l
>

     

WWII!“WNHIWNHIHIWW1IIHIWIHIWHINI

.CDNU'I

J
fll/MICHIGAN STATE UNIVERSITY LIBRARIES

l/f/f/I/I/llllll/l/l/fl/ll////I///I///I///I//I/I//////I//l L

3 1293 01020 0792

This is to certify that the
thesis entitled

THE ROLE OF OUTPUT IN
ACQUISITION 0F SECOND LANGUAGE SYNTAX

presented by

Elizabeth Harriett Glew

has been accepted towards fulfillment
of the requirements for

M o A . degree in German

 

 

améafl

Major professor

Datew/fy' /'0//f1 ‘71

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

 

LIBRARY
Michigan State
University

 

 

 

PLACE N RETURN EOXtoromovothiI mum your record.

TO AVOID FINES Mum on or baton duo duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

: I I |
- [:3-
-
| l |

MSU to An Animation ActioNEquni Oppomrnity Instituion
mm:

 

 

 

THE ROLE OF OUTPUT
IN ACQUISITION OF SECOND LANGUAGE SYNTAX

BY

Elizabeth Harriett Glew

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

1994

ABSTRACT

THE ROLE OF OUTPUT
IN ACQUISITION OF SECOND LANGUAGE SYNTAX

BY
Elizabeth Harriett Glew

Although most language instructors believe that
students need practice, research findings on the role of
output in SLA have been mixed. Some studies have failed to
demonstrate that acquisition is aided by output, but others
have shown that output does aid acquisition. I hypothesize
that output does promote acquisition of target-like
syntactical structures

Subjects were two groups of beginning German students:
one whose instruction included output and another whose
instruction did not. The subjects’ acquisition of six
grammatical structures was tested using two tasks,
grammaticality judgments and elicited imitation.

While no significant differences were found between the
two groups’ performance on the elicited imitation task, t-
test results show that the [+output] group’s grammaticality
judgments were significantly more accurate than the [-
output] group’s for five of the six sentence types tested.
Results of the grammaticality judgment task indicate that

output aids acquisition of L2 syntax.

Copyright by

Elizabeth Harriett Glew

1994

Dedicated to my grandfather

Horace H. Glew

a devoted and pioneering teacher

iv

ACKNOWLEDGMENTS

I am grateful to my advisor, Dr. Tom Lovik for his
helpful feedback, encouragement, and guidance. I am also
grateful to the members of my committee; Dr. George Peters,
Dr. Pat McConeghy, and Dr. Anne White for their kind
assistance.

Special thanks to Dr. Sue Gass, without whose generous
advice I could not have designed my research and also to
Karin Heinze and Wally Sue and their students, who
generously donated class time to the study.

I am deeply grateful to my parents for their unfailing
support, encouragement, and sound advice at all hours.

Special thanks to my partner and statistical consultant
Carla Monestere, whose expertise and support were essential

to completion of this project.

TABLE OF CONTENTS

List of Tables

I. The Problem

II. Method

A. Subjects

B. Measures

C. Evaluation of Data

III. Results

A. Elicited Imitation Task

B. Grammaticality Judgment Task

IV. Discussion

Endnotes
Appendix
Appendix

Appendix
Subjects

Appendix

Appendix

Bibliography

Test Sentences Grouped by Type
Times Tested Structures Were First Presented

Demographic Questionnaires Administered to

Instructions to Subjects

Explanation of Grammaticality Judgment Task

vi

vii

17
17
24
34
36
37
39
42
47
48

50

51
57
58

59

LIST OF TABLES
Table 1: Subjects' responses to demographic survey

Table 2: Elicited Imitation Scores of [+output] and
[-output] Groups

Table 3: Grammaticality Judgment Scores [+/-output] Groups

Table 4: Times Tested Structures Were First Presented

vii

I. The Problem.

In recent years, questions about what is necessary,
sufficient, and helpful for second language acquisition
(SLA), have been the focus of considerable discussion among
language teachers and SLA researchers. The role of input
(learner exposure to the target language) in SLA has been
given particular attention. The role of output (learner
production of the target language) has also been examined,
but less intensively.

Krashen's ideas about second language acquisition,
particularly his input hypothesis, have had a dramatic
effect on the ways teachers and researchers look at SLA.
According to this hypothesis, "We acquire by understanding
language that contains structure a bit beyond our current
level of competence (1+1). This is done with the help of
context or extra-linguistic information" (1982, 21). In
Krashen's view, it is unnecessary to practice speaking or
writing in order to become skilled at communicating in a
second language. He believes that output (learner
production of the language being acquired) may be indirectly
helpful because it causes more input to become available,
but that it is not necessary in order for acquisition to
take place. In his words,

According to the Input Hypothesis, speaking is not

absolutely essential for language acquisition. We

acquire from what we hear (or read) and

understand, not from what we say. The Input

1

2

Hypothesis claims that the best way to teach

speaking is to focus on listening (and reading)

and spoken fluency will emerge on its own (Krashen

& Terrell 56).

Krashen (1985, 18) extends this claim, maintaining that
ability to write is developed by reading.

Krashen's hypotheses about SLA, including his Input
Hypothesis (as well as Natural Order, Acquisition/Learning,y
Monitor, and Affective Filter), have been strongly
'criticized by a number of scholars (Higgs & Clifford,
McLaughlin) for circularity, vagueness, self-contradiction,
unsupported assertions, metaphorical nature, and lack of
falsifiability. McLaughlin, for example, one of Krashen's
most energetic critics, points out problems with each aspect
of Krashen’s theory: The impossibility of determining
objectively what constitutes acquisition and what is
learning makes it impossible to test the acquisition-
learning hypothesis and some data suggest that learning can
become acquisition. Tests on Monitor use have failed to
show predicted improvement, even when there is focus on
form, sufficient time, and knowledge of rules--all the
conditions supposedly necessary for Monitor use. And, "in a
real sense, the Monitor Hypothesis is untestable" (30)
because it is impossible to determine the source of an
utterance. Research results only partially support the
Natural Order Hypothesis, with differences apparent between

cross-sectional and longitudinal studies and between

3
learners with different L1 backgrounds or learning
strategies. The Input Hypothesis cannot be tested because
of the impossibility of determining exactly what input is
comprehensible to a learner or what a learner’s "i" is, and
there is some "evidence that first- and second-language
learners acquire structures that are neither understood nor
due to be acquired next" (40). Furthermore, cross-cultural
research indicates that some children learn their first
language without "comprehensible input" as defined by
Krashen. The Affective Filter Hypothesis does not explain
how such a filter would work or why L1 affects SLA, and
there is no evidence that nervous or anxious people are
necessarily poor language learners. In summary, McLaughlin
specifies four criteria for evaluating theories:

(1) the theory must have definitional precision

and explanatory power, (2) the theory must be

consistent with what is currently known, (3) the

theory must be heuristically rich in its

predictions, and (4) the theory must be

falsifiable (55).

He asserts that "Krashen's theory fails at every juncture"
(56).

In spite of this type of attack from the scholarly
community, Krashen's ideas have become widely popular among
teachers, perhaps in part, as McLaughlin suggests, because
they are presented "in a way that makes them readily

understandable to practitioners" (19) and because much of

4
what Krashen claims (the affective filter hypothesis, for
example) is consistent with "common sense" based on
classroom experience.

Research in both first and second language acquisition
has shown that the input learners receive is important to
the acquisition process and that no acquisition occurs
without input (Larsen-Freeman & Long 128-145). The "strong
version," of the input hypothesis - that input is sufficient
for acquisition - is not entirely supported by research
results. Although a few cases of first language acquisition
in the absence of output (children physically unable to
speak, but mentally able to use language) have been
documented, even Krashen believes that it is probably better
for children acquiring their first language to have
opportunities for output (1982, 60).

While it may be the case that second language
acquisition can occur in the absence of output, it seems
intuitively obvious that this is not the optimal condition
for SLA. Even if the only benefit of output is indirect, as
Krashen believes, we should not hastily discount its value.
Research on the utility of output for SLA has been less
extensive than that related to input and findings are mixed
(Larsen-Freeman & Long 130-132). Some studies have failed
to demonstrate that acquisition is aided by output (Sato
1986, Match), but others have shown that productive practice
does aid acquisition (Swain, Stevens, Terrell, Osguthorpe &

Chang). No studies have shown output to hinder acquisition

5
and even indirect benefits can enhance learning. It seems
most likely that output is differentially useful, that is,
that output is, as Sato (1990, 84-91) has argued, more
beneficial to some aspects of SLA than to others. More
research needs to be done to determine how output functions
in SLA. This study is a small step in that direction.

