Professional Documents
Culture Documents
Heidi Byrnes
Georgetown University
[Ed Note: This paper is reprinted from Lan- It is worth emphasizing that these steps
guage Laboratory, the journal of the Language are far from complete and by no means
Lab Association ofJapan, and is a revised version universally accepted. In addition, a caution-
ofthe keynote address presented on July 31, 1990 ary note is appropriate: their presupposi-
at the 30th anniversary meeting of the Lan