of CALL has a positive impact on the
attitudes of participants.

In a recent quasi-experimental study, Gail Robinson (1990)
argued that students demonstrated positive attitudes toward a CALL
program which cognitively challenged them as opposed to other
aspects of the program which merely gave explicit right/wrong
feedback or automatically supplied the'correct answer. Robinson
stated that the level of student control and interactivity influences
how students react to CALL. These results are difficult to generalize,
however, as Robinson does not fully describe her procedures; she

also fails to describe or even list the quantity of the participants in

24

her study.

Despite these shortcomings, Robinson (1990) makes a valid
observation of the participants, warning of a phenomenon in social
learning theory called learned helplessness. When the computer
overwhelms a student by a vast assortment of beeps and error
messages, and gives that student no locus of control by utilizing
specific guides which show learners how to prevent future errors, a
sense of learned helplessness can develop. This can have the effect
of “...decreasing their judgments about their language ability, and of
depressing subsequent motivation to continue” (G. Robinson, 1990, p.
157). By providing subjects with detailed feedback, which both
informed them about the ungrammaticality of their responses and
pointed to that which is grammatical, the study aimed to avoid this
phenomenon.

In another study, Stevens (1989b) surveyed a sample of
Arabic-speaking science majors at Sultan Qaboos University, most of
whom had recently used computers for the very first time, to assess
their attitudes toward using CALL software in their study of English.
He created a questionnaire to answer four basic research questions:

0 Do students enjoy using computers to study English?

0 Do students find computer use to be an easy or a daunting
task?

0 Do students feel that they are making satisfactory progress in
learning English by using the computers at the Student
Resource Centre?

25

0 Do students show increasingly positive attitudes with
increasing exposure to computers?

Stevens’ results indicate a resounding yes on all four questions.
However, one concern with the study, which Stevens himself is apt to
point out, was the quasi-experimental design of the project. It is
nearly impossible to prove hypotheses in a qualitative study. Also, it
would be unwise to generalize the results, as only science students
were used in the project; perhaps students who tend to think
analytically (such as science majors) enjoy using CALL software,
whereas global thinkers may not. Regardless, Stevens’ study may be
useful in identifying a trend toward positive attitudes toward CALL.

Finally, Brandi (1995) conducted a study that demonstrated
positive learner attitudes when participants were able to choose the
kind of feedback they received. He identified 21 students from two
first-year sections of German as High Achievement (HA) and Low
Achievement (LA) students. Brandi accomplished this by
administering the American Association of Teachers of German
(AATG) Placement Exam, and excluding from the study all students
who scored between 76% and 83%. Brandi revised his German
Passive Voice Tu tor to allow subjects to choose the following methods
of feedback for wrong answers: metalinguistic, right/wrong, error
feedback (in which all incorrect words were highlighted by the

program), and being shown the correct answer.

26

After treatment, Brandi conducted interviews with subjects to

discover their attitudes in using the software. He structured the

questions on a three-point ordinal scale (as interpreted by the

interviewer), ranging from 0 (not at all) to 2 (very much). The

results of several of the questions are visible in Table 1.

Table 1. Means (Standard Deviations) from Brandl’s follow-up

interviews (p. 205).

 

 

 

 

 

Statements used by interviewer High Achievers Low Achievers
to Assess Student Opinion n=8 n=10
The program was helpful. 2.00 (0.00) 1.88 (0.35)
I liked to explore errors. 2.00 (0.00) 1.50 (0.93)
The grammatical feedback was helpful. .88 (0.83) .43 (0.79)
I felt I had understood the forms and 1.63 (0.52) .50 (0.76)
concepts of passive voice before doing

the exercises.

I felt I had understood the forms and 2.00 (0.00) 1.50 (0.53)
concepts of passive voice better

afterwards. '

The exercise helped me a lot to practice 2.00 (0.00) 1.71 (0.49)
forms.

The exercise helped me a lot to think 2.00 (0.00) 2.00 (0.00)
about my mistakes.

Ifelt frustrated. 1.60 (0.89) 1.29 (0.95)

 

Although subjects did not feel they had benefited much from

the metalinguistic feedback, Brandl’s results show a marked

perceived improvement after treatment, both in the HA and LA

groups. While the high rate of frustration among learners is a cause

of concern, the learners’ high perceptions of improvement support

the third hypothesis. CALL software, if well-designed, can yield

positive attitudes among its users.

 

Chapter 3

METHODS AND PROCEDURES

Institutinnalandflassmmfietfings

This experiment took place at Michigan State University, a
large public university enrolling approximately 40,000 students. The
pre-test and post-test portions of the experiment were conducted in
the subjects’ normal German classrooms, which were nearly identical
in terms of size and the equipment contained therein. All German
classrooms were located in the same building. The computerized
portion was conducted in the Macintosh-equipped language
Learning Center in the Old Horticulture building. Many of the
students had no familiarity with the lab prior to the experiment.
With every experimental group, I gave a brief introduction to the lab
facility prior to treatment.
Whats

The subjects of the experiment were enrolled in four sections
of German 102, a second-semester continuation of beginning-level
German, totaling 83 students. I randomly assigned each of the four
participating course sections to a specific group. Although
predominately freshman and sophomore students fill these sections,
upperclassmen also participated in this study; see table 2 for the

percentages in each section.

27

28

Table 2. Class percentages in each group.

 

Control Metalinguistic Right/Wrong Reflection Total

 

Freshmen 36% 61% 33% 37% 43%
Sophomores 14% 6% 40% 25% 20%
Juniors 21% 17% 27% 25% 22%
Seniors 29% 10% 0% 13% 13%
Graduate Sts. 0% 6% 0% 0% 2%

 

As is clear from table 1, there was a disproportionate number
of freshmen in the metalinguistic group. Although I used intact
classes in this study, I believe that the class standing of participants
had little or no effect on the overall results.

Subjects were given instruction on relative clause formation in
their German classes approximately two weeks prior to the pre-test.
A total of 63 participants remained in the study after attrition: 14
students participated in the control group, 18 subjects made up the
metalinguistic group, 15 subjects participated in the right/wrong
group, and the reflection group consisted of 16 participants. The
pre- and post-attrition numbers for each group are listed below in

table 3.

Table 3. Rates of attrition for each group.

 

 

Group Pre-Attrition Post-Attrition
Control 17 14
Metalinguistic 24 18
Right/W rong 21 15
Reflection 21 16

 

As noted in the table 4, below, instructor A taught both an

29

experimental and a control section. All instructors were teaching
assistants enrolled in the German graduate program at Michigan
State University, had less than three years teaching experience, and
were under the age of thirty. However, Instructor B is a native
speaker of German, whereas the other two instructors have English

as their L1.

Table 4. Instructors and groups.

 

 

Section Instructor Group
1 A Control
2 A Metalx feedback
3 B Right /W rong feedback
4 C Metalx feedback+reflection

 

WW

Following procedures prescribed by the University Council on
Research Involving Human Subjects (UCRIHS), all subjects signed a
consent form prior to the pre-test. I adhered to all guidelines
stipulated by this committee, as described in the consent form itself.
The complete participant consent form can be found in Appendix F.
Method
Pre-test

After completing the consent form, a standardized paper-and-
pencil pre-test (Appendix A) was administered to all subjects, who
entered their student number and section number on the test for

identification purposes. The test contained ten multiple-choice and

30

ten cloze items, and aimed to evaluate subjects’ preexisting
knowledge of relative pronouns, in order to make certain) that the
groups demonstrated roughly equal initial ability. The cloze items
were added for the purpose of testing not only recognition of the
grammatical forms, but also the production of those forms; I wished
to ascertain if the treatment, which contained only recognition items,
had any effect on production ability. Upon completion of the pre-
test, subjects were required to complete a brief survey form about
their attitudes toward computer use; I needed to ascertain pre-
treatment attitudes before making any claims about attitude
improvement. This form can be found in Appendix D.
Treatment

All three experimental groups underwent two separate
treatments in the Macintosh lab. The two treatment sessions
consisted of two separate HyperCard stacks, which I programmed
with HyperCard 2.3. Each stack featured thirty multiple-choice items
designed to improve subjects’ ability to accurately select all types of
German relative pronouns, with the exception of the genitive, as
subjects are not taught the genitive case at Michigan State until the
second year of instruction.

I chose relative pronouns as the target structure in this study
for two reasons. First, from a practical standpoint, relative clauses
are taught during the second semester of beginning-level German,

and I therefore knew that all students had exposure to this structure.

31

More importantly, it is important for students to discover the way
relatives pronouns in German contrast with those in English; unlike
English, correct relative pronoun selection in German necessitates a
base knowledge of the German gender and case system. I provide
examples of this in the following paragraphs. While the word order
of German relative clauses tends to give English speakers the most
trouble (Kufner, 1962), I avoided this problem by providing the base
sentence structure, thus giving my subjects the sole responsibility of
selecting the correct pronoun.

I chose to use only declensional relative pronouns when
designing the treatments. Other relative pronouns, such as wer (wen,
wem) , as well as wo- compounds, were omitted from the study,
since they are not taught in the students’ first year at MSU, at least
not within the context of the relative clause. In addition, the
declensional relative pronoun is more commonly used3 .

For those with less familiarity with the formation of relative
pronouns in German, Zorach & Melin (1994) list a series of steps
which students can follow to assist them in producing the correct
German relative pronoun in a particular context. First, students must
find the antecedent to which the relative pronoun refers. Then they
need to determine the number and gender of the antecedent, since
this determines number and gender of the relative pronoun. Finally,

the students must determine the function (subject, object, etc.) of the

2'Hellinger (1977) notes that the use of wo-compounds is waning in modern
German.

 

32

relative pronoun within the relative clause, since function
determines case. The declensional forms of relative pronouns are

listed in table 5, below.

Table 5. Declensional forms of relative pronouns.

 

 

 

 

 

 

 

SIN GU LAR PLURAL
masculine neuter feminine maac/ ntr/ fem
N om inative ..., der ..., das ..., die ..., die
Accusative ..., den ..., das ..., die ..., die
Dative ..., dem ..., dem ..., der ..., denen

 

 

 

While only indirectly bearing on my hypotheses, clause type is
also a factor which should be kept in mind during treatment design.
The inclusion or exclusion of specific clause types in the treatment
can have drastic effects on the results. Keenan & Comrie (1977)
demonstrated this in their theory of the Accessibility Hierarchy, a
means of expressing the relationship between different kinds of
relative clauses. The hierarchy is outlined here:

Subject Clause > Direct Object Clause > Indirect Object Clause >
Object of a Preposition Clause > Genitive Clause > Object of a
Comparison

In their study of linguistic differences between languages,
Keenan & Comrie made the following discoveries:
0 All languages have subject relative clauses.

0 If a certain language has a kind of relative clause, it will also
have every clause type that is further left on the hierarchy (i.e.
if a language permits a genitive clause, it will also have a direct
object clause).

33

0 Clause types become increasingly difficult for L2 learners as
one goes further to theright in the hierarchy (i.e. indirect
object clauses are more difficult to acquire than subject
clauses).

Several scholars have tested this theory empirically, most
notably Gass ( 1979), as well as Aarts & Schils (1995). With the
exception of genitive clauses, which Gass assesses to be further to the
right on the hierarchy than Keenan & Comrie assert, they have found
the Accessibility Hierarchy to be essentially accurate.

German permits all of the above relative clause types, with the
exception of object of a comparison; this particular clause type is also
questionable in English (“David is the man that I’m smarter than”).

In addition, no genitive clauses were used in the study, as German
students at MSU receive no classroom instruction on the genitive case
until their second year. Although participants may have had
exposure to the genitive case elsewhere, it is not the purpose of this
study to assess the effect of instruction, but rather to determine how
feedback type can improve the accurate selection of relative
pronouns types to which subjects have already been exposed. The
subjects in this instance are not necessarily familiar with genitive
clauses.