Another aspect of the problem is how to define and
measure acquisition. Krashen and Terrell offer a definition
of "acquiring," stating that "acquiring a language is
'picking it up,’ i.e., developing ability in a language by
using it in natural, communicative situations (18)." This
definition is surprisingly vague and anecdotal and, since
their purpose is to differentiate between learning and
acquisition, rather than to indicate precisely what
acquisition is, it is not of much use in measuring
acquisition.

SLA researchers generally define acquisition
operationally, depending on what exactly they are examining.
For example, researchers studying morpheme acquisition
typically consider a morpheme to have been acquired when a
speaker supplies it in 80-90% of obligatory contexts
(Larsen-Freeman & Long 300). Acquisition can also be
measured relatively, by comparing how target-like learners'
skills are. Such measurements of relative acquisition are
sufficient for the purposes of much SLA research. For the
purposes of this study, acquisition is defined as the

ability to recognize and imitate correct use of selected

6
syntactical structures in limited contexts. It will be
measured relatively, by comparing the [+output] group's
ability to recognize and imitate correct use of the selected
syntactical structures to the [-output] group’s.‘

Syntactical structures are particularly interesting in
light of Swain’s (1985) finding that immersion students, who
received copious comprehensible input developed native-like
skills in several areas, (listening comprehension and
‘"productive discourse competence", for example) but did not
achieve native-like morphological and syntactical
grammaticality. It appears that, while comprehensible
input may be sufficient for some aspects of SLA, written and
or oral output (learner production of target language) may
be necessary for thorough acquisition of L2 syntax.

While some preliminary research on the role of output
in SLA has been conducted, its role in development of L2
syntax has yet to be investigated specifically (Larsen-
Freeman & Long 130-131). It is my hypothesis that output
promotes development of target-like syntactical structures.
The "target language" in this case is Modern High German, as
presented in the subjects' textbooks. This hypothesis will
be supported if the [+output] group’s use of syntactical
structures is more target-like than the [-output] group’s
use Of syntactical structures.

Tracey Terrell is one researcher who has begun to
investigate possible roles of output in SLA. _In his 1986

article, which attempts to define acquisition by introducing

7
the concepts of binding and access, he distinguishes himself
from (but carefully does not argue with) Krashen regarding
the role of output in SLA. He describes binding as "the
cognitive and affective mental process of linking a meaning
to a form" (214) and access as "the production of an
appropriate form to express a specific meaning in an
utterance" (215). Access, he argues, "does not follow
automatically from binding" (215). A number of other
factors, such as the salience and frequency of the forms in
input, and the phonological or morphological complexity of
the forms may also affect ease of access. Most importantly,
Terrell hypothesizes that "facility in access is also
related to frequency of opportunities to access a specific
form in a meaningful context" (215).

As a test of this hypothesis, Terrell experimented with
students learning German. Two groups of students
participated in a TPR lesson. The content of the lesson was
identical for the two groups, but one group of students was
asked to give commands to their peers after having responded
to them as given by the instructor, while the other group of
students only responded to commands from the instructor.
When the students were tested on their ability to access the
forms taught in the TPR lessons, both groups of students
were able to access the forms, but the students who had
practiced accessing the forms, were able to do so an average
of .55 seconds faster than the students who had not

practiced accessing the forms.

8

Based on this research, Terrell draws the conclusion
that comprehensible input and opportunities to access forms
are both necessary for acquisition. He defines acquisition
of a form as "the process which leads to the ability to
understand and produce that form correctly in a
communicative context"(213) and states that he considers a
form to have been acquired "only when it can be both
understood and produced" (213). His conclusion is not
entirely warranted by his test results, since all the
students (both those who had opportunity to access the forms
and those who did not) were able to access the forms tested,
but his results do show clearly that practice in accessing
bound forms is beneficial to students. He argues that "the
student must have real two-way conversational experiences
for complete acquisition to take place" (220). He does not
address the possibility of providing students with
opportunities to access bound forms in other settings, such
as a language lab or oral drill.

Similarly, Swain argues for the importance of what she
calls "interactional, meaning-negotiated conversational
turns" (247) in second language acquisition. Her study of a
group of French immersion students found that, although they
had been receiving large amounts of comprehensible input in
French for seven years and performed as well as native
speakers on tests of listening comprehension and nearly as
well as natives on tests of productive discourse competence,

"the target system ha[d] not been fully acquired” (246).

9

Swain measured her subjects’ acquisition of French as a
second language by comparing their performance on tests of
grammatical, discourse, and sociolinguistic competence to
- native speakers’ performance. Although the immersion
students’ performance was similar to native speakers’
performance on the measures of discourse competence (film
retelling, argumentation, narratives, letters), they did not
perform at near-native levels on tests of grammatical or
sociolinguistic competence (structured interview, requests,
suggestions, complaints, directives). Swain argues that
this deficiency is attributable, not to insufficient or
defective input, but to insufficient ”opportunity for
meaningful use of linguistic resources" (248) and the
resulting paucity of ”comprehensible output,” which she
defines as "being pushed toward the delivery of a message
that is not only conveyed, but that is conveyed precisely,
coherently, and appropriately" (249). Swain also
hypothesizes that comprehensible output, which "forces the
learner to pay attention to the means of expression needed
in order to successfully convey his or her own intended
meaning" (249) may help learners progress from semantic to
syntactic processing of their second language. Semantic
processing alone, she argues, often suffices for
comprehension, but not for accurate production.

Swain’s hypothesis is supported by Stevens’ 1982
study comparing two groups of French immersion students; one

in a traditional teacher-centered program, in which French

10

was the classroom language for 21% hours per week and one in
an "activity centered" program, in which French was the
classroom language for 133 hours per week. All the
students were given standardized, normed language and math
tests used in Francophone schools, which allowed subjects to
be compared with native speakers, a French-language
adaptation of the multiple-choice California Reading Test
consisting of 90 vocabulary questions and 29 comprehension
questions, and a listening comprehension test in which they
answered multiple-choice questions about a story they heard
read on tape. Sixty percent of the students’ speaking
skills were evaluated in individual taped interviews using
picture description and question/answer formats.

In the French language tests, a comparison of the

two immersion programme groups reveals significant

differences only in 2 of the written measures: the

Reading Test and the te§t_de_rendement_en_fzangais

in favour of the TC [teacher centered] immersion

group. Both these tests are instruments which

evaluate academic type skills, and are affected by

both the students IQ level and formal teaching

methods. There was no statistically significant

difference between the two groups on the other

tests, i.e. aural comprehension, math, and oral

production. This finding is of particular

interest because of the shorter length of time

spent by the AC [activity centered] immersion

11

group in French studies and, also, because two of

the schools in the AC programme had never studied

math in French (7).

The students in the activity-centered program spent a large
part of their class-time working independently or in student
groups and therefore had greater opportunity (and
motivation) to use their language in meaningful ways than
did the students in the teacher-centered program. They were
encouraged and given opportunities to discuss and write
about their projects in French. In short, they produced
more output. Stevens cites this as well as the greater
motivation to learn provided by an activity-centered
classroom as the most important factors accounting for these
students’ acquiring approximately equal language skills with
less instructional time.

Liming also points to the need for practice when
learning a new language. He was a successful learner of
English as a second language and a professor of English in
China. Yet, upon immigrating to Canada, he found his skills
inadequate for some basic communication needs, even after
living in an English-speaking environment (and receiving
comprehensible input) for 18 months. He had difficulty, for
example, distinguishing between various forms of greeting
such as "good morning" "hello" and "hi." He describes
hearing various forms of greeting for over a year and
reading about differences of register among the greetings in

question, but mastering their use "only by practicing such

12
greetings in the authentic communication context" (22).
Other embarrassments he encountered involved misusing
English idioms such as telling a more senior professor that
"your face is glowing with radiance" (10) when he really
meant to say that the professor was looking particularly
well that day and being unable to get change for the bus
because he asked someone to "exchange" rather than "change"
his bill (10).

Based on these and other experiences, recorded in a
diary study, Liming describes three functions of output in
advanced learners’ acquisition of socially appropriate
communication skills. He found that, as Swain predicted,
production of comprehensible output aided his development of
syntactic processing skills. He also found Krashen’s idea
that learners’ efforts to produce comprehensible output will
result in their receiving greater quantities of
comprehensible input to be correct, but he does not believe
that this is the only function of output. In Liming’s
experience, learner output is also necessary to stimulate
"negative reactions and/or corrective feedback from speakers
of the target language"(12), without which acquisition
beyond a basic level is difficult (if not impossible).

Of course, such a study, relying as it does on a single
subject’s self-reports of learning, cannot be generalized to
explain language acquisition in a broader sense. It also
relies on chance to select the situations to be dealtwith

and language forms to be learned. While this is realistic,

13
it also means that the study cannot be replicated and that
the researcher has no control over what is examined.

In contrast to Liming’s personal, practical
approach, Jarvis (1983) presents theoretical arguments that
SLA researchers should be informed by (but not dependent on)
the psychology of learning. He asserts that second language
acquisition behavior is very similar to problem solving
behavior. He cites Gagné’s description of problem-solving
behavior and points out similarities to language-learning
behavior.

The learners are placed in a problem situation, or

find themselves in one. They recall previously

acquired rules in the attempt to find a

"solution." In carrying out such a thinking

process, the learners may try a number of

hypotheses and test their applicability. When

they find a particular combination of rules that

fit the situation, they have not only " solved the

problem" but have also learned something new,...in

the sense that the individual’s capability is more

or less permanently changed (156-7).
Jarvis sees each new utterance as a "problem situation" for
second language learners since each one either presents or
necessitates creation of a novel combination of linguistic
elements (except in the case of repetitive drills).
Learners draw on what they already know to decode and encode

messages and, when they understand or create a meaningful

14
message, they add to their repertoire accordingly.

Jarvis argues that, since SLA is like problem solving,
and since problem solving skills are enhanced by practice in
solving problems, language teachers, like those who teach
problem solving skills, should provide their students with
opportunities for "substantial practice that is
sufficiently, but not excessively varied in nature"(399).
New material should be practiced in as many combinations
with other new and old material as possible, but learners
should be able to "decode" novel utterances. That is,
students should be exposed to large number and variety of
"problems" but none should be insoluble, since that would
lead only to frustration.

Jarvis appears to agree with Krashen that input is not
useful unless it is comprehensible input. Indeed, his
description of useful problems is reminiscent of Krashen’s
description of i+1. However, Jarvis disagrees with Krashen
about the sufficiency of comprehensible input. Jarvis
believes that second language acquisition results from
solving numerous language "problems", rather than from
exposure to sufficient comprehensible input. He believes
comprehensible input to be necessary, but not sufficient.
In this respect, Jarvis agrees with Omaggio Hadley, whose
first two hypotheses of methodology and proficiency
emphasize the need for students to have varied opportunities
for practice (79-83).

Jarvis’ assertions, while reasonable, are based only on

15

his own observations and thinking. In order for them to be
supported fully, empirical research would have to be carried
out to determine how similar L2 acquisition processes really
are to other problem-solving processes. Actual similarity
cannot be deduced with certainty from apparent similarity.

Although there are substantial differences between the
processes of first and second language acquisition, the
similarities between the two are undeniable and studies of
how children acquire their first language can shed some
light on the processes of second language acquisition. In
her 1978 book, Hatch uses discourse analysis to examine how
communicative interaction contributes to both first and
second language acquisition. Hatch’s discourse analysis of
learner conversations led her to conclude that language
structures evolve

out of discourse...not because of some magic

acquisition device which operates automatically on

input but because of the conscious desire of the

child to say something, to talk about something

(405).
Hatch concludes that the development of syntactic structures
is based on conversational interaction, not vice versa. If
these findings are reliable, then output, in the form of
conversational interaction, plays a vital role in language
acquisition.

Studies by the Heidelberger FOrschungsprojekt, a

group of researchers examining naturalistic acquisition of

16
German as a second language by Spanish- Italian-, and
Turkish-speaking Gastarbeiter (guest workers), also point to
the importance of conversation for development of L2 syntax.
The group analyzed speech samples from 48 foreign workers.
When the relationships between the subjects’ acquisition and
several non-linguistic factors were examined, it was found
that "contact with Germans during leisure time" (18) had a
greater effect on subjects’ acquisition of German syntax
than any of the other factors--age at time of immigration,
contact with Germans at place of work, professional training
in the country of origin, education, or duration of stay [in
Germany]. The authors do not offer an explanation of why
leisure-time contact is more beneficial than contact at work
(which was third of the six factors, more important than
training, education, or duration of stay). Perhaps it is
related to the nature of the jobs most commonly held by
guest workers, which could limit the amount of German
spoken. It may also depend on affective factors which are
difficult to quantify. In any case, while the authors do
not define what "contact" Consisted of, it seems safe to
assume that contact with Germans was salient because of the
subjects’ interactions with the native speakers, not because

merely being in their presence provided some benefit.

II. Method
A. Subjects

Subjects were 22 beginning German students in two
groups: one group ([+output]) whose instruction included
output and another ([-output]) whose instruction did not
include output. For the purposes of this study, output is
defined as learners’ written or oral production of German.
Subjects were drawn from two MSU German classes, GRM 101 AND
GRM 400. Each class had approximately 20 students in it,
who met with their instructor four to five hours per week.

Eleven students (6 male and 5 female) in GRM 101 formed
the [+output] group. This primarily undergraduate beginning
German class is focused on the development of communicative
competence and devotes a great deal of instructional time to
conversational practice. Students in this class are also
assigned regular written exercises and compositions in
German. Less than 20% of class time is typically spent on
grammar instruction. Students are expected to study the
grammar lessons on their own and they spend most of their
class time on communicative activities, such as role plays,
interviews, and small group discussions, which are designed
to foster the development of communication skills. These
activities incorporate the grammatical structures assigned
for study. The GRM 101 students generate large amounts of
both written and oral output.

The [-output] group consisted of 11 students (7 male

and 4 female) from German 400, a beginning class for

17

18
graduate students who need to learn to read German. This
class is focused solely on the development of the reading
skill. Students receive only input (written and aural) and
spend no time practicing German speaking or writing Skills
except for brief instruction in the sounds of spoken German
during the first week of class. Their source of aural input
is the instructor’s reading of textbook exercises to them.
They receive written input from their textbook exercises and
from other German readings relevant to their fields of
study. Homework and class discussions are in English. GRM
400 students spend about one third of their class time on
grammar topics and about two thirds on reading practice.
After they have been introduced (in the first nine weeks) to
all aspects of German grammar which they are likely to
encounter in reading, they spend the rest of the term
practicing reading German--receiving a great deal of input
but producing no output.

At first glance, it seems that the very differences
which make these two groups interesting as research subjects
would be likely to skew the results of a test for syntax in
favor of the GRM 400 students, since they receive more
grammar instruction. However, the most significant
difference in grammar instruction between the two classes is
that the GRM 400 students are presented with a greater
number of grammatical structures than the GRM 101 students
are. The amount of time spent on each grammatical structure

is not greater for the GRM 400 students than it is for the

19
GRM 101 students. The other important difference between
grammar instruction for the two groups is that GRM 400
students receive explanations of grammar in English, while
GRM 101 students practice using structures about which they
have read an English description in their textbook. They
receive little or no formal grammar instruction in English
in class.

All attempts were made in the selection of subjects for
the two groups to minimize the variables and to establish
uniformity. All subjects indicated that German was not
their first language nor had it been spoken in their
childhood homes. None had had more than one semester (or
one year in high school) of German instruction prior to
enrolling in GRM 400 or GRM 101, and none had spent more
than three months in a German speaking country prior to
testing. Students who did not meet these criteria were
excluded from the pool of subjects.

Table 1 shows that analysis (using a Mann-Whitney U
test) of the subjects' responses to demographic
questionnaires (Appendix C) revealed no statistically
significant differences in attitude, tiredness, or stress
level between the two groups. When asked about their
attitude toward learning German, 90.9% of all respondents
said they were either "very interested" or "somewhat
interested" in learning German. There were no statistically
significant differences between how tired or stressed

students in the two groups reported feeling.

20

Although one might expect a group of graduate students
to be better students, on average, than a group of
undergraduates, this does not appear to be the case for the
subjects of this study. The undergraduates in the [+output]
group were asked to report their current grade point average
(GPA) and graduate students in the [-output] group were
asked to report their undergraduate GPA. There was no
statistically significant difference between the [+output]

group’s and the [-output] group’s reported GPAs.