Upon reexamination of the treatment and test questions, it
quickly became clear that subjects had been tested most on the
clause form in the study that was the most difficult to acquire: object

of a preposition. Out of the sixty relative clauses in the two

34

treatment sessions, 17 were subject clauses, 1 1 were direct object
clauses, only 3 indirect object clauses appeared, and 29 object of a
preposition clauses were presented. Out of the forty post-treatment
test questions, 1 1 were subject clauses, 6 direct object clauses
appeared, only one indirect clause was given, and 2 2 of the 40
questions dealt with object-of-a-preposition clauses. The

percentages of each clause type are listed in Figures 3 and 4.

 

   
   

     

Object Object
of a of a

prep. Subject Prep. Subject
49% 28% 54% 28%

 

 

 

 

 

Direct
Ob'ect _
Indirect 1 3% indirect
Object Object
5% 3%
Figure 3. Treatment questions. Figure 4. Post-treatment test questions.

I was unaware of the Accessibility Hierarchy when designing
this study, and as a result the treatment of clause types is clearly
unbalanced. Nonetheless, the Accessibility Hierarchy proposes that
subjects would generalize their knowledge of object of a preposition-

type clauses to easier clause forms, and I anticipated that my results

35

would corroborate this. The groups’ accuracy of each clause type on
the post-tests is discussed in chapter 5. My expectation that subjects
would generalize their knowledge to easier clause types is supported
by a study which is discussed in detail in the previous chapter;
Doughty (1991) treated subjects only with object of a preposition
clauses. When she gave a post-test, which covered an array of
relative clause forms in English, she discovered that the subjects had
generalized this knowledge to the easier relative clause forms.

I included no cloze items in the treatments, due to the technical
difficulty involved with supplying metalinguistic feedback on such
exercises. In addition, it was feared that the inclusion of cloze
exercises in the treatment would unnecessarily frustrate the
subjects. Even with guided feedback, when subjects continually
answer incorrectly, they could possibly become discouraged, and
their responses could thus become more and more haphazard. Cloze
items were, however, part of the pre- and post-tests for the purpose
of determining whether subjects can generalize the acquired
knowledge during treatment to new contexts. Subjects were
required to proceed in a linear fashion through the exercise.
However, the material was presented in no particular order in regard
to case.

When constructing the treatments, care was taken to make the
program aesthetically pleasing, while still keeping it visually simple.
Higgins & Johns (1984) suggest that visual rewards for a correct

36

response, such as a graphic fireworks display, or gushing textual
congratulations (“Wowl You’re a geniusl”) can easily become at best
tiresome, at worst condescending. Taking this idea even further,
Stevens (1989a) suggests that the “wow-effect” of these visual
rewards is always disturbing, stating that eventually “...the novelty
wears off, and student users do not spend any more time with
redundant and repetitive courseware than they do with the books on
which such courseware is based” (p. 32).

For every question, subjects had four possible answers from
which to choose, and only one response was correct. For every
wrong answer, subjects in the metalinguistic group were supplied
with metalinguistic feedback (in English), consisting of grammatical
rules and additional clues as to why the chosen answer was incorrect.
When subjects in the reflection group entered incorrect responses,
they were instructed to postulate and then enter their thoughts into
the computer, using a special text field within HyperCard, as to why
their answer may have been erroneous. Upon returning to the
exercises, these subjects also received metalinguistic feedback.
Subjects in the right/wrong group, upon selecting a response, were
informed only whether the chosen response was right or wrong, with
no other corrective feedback of any kind, and were asked to try
again if the answer was deemed incorrect. The control group, which
was added for the purpose of controlling a test repetition effect,

received no treatment.

37

To illustrate the above conditions within the context of the
program, consider the following example from the second treatment

session:

 

1. Jim Morrison ist ein Rocksanger, an Drogen gestorben

ist. (Jim Morrison is a rock star who died from drugs)

a. dem
b. der
c. den
(1. das

In this case, the second option “der” is the correct response.
Let us suppose that a subject chose (c), which is incorrect. The actual
computer responses for each of the three groups are listed below.
The metalinguistic feedback for the responses on all questions is

listed in Appendices G and H.

Metalinguistic Group: “I'm afraid that's incorrect. Since the
relative pronoun is the subject, the nominative is needed in this
instance, not the accusative. Try again.”

Right/Wrong Group: “I’m sorry, that’s incorrect. Give it another
07-”

Reflection Group: “I'm sorry, that's incorrect. Please reflect on
why you think your answer may have been incorrect, then click on
the button marked 'reflect'. This will bring you to a page where you
can write down your thoughts.” (Upon returning to the program, this
group received the same feedback as the metalinguistic group).

Subjects in all treatment groups were allowed to select another

answer if their previous choice was incorrect, and were allocated an

38

infinite number of ‘tries’. Subjects were not allowed to proceed to
the next item without having first ascertained the correct response
for the current item.

Post-tests

A standardized paper-and-pencil post-test was administered
immediately after the second treatment. It was similar to the pre-
test in both form and content. This test is in Appendix B.

I also administered a delayed post-test approximately two
weeks following the immediate post-test in order to examine
subjects’ retention of the information following treatment (Appendix
C). Its form was similar to that of its predecessors.

Upon completion of the delayed post-test, subjects were
required to fill out a survey form. All questions were on a five-point
ordinal scale, with responses ranging from “strongly disagree” to
“strongly agree”. This survey form is available in Appendix E.
W125

There are two dependent variables in this experiment. The
first, namely the improvement in German relative pronoun selection,
was determined by scores on the standardized post-tests. These
scores were compared with scores from the pretest to determine
improvement.

The second dependent variable, subject attitudes, was assessed
with two survey forms which accompanied the pretest and the

delayed post-test, and which questioned subjects on their enjoyment

39

and perceived benefit from the project.
W

This study used the quasi-experimental model of a Control
Group Pre-test - Post-test design. The employment of a pretest
aided in assuring that any inferences made about the groups after
treatment were not the result of preexisting variables (Hatch and
Lazarton, 1991).

The algebraic design for the groups, with X representing the

pre— and post-tests and T representing the treatments, is as follows:

- Table 6. Algebraic design of the treatments.

 

Metalx group X1 T1 X2 X3
Metalx+reflection group X1 T2 X2 X3
Right / W rong group X1 T3 X2 X3
Control group X1 T0 X2 X3

 

The design reflects a standard pre-test, post-test, and delayed
post-test, with varying treatments. All subjects received the exact
same pretest and post-tests, while their treatments varied.
DataLnllecticancedures

I administered all pre-test, post-test, and survey material, and
also collected this data. All software data, namely the text files from
subjects in the metalinguistic+reflection group, were collected after
they finished the exercise and exited the program.

liming
The pre-test took place approximately three days prior to the

40

first treatment. Twenty minutes were allotted for subjects to read
and sign the consent form, complete the pretest, and fill out the
survey. Each treatment lasted approximately twenty minutes, except
in the case of the metalinguistic+reflection group, who had 35
minutes to complete the treatment, due to the additional time
required to reflect and report on incorrect responses. For most
subjects, exactly one week transpired between treatments. However,
some subjects who missed one of the treatments or post-tests agreed
to come in at an alternate time and make up the treatment. The
immediate post-test took place on the day after the second
treatment; subjects had fifteen minutes to complete it. The delayed
post-test followed two weeks after the immediate post-test. Subjects
had fifteen minutes to hand in the delayed post-test along with the

final survey.

Chapter 4

RESULTS

In this chapter, I have outlined the results of my data analysis.
The results section is divided into three portions. The first portion
summarizes the data from the pretest and post-tests; its purpose is
to chart the improvement of subjects both within and between
groups, to see if hypotheses 1 and 2 are verified or not. The second
portion, which directly pertains to hypothesis 3, charts the outcome
of the survey data taken from participants’ questionnaires. In the
third portion, I analyze the reflective writings of the reflection group
for clues about the hidden motives and thought processes behind
subjects’ choices.
W

The mean scores of all groups on the pre—test were roughly
equivalent, as outlined in table 7. Subjects’ recogmtion of the target
forms were clearly stronger than their Winn of them. This is to
be expected. Several researchers have noted that input, under the
right conditions, can lead to intake (recognition), where it is available
for hypothesis testing and then output (Long, 1981; Schachter, 1983,
VanPatten & Cadierno, 1993). In addition, it is Tomlin & Villa’s

(1994) assertion that a target structure must be detected

41

42

(recognized) before it is available for further processing (hypothesis
formation, and ultimately, production). All numbers represent point
scores: ten points were possible in the recognition section, as well as
ten points in the production section, thus giving the pre-test (as well

as both post-tests) a total of twenty points.

Table 7. Mean pre-test scores of all groups.

 

 

Group Recognition Production Total
Control 5.36 3.93 9.29
Metalinguistic 5.61 4.11 9.72
Right/W rong 5.13 3.6 8.73
Reflection 5.81 4.38 10.19

 

As is clear in the above table, subjects were much more able to
recognize the target structure than to produce it, and this relatively
weak production ability appears to have negatively affected their
overall scores. The total mean score for the groups ranged from 8.73
(Right/Wrong) to 10.19 (Reflection); even the highest-scoring group
achieved only roughly half of the total possible points.

Chi-square analyses

In analyzing the data, a number of statistical tests based on the
non-parametric chi square measure were performed in order to
determine the relationship between groups in regard to post-
treatment performance. In statistics, a chi-square test is used to test
the relationship of nominal variables. Nominal data is data that can

be separated into categories, e.g. subjects who improved versus those

43

who did not. These are listed as frequencies. Without the use of
statistical analyses, frequency data alone may be misleading, since
there are several factors involved in examining differences in
variables. Therefore, a chi—square test is necessary in determining
how different these frequencies must be before any relationship
between variables can be claimed.

On each of the two test sections, recognition and production, I
deemed an improvement of +3 to be significant for the purposes of
this study. In either of the sections, if a subject improved by three
points, she/ he was labeled a “developer”. “Non-developer” indicates
a subject who improved by fewer than three points, or who
experienced no improvement whatsoever. It is important to note
that the +3 score improvement is an arbitrary figure" , which I used
only to classify the subjects as either developers or non-developers.
In each group, the number of developers versus non-developers was
listed, and a chi~square analysis was used to determine significance.
It is important to note that statistical significance on a chi-square
only indicates a relationship between variables; it in no way implies
a cause-effect relationship.

Only one chi-square test demonstrated a significant result. In
the production aspect, the number of developers between the pre-
test and the immediate post-test differed significantly. In this
particular analysis, I merged the metalinguistic group and reflection

‘Several recent SLA studies assign arbitrary measures to indicate development (Mackey &
Philp, 1998).

 

44

group for testing purposes. As previously mentioned in chapters 2
and 3, the reflection group had access to the same metalinguistic
feedback as the metalinguistic group. I aimed to test the relationship
between those who received metalinguistic feedback and those who
did not. The number of developers and non-developers for this test
are listed in table 8. The result of the chi-square indicated a P-value
of 0.029, thus establishing significance. The statistical results of the
chi-square can be found in Appendix G.

Table 8. Developers vs. non-developers for chi-square test: production, pre-

test to immediate post-test. (C=Control, Mx=Metalinguistic group,
R / W=Right / Wrong group)

 

Groups Developers Non-Developers Total

 

C 2 12 14
Mx+Refl 17 17 34
W 9 6 1 5
Total 28 35 63

 

Therefore, those groups which read the metalinguistic feedback
cultivated significantly more developers than the groups which had
no such feedback.

ANOVA Analyses

In another set of analyses, I chose to examine the between-
group relationships of all subjects by utilizing a one-way ANOVA
test. I used a one-way ANOVA, in contrast with other analyses of
variance, because I had only one dependent variable and one

independent variable for testing my first two hypotheses. The

45

purpose of an ANOVA test is to explain the variance of a dependent
variable (in this case, accurate relative pronoun selection, as
measured by the pretest and post-tests) in terms of one
independent variable (the kind of feedback received) with three or
more levels (treatment groups). An ANOVA determines whether the
variation between these groups outweighs within-group variation. I
ran four separate procedures, as noted in table 9, below. As with the
chi-square, a one-way ANOVA procedure can only spot differences
between the groups. It cannot report exactly where such differences
lie; a post-hoc multiple-range test (such as the Tukey test) is

required to pinpoint the source of differences.

Table 9. Kinds of ANOVA procedures run.

 

Pre to Imm. Post-test Recognition
Pre to Imm. Post-test Production
Pre to Delayed Post-test Recognition
Pre to Delayed Post-test Production

 

In each test, a single score was given to each subject: in each
section of the two test sections (recognition and production) all
subjects’ pre-test scores were subtracted from their post-test scores
(either immediate or delayed, depending on the test). For example, if
a subject scored 4 points on the recognition section of the pre-test,
then achieved 7 points in the same section on the immediate post-
test, her/ his improvement score would be 3 from pretest to

immediate post-test in the recognition section. The numbers, means,

46

and standard deviations of the results are outlined in tables 10, 1 1,
12, and 13, below. A column graph, which can be viewed in figure 5,
makes the differences more easily detectable. The treatment groups
make up the X-axis; the Y-axis consists of score improvement in the
recognition and production sections between two tests: pretest to
immediate post-test (P-l) or pretest to delayed post-test (P-D).

Table 10. Score improvement in the recognition section, pretest to
immediate post-test.

 

 

Group N Mean StDev
Control 14 0.071 1.730
Metalinguistic 18 1.000 1.749
Right/W rong 15 1.400 1.844
Reflection 16 1.187 1.759

 

Table 11. Score improvement in the production section, pretest to
immediate post-test.

 

 

Group N Mean StDev
Control 14 1.429 3.204
Metalinguistic 18 2.889 2.111
Right/W rong 15 2.533 1.995
Reflection 16 2.250 2.840

 

Table 12. Score improvement in the recognition section, pretest to delayed

 

 

post-test.

Group N Mean StDev
Control 14 0.714 1.590
Metalinguistic 18 1.833 1.855
Right/W rong 15 1.800 1.897
Reflection 16 1.688 1.740

 

47

Table 13. Score improvement in the production section, pre—test to delayed
post-test.

 

Group N Mean StDev
Control 14 2.214 2.887
Metalinguistic 18 2.278 2.244
Right/Wrong 15 1.600 1.639
Reflection 16 1.312 2.845

 

_ Control Mx ' R/W _ Reflection

Figure 5. Improvement in test scores between groups and between tests.
(Mx=Metalinguistic, R / W=Right/ Wrong).

Although some post-treatment improvement took place in all
groups (even control), none of the tests showed any significant
variation between groups. In the following chapter, I shall explore
potential procedural limitations that may have influenced the results,
as well as suggestions for future research.

W

As previously stated, the pretest and post-test surveys

provided valuable information about subjects’ attitudes toward using

the computer, both in general and within a CALL context. The

48

survey forms, which can be found in Appendices D and E,
predominately used a five-point ordinal scale to determine the
attitudes of subjects on a variety of related topics. A lower score
indicates agreement, a higher score disagreement. The pre-test
survey assessed initial attitudes about computers and their
applications in the classroom, whereas the post-test survey examined
subjects’ own assessment of their attitudes toward the experiment
itself.

The control group did not receive a post-test survey, as they
did not undergo treatment, and their pre—test surveys were
administered solely to collect biodata. Each survey contained
approximately 12-14 questions, but due to flaws in the design of
these questionnaires, I could obtain few reliable results.

Despite this limitation, subjects gave interesting responses to a
question on the post-test survey which asked them to assess their
own post-treatment changes in attitude: “My attitudes toward using
a computer in the second language classroom have improved since
this study began.” All reported slightly improved attitudes, as
outlined in table 14 (with a score of 3 indicating a neutral stance, and
a lower score indicating improved attitude). In addition, the range of

responses that participants gave is also included.

49

Table 14. Post-treatment attitude improvement among groups.

 

 

Group Impr. Attitude Range
Metalinguistic 2.22 1-3
Right/W rong 2.27 1-4
Reflection 2.38 1-3

 

Although the mean scores on this question were roughly
equivalent, the range of given responses demonstrated that the
right/wrong group was the only group which contained subjects who
disliked the courseware.

Regardless of whether subjects actually gain additional
linguistic competence from participating in the treatment sessions,
monitoring their pemeiyed improvement is also a valuable resource
in determining the effectiveness of any courseware package. The
perception of improvement and a belief in the effectiveness of the
software, in addition to enjoyment, can be a powerful motivational
factor in continuing to use it.

As such, the next question addresses subjects’ perceptions of
their own improvement and current competency specifically in the
area of German relative clauses. For, in addition to the enjoyment of
and actual linguistic gain from the courseware, its effectiveness
should also be determined by the perceived gain. If students feel
that they have not advanced in some way after undergoing
treatment, what will prevent them from losing interest in the

software? Listed under question #3 on the post-test survey form,

50

subjects were asked to respond to the statement, “My mastery of
German relative clauses has greatly improved since this study

began.” The results are listed below in table 15.

Table 15. Subjects’ self-assessed improvement in relative pronoun selection.

 

 

Group Perceived Benefit Range
Metalinguistic 2.17 1-3
Right /W ron g 2.87 2-5
Reflection 2.38 2-4

 

The difficulties experienced by the right/wrong group also
seem apparent in this question. For not only is their mean response
significantly higher (more negative) than the other two treatment
groups, a few subjects flatly disagreed with the assertion that any
improvement took place at all, as is clear by the R/W group’s range
of responses. By the same token, the reflection group’s response
range also included dissenting opinions, although all groups’ mean
scores were essentially positive. A column graph charting subject
responses to both questions is located in figure 6, below. The graph

numbers have been reversed to facilitate viewing.

51

 

 

l

 

l
FIESTAmtude l
Lljgceived Benefit l

. l
Mx R/W Refl. l
l

 

 

i._. ._._._.- 7*. _-7

Figure 6. Post-treatment attitude and perceived benefit on an ordinal scale of
1 to 5. (Mx=Metalinguistic group, R/ W=Right/ Wrong group).

W

In the previous two portions, I have cited the results of both
the testing and survey data, to see how the treatment affected the
subjects’ performance and attitudes. Although the reflection data
does not pertain directly to my three hypotheses, an examination of
the data does reveal some interesting details about subjects’ thought
processes during treatment (possibly pointing to aspects about the
courseware that could be improved), as well as their feelings about
the courseware, thereby indirectly pertaining to my hypotheses.

Despite this, the software data from the reflection group itself
will only be briefly noted here; the act of reflection and production in
itself is the most important aspect for the purposes of this study.
However, the students were told that the software data would be
collected, with the anticipation that subjects would exercise greater

care in the formulation of their written responses.

52

In the reflection treatment, students were instructed to list the
item number of the problem they were writing about, and to give a
clear, concise possible reason for their incorrect answer. Although
most complied with these instructions, some item numbers were
missing, and a few wrote explanations that were either too brief or
so convoluted that they were impossible to decipher. A few subjects
wrote nothing at all.

In the first treatment session, 17 out of 21 subjects correctly
identified at least one of their errors in the reflection portion. In the
second, 1 1 of 19 correctly identified at least one error. In the first
treatment session, only 8 subjects responded with “I don’t know,”
and in the second session, this number dropped to 5. Several of the
subjects correctly identified at least half of their errors, possibly
selecting their responses less impulsively in the future.

5111111113131

To briefly summarize the results, several statistical analyses
were carried out in order to chart between-group improvement on
post-treatment test scores. In the production aspect, from pretest
to immediate post-test, there were significantly more developers in
the groups that received metalinguistic feedback as opposed to those
that did not, as revealed by a chi-square test. I also ran several
ANOVA tests to determine the difference in mean improvement
between groups, but there appeared to be no significant difference.

While the mean attitude improvement between groups appeared to

53

be roughly similar, the right/wrong group contained participants

who believed they had not benefited at all from the software.

Despite this, all experimental groups’ mean attitude improvement, as

well as perceived benefit, were positively slanted.

Table 16. Summary of the results in regard to the three hypotheses.

 

. r

 

" Hypothesis 1. Students who

utilize a computerized program
focusing on relative pronouns,
and featuring metalinguistic
feedback, will score significantly
higher on a post-test which tests

, them on these forms than

students who only receive

- right/wrong feedback and are not

instructed to reflect on their
responses.

Hypothesis 2. Students who only
receive indications of

correct [incorrect responses from
the computer and who are
required to reflect on their
incorrect responses, so that they
may determine for them selves
why the responses are incorrect,
and then receive metalinguistic
feedback, will score higher than
subj ects who only receive
metalinguistic feedback.
Hypothesis 3. $qu ects in the two
groups who receive metalinguistic
feedback will report both greater
enjoyment and greater perceived
benefit of their CALL experience
on a standardized post-test survey
than those students in the

' right/wrong group.

 

Result: Not supported.

Only one of the statistical
measures carried out reported
any significant difference
between those groups which
received m etalinguistic feedback
and those which did not. In
general, there was no signifieant
difference between groups in
regard to degree of improvement.
Result: N of supported.

In fact, those subjects in the
reflection group did not even
show as great of a tendency
toward improvement as the

right/wrong group.

Result: Supported, but results are
difficult to generalize. -

Subjects who received
metalinguistic feedback
demonstrated a greater attitude
improvement than those in the
right/wrong group, but these
results are a) quantitative and b)
from a flawed survey design.

 

 

Chapter 5

DISCUSSION

This chapter is divided into four sections. The first two
sections address hypotheses 1 and 2. In section 1, I discuss the
implications of the pretest and post-test results for these
hypotheses. Although the data obtained did not support them, I did
gain valuable insight into the manner in which task type can
influence acquisition; it is clear from the results that students who
can practice the recognition of a target structure are then able to
more accurately produce that structure. In addition, subjects seemed
able to generalize their knowledge to easier clause types, which lends
support to the Accessibility Hierarchy. However, to compile further
data which might support the original hypotheses, I would
recommend future research be carried out in this study, redesigned
to constrain variables which I had not taken into account. I discuss
these factors as they relate to hypotheses 1 and 2 in the second
section. Sections 3 and 4 of this chapter address hypothesis 3. In
the third section, I examine the results of the survey data within the
context of improvement in attitude and perceived benefit, to see if
hypothesis 3 is supported. The fourth section, which discusses the

reflection data, also applies to hypothesis 3, albeit to a lesser extent.

54

55

oi'D , ,.,.._. 'o-- :-

Despite the general improvement of all groups, hypothesis 1
does not seem to be supported by the results. With the exception of
the control group, which lagged behind the others, none of the groups
experienced a significant between-group difference in improvement.
The significant result of the production chi-square from pretest to
post-test does little to establish whether the metalinguistic feedback
brought about greater improvement in production; the right/wrong
group also had a high number of deveIOpers (60% of the total N, as
indicated in table 6). Thus, for the groups which received
metalinguistic feedback, the salience of the feedback did not lead to
improvement that exceeded that of the right/wrong group. However,
some valuable observations can be drawn from the results.