Table 1 Subjects’ responses to demographic survey_

     
    
     

  

[+output]
group
mean rank_

[-output]
group
mean rank_

 

 

 

 

 

 

 

 

 

 

attitude 11.18 11.82 '-.266 .7903 IN)
tiredness 10.64 12.36 -.652 .5145 no
stress 11.77 11.23 -.252 .8009 In)
[undergrad 13.59 9.41 -1.531 )1140 no
GPA =====

 

A 2 value greater than $1.96 indicates that the two
medians are different at the .05 significance level (Bruning
& Kintz, 204), which means that the probability (p) of the
results occurring by chance is less then 5%. Since none of
these 2 values (table 1) are greater than $1.96, we know
that the differences between the two groups are not
statistically significant.

Other differences between the two groups, such as level

of maturity, purpose(s) for learning German, teaching

21

methods, and emphasis on structure in course goals, which
might appear to make them unsuitable for comparison, are in
fact not problematic. While the effect of maturity on SLA
has not been extensively studied, perhaps because it would
be difficult to define maturity operationally and to measure
it, the effects of age have been investigated. While there
is some evidence supporting the existence of a critical
period for second language acquisition}, it is not
conclusive.

The available evidence does not consistently

support the hypothesis that younger second

language learners are globally more efficient and

successful than older learners. Nor is it

possible to conclude from the evidence that older

second language learners are globally more

efficient and successful than younger learners

(Singleton, 137).
In any case, the largest age-related differences in
acquisition have been found between children beginning SLA
before and after age six, while differences found between
younger and older adults have been minimal (Larsen-Freeman
and Long, 158). The students in GRM 400 and in GRM 101 are
all (obviously) well past the age of six and furthermore,
are all well past the onset of puberty, the end of
Lenneberg’s critical period. Although it is not possible to
determine the effect (if any) of maturity, the slight age

difference, at least, does not decrease the suitability of

22
these two groups as test subjects.

Although data on students’ purposes for learning German
were not ecllected, we know that most GRM 101 students take
the course to fulfil requirements for their major or their
college. Since graduate students very rarely have the
luxury of taking "elective" courses merely because they find
them interesting, it seems reasonable to assume that the GRM
400 students were also fulfilling program requirements. The
two groups of students probably had similar purposes for
learning German. I assume therefore that purpose for
learning German is not sufficiently different for the two
groups to be problematic.'

The difference in teaching method is essential to this
study, since it is this which creates environments in which
students do or do not produce target-language output. The
difference in emphasis on structure in course goals is a
similar case. The GRM 101 students produce large amounts of
output precisely because the main goal of their course is to
develop their L2 communication skills, while the GRM 400
students produce no output because the main goal of their
course is to develop the reading skill. Because they have
only one semester in which to acquire the skill they need,
they must cover all necessary grammatical structures in that
semester. GRM 101 students are part of a program lasting at
least two semesters, allowing grammatical structures to be
introduced more gradually. They are, however, held

accountable for structures that have been introduced just as

23
the GRM 400 students are. To that extent, emphasis on
structure is similar in the two courses. Naturally, only
structures covered by both classes were tested.

There are certainly more differences between the two
groups of students than is consistent with a truly
experimental research design, which would require random
assignment of subjects to groups and precise control of
their treatment. However, no two "real" groups of students
will be free of such disturbing variation. .Recruiting and
screening subjects for a laboratory type experiment would
present another set of practical and theoretical problems.
For this reasOn, a quasi-experimental design was chosen.
Although internal validity is somewhat compromised by using
such a design, external validity (the degree to which the
treatment effect can be generalized) is good because the
conditions are representative of those present in real
education situations (Seliger & Shohamy, 148-9).

Quasi-experimental research is more likely to have

external validity because it is conducted under

conditions closer to those normally found in
educational contexts. For this reason, research
conducted under a quasi-experimental design format

is often less likely to meet resistance from the

’consumers’ of research, such as language

teachers. Furthermore, since these designs are

less intrusive and disruptive than others it is

easier to gain access to subject populations and

24

thus easier to conduct such research (Seliger &

Shohamy, 149).
Swain’s study is a good example of the utility of quasi-
experimental research designs. Her subjects were children
in two different schools, in different cities, with
different teachers and very different programs of
instruction. Nevertheless, her results are interesting and
her work is widely respected and is frequently cited by
other researchers”. The subjects in the present study are
at least as similar to each other as Swain's subjects were
and the present study uses a similar quasi-experimental
.design, but is simpler because it measures only one aspect

of subjects’ acquisition, rather than several.

B. Measures
The subjects’ acquisition of the following six
syntactical structures was tested:
1. main clause final placement of verbal prefixes
(Der Film fangt in 5 Minuten en.)
[The film starts in 5 minutes up.]
The film starts up in 5 minutes.
2. subject verb inversion in main clauses
(Am Sonntag bring; Themee das Bier zur Party.)
[On Sunday brings Thomas the beer to the party.]

Thomas will bring the beer to the party on Sunday.

25

3. verb position in sentences with present tense modal
verbs

(Herr Schmidt mufi Spanisch leznen.)

[Mr. Schmidt must Spanish to learn.]

Mr. Schmidt must learn Spanish.
4. verb final word order in dependent clauses

(Tanja trinkt Wasser, weil sie durstig 15;.)

[Tanja drinks water because she thirsty is.]

Tanja drinks water because she is thirsty.
5. verb placement in main clauses in German Perfekt

(Sonja Lee in den Alpen gegengezg.)

[Sonja went in the alps hiking.]

Sonja went hiking in the alps.
6. use Of dative articles with indirect objects

(Stefan schenkt ger Sekretarin Blumen.)

Stefan gives the secretary flowers.
These six syntactical structures are well suited to this
investigation since they contrast with the corresponding
English structures and are easy to test. Because they are
not similar to English structures, it is clear that the
' students’ mastery of new German material, not transfer from
English, was being tested. These structures are also useful
because they are not at all ambiguous. Subjects either
placed the words correctly or they did not. They either
used the dative article or they did not. There was no
middle ground to muddy the waters. All six structures had

been taught to both groups of students prior to data

26
collection. All test sentences were constructed using
vocabulary items that had been presented to both groups
prior to testing. Appendix A contains a complete list of
the thirty sentences used (twenty four test sentences and
six distractors).

The structures were tested using two tasks:
grammaticality judgments and elicited imitation. A
grammaticality judgment task requires subjects to give their.
opinions about the grammaticality of sentences. The results
show how well subjects are able to distinguish grammatical
from ungrammatical sentences in their L2. An elicited
imitation task requires subjects to mimic spoken L2
sentences (or other utterances) which are slightly too long
to be retained in short-term memory. The length of the
utterances forces the subjects to process (or mentally
"digest") them before repeating them. This means that the
subjects’ responses will reflect the rules of the subjects'
interlanguage(s), allowing a researcher a glimpse of what
those rules are. Although it is theoretically possible to
construct entire grammars of subjects’ interlanguages, this
study is interested merely in determining to what extent
subjects’ interlanguage grammars were consistent with target
language grammar.

The written grammaticality judgments were intended to
favor slightly the GRM 400 ([-output] group) students (whose
instruction emphasiZed reading) without being prohibitively

difficult for the GRM 101 ([+output] group) students. The

27
elicited imitations were intended to favor slightly the GRM
101 ([+output] group) students (whose instruction emphasized
speaking and listening) without being prohibitively
difficult for the GRM 400 ([-output] group) students. One
half of each class performed each task first, so that any
differences would not be attributable to the order of the
tasks.

There were four test sentences (two grammatical and two
ungrammatical) for each structure as well as a total of six
distractor sentences unrelated to the structures under
investigation. The following sentences were used for both
the written grammaticality judgment task and the oral
elicited imitation task, but were presented in random order

without marking of ungrammatical sentences.

TYPe 1

 

Der Konferenz findet heute statt.

[The conference takes today place.]
The conference takes place today.
Der Film fangt in 5 Minuten an.
[The film starts in 5 minutes up.]
The film starts up in 5 minutes.
*"Star Trek" anfangt in 10 Minuten.
*"Star Trek upstarts in 10 minutes.
*Das Seminar stattfindet heute.

*The seminar placetakes today.

Typez

28

eggjeet ZGLQ inversien in main elagses
Am Sonntag bringt Thomas das Bier zur Party.

[On Sunday brings Thomas the beer to the party.]
Thomas will bring the beer to the party on Sunday.
Dieses Jahr studiert Tina an der Universitat.
[This year studies Tina at the university.]
Tina is studying at the university this year.
*Am Montag Michael bringt das Buch zur Klasse.
[On Monday Michael brings the book to class.]
*On Monday brings Michael the book to class.
*Im Winter Maria studiert an der Hochschule.
[In the winter, Maria studies at the college.]
*In the winter, studies Maria at the college.
Type3

- . up Ttio: j se t-_ es f p_e e1, tens- flOQOV
Herr Schmidt muB Spanisch lernen.

[Mr. Schmidt must Spanish to learn.]
Mr. Schmidt must learn Spanish.
Frau Braun kann kaltes Bier trinken.
[Ms. Braun kann cold beer to drink.]
Ms. Braun can drink cold beer.
*Frau Schulz muB lernen Mathematik.
[Ms. Schulz must to learn mathematics.]
*Ms. Schulz must mathematics learn.
*Herr Becker kann trinken warme Milch.
[Mr. Becker can to drink warm milk.]

*Mr. Becker can warm milk drink

29

Type4

e ' de c a
Tanja trinkt Wasser, weil sie durstig ist.
[Tanja drinks water because she thirsty is.]
Tanja drinks water because she is thirsty.
Marc kommt nach Hause, wenn er Sommerferien hat.
[Marc comes to home when he summer vacation has.]
Marc comes home when he has summer vacation.
*Emma trinkt Kakao, weil das Wetter ist kalt.
[Emma drinks cocoa because the weather is cold.]
*Emma drinks Cocoa because the weather cold is.
*Kurt kommt ins Theater, wenn er hat 15 Dollar.
[Kurt comes to the theater when he has 15 dollars.]

*Kurt comes to the theater when he 15 dollars has.

 

Sonja ist in den Alpen gewandert.

[Sonja went in the alps hiking.]

Sonja went hiking in the alps.

Erik hat in Berlin gelebt.

[Erik has in Berlin lived.]

Erik has lived in Berlin.

*Stephanie ist gewandert in den Appalachians.
[Stephanie went hiking in the Appalachians.]
*Stephanie went in the Appalachians hiking.
*Albert hat gelebt in New York.

[Albert has lived in New York]

30
*Albert has in New York lived.
Type6
WW
Christine gibt dem Mann ein Buch.
[Christine gives (to) the man a book.]
Christine gives a book to the man.
Stefan schenkt der Sekretarin Blumen.
[Stefan gives (to) the(fem.) female secretary flowers.]
Stefan gives flowers to the female secretary.
*Frau Graf gibt den Mann die New York Times.
Ms. Graf gives the man the New York Times. (no indirect
object)
*Monika schenkt dem Lehrerin einen Apfel.
Monika gives (to) the(masc.) female teacher an apple.
Type 7
511W
Das Wetter ist sehr schén.
The weather is very nice.
Herr Lenz besucht Frau Frisch in Berlin.
Mr. Lenz visits Ms. Frisch in Berlin.
Annette treibt gern Sport.
Annette likes to engage in sports.
Herr Muller findet 10 Dollar.
Mr. Muller finds 10 dollars.
Frau Schiller hat ein Auto.
Ms. Schiller has a car.

Peter kennt den Mann.

31

Peter knows the man.

Subjects were informed that the researcher was studying how
people learn German, but were not told the exact nature of
the study or what structures were being tested. The six
distractor sentences were included to lessen the possibility
of subjects identifying the test structures, which could
affect their responses. The same sentences were used for
both the grammaticality judgment and the elicited imitation
tasks so that neither would accidentally be more difficult
than the other. The items were presented in a different
random order for each task. Similar test formats have been
used successfully by other SLA researchers (Gass, in press).
In retrospect, it was an error to use the ungrammatical
sentences in the elicited imitation task, as they do not
yield reliable data on subjects’ competence.

The grammaticality judgment task was administered by
the students’ own teacher in their usual classroom.
Students recorded their responses on machine-readable
"bubble sheets." Subjects were instructed to "Please give
yeer_epinien of the acceptability of each sentence according

to the following scale.

32

1 This is definitely an acceptable German sentence.

2 This is probably an acceptable German sentence.

3 I have no idea at all.

4 This is probably not an acceptable German sentence.

5 This is definitely not an acceptable German sentence."

Each student received an instruction sheet (Appendix D) and
the instructor read an explanation of the task (Appendix E)
to them before they began.

The most accurate response to grammatical test sentences was
1 and the most accurate response to ungrammatical test
sentences was 5. Subjects’ responses to the distractor
sentences were not evaluated.

The elicited imitation task was administered by the
researcher in the language lab and answers were recorded on
cassette tapes which were labelled with code numbers. After
collection, these code numbers were covered with randomly
assigned numbers so that it was not apparent to which group
any tape belonged. This prevented the researcher from
inadvertently skewing the results while coding the
responses.

The thirty test sentences (Appendix A) used for the
elicited imitation task were the same as those used for the
grammaticality judgment task, (see above, pp.27-30) but were
presented in a different random order. Responses to the
grammatical stimulus sentences were coded 1 for incorrect or
2 for correct. Responses to the ungrammatical stimulus

sentences were not evaluated since they would not provide ‘

33
any information about the subjects’ ability to imitate
correct German. In fact, rather than merely repeating the
ungrammatical sentences, some subjects gave grammatical
responses to the ungrammatical stimuli. But it was not
possible to determine whether this occurred because subjects
were correcting the ungrammatical stimulus sentences or
because subjects were making more or less random errors.
Most likely, some subjects made corrections and others made
mistakes, and it is impossible to distinguish between the
two. Since the subjects were instructed to imitate the
sentences they heard as accurately as they could, some
subjects probably imitated some ungrammatical sentences
accurately (repeating the ungrammatical sentences) even
though they knew them to be ungrammatical (and may have
known what the corresponding grammatical sentence would be).
Whether this occurred (and if so, where) cannot be
determined from data collected. Therefore, responses to
ungrammatical stimulus sentences could not give accurate
information about the respondents’ knowledge of German and
these sentences were excluded from evaluation.

Students also responded to a questionnaire (Appendix C)
asking about their prior experiences with German and other
foreign languages and about their feelings regarding their
current class and the testing situation. These answers were
also recorded on machine readable answer sheets and were
used to eliminate students with prior German knowledge from

the subject pool and to compare the two groups.

34

C. Evaluation of Data
1. Grammaticality Judgment

In order to facilitate evaluation of the results,
responses to all ungrammatical items were re-coded
(5=1...1=5) after collection to match the grammatical items.
Thus the lower a subject scored, the more accurate the
subject’s responses were. Mean scores for each group were
calculated for each type of sentence and for all types taken
together. The means for the two groups were compared and
the differences between them were tested for significance

using a t-test.

2. Elicited Imitation

A response was coded correct if the structure under
investigation was accurately imitated, even if the subject
made other errors. A response was coded incorrect if the
structure under investigation was not accurately imitated.
Since phonetic accuracy was not a topic of investigation,
mispronounced or only partly articulated words were
considered to have been imitated accurately if they were
otherwise accurate. FOr example, if a respondent imitated
the sentence "Christine gibt dem Mann ein Buch." by saying
"Christin gib dem Mann das Buch.", the response would be
coded correct since the structure under investigation, dem,
was imitated accurately and the mistakes in the noun, verb,
and other article are irrelevant. Similarly, if a subject

responded to "Die Konferenz findet heute statt." with "Der

35
Konferenz findet heute." the response was coded incorrect
since the structure under investigation, the separable
prefix, was not accurately imitated. But if a subject
responded "Duh Konfrens find heute sta." the response was
coded correct since the verb prefix was present and placed
correctly, although pronunciation was faulty. Since
phonetics and verbal morphology are not topics of this
study, the subject’s mispronouncing the verb and omitting
its ending are not relevant.

After all responses had been coded, mean scores for
each group were calculated for each type of sentence and for
all types taken together. The means for the two groups were
compared and the differences between them were tested for

significance using a t-test.

III. Results

Test results were calculated for each of the six types
of test sentence and for all 24 test sentences taken
together. Responses to the six distractor sentences were

not included in the data. The six test sentence types were:

Type 1: main clause final placement of verbal prefixes
(Der Film fangt in 5 Minuten en.)
The film starts in 5 minutes up.
The film starts up in 5 minutes.
Type 2: subject verb inversion in main clauses
(Am Sonntag bringt Themee das Bier zur Party.)
On Sunday brings Thomas the beer to the party.
Thomas will bring the beer to the party on Sunday.
Type 3: verb position in sentences with present tense modal
verbs
(Herr Schmidt mufi Spanisch lernen.)
Mr. Schmidt must Spanish learn.
Mr. Schmidt must learn Spanish.
Type 4: verb final word order in dependent clauses
(Tanja trinkt Wasser, weil sie durstig 15;.)
Tanja drinks water because she thirsty is.
Tanja drinks water because she is thirsty.
Type 5: verb placement in main clauses in German Perfekt
(Sonja Le; in den Alpen gexendezt.)
Sonja went in the alps hiking.

Sonja went hiking in the alps.

36

37
Type 6: use of dative articles with indirect objects
(Stefan schenkt fie: Sekretarin Blumen.)
Stefan gives (to) the(fem.) female secretary flowers.
Stefan gives flowers to the female secretary.
See Appendix A for a complete list by type of all thirty

test sentences.

A. Elicited Imitation Task

Mann-Whitney U tests revealed no significant
differences between the [+output] group’s and the [-output]
group’s performances on the elicited imitation task. Table
2 shows the results of this analysis. Since correct
responses were coded 2 and incorrect responses were coded 1,
the highest median scores reflect the most accurate

responses .

Table 2

Elicited Imitation Scores of [+outputj and [:output] Groups

 

 

 

 
 
 
  

  
  
    

 
 
  

 

 

 

 

 

 

 

 

 

 

 

 

 

Item Type Median Median 2 value

Score Score

[+output] [-output]
All Items 12.09 9.80 -.8525 p<.3939 no
(1-6)

LType 1 11.18 10.80 -.1626 p<.8709 no
Type 2 11.64 10.30 -.5276 p<.5977 no
Type 3 12.00 9.90 -1.5199 p<.1285 no
Type 4 6.31 6.88 -.3062 p<.7595 no

[[Type 5 11.50 10.49 -1.0488 p<.2943 no

[Type 6 9.86 12.25 -1.0555 p<.2912 no

 

 

 

The Mann-Whitney U tests the hypothesis that the medians of

38
the two groups are equal. A z value greater than $1.96
indicates that the two medians are different at the .05
significance level (Bruning & Kintz, 204), which means that
the probability (p) of the results occurring by chance is
less than 5%. Since none of these 2 values (table 2) are
greater than i1.96, it is clear that the differences between
the two groups’ performance on the elicited imitation task
are not statistically significant.

The inconclusive results from this task are probably
attributable to the small number of items which could be
analyzed. The grammaticality judgment and elicited
imitation tasks contained the same sentences so that neither
task would inadvertently be made more difficult than the
other. This was important since each task was intended
slightly to favor one of the groups and having one task more
difficult than the other would have skewed the results.

Only after data had been collected did it become apparent
that the ungrammatical sentences essential to the
grammaticality judgment task would not yield useful data
from the elicited imitation task. Discarding the responses
to ungrammatical stimulus sentences from the elicited
imitation data analyzed left each of the six sentence types
represented in the data by only two items. Differences may
have existed between the two groups, but there simply were

not enough responses to make any such differences apparent.

B. Grammaticality Judgment Task

T-test results show that the [+output] group’s
grammaticality judgments were significantly more accurate
than the [-output] group’s overall (all six sentence types
combined) and for five of the six individual sentence types.
Table 3 shows the results of the grammaticality judgment
task for each sentence type. The scoring system was
constructed so that lowest mean scores indicate the greatest
accuracy.
Table 3

Grammaticality Judgment Scores of [+output] and [-output]
Groups

 

      

 

 

 

 

 

 

 

 

 

 

 

Mean Mean t
Score Score value
-output ___
All 53.23 71.18 -4.00 p<.001 19.78 yes
1 8.73 11.82 -2.08 p<.050 18.78 yes
2 8.09 12.64 -3.50 p<.003 16.03 yes
3‘ 5.82 8.64 -2.50 p<.021 19.99 yes
4 11.27 12.18 -.66 p<.516 19.72 no
5 8.91 12.82 -3.03 p<.008 15.05 yes
6 10.46 13.09 -2.43 p<.026 17.24 yes

 

 

 

 

A t value greater than or equal to i1.96 indicates that the
difference between the means is significant at the .05
level, which means that the probability (p) of the results
occurring by chance is less than 5%.

The mean scores of the [+output] group are lower (more
accurate) than the mean scores of the [-output] group on all

sentence types. The difference between the mean scores of

39

40
the two groups was not statistically significant for type
four sentences (sentences containing dependent clauses).
For all other sentence types, the t values are greater than
i1.96. Therefore, it is clear that the [+output] group was
significantly better than the [-output] group at
distinguishing grammatical from ungrammatical German
sentences except when the sentence contained a dependent
clause (sentence type four).

The non-significance of the results obtained for
sentence type four may be due to the difficulty of the
dependent clause structure for the students. Neither group
did well at distinguishing grammatical from ungrammatical
dependent clause sentences. Sentences containing dependent
clauses were also the most difficult of the six types for
both groups to imitate, as shown by the low scores obtained
for type four sentences on the elicited imitation task (6.31
for the [+output] group and 6.88 for the [-output] group).
Although most students are able to understand the concept of
verb final word order in dependent clauses, most find it
quite difficult to apply this knowledge and consistently
place the verb after the subject, as in English. It is
unclear why students find this construction so problematic,
but it may be due to the lack of similar English
constructions since many beginning students tend to rely on
their L1 grammar and "paste" German words onto essentially
English sentences. This strategy of L1 transfer is often

successful, especially for very basic communication needs,

41
since English and German do have many similarities.
However, in this case, the strategy appears to be counter-
productive, delaying rather than facilitating students’
acquisition of the target structure. In any case, further
studies of the acquisition of German verb-final syntax by

English speakers are still needed.

IV. Discussion

Do second language learners in fact "acquire by
understanding language that contains structure a bit beyond
[their] current level of competence (i+1)" as Krashen (1982,
21) believes? Is comprehensible input all that is needed
for SLA, or is something else required as well? We know
that input is necessary, but is it sufficient? Even if
input is sufficient, it does not necessarily follow that
"input-only" is the best way for learners to acquire a
second language. What is the role of grammar instruction?
Of output? A great deal of research will be needed before
these questions can be answered definitively. This study,
comparing acquisition of L2 syntactical structures by
learners who produced output and learners who did not is, so
to speak, a drop in the research bucket. If the [+output]
group proved better able to recognize and imitate correct
use of L2 syntactical structures than the [-output] group,
then the hypothesis that output aids acquisition of L2
syntax would be supported.

This hypothesis is, in fact, partially supported by the
data from the study. The [+output] group performed
significantly better than the [-output] group on the
grammaticality judgment task, even though this task was
designed slightly to favor the [-output] group. Only for
sentence type 4 (word order in dependent clauses) were the
[+output] group’s responses not significantly more accurate

than the [-output] group’s (t=-.66, p<.516). There was no

42

43
significant difference between the two groups for this type
of sentence. The reason for this is not clear from the
data. It may well be because, in order for output to aid
acquisition, learners must produce specifically the
structures to be acquired, not just the target language in
general. That is, it may be the case that in order for
output to aid acquisition of (for example) verb final word
order in dependent clauses, learners must produce target
language dependent clauses with verb final word order. If
this is the case, then it is not surprising that the
[+output] group performed similarly to the [-output] group
only on sentence type four. Most beginning students find
dependent clauses difficult and tend to avoid producing them
if possible. This could be a factor simply of the length of
utterance required to use these sentences or of some more
complex phenomenon. Since the students’ output was not
manipulated to ensure that they produced all of the
structures under investigation equally, it seems likely that
their output would have included dependent clauses less
frequently than it included the other test structures, which
generally seem to be somewhat less problematic for beginning
students. The hypothesis that output aids acquisition only
if students produce the specific structures to be acquired
could be tested using three groups of subjects; a [-output]
group, a [+output] group similar to those used for this
study, and a [+ manipulated output] group, whose production

would be guided to ensure that they practiced equally all of

44
the structures under investigation. If test results showed
significant differences between the three groups, it would
be possible to conclude that not only output, but type of
output is an important factor in acquisition.

Alternatively, it may be the case that output is more
beneficial in acquiring some structures than others.
However, this does not seem highly plausible. If output is
beneficial in acquiring syntax, it seems reasonable that it
would be beneficial in acquiring all aspects of syntax, not
just a few. This hypothesis could also be tested by
controlling subjects’ output, allowing various groups to
produce various combinations of structures. More research
will be needed before the exact mechanism by which output
contributes to SLA can be understood, but the results of
this study do indicate generally that output is beneficial
to acquisition of L2 syntax.

The results of the grammaticality judgment task
(detailed in the previous section) support Swain’s (1985)
argument that output is necessary if L2 learners are to
perform at near-native levels grammatically and are
consistent with her hypothesis that output may help learners
progress from semantic processing (which is largely adequate
for comprehension) to syntactic processing of their L2,
which is needed for accurate production. Swain’s idea that
output aids development of sociolinguistic competence falls
outside the realm of this study and should also be examined

experimentally.

45

The results of the grammaticality judgment task are
also consistent with Stevens’(l982) argument, based on her
study of elementary immersion students, that output is an
important factor aiding the SLA process, and with the
results of the Heidelberger Forschungsprojekt (1978) study,
which found that contact with Germans was the most important
factor in predicting guest workers’ acquisition of German
syntax.

Of course, much more study will be required before the
function of output in SLA is thoroughly understood. Many
questions remain. 393 does output assist acquisition of a
L2 and of L2 syntax in particular? Is output more helpful
for acquisition of some structures than of others? What is
the role of output in other aspects of second language
acquisition, such as development of target-like
pronunciation and acquisition of vocabulary items? Does
output affect listening and readingcomprehension and if so,
how? What is the role of written output? However, taking
the findings of the present study, in conjunction with
earlier data, into consideration, it is clear that output
plays a significantly more important role in SLA than the
one it is allotted by Krashen’s theory.

This finding is of interest to SLA theorists and
researchers as well as to designers of L2 curricula and
materials, and to L2 instructors. Since it appears that
output does aid acquisition of L2 syntactical structures,

second language learners will be best served if materials

46
and curricula for SLA provide sufficient opportunities for
output. Textbooks and other instructional materials should
include activities designed to generate learner production
of the target language, perhaps of specific structures.
Instructors should plan their lessons to encourage output as
well as to provide plenty of input. Instruction for
beginning foreign language teachers and continuing education
for experienced teachers should include information about
the importance of output in SLA. Teachers should receive

training in ways of incorporating output in their curricula.

NOTES

1 Based on his studies of patients with brain lesions and of
deaf children, mentally retarded children, and data pertaining to
so-called "wolf children," Lenneberg (1967) postulated a
biologically-based critical period for language acquisition
beginning at approximately age two and continuing until the onset
of puberty. He believed that changes in the brain around the
time of puberty, particularly a decrease in brain "plasticity"
and a corresponding increase in "cortical specialization" made
acquisition of a first or native language impossible after
puberty. The idea of a critical period for human language
acquisition, which was widely accepted through the mid-seventies,
has been viewed with increasing skepticism since that time.

Close re-examination of Lenneberg’s (and other) data by Krashen
(among other researchers) has cast doubt on the reliability of
Lenneberg’s conclusions and of the critical period hypothesis.
For a comprehensive discussion of age factors related to language

acquisition, see Singleton (1989).

2 The S99ial_§gienges_gitatign_lndex lists seventeen

citations of the relevant article in the 1991-1993 volumes.

3 "Degree of freedom is a measure of the extent to which the
values of the observations are free to vary and still satisfy
some constraint" (Clayton, 107). The constraints are the means

and the numbers of people sampled.

47

APPENDICES

Appendix A

Appendix A

Test sentences grouped by type.
Subjects were presented with sentences in random order
without asterisk marking of ungrammatical sentences.
Type 1
main clause fine; pleeement of verQal prefixes
Die Konferenz findet heute statt.
Der Film fangt in 5 Minuten an.
*"Star Trek" anfangt in 10 Minuten.
*Das Seminar stattfindet heute.
Typez
' n 'n ' 1 es
Am Sonntag bringt Thomas das Bier zur Party.
Dieses Jahr studiert Tina an der Universitat.
*Am Montag Michael bringt das Buch zur Klasse.

*Im Winter Maria studiert an der Hochschule.

Type 3

er! .00_ .t.” 1 _‘I ‘I ‘- r, ! .9,‘-‘l ‘l.‘ ’l'!-. 7..

Herr Schmidt muB Spanisch lernen.
Frau Braun kann kaltes Bier trinken.
*Frau Schulz muB lernen Mathematik.

*Herr Becker kann trinken warme Milch.

Type 4
verb final word order in dependen; elegses

Tanja trinkt Wasser, weil sie durstig ist.

Marc kommt nach Hause, wenn er Sommerferien hat.

48

49

*Emma trinkt Kakao, weil das Wetter ist kalt.
*Kurt kommt ins Theater, wenn er hat 15 Dollar.
Type 5

er e ' a' s ' e
Sonja ist in den Alpen gewandert.

Erik hat in Berlin gelebt.

*Stephanie ist gewandert in den Appalachians.

*Albert hat gelebt in New York.

Type 6
use ef dative ertielee 11th indireet ebjects

Christine gibt dem Mann ein Buch.

Stefan schenkt der Sekretarin Blumen.
*Frau Graf gibt den Mann die New York Times.
*Monika schenkt dem Lehrerin einen Apfel.
Type 7

distragters

Das Wetter ist sehr schon.

Herr Lenz besucht Frau Frisch in Berlin.
Annette treibt gern Sport.

Herr Mfiller findet 10 Dollar.

Frau Schiller hat ein Auto.

Peter kennt den Mann.

Appendix B

Appendix B

Times Tested Structures Were First Presented

Table 4 Times Tested Structures Were First Presented

 

 

 

 

 

 

 

objects

 

 

 

Structure GRM 101 GRM 400
[+output] [-output]

Type 1
main clause final placement of verbal week 4 week 3
prefixes
Type 2
subject verb inversion in main clauses week 4 week 2
Type 3
word order in sentences with present week 8 week 5 "
tense modal verbs
Type 4
word order in dependent clauses week 9 week 6
Type 5
verb placement in main clauses in week 11 week 3
German Perfekt
Type 6
use of dative articles with indirect week 13 week 1

 

 

50

Appendix C

Appendix C

Demographic questionnaires administered to subjects.

QUESTIONNAIRE FOR GERMAN 101 STUDENTS 14 April 1993

1. What German class are you taking now?

1.
2.

German 101
German 400

2. What is your sex?

1.
2.

male
female

3. What ethnic group do you consider yourself a part of?

(fl-thH

m-thI—l

6. Which of

m-wat-I‘

7. How many

White / Caucasian

Black / African American
Asian, Pacific Islander
Native American

other

your native or first language?

. English

. German

. an Asian language (Mandarin, Japanese, etc.)
. a Romance language

. another language. What?

 

your approximate GPA?

3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49
below 2.0

the following statements best describes your attitude?
I am very interested in learning German.

I am somewhat interested in learning German.

I am indifferent about learning German.

I would rather not learn German.

I hate learning German.

semesters of college German had you taken before starting

this class?

m-tho—t

none (or less than one)

. one
. two

three

. more than three

51

10.

11.

52

How many years of high school German had you taken before starting
this class?

none (or less than one)

one

. two

. three

. more than three

(fl-thl-fi

How many college semesters or high school years of instruction in
a language or languages other than German had you had before
starting this class? (Please do not count instruction in your
native language.) '

1. none (or less than one)

2. one

3. two

4. three

5. more than three

What language(s) did you take?

 

How many linguistics courses have you taken?
1. none.
2. one introductory class in high school.
3. one introductory class in college.
4. more than one introductory class.

How many months did you spend in a German-speaking country before
starting this class? (Please do not include time spent in an
English-speaking environment. Use the space at bottom to explain
if necessary.)

1. less than one month (or never been to a German-speaking

country)

12.

13.

14.

15.

. one to three months
. four to six months
seven to nine months
ten months or more

‘31-th

When you were a child, was German spoken in your home?

. No, never.

. Yes, but infrequently. I didn’t understand it.
Yes, I understood some, but never spoke it myself.
. Yes, I understood it and spoke it some.

Yes, it was the/a primary language used at home.

I—I

U'I-hWN

Please rate how tired you are feeling today.
1 = not tired » 5 = very tired

Please rate how stressful you find this situation.
1 = not stressful » 5 = very stressful

How many college credits will you have completed by the end of
this semester?

1. l - 27

2. 28 - 55

53

3. 56 - 87
4. 88 or more

If you would be willing to participate in a short interview related to
this study, please write your name and phone number here.

Thank you very much for contributing to this research!

Please turn the answer sheet over and begin your responses at nulber 57.

54

 

 

QUESTIONNAIRE FOR GERNAN 400 STUDENTS 14 April 1993
1. What German class are you taking now?
‘1. German 101
2. German 400
2. What is your sex?
1. male
2. female
3. What ethnic group do you consider yourself a part of?
1. White / Caucasian
2. Black / African American
3. Asian, Pacific Islander
4. Native American
5. other What?
4. What is your native or first language?
1. English
2. German
3. an Asian language (Mandarin, Japanese, etc.)
4. a Romance language (Spanish, French, Italian)
5. another language. What?
5. What was your approximate undergraduate GPA?
l. 3.5 - 4.0
2. 3.0 - 3.49
3. 2.5 - 2.99
4. 2.0 - 2.49
5. below 2.0
6. Which of the following statements best describes your attitude?
1. I am very interested in learning to read German.
2. I am somewhat interested in learning to read German.
3. I am indifferent about learning to read German.
4. I would rather not learn to read German.
5. I hate learning to read German.

7. How many semesters of college German had you taken before starting
this class?

aha-wron—

. none (or less than one)
. one

. two

. three

. more than three

8. How many years of high school German had you taken before starting
this class?

1.
2.
3.
4.

none (or less than one)
one

two

three

55

5. more than three

 

9. How many college semesters or high school years of instruction in
a language or languages other than German had you had before
starting this class? (Please do not count instruction in your
native language.)

1. none (or less than one)
2. one
3. two
4. three
5. more than three
What language(s) did you take?
10. How many linguistics courses have you taken?
1. none.
2. one introductory class
3. more than one introductory class.

11. How many months did you spend in a German-speaking country before

starting this class?
(Please do not include time spent in an English-speaking
environment. Use space at bottom to explain if necessary.)
1. less than one month (or never been to a German-speaking
country) '
2. one to three months
3. four to six months
4. seven to nine months
5. ten months or more
12. When you were a child, was German spoken in your home?
1. No, never.
2. Yes, but infrequently. I didn’t understand it.
3. Yes, I understood some, but never spoke it myself.
4. Yes, I understood it and spoke it some.
5. Yes, it was the/a primary language used at home.

13. On a scale of one to five, please rate how tired you are feeling
htoday.

1 = not tired a 5 = very tired

14. On a scale of one to five, please rate how stressful you find this
situation.

1 = not stressful » 5 = very stressful
15. How many graduate credits will you have completed by the end of

Please write your approximate current GPA here.

this semester?

1. 1 - 12

2. 13 - 24
3. 25 - 36
4. 37 - 48

5. more than 48

 

56

If you would be willing to participate in a short interview related to
this study, please write your name and phone number here.

Thank you very much for contributing to this research!

Please turn the answer sheet over and begin your responses at number 57.

Appendix D

Appendix D

Instructions to Subjects
INSTRUCTIONS

please do not write an instruction sheets

Please give your opinion of the acceptability of each sentence according
to the following scale.

This is definitely an acceptable German sentence.
This is probably an acceptable German sentence.

I have no idea at all.

This is probably not an acceptable German sentence.
This is defihitely not an acceptable German sentence.

019me

For example if you thought the sentence "My dog have fleas." was -
probably not acceptable, you would mark "4" on the answer sheet.

Please indicate on the questionnaire nnx you think an unacceptable
sentence is not acceptable. What don’t you like about it? For example,
if you didn’t like the verb in this sentence, you might change it to
"has". .
has
My dog have fleas.

Or, if you didn’t find "My" acceptable, you might change it to "mine".
Mine
My dog has fleas.
It is not important that you completely understand the sentences, just
indicate whether they seem like acceptable German sentences to you or
not. There are no right on wrong nnsnens to the questions on this
opinion survey.

Please work quickly and dn NOT go back and change any of your responses.

57

 

Appendix E

Appendix E

Explanation of Grammaticality Judgment Task

read to subjects by instructor.

Please check to be sure that your white answer sheet has the same
ID number as your computer answer sheet.

Please turn the computer answer sheet over and begin the next part
at number 57.

On a scale of one to five, please rate how acceptable each
sentence on the white page is to you. One is the most acceptable and
five is the least acceptable.

If you rate a sentence four or five, please indicate why on the
white survey. There is an example of how to do this on the orange
instruction sheet.

You will have approximately fifteen minutes for this part.

Please read the orange instruction sheet and begin giving your

opinions now.

58

BIBLIOGRAPHY

BIBLIOGRAPHY

Bruning, James L. and B.L. Kintz. Computational Handbook of Statistics.
Glenview, IL: Scot, Foresman & Co., 1968.

Clayton, Keith N. An Introduction to Statistics for Psychology,and
Education. Columbus: Charles E. Merrill, 1984.

Cummins, Jim. "The Cross-Lingual Dimensions of Language Proficiency:
Implications for Bilingual Education and the Optimal Age Issue.
TESOL Quarterly 14 (1980): 175-88.

Gass, Susan M. "The Reliability of L2 Grammaticality Judgments."
Research Methods in Second Language Acquisition. Elaine Tarone,
et. al., eds. Hillsdale, NJ: Erlbaum, in press.

Gagné, Robert M. The Conditions of Learning, 3rd ed. NY: Holt,
Rinehart, 1977.

Hatch, Evelyn. "Discourse Analysis and Second Language
Acquisition." Seeond Language Acquisition: A Book of Readings.
Evelyn Hatch, ed. Rowley: Newbury House, 1978. 23-45.

Hatch, Evelyn. Psycholinguistics: agSecond Language Perspective.
Rowley, MA: Newbury House, 1983.

Heidelberger Forschungsprojekt. "The Acquisition of German Syntax by

Foreign Migrant Workers." Linguistic Yariation: Models and
Methods. David Sankoff, ed. NY: Academic Press, 1978.

 

Jarvis, Gilbert A. "The Psychology of Second-Language Learning: A

Declaration of Independence." Modern Language qurnal 67.4
(1983): 393-402.

Krashen, Stephen D. Principles and Practice in Seeond Language
Acquisition. New York: Pergamon, 1982.

Krashen, Stephen D. The Input Hypothesis: Issues and Implications. NY:
Longman, 1985.

Krashen, Stephen D., and Tracy D. Terrell. The Natural Approach.
Edgewood Cliffs, NJ: Alemany Press, 1983.

Larsen—Freeman, Diane and Michael H. Long. An Introduction tn Second
Language Acquisition. Applied Linguistics and Language Study.
NY: Longman, 1991

Lenneberg, Eric H. Biological Foundations of Language. NY: Wiley, 1967.

59

 

60

Liming, Yu. "The Comprehensible Output Hypothesis and Self-directed
Learning: A Learner’s Perspective." TESL Canada Journal 8.1
(1990): 9-26.

McLaughlin, Barry. Theories of Second Language Learning.
Baltimore: Edward Arnold, 1987.

Oller, John W. "Language as Intelligence?" Language Learning 31 (1981):
465-92.

Omaggio Hadley, Alice. Teaching Language in Context, 2nd ed. Boston:
Heinle & Heinle, 1993.

Osguthorpe, Russell T. and Linda Li Chang. "The Effects of Computerized
Symbol Processor Instruction on the Communication Skills of
Nonspeaking Students." Augmentative and Alternative Communication
4.2 (1988): 23-34.

Sato, Charlene. "Conversation and Interlanguage Development: Rethinking
the Connection." Talking to Learn: Conversation in Second Language
Acquisition. R. Day, ed. Rowley: Newbury House, 1986.

Sato, Charlene J. lne Syntax of Conversation in Interlanguage
Development. TObingen: GUnter Narr, 1990.

Seliger, Herbert W. and Elana Shohamy. Second Language Research
Methods. Oxford: Oxford University Press, 1989.

Singleton, David. Language Acquisition: The Age Factor. Multilingual
Matters 47. Philadelphia: Clevedon, I989.

Swain, Merrill. "Communicative Competence: Some Roles of Comprehensible
Input and Comprehensible Output in its Development." Input in-
Secgnd Language Acquisition. Susan M. Gass and Carolyn G. Madden,
eds. Rowley: Newbury House, 1985. 235-253.

Stevens, Florence. Activity Centred Approaenes to Second Language
Learning. ERIC, 1982 (ED 221 066).

 

Terrell, Tracey 0. "Acquisition in the Natural Approach: The

Binding/Access Framework." Modern Language Journal 70.3 (1986):
213-227.

Wood, Gordon. Fundamentals of Psychologiea! Researgh. Boston: Little,
Brown, and Co., 1981.

HIGAN STATE UNI

930||1| III”!IIIWHIHI