As is clear from the post-test results, virtually all subjects
displayed some form of improvement after treatment (as
demonstrated in figure 3). Even the scores of the control group
indicated a slight trend toward improvement, which may
demonstrate that exposure merely through testing can lead to some
marginal improvement in relative pronoun selection. This
phenomenon, called the “testing effect,” refers to the process by
which testing alone can produce learning (Kuo & Hirschman, 1997).
However, this had little effect on the overall results; as the gains of
the experimental groups were generally far above those of the

control group, the treatments clearly facilitated improvement.

56

Although none of the groups performed significantly better than the
other groups on the post-tests (as determined by the ANOVA), every
group demonstrated a trend toward improvement (consult tables 8-
1 1 for means and standard deviations). It would appear from these
analyses that the kind of feedback subjects received in the study
played only a marginal role in their improvement in relative
pronoun selection.

However, as indicated by the significant result on the
production chi-square test from pretest to immediate post-test,
there was a much larger number of developers in the groups which
received metalinguistic feedback (metalinguistic and reflection
groups) as opposed to those which did not (as shown in Appendix G),
although these groups were not given production-oriented tasks
during treatment. However, the right/wrong group also received no
production tasks in treatment, and as preViously stated, they also
had a large percentage of developers. With these factors in mind, it
may be inferred that subjects did generalize their recognition
experience in the two treatment sessions to another context, namely
the cloze sections of the post-tests. Despite Sharwood Smith’s (1991)
notion that the internalization of negative evidence does not
necessarin lead to the correct production of that target form, the
negative feedback clearly helped many subjects improve their
production ability of relative pronouns in this case. However, it is

also important to note the subjects’ production scores on the pre-test

57

were much lower than their recognition scores, so the production
scores had considerably more room for improvement.

As for the gains of the reflection group, the results of the study
do not support hypothesis 2, either. Giving subjects time to reflect
seemingly did nothing to improve their relative pronoun selection
capabilities beyond that of the metalinguistic group; in fact, the
reflection group was outperformed by the two other experimental
groups in almost every area. It was anticipated that giving subjects
an opportunity to reflect would help lessen the increased capacity for
error that accompanies a high level of awareness (Tomlin & Villa,
1994). However, perhaps asking subjects to reflect after entering a
response simply did not have the desired effect of thoughtful
deliberation about their errors. It may have been more fruitful to
require reflection More entering any response at all, as was the case
in Meredith’s (1978) study.

Although Keenan & Comrie’s (1977) Accessibility Hierarchy
only indirectly pertains to my hypotheses, I tallied the percentages
of correctly answered items for each clause type in both of the post-
tests. The results can be found in tables 17, 18, and 19.

Key: S=Subject clause, DO=Direct Object clause, IO=Indirect Object clause,
OP=Object of a Preposition clause

58

Table 17. Percentages of correctly answered items in the metalinguistic group.

 

S DO IO OP
1mm . Post 82% 44% n / a 64%
Delayed Post 78% 69% 67% 66%

 

 

Table 18. Percentages of correctly answered items in the right / wrong group.

 

S DO IO OP
Im m. Post 76% 40% n / a 59%
Delayed Post 73% 65% 67% 54%

 

 

Table 19. Percentages of correctly answered items in the reflection group.

 

5 DO IO OP
Im m . Post 76% 53% n / a 67%
Delayed Post 69% 67% 69% 65%

 

 

Although most of the treatment items were object of a
preposition clause types, it appears from the above tables that
subjects in all treatment groups generalized what they learned from
the treatments to the easier clause formss . The only clause type that
differs from the Accessibility Hierarchy in this case are direct object
clauses. One possible reason for this stems from the nature of these
clauses. While direct object relative pronouns in German generally
require the accusative, both post-tests each had an item which
contained the verb “helfen” (to help); although these are technically
direct object clauses (and counted as such in the above tables), this

particular verb requires the dative. Most of the subjects missed

 

5 Since testing the accuracy of the Accessibility Hierarchy was not originally part of this study, the
distribution of clause types were uneven, and any such claims are thus impossible to
demonstrate.

59

these two items, even though this verb is also found in both
treatment sessions.

The total percentage of correctly answered items for all groups
on both post-tests can be found in figure 7, below. This clearly
shows a far more even distribution of clause types than in either the

treatment sessions or the post-tests.

 

Object of a
Prep. Subject

  
  

24% 28%

 

 

Object Object
26% 22%

 

 

 

Figure 7. Clause type percentages of the total correct responses on the
immediate and delayed post—tests.

WWW

In the following paragraphs, 1 make recommendations for
future research, discussing procedural limitations which may have
impeded post-treatment improvement for the groups which received
metalinguistic feedback. A new study, designed to take these
limitations into account, may help to determine if those factors had
an effect on the results in this study.

When examining subjects’ reflection data, it became clear that a

few subjects experienced problems understanding the grammatical

60

explanations from the metalinguistic feedback. As a researcher, I
should not have assumed that students’ are 1) knowledgeable about
grammar to begin with, and 2) familiar with grammatical
terminology. As Garrett (140) points out, it is often assumed that
“knowing grammar” means a familiarity with the labels that linguists
attach to grammatical structures. Garrett also asserts that the
knowledge of these terms will not in itself lead to acquisition.
Although I agree that this assertion is essentially true, I also
acknowledge that subjects needed a base knowledge of both the
grammatical content of relative clauses, as well as the terminology
used in the treatment, to fully understand the metalinguistic
feedback.

As for the first concern of grammatical knowledge, I knew that
the grammar of relative clauses had been taught in subjects’ German
classes. The implementation of additional instructional materials in
the treatment, in order to help subjects attain a base grammatical
knowledge of relative clauses, would have tainted the collected data.
The research instrument had not been designed to test the effect of
instruction, and I would have had no means of determining if
improvement had been a result of the negative feedback or the
additional instruction. Although such a pretreatment lesson would
be highly desirable in a non-experimental context, it was unfeasible
for this study.

However, in regard to the use of grammatical terminology, it

61

would have been useful to include a separate help section within the
software to identify these terms. This could be accomplished in two
different ways, each with its own set of advantages and
disadvantages. When looking over the metalinguistic feedback for a
particular response, hypertext buttons could be added so that when
the term is clicked on, a brief definition appears in a pop-up window.
In addition, an introductory tutorial would serve as a valuable
resource not only for providing definitions, but also examples of
what these terms represent in a given context. Hypertext pop—up
windows provide explicit information, including, but not limited to,
such items as definitions of grammatical terms, examples, and hints
about the current question. Tutorials can provide all of the above as
a single unit, although much of this information may be less relevant
to the task at hand. The advantages and disadvantages of both
approaches are listed below.

A tutorial would certainly be easy to implement, but subjects
may have a tendency to skip over this introduction, possibly with no
means of returning to it for reference when actually working through
the exercises, when the information is particularly pertinent. Pop-up
windows would eliminate this concern, but due to lack of space on
the screen, such windows may not elaborate enough to be truly
useful to the learner. Also, the addition of such hypertext links for
each and every line of feedback would be a daunting and time-

consuming task for the software designer. Perhaps an ever-present

 

62

button that, upon clicking, takes the user to a help screen would be a
solution that offers the best of these two dimensions.

Another pitfall to be avoided by future research is the
limitation of two treatment sessions for all groups. As Hatch &
Lazarton (1991) state, “We know that even though we offer students
instruction and they are able to perform well during the instruction
period, they may not truly internalize the material unless it is
recycled over a fairly long period of time” (pp. 93-4). It would have
been highly desirable in this study to have at least three treatment
sessions, but unfortunately, this was not possible; since all testing
and treatment was completed during the subjects’ normal German
classes, only a certain amount of time could be devoted to the
experiment. While the project could have been designed to avoid
this problem by scheduling testing and treatment outside of class
time, the participant count would almost certainly have suffered.
More treatment sessions could possibly contribute to more variation
between treatment groups, as each kind of treatment would have the
possibility of reaching the “ceiling” of its natural effectiveness.

Also, despite the fact that the uncontrolled variable of intact
classes did not create a major problem in the study, students did
score higher on the pre-test in some sections than in others (refer to
table 5 in the previous chapter). Although it was unavoidable in this

case, it is recommended that future samples be truly random.

63
WW

As demonstrated in tables 12 and 13 (as well as figure 4) of
the previous chapter, all experimental groups demonstrated a slight
post-treatment improvement in attitude and perceived benefit
toward CALL. Chapelle and Jamieson (1989), citing various early
studies, found that “questionnaire results typically - but not always -
indicate favorable attitudes toward CALL” (p. 52). This tendency
could indicate why subjects demonstrated generally positive
attitudes toward the treatments.

However, it is important to keep in mind that the survey
results are qualitative, and such qualitative data make it difficult to
support empirical hypotheses, although the knowledge gained is still
valuable. Also, due to the flawed design of the survey materials,
most of the data could not be salvaged; it is nearly impossible to
generalize any results obtained from only two questions.

Attitude Improvement

As indicated by table 12, the mean attitude improvement of all
experimental groups was nearly equivalent (varying by less than
1/5 of one point), and all were generally positive. The right/wrong
group’s score was only slightly higher (more negative) than that of
the two metalinguistic groups. The right/wrong group’s range of
responses seems consistent with this result, as it was the only group
containing subjects who disagreed with the statement that they had

improved, although no subjects strongly disagreed (as indicated by a

64

S on the ordinal scale). Some subject’s distaste of the software could
be attributed to the fact that they, unlike the other two experimental
groups, had no tools by which to ascertain the correct answer, other
than haphazardly pressing buttons until they uncovered the correct
response.

Perceived Benefit

Regardless of whether subjects actually experienced
improvement from participating in the treatment sessions,
monitoring their mm improvement was also a valuable
resource in determining the effectiveness of the CALL software. The
perception of improvement and a belief in the effectiveness of the
software, in addition to enjoyment, can be a powerful motivational
factor in continuing to use it (Higgins, 1988).

In the perceived benefit section, there was slightly greater
variation between groups than in attitude improvement. Again, the
right/wrong group lagged behind the other two experimental groups.
A few subjects even strongly disagreed with the survey statement
that they had improved at all. Once more, this is likely due to the
fact that these subjects were given no amount of aid in the
recognition of their own errors. And thus, having no means of
understanding why a given response was ungrammatical, they were
left with only a single option when making an error: to randomly

press buttons until they had stumbled onto the correct response.

65

WW

Upon examination of the reflection data, it was clear that while
some subjects had been haphazardly pressing buttons with no
reasoning behind their choices, others found the reflection exercise a
valuable tool in discovering their own mistakes. Despite this, a few
subjects used the reflection portion of the software to vent their
frustrations about the program and about relative pronoun selection
in general. These emotional proclamations, such as, “Confuzzled...”
and, “Basically, I just suck at German,” are perhaps more telling
about the subjects’ difficulties with the treatment than the survey
forms, which showed that most of the subjects appeared to find the
treatment sessions an enjoyable and beneficial experience. This is
perhaps an indication of the more active role that the reflection
group was required to take in its own feedback; this increased level
of participation also meant that more time was needed for them to
complete the treatment. It is therefore possible that the subjects in
this group simply felt overwhelmed by the task, a possibility that
will have to be addressed in future versions of the software.

Nonetheless, the subjects did seem to grow more confident as
their experience with the program increased, as indicated by the
fewer “I don’t know”-type responses in the second treatment session.
It is uncertain, however, whether their ease can be attributed to
more experience with the program (technological expertise) or to

experience with relative pronoun selection (grammatical knowledge).

CONCLUSIONS
In conclusion, I have returned to the specific research

questions outlined in chapter 1. The questions and responses are

listed in table 20, below.

Table 20. Research questions and responses.

 

‘ 1. To what extent does Response: None of the research

the reception of findings indicated that supplying
-‘ metalinguistic feedback subj ects with metalinguistic feedback
influence subject boosted their performance on a post-
performance? test over that of the non-
metalinguistic groups.

l

2. How does allowing Response: Giving subject reflection
the subjects extra time time after responding seem ed to have
to reflect upon their no effect. Perhaps forcing subjects to

incorrect responses wait before responding would have
affect overall been more effective.

performance?

3. Is there a Response: The groups which received

relationship between metalinguistic feedback felt they had
‘ the kind of feedback benefitted more from the treatment

received and the than subjects in the right/wrong
subjects’ perceived group. However, these results are
benefit from the CALL difficult to generalize due to flaws in
exercise? the survey design.

4. Is there a Response: Subjects who received

relationship between m etalinguistic feedback generally
the kind of feedback reported greater attitude

 

 

 

received and the improvement, although the same
sutj ects’ overall limitation as in #3 applies here, also.
enjoyment of the

CALL exercise?

 

However, due to possible procedural flaws, other research

questions come to light, which may aid in explaining the subjects’

66

67

lack of anticipated improvement:

0 How might the addition of more elaborate means of feedback
(i.e. help sections, tutorials, etc.) contribute to higher levels of
improvement?

0 Would additional treatment sessions help establish more score
variation between groups?

0 How does the kind of task contained in the CALL courseware
influence subjects’ acquisition of the target structure?

0 How does the proportion of relative clause types affect
subjects’ overall improvement?

Dealing with the above questions could serve to determine if
metalinguistic feedback and additional reflection can truly lead to
improvement in relative pronoun selection. This study clearly merits
further pursuit, but with larger numbers so that the results can be
more easily generalized.

In addition, while conducting the study, I made two discoveries
which, although only indirectly related to my hypotheses, could
easily be utilized in future such studies. First, through measuring the
percentage of correct answers for each clause type on both the
immediate and delayed post-tests, I have found the Accessibility
Hierarchy to be essentially accurate. With more careful manipulation
of the clause types during treatment, a similar result could be more
easily generalized. Also, I found that subjects who only have
exposure to recognition tasks during treatment can adapt this

knowledge, using it in more a productive manner later.

68

These are only two of the many additional items that arose
when trying to answer my four specific research questions. By
pursuing this study further, I could incorporate these additional
questions in determining not only whether the kind of feedback
leads to greater improvement, but also how learners use the
feedback they are given, possibly pointing to learning strategies that

are employed by students in their foreign language courses.

APPENDICES

APPENDIX A

APPENDIX A

W
Please choose the correct response:
1. Das ist die Frau, ich gestern geholfen habe.

a) die b) das c) der d) dem

2. lch habe dem Hund, vorgestern hier war, einen Ball gegeben.
a) der b) die c) dem d) den

3. Das ist die Firma, bei ich seitfiinf Jahren arbeite.
a) dam b) der c) die d) denen

4. Mattias und Kirsten sind die Freunde, mit ich am liebsten zusammen bin.
a) der b) die 0) denen d) dem

5. DeutschistderKurs,fur ichoftlernen mul3.
a) dem b) das c) der d) den

6. lch bin ein Mann, Sport nichtgerntreibt.
a) den b) der c) dem d) die

7. Also bist du die Frau, von ich so oft gehbrt habe!
a) der b) die c) dem d) das

8. Das ist die Stelle, um ich mich bewerbe.
a) die b) das c) der d) den

9. Maine Firma ist das Gebaude, neben dem Postamt steht.
a) die b) der c) dem d) das

10. Deine Freundin ist die Frau, ilber ich so viel gehdrt habe.
a) der b) die 0) den d) dem

69

70

APPENDIX A

Please enter the correct relative pronoun:

Mattias: Hallo Kirsten! Bist du die Frau, die Stelle als Lehrerin
bekommen hat?

Klrsten: Ja! Herr Adler ist mein Chef. Er ist der Professor, bei ich
Ietztes Jahr studiert habe.

Mattias: lch finde, er ist etwas streng. Er war es, mir Ietztes
Semester eine ‘F' in deutsch gegeben hat.

Kirsten: Also bist du es, er die ‘F’ gab. lch dachte, Susanne war
die Studentin, durchgefallen ist.

Mattias: Nein, das war ich. Aber ich bin nicht der Einzige, ihn
nicht mag. Er ist ein Professor, iiber man immer Horrorgeschichten
hon.

Kirsten: Das ist nicht fair von dir. Letztes Semester hast du das Buch nicht
gekauft, du fur den Kurs lesen muBtest.

Mattlas: Der Kurs, fiir ich kein Buch gekauft habe, war Chemie.

lch habe alles flir meinen Deutschkurs gelesen.
Klrsten: Aber trotzdem hast du nicht viel gelernt. Deutsch ist ein Fach, in
man viel arbeiten muBl

APPENDIX B

APPENDIX B

E | |' il l I' | E I-
Part 1 (10 Points). Please choose the correct response:

1. Julia ist die Studentin, ich mitderHausaufgabe helie.

a) der b) das 0) die (1) dem

2. lch habe meinem Sohn, so brav war, ein Spielzeug gekauft.
a) dam b) die 0) der d) den

3. Das istdas Haus, in sie wohnt.
a) denen b) der c) die d) dem

4. Otto und Maria sind die Leute, mit wir immer Karten Spielen.
a) der b) denen c) die d) dem

5. Katie ist die Freundin, filr du das Buch gekauft hast, oder?
a) dem b) die 0) der d) den

6. Thomas isteinjunger Mann, gern Bier trinkt.
a) den b) die 0) dem d) der

7. Siemens ist die Firma, bei Karl sechs Jahre lang gearbeitet hat.
a) der b) die c) dem d) das

8. Theater isteine Karriere, an ich immer gedacht habe.
a) die b) das c) der d) den

9. Meine Firma ist das Gebaude, neben dem Postamt steht.
a) die b) der 0) dem d) das

10. Das Auto ist ein Geschenk, fiber ich michfreue.
a) der b) das c) den d) dem

71

72

APPENDIX B

Part 2 (10 Points). Karl and Thomas are kicking around a soccer ball in the
park, when they see someone they know. Please enter the correct relative
pronoun:

Karl: He, Thomas, du tragst tolle Schuhe! Wo hast du sie gekauft?

Thomas: Dies sind die Schuhe, ich von meinen Eltem zum
Geburtstag bekommen habe.

Karl: Cool. Sind sie die Schuhe, von Michael Jordan immer im
Fernsehen spricht?

Thomas: Ja. Er ist der einzige Mann auf der Welt, boner springen
kann als ich.

Karl: Ach, wirklich? Dann milssen sie tolle Schuhe sein. He, sag mal, ist das
nicht das Madchen, mit wir gestern im gesprochen haben?
Thomas: lch glaube schon. Sie heiBt Katja, oder? Ja, sie ist bestimmt das
Madchen, gestern hier war. Hallo, Katja! (Sie kommt).

Katja: Hallo, Jungsl

Karl: Sag mal, bist du nicht die Frau, mit wir gestern gesprochen
haben?

Katja: Doch! Der Park ist eine Ort, in ich mich immer wohl fiihle.
Thomas: Warum muBtest du gestern so friih weggehen?

Kaila: lch muBte fur melne Mutter einkaufen, im Moment krank ist.
Karl: Was hat sie denn?

Katja: Sie hat die Grippe. Es ist eine Krankheit, nicht leicht
weggeht.

Thomas: Schade. Die Grippe ist eine Krankheit, von man viel

hort, aber ich habe sie nie gehabt.
Katja: Du hast Gltick gehabtl

APPENDIX C

APPENDIX C

Belatiyeclauaenelaxedflestztest
Part 1 (10 Points). Please choose the correct response:
1. Frau Schmidt ist die Professorin, immer so viele Hausaufgaben gibt.
a) der b) das c) dem (1) die

2. Das istdas Kind, du den Ball gegeben hast.
a) dem b) die 0) der d) den

3. Der Koffer, in ich meine Andenken von Berlin habe, ist im Keller.
a) denen b) dem 0) die d) der

4. Gabi und Klaus sind die Freunde, von du immersprichst.
a) der b) denen 0) die d) dem

5. Annas Bruder, fur Anna einen Bierkrug gekauft hat, heiiit Jeff.
a) dam b) die c) der d) den

6. Dieter istein Sportler, am liebsten FuBball spielt.
a) der b) die 0) dem d) den

7. Das ist das Restaurant, in ich mein erstes ‘Hamburger’ gegessen habe.
a) der b) die 0) dem d) das

8. Katrin ist die Frau, an ich immerwiederdenke.
a) die b) das c) der d) den

9. Der Tisch, auf die Teller stehen, ist braun.
a) die b) der c) dem d) das

10. Das Theaterstuck, wir sehen wollten, ist ausverkauft.
a) der b) das c) den d) dem

73

74

APPENDIX C

Part 2 (10 Points).

Gabl: Hallo, Klaus. Wie geht’s dir?

Klaus: Gut. Die Firma, bei ich arbeite. hat vielleicht eine bessere
Stelle fiir mich - als Programmierer.

Gabi: Klaus, das ist die Stelle, du immer wolltest!

Klaus: Ja, aber es gibt ein Problem, mir die Stella kcsten konnte.
Mein Chef, ein Idiot ist, will nicht, daB ich die Stelle bekomme.
Diese Stelle ist eine Gelegenheit, auf ich immer gewartet habe. Hilf
mirl

Gabl: Dein Chef, ich Ietztes Jahr bei einem Projekt geholfen habe,

ist sicher ein dummer Mann. Aber er mag mich. Vielleicht kann ich mit ihm
reden.

Klaus: Ach, Gabi, du bist die tollste Frau, ich jemals kennengelernt
habe! Computerprogrammierer ist ein Beruf, von ich immer
getraumt habe.

Gabi: lch werde morgen mit ihm sprechen. Willst du jetzt ins Kino gehen. Was
war der Film, iiber wir gestern gelesen haben?

Klaus: “SOphies Abenteuer". Das ist der Film, in die Hauptfigur,
Sophie, eine Fleise nach Deutschland macht. Gehen wirl

APPENDIX D

APPENDIX D

W

Please answer the following questions:

1. Briefly describe your prior experience with German (high school, study
abroad, tutoring, etc).

2. How much time do you spend studying German per day?

 

3. Please circle your class standing: Fr. Soph. Junior Senior Grad

4. Please briefly describe your prior experience using computers, particularly in
an educational setting:

5. What is your native language?

 

Please read the following statements, and circle the number that best expresses
your views. ’

Key: 1=strongly agree, 2=agree, 3=neutral, 4=disagree, 5=strongly disagree

6. I feel that computers can be beneficial in learning. 1 2 3 4 5
7. Computer use is essential to my everyday life. 1 2 3 4 5
8. I enjoy using a computer. 1 2 3 4 5
9. Computers are only good for video

games and possibly word processing. 1 2 3 4 5

10. Learning with the aid of a computer is more
beneficial than learning by more traditional methods. 1 2 3 4 5

11. Learning with the aid of a computer is more
enjoyable than learning by more traditional methods. 1 2 3 4 5

12. I dislike using computers because

they're difficultto operate. 1 2 3 4 5
13. Computers are impersonal. 1 2 3 4 5
14. Iuseacomputer at least four times per week. 1 2 3 4 5

75

APPENDIX E

APPENDIX E

W
1. Do you speak any other languages than your native language?
Which?

Please read the following statements, and circle the number that best expresses
your views.

Key: 1=strongly agree, 2=agree, 3=neutral, 4=disagree, 5=strongly disagree

2. I enjoyed working with the computer in this study. 1 2 3 4 5
3. My mastery of German relative clauses has

greatly improved since this study began. 1 2 3 4 5
4. After working in this study, I feel that computer use

should be commonplace in the language classroom. 1 2 3 4 5
5. I think computers could be useful in learning

many different grammatical forms. 1 2 3 4 5

6. My attitudes toward using a computer in the second
language classroom have improved since this study began. 1 2 3 4 5

7. My attitudes toward using computers in general

have improved since this study began. 1 2 3 4 5
8. I found the program we used to be interesting and helpful. 1 2 3 4 5
9. The use of the computer did not much help my

understanding of German relative clauses. 1 2 3 4 5
10. The study has worsened my attitude toward

using a computer in the classroom. 1 2 3 4 5
11. I did not enjoy using a computer in this study. 1 2 3 4 5
12. I consider myself to be a good German student 1 2 3 4 5

76

APPENDIX E

APPENDIX E

Cmsmtlicrm

This study will be testing the role of feedback, to see which kinds of feedback
enhance performance and which do not. As a subject in this study, you will be
required to complete some exercise and testing materials on the subject of
German relative clauses. The data collected will be analyzed as part of the
investigator’s master’s thesis. By signing this form, you consent to participate in
the study, and give your permission for the investigator to use data obtained
from your participation.

Please read over the following items before signing the form.
Be advised that...

- no time outside of your normal class time will be required of you for this
study. In-class time spent on this study will be approximately 2.5 hours.

- you will be identified by your student numbers and by section number
only, which will be known only to the investigator. All information will be
held strlctly confldentlal.

- by signing this consent form, you acknowledge that you are a willing
participant in the experiment and thus give your permission for the
experimenter to use the data collected from your participation.

- participation in the study is entirely voluntary, and that refusal (or
willingness) to take part in the study will In no way affect your grade in
this course. You may choose to discontinue your participation at any
time.

- no beneficial effects of the treatment are guaranteed.

- you will not receive a grade for your performance on these tests. If you
wish to know how you scored on the tests, please contact the investigator
at one of the numbers below.

I fully understand the conditions of my involvement in the study, and I wish to
participate.

Signature:

 

Date:

 

If you have any questions, comments, or concerns, contact Daniel Park at
353-7619 (office), 346-4907 (home), or e-mail him at: parkdani@pilot.msu.edu

77

APPENDIX G

 

APPENDIX G

W
Production: Pre-test - Immediate Post-test
Expected counts are printed below observed counts
Ct =Developers

CZ=Non-Developers

Table 21. Chi-square figures, pretest to immediate post-test.

 

Chi-Sq = 2.865 + 2.292 + 0.236 + 0.189 + 0.817 + 0.653 = 7.052
DF :2, P-Value = 0.029

78

APPENDIX H

APPENDIX H

Metalinguistinieedhack
Eirstlreatmentfiessicn

Before the actual feedback, a random response generator gives the subject an
indication that the response was incorrect, i.e. “I'm sorry, that is incorrect". After
the metalinguistic feedback, the random generator gives the subject some
indication to try again, i.e. “Please try again.” If the response was correct, the
random generator provided a congratulatory message and asked the subject to
proceed.

Key: S=Subject Clause, DO=Direct Object Clause, l0=lndirect Object Clause,
0 P=Object of a Preposition Clause.

1. let das die Frau, __ wir gestern gesehen haben? (DO)
(Is that the woman we saw yesterday?)
a. der - “Sehen” is not a verb which requires dative.
b. dem - The antecedent “Frau” is feminine, and “sehen” is not a verb which
requires dative.
c. die - (correct response)
d. das - The antecedent “Frau” is feminine, not neuter.

2. Der Mann, __ du gestern kennengelernt hast, is Julias Freund. (DO)
(The man you met yesterday is Julia's boyfriend.)

a. das - The antecedent “Mann" is masculine, not neuter

b. den - (correct response)

0. der - "Du" is nominative, because it is the subject within the relative
clause. The relative pronoun, as the direct object, should be in the
accusative.

d. dem - The relative pronoun, as the direct object, should be in the
accusative.

3. Das ist der Bus, auf __ wir seit zwei Stunden gewartet haben. Endlich
kommt erl (OP)
(That’s the bus for which we’ve been waiting for the past two hours. It’s
finally comingl)
a. dam - Since “dem” indicates the dative case, that would mean that you
were waiting on top of the bus!
b. der - The verb “warten+auf" (to wait for) requires the accusative.
c. das - The antecedent “Bus" is masculine, not neuter.
d. den - (correct response)

79

 

80

APPENDIX H

4. Ania liest das Buch, von _ Kirsten mir so viel erzahlt hat. (OP)

(Anja is reading the book that Kirsten told me so much about.)

a. dem - (correct response)

b. das - The preposition “von” requires the dative.

c. der - Although you’re right that the preposition “von” is dative, the
antecedent “Buch” is neuter, not feminine.

d. den - The antecedent “Buch” is neuter, not masculine. Also, the
preposition “von” requires the dative.

5. “Shine" ist der Film, in _ Geoffrey Flush die Holle des Pianisten David
Helfgott spielt. (OP)
(“‘Shine" is the film in which Geoffrey Rush plays the role of pianist David
Helfgott.)
a. den - Although “in” is a two-case preposition, in this instance the dative is
required because there is no indication of motion.
b. der - Although you are correct that the relative pronoun should be dative,
I'm afraid that the antecedent “Film” is masculine, not feminine.
c. dem - (correct response)
d. denen - Although you are correct that the relative pronoun requires the
dative, I’m afraid that the antecedent “der Film” is singular, not plural.

6. Karin ist die Studentin, __ ich immer mit den Hausaufgaben helfe. (DO)

(Karin is student whom I’m always helping with homework.)

a. der - (correct response)

b. die - You're probably forgetting that “helfen” is a verb which requires
dative.

c. dem - Although you’re correct that the relative pronoun should require
the dative, the antecedent “Frau” is feminine.

d. denen - Although you’re correct that the relative pronoun should be
dative, the antecedent “Frau” is singular, not plural.

7. Deutsch ist das Schulfach, mit __ ich die meisten Probleme habe. (OP)
(German is the subject with which I have the most problems.)

a. denen - While it is true that "mit" is a preposition that requires dative, the
antecedent "Schulfach" is singular, not plural. "Mit denen" would indicate
the plural.

b. dem - (correct response)

0. das - The preposition “mit" requires the dative.

cl. den - That's incorrect, for two reasons: First, the preposition “mit" requires
the dative. Second, the antecedent “Schulfach” is neuter, not masculine.

81

APPENDIX H

8. Disses Fahrrad was das Geschenk, fiber __ ich mich als Kind am meisten
gefreut habe. (OP)
(This bicycle was the gift that made me the most happy as a child.)
a. die - The antecedent “Geschenk” is neuter, not feminine.
b. den - Although it"s true that the verb "sich freuen + fiber" requires the
accusative, the antecedent "Geschenk" is neuter, not masculine.
c. dem - The verb "sich freuen + fiber" (to be happy about) requires the
accusative case.
d. das - (correct response)

9. Dr. Lovik, __ mein Lieblingsprofessor ist, hat ein neues Buch geschrieben.

(8)

(Dr. Lovik, who is my favorite professor, wrote a new book.)

a. dam - Since the relative pronoun is the subject, the nominative is
required, not dative. Please try again.

b. das - The antecedent "Professor" is masculine, not neuter.

c. den - Since the relative pronoun is the subject, the nominative is
required, not accusative.

d. der - (correct response)

10. Die Frau, __ mir disses Hemd geschenkt hat, ist meine Freundin Ute. (S)
(The woman who gave me this shirt is my girlfriend Ute.)

a. der - The relative pronoun is the subject of the relative clause, and is
therefore nominative.

b. die - (correct response)

0. dem - That is incorrect, for two reasons. First, the relative pronoun is the
subject of the relative clause, and is therefore nominative. Second, the
antecedent "Frau" is feminine.

d. das - Your response is incorrect, for two reasons. First, the relative
pronoun is the subject of the relative clause, and is therefore nominative.
Second, "Frau" is feminine, not neuter.

11. Herr Young ist ein Mann, _ immer punktlich ist. (S)
(Mr. Young is a man who is always on time.)

a. der - (correct response)

b. denen - First, the relative pronoun is the subject, and therefore requires
the nominative, not dative. Second, the antecedent "Mann" is singular,
not plural.

c. dem - The relative pronoun is the subject, and therefore requires the
nominative, not dative.

d. den - The relative pronoun is the subject, and therefore requires the
nominative, not accusative.

12.

13.

14.

15.

82

APPENDIX H

Das Kind, _ letzte Woche hier war, ist meine Nichte. (S)
(The child who was here last week is my niece.)

a. dem - The relative pronoun is the subject, and therefore requires the
nominative, not dative.

b. der - The antecedent "Kind" is neuter, not masculine.

c. das - (correct response)

(1. die - The antecedent "Kind" is neuter, not feminine.

Der Tisch, auf __ die Teller stehen, ist wackelig. (OP)
(The table, upon which the plates are set, is wobbly.)

a. den - That would mean that the dishes climbed up onto the table by
themselves! Since there is no motion, the verb "stehen + auf" requires
the dative.

b. das - The antecedent "Tisch" is masculine, not neuter.

c. der - The verb "stehen + auf" requires the dative, not the nominative.

d. dem - (correct response)

Main BMW ist ein Auto, __ sehr teuer ist. (8)
(My BMW is a car that is very expensive.)

a. das - (correct response)

b. der - The antecedent "Auto" is neuter, not masculine.

c. den - First, the relative pronoun is the subject, and is therefore
nominative, not accusative. Second, the antecedent "Auto" is neuter, not
masculine.

d. dem - The relative pronoun is the subject, and is therefore nominative,
not dative.

Das Madchen, __ ich gestern gekilfit habe, ist Roberts Freundin. (DO)
(The girl whom I kissed yesterday is Robert’s girlfriend.)

a. die - This one was tough. Although the antecedent "Madchen" refers to a
female, the -chen ending makes it neuter and not feminine.

b. der - The antecedent "Madchen" ist neuter, not masculine.

c. das - (correct response)

d. dem - Since the relative pronoun is the direct object within the relative
clause, the accusative is needed in this instance.

 

16.

17.

83

APPENDIX H

Die Stadt, aus _ Jiirgen kommt, heiBt Dresden. (OP)
(The city that Jilrgen is from is called Dresden.)

a. dem - Although you"re correct that the preposition "aus" requires the
dative, the antecedent "Stadt" is feminine.

b. der - (correct response)

0. die - The preposition "aus" requires the dative.

d. denen - Although you"re correct that the preposition "aus" requires the
dative, the antecedent "Stadt" is singular, not plural.

Der Junge, fur _ wir das Spielzeug gekauft haben. ist unser Sohn David.

(0")

18.

19.

(The boy for whom we bought the toy is our son David.)
a. der - The preposition "fur" requires the accusative, not the nominative.
b. die - The antecedent "Junge", although it sports an -e ending, is
masculine, not feminine.
c. den - (correct response)
d. dem - The preposition "fur" requires the accusative, not the dative.

Die Leute, mit __ wir ins Kino gegangen sind, waren nett. (OP)
(The people with whom we went to the movies were very nice.)

a. der - A good guess, and while you"re correct that the preposition "mit"
requires the dative, the antecedent "Leute" is plural, not feminine.

b. die - The preposition "mit" requires the dative.

c. dem - While you"re correct that the preposition "mit" requires the dative,
the antecedent "Leute" is plural, not masculine or neuter.

d. denen - (correct response)

Das ist die Frau, mit _ ihr gestern gesprochen habt, oder? (OP)
(T hat’s the woman you spoke with yesterday, right?)

a. der - (correct response)

b. denen - While it"s true that "mit" is a preposition which requires dative,
the antecedent "Frau" is feminine singular, not plural.

c. dem - Although you"re right that "mit" is a preposition that requires dative,
the antecedent "Frau" is feminine, not masculine or neuter.

(1. die - "Mit" is a preposition that requires dative.

 

84

APPENDIX H

20. Die Frau, __ ich Blumen mitgebracht habe, heiBt Nancy. (IO)

(The woman whom I brought flowers is Nancy.)

a. die - Since the relative pronoun is the indirect object within the clause,
the dative is required.

b. der - (correct response)

0. dem - Although you"re correct that the relative pronoun requires the
dative, the antecedent "Frau" is feminine, not masculine or neuter.

d. denen - While it"s true that the relative pronoun requires the dative, the
antecedent "Frau" is feminine singular, not plural.

21. Die Tante, bei __ Katja als Kind gewohnt hat, kommt morgen zum Besuch.
(OF)
(The aunt with whom Katja lived as a child is coming for a visit.)
a. dem - While you"re correct that "bei" is a preposition that requires dative,
the antecedent ”Tante" is feminine, not masculine or neuter.
b. die - The preposition "bei" requires the dative case.
0. denen - Although it"s true that "bei" is a preposition that requires dative,
the antecedent "Tante" is feminine singular, not plural.
d. der - (correct response)

22. lch habe dem Kind, __ so nett war, sine Tafel Schokolade gekauft. (S)

(I gave the child, who was so nice, a bar of chocolate.)

a. der - The antecedent "Kind" is neuter, not masculine, regardless of the
child"s actual gender.

b. das - (correct response)

0. dem - Within the relative clause, the relative pronoun is the subject of the
sentence, and therefore requires the nominative.

d. die - The antecedent "Kind" is neuter, not feminine, regardless of the
child"s actual gender.

23. Also seid ihr die Studenten, ilber __ Susanne so oft spricht. (OP)
(So you’re the students Susanne talks about so often.)
a. denen - Although you"re correct that the antecedent "Studenten" is plural,
the verb "sprechen + iiber" requires the accusative, not dative.
b. das - The antecedent "Studenten" is plural, not neuter singular.
0. die - (correct response)
d. der - The antecedent "Studenten" is plural, not masculine singular.

 

24.

25.

26.

85

APPENDIX H

Schau mall Da ist der Fernseher, _ ich liberal! gesucht habe. (DO)
(Look! There's the television I’ve been looking everywhere for.)

a. den - (correct response)

b. das - The antecedent "Fernseher" is masculine, not neuter.

c. der - Since the relative pronoun is the direct object within the relative
clause, the accusative is required, not the nominative.

d. dem - Since the relative pronoun is the direct object within the relative
clause, the accusative is required, not the dative.

Torsten hat den Fufiball verloren, __ ich ihm geschenkt habe. (DO)
(T orsten lost the football I gave him.)

a. der - Since the relative pronoun is the direct object within the relative
clause, the accusative is required, not the nominative.

b. das - The antecedent "Fuliball" is masculine, not neuter.

c. den - (correct response)

d. dem - Since the relative pronoun is the direct object within the relative
clause, the accusative is required, not the dative.

“Buddenbrooks” heiBt das Buch, __ ich Ietztes Semester gelesen habe.

(DO)

27.

(“Buddenbrooks” is the name of the book I read last semester.)

a. dem - Since the relative pronoun is the direct object, the accusative is
required, not the dative.

b. der - First, the antecedent "Buch" is neuter, not masculine. Second, since
the relative pronoun is the direct object, the accusative is required, not
the nominative.

0. die - The antecedent "Buch" is neuter, not feminine.

d. das - (correct response)

“Kaspar Hauser" heiBt der Film, von __ wir so viel gehbrt haben. (OP)
(“Kaspar Hauser” is the name of the film we’ve heard so much about.)

a. dem - (correct response)

b. den - "Von" is a preposition that requires the dative, not accusative.

c. der - "Von" is a preposition that requires the dative, not nominative.

d. denen - Although you"re right that "von" is a preposition that requires the
dative, the antecedent "Film" is masculine singular, not plural.

86

APPENDIX H

28. Ohio State ist die FuBballmannschaft, gegen _ wir am Samstag spielen.
(0?)
(Ohio State is the football team we’re playing against on Saturday.)

a. der - The preposition "gegen" requires the accusative, not the dative.

b. die - (correct response)

c. denen - First, the preposition "gegen" requires the accusative, not the
dative. Second, the antecedent "Fuliballmannschaft" ist feminine
singular, not plural.

d. das - The antecedent "Fufiballmannnschaft" is feminine, not neuter.

29. Herr Schmidt ist der Professor, bei __ Anna lezten Sommer gearbeitet hat.
(0")
(Dr. Schmidt is the professor with whom Anna worked last summer.)
a. der - The preposition "bei" requires the dative, not the nominative.
b. denen - The antecedent "Professor" is masculine singular, not plural.
c. dem - (correct response)
d. den - The preposition "bei" requires the dative, not the accusative.

30. Erik ist der Student, _ Helativsatze sehr leicht findet! (S)
(Erik is the student who finds relative clauses very easy!)

a. den - No, that would mean that relative clauses find Erik to be easy!
Since we want Erik to be the subject, the relative pronoun should be
nominative.

b. der - (correct response)

0. die - No, the antecedent "Student" is masculine, not feminine.

d. dem - Since the relative pronoun is the subject, the nominative is
required, not dative.

APPENDIX I

APPENDIX I

Metannsuisticfieedhack
SecondJ‘reatmenLSessicn

Before the actual feedback, a random response generator gives the subject an
indication that the response was incorrect, i.e. “I’m sorry, that is incorrect”. After
the metalinguistic feedback, the random generator gives the subject some
indication to try again, i.e. “Please try again.” If the response was correct, the
computer provided a congratulatory message and asked the subject to proceed.

Key: S=Subject Clause, DO=Direct Object Clause, IO=Indirect Object Clause,
0 P=Object of a Preposition Clause.

1. Jazz ist die Musik, __ ihm am besten gefallt. (S)
(Jazz is the kind of music he likes best.)
a. das - The antecedent "Musik" is feminine, not neuter.
b. dem - Since the relative pronoun is the subject, the nominative is
required in this instance, not the dative.
c. die - (correct response)
d. der - Because the relative pronoun is the subject, the nominative is

required, not the dative.

2. Anja ist die Frau, __ ich liebe. (DO)

(Anja is the woman I love.)

a. die - (correct response)

b. der - Since the relative pronoun is the direct object, the accusative is
required, not the dative.

c. das - The antecedent "Frau" is feminine, not neuter.

d. denen - First, die antecedent "Frau" is feminine singular, not plural.
Second, since the relative pronoun is the direct object, the accusative is
required, not the dative.

3. Georg ist der Freund. _ ich immer mit den Hausaufgaben helfe. (DO)
(Georg is that friend of mine whom I'm always helping with his
homework.)
a. den - "Helfen" is a verb which requires the dative, not the accusative.
b. der - "Helfen" is a verb which requires the dative, not nominative.
c. das - The antecedent "Freund" is masculine, not neuter.
d. dem - (correct response)

87

88

APPENDIX I

4. Jim Morrison ist ein Rocksanger, __ an Drogen gestorben ist. (S)

(Jim Morrison is a rock star who died from drugs.)

a. darn - Because the relative pronoun is the subject, the nominative is
required, not the dative.

b. der - (correct response)

c. den - Since the relative pronoun is the subject, the nominative is needed
in this instance, not the accusative.

d. das - The antecedent "Rocksanger" is masculine, not neuter.

5. Der Fernseher, __ ich vor zwei Tagen gekauft habe, ist schcn kaputt. (DO)

(The television I bought two days ago is already broken.)

a. dem - As the relative clause is the direct object, we require the accusative
here, not the dative. .

b. die - The antecedent "Fernseher" is masculine, not feminine.

c. den - (correct response)

d. der - Since the relative clause is the direct object, the accusative is
required, not the nominative.

6. Das Land, aus __ er kommt, heiBt Schweden. (OP)

(The country he comes from is called Sweden.)

a. der - Although you"re correct that the dative is required, the antecedent
"Land" is neuter, not feminine.

b. das - The preposition "aus" requires the dative, not the accusative.

c. dem - (correct response)

at denen - Although you”re correct "aus" requires the dative, the antecedent
"Land" is neuter singular, not plural.

7. Das Madchen, __ das griine Kleid tragt, ist Bernhards Tochter. (S)

(The girl who is wearing the green dress is Bernhard"s daughter.)

a. die - Although it"s easy to assume that "Madchen" would be feminine, the
suffix -chen makes any noun neuter.

b. dem - Since the relative pronoun is the subject, the nominative is
required, not the dative.

c. der - The antecedent "Madchen" is neuter, not masculine.

d. das - (correct response)

89

APPENDIX I

8. Ich habe den Film gesehen, von __ Katja so oft spricht. (OP)
(I saw the film that Katja talks about so often.)
a. dem - (correct response)
b. der - The preposition "von" requires the dative, not the nominative.
c. den - With the preposition "von", the dative is required, not the
accusaflve.
d. das - The antecedent "Film" is masculine, not neuter.

9. lch haben die Note in Deutsch bekommen, __ ich verdient habe. (DO)
(I received the grade in German that l deserved.)

a. die - (correct response)

b. denen - First, as the relative pronoun is the direct object, the accusative is
required, not the dative. Second, the antecedent "Note" is feminine
singular, not plural.

c. der - Since the relative prounoun is the direct object, the accusative is
required, not the dative.

d. das - The antecedent "Note" is feminine, not neuter.

10. Das Madchen, _ du das Spielzeug gegeben hast, helBt Annemarie. (IO)

(The girl to whom you gave the toy is named Annemarie.)

a. das - The relative pronoun, as the indirect object, requires the dative, not
nominative or accusative.

b. dem - (correct response)

c. der - The antecedent "Madchen" is neuter, not masculine.

d. den - First, the antecedent "Madchen" is neuter, not masculine. Second,
since the relative pronoun is the indirect object, the dative is needed, not

the accusative.

11. Die Stadt, durch _ wir gefahren sind, heiBt Regensburg. (OP)
(The city we drove through is called Regensburg.)
a. denen - First, the antecedent "Stadt" is feminine singular, not plural.
Second, the preposition "durch" requires the accusative, not the dative.
b. der - The preposition "durch" needs the accusative, not the dative.
c. das - The antecedent "Stadt" is feminine, not neuter.
d. die - (correct response)

90

APPENDIX I

12. Friihling ist die Jahreszeit, in _ die VOgeI singen. (OP)

(Spring is the season in which the birds sing.)

a. der -

b. die - Since the sentence indicates no motion within a fixed point in time,
the relative pronoun requires the dative, not the accusative.

c. denen - Although you"re correct that the dative is needed, the antecedent
”Jahreszeit" is feminine singular, not plural.

d. das - The antecedent "Jahreszeit" is feminine, not neuter.

13. Der Junge, __ mit Julia zur Schule geht, heiBt David. (S)

(The boy who goes to school with Julia is named David.)

a. den - The relative pronoun, as the subject, requires the nominative, not
the accusative.

b. die - The antecedent "Junge" is masculine, not feminine.

c. der - (correct response)

d. dem - Since the relative pronoun is the subject, the nominative is
required, not the dative.

14. Die Frau, mit __ ich gestern im Restaurant war, ist meine Freundin. (OP)

(The woman with whom I was at the restaurant yesterday is my girlfriend.)

a. denen - Although you"re right that "mit" requires the dative, the
antecedent "Frau" is feminine singular, not plural.

b. der - (correct response)

c. das - The antecedent "Frau" is feminine, not neuter.

d. dem - Although you"re correct that "mit" requires the dative, the
antecedent "Frau" is feminine, not masculine or neuter.

15. Dirk hat dem Jungen, __ so brav war, eine Tafel Schokolade gegeben.

(8)

(Dirk gave a bar of chocolate to the boy who was so well-behaved.)

a. das - The antecedent "Junge" ist masculine, not neuter.

b. dem - The relative pronoun, as the subject, requires the nominative, not
dative.

c. der - (correct response)

d. den - Since the relative pronoun is the subject, the nominative is
required, not the accusative.

16.

91

APPENDIX I

Die Frau, _ wir den Pulli geschenkt haben, arbeitet in unserem Baro.

('0)

17.

18.

(The woman to whom we gave the sweater works in our office.)

a. die - Since the relative pronoun is the indirect object, the dative is

required, not the nominative or accusative.

b. denen - Although you"re right that the dative is required, the antecedent
"Frau" is feminine singular, not plural.

c. dem - Although you"re correct that the dative is required, the antecedent
"Frau" is feminine, not masculine or neuter.

d. der - (correct response)

Das ist der Mann, mit __ ihr letzte Woche gesprochen habt. (OP)
(T hat"s the man you (pl.) spoke with last week.)

a. dem - (correct response)

b. der — The preposition "mit" requires the dative, not the nominative.

c. denen - Although you"re right that "mit" requires the dative, the
antecedent "Mann" is masculine singular, not plural.

d. den - The preposition "mit" requires the dative, not the accusative.

Die Professorin, _ mir eine 4.0 gegeben, ist jetzt Ieider krank. (S)
(The professor (f.) who gave me a 4.0 is unfortunately ill now.)

a. die - (correct response)

b. das - The antecedent "Professorin" is feminine, not neuter.

c. den - The antecedent ”Professorin" is feminine, not masculine.

d. der - Since the relative pronoun is the subject, the nominative is required,
not the dative.

19. Das Buch, von _ ich so viel gehdrt habe, habe ich endlich gelesen.
(OP)

(I finally read the book that I had heard so much about.)

a. das - The preposition "von" requires the dative, not the accusative.

b. der - Although you"re correct that "von" requires the dative, the
antecedent "Buch" is neuter, not feminine.

c. denen - Although you"re right that "von" requires the dative, the
antecedent "Such" is neuter singular, not plural.

d. dem - (correct response)

 

20.

92

APPENDIX I

Der Stuhl, auf _ Jurgen sitzt, ist bIau. (OP)
(The chair, upon which Jurgen is sitting, is blue.)

a. den - Since there is no motion, the preposition "auf" requires the dative,
not the accusative.

b. denen - Although you"re right that the dative is required, the antecedent
"Stuhl" is masculine singular, not plural.

c. dem - (correct response)

d. das - The antecedent "Stuhl" is masculine, not neuter.

21. Der Stuhl, auf __ ich mich hinsetze, ist grun. (OP)
(The chair, upon which I seat myself, is green.)
a. der - Since there is an indication of motion, the preposition "auf" requires
the accusative, not the nominative.
b. den - (correct response)
c. dem - Considering that there is an indication of motion, the preposition
"auf" requires the accusative, not the dative.
d. die - The antecedent "Stuhl" is masculine, not feminine.
22. Die Stadt, in __ der Eiffelturm steht. heiBt Paris. (OP)
(The city in which the Eiffel Tower is located is called Paris.)
a. die - Since there is no motion, the preposition "in" requires the dative, not
the accusative.
b. dem - Although you"re correct that the dative is required, the antecedent
"Stadt" is feminine, not masculine or neuter.
c. der - (correct response)
d. denen - Although you"re right that the dative is needed, the antecedent
"Stadt" is feminine singular, not plural.
23. lch habe das Theaterstuck gesehen, fiber __ Hannes geschrieben hat.
(OF)

(I saw the play that Hannes wrote about.)
a. das - (correct response)
b. die - The antecedent "Theaterstuck" is neuter, not feminine.
c. den - The antecedent "Theaterstiick" is neuter, not masculine.
d. dem - The verb "schreiben + iiber" (to write about) requires the
accusative, not the dative.

93

APPENDIX I

24. Der Mann, bei __ ich wohne, arbeitet als Lehrer. (OP)
(I man I live with work as a teacher.)
a. der - The preposition "bei" requires the dative, not the nominative.
b. denen - Although you"re correct that the dative is required, the
antecedent "Mann" is masculine singular, not plural.
c. den - The preposition "bei" requires the dative, not the accusative.
d. dem - (correct response)

25. Es war meine Frau, fur _ ich das Auto gekauft habe. (OP)
(It was my wife for whom I bought the car.)
a. die - (correct response)
b. den - The antecedent "Frau" is feminine, not masculine.
c. der - The preposition "fur" requires the accusative, not the dative.
d. das - The antewdent "Frau" is feminine, not neuter.

26. Mein Bruder, _ sehr reich ist, hat gestern einen BMW gekauft. (S)

(My brother, who is very rich, bought a BMW yesterday.)

a. den - The relative pronoun, as the subject, requires the nominative, not
the accusative.

b. der - (correct response)

0. das - The antecedent "Bruder" is masculine, not neuter.

d. dem - Since the relative pronoun is the subject, the nominative is
required, not the dative.

27. Sonja hat den fremden Mann, _ sie gekiiBt hat, ins Gesicht geschlagen.

(S)
(Sonja punched in the face the strange man who had kissed her.)
a. das - The antecedent "Mann" is masculine, not neuter.
b. die - The antecedent "Mann" is masculine, not feminine.
c. der - (correct response)
d. dem - Since the relative pronoun is the subject, the nominative is
required, not the dative.

28. Er kauft den Computer, __ am schnellsten ist. (S)

(He will buy the computer that is fastest.)

a. den - The relative pronoun, as the subject, requires the nominative, not
the accusative.

b. dem - Since the relative pronoun is the subject, the nominative is
needed, not the dative.

c. der - (correct response)

d. das - The antecedent "Computer" is masculine, not neuter.

29.

30.

94

APPENDIX I

Das sind die Freunde, mit __ ich in der Kneipe war. (OP)
(Those are the friends with whom I was in the bar.)

a. der - Although you"re correct that "mit" requires the dative, the antecedent
"Freunde" is plural, not feminine singular.

b. dem - Although you"re right that the dative is needed, "Freunde" is plural,
not masculine or neuter singular.

0. die - The preposition "mit" requires the dative.

d. denen - (correct response)

Der Schauspieler, __ sehr nervos war, hat nicht so gut gespielt. (S)
(The actor, who was very nervous, didn"t perform so well.)

a. die - The antecedent "Schauspieler" is masculine, not feminine. A
female actor would be a "Schauspielerin".

b. der - (correct response)

c. dem - Since the relative pronoun is the subject, the nominative is
required, not the dative.

d. den - The relative pronoun, as the subject, requires the nominative, not
the accusative.

 

LIST OF REFERENCES

 

REFERENCES

Aarts, F., & Schils E. (1995). Relative clauses, the accessibility
hierarchy, and the contrastive analysis hypothesis.” W 47-
63.

Bradln. C. (1996). W
teachers. Unpublished glossary.

Brandl, K. (1995). Strong and weak students’ preferences for
error feedback: Options and preferences. W
12, 1 94-2 1 1 .

Carroll, 8., & Swain, M. (1993). Explicit and implicit negative
feedback. An empirical study of the learning of linguistic
generalizations. WWW 357-
386.

Chapelle, C., & Jamieson, J. (1989). Research trends in CALL. In
M. Pennington (Ed). W
,QLtheAn, Houston, TX: Athelstan.

Doughty, C. (1991). Second language instruction does make a
difference: Evidence from an empirical study of SL relativization.
MW 43 1-469-

Fotos, 8. (1993). Consciousness raising and noticing through

focus on form: Grammar task performance versus formal instruction.

W 385-407-

95

 

96

Garrett, N. (1986). The problem with grammar: What kind can

the language learner use? W 133-49.
Garrett, N. (1991). Technology in the service of language

learning: Trends and issues. MW 74-101.

Gass, S. (1979). Language transfer and universal grammar
relations. W 327-344.

Gass, S. (1997). W
immer, New Jersey: Iawrence Erlbaum.

Gass, S., & Selinker, L. (1994) W
W New Jersey: iawrence Erlbaum.

Hellinger, M. (1977). WWW
Tiibingen: Niemeyer.

Higgins, J. (1988). W london:
longman Group UK Ltd.

Higgins, J., & Johns, T. (1984). W
Collins ELT/Addison—Wesley.

Keenan, E., & Comrie, B. (1977). Noun phrase accessibility and

universal grammar. WM 63-99.

Kufner. H. (1962). W
German. Chicago: University of Chicago Press.

Kuo, T.-M., & Hirschman, E. (1997). The role of distinctive

perceptual information in memory: Studies of the testing effect.

WWW 188-201.
Long, M.H. (1981). Input, interaction, and second language

 

97

acquisition. In H. Winitz (Ed.), WW

..c. 004301... i‘\:. out-1'" o ‘r‘ '259-
278.

Long, M.H. (1996). The role of the linguistic environment in
second language acquisition. In W.C. Ritchie and T.K. Bhatia (Eds),
01.00.010t0.o.l'.e"-.' IO! 0 i‘q0f0_-.I-°.o.."
acquisition, New York: Academic Press.

Long, M.H. (1997, February). MW
syimhusjesign. Workshop presented at the University of
Pennsylvania.

Mackey, A., & Philp, J. (1998). Conversational interaction and
second language development: Recasts, responses, and red herrings?

WW
Meredith, RA (1978). Improved oral test scores through

delayed response. Modflnlanzuazelonmmz. 321-327.
Nagata, N., & Swisher, M.V. (1995). A study of consciousness-
raising by computer: The effect of metalinguistic feedback on second

language learning. W 337-47.

Pennington, M. (1996). The power of the computer in language
education. In M. Pennington (Ed.), W (pp. 1-13).
Houston, TX: Athelstan.

Robinson, G. (1990). Effective feedback strategies in CALL. In

P. Dunkel (Ed.), WW
WW(PD-155-167)- New York: Newbury

98

House.
Robinson, P. (1995). Attention, memory and the ‘noticing’

hypothesis. W 283-322.
Schachter, J. (1983). A new account of language transfer. In S.

Gass & L. Selinker (Eds). languagemnsfenmJangnageleammg (pp.
98-1 1 1). Rowley, MA: Newbury House.

Schmidt, R. (1993). Awareness and second language
acquisition. AnnuaLReyiewgfiApniiediingiiisncm 206-226.

Sharwood Smith, M. (1991). Speaking to many minds: On the
relevance of different types of language information for the L2
learner. WWW 118-132.

Stevens, Vance. (1989a). A direction for CALI: From
behavioristic to humanistic courseware. In M. Pennington (Ed.),

I" ...o' . u on- ' ' u‘ .. - o r‘ u. Houston,

TX: Athelstan.

Stevens, Vance. (1989b). Student attitudes toward CALL in a
self—access centre. W 4-5.

Tomlin, R.S., & Villa, V. (1994). Attention in cognitive science
and second language acquisition. SSLA,_1_6_, 183—203.

Van der Linden, E. (1993). Does feedback enhance computer-
assisted language learning? W 61-65.

VanPatten, B., & Cadierno, T. (1993). Explicit instruction and
input processing. WW1 225-
243.

 

99

Warschauer, M. (1997). Computer-assisted language learning:

An introduction. In S. Fotos (Ed.), Mulnmedialangnageleaching
(pp. 3-20). Tokyo: logos International.

Zorach. 0. & Melin. C. (1994). Englishgrammanfomudenm
German (3rd ed.). USA: Olivia Hill Press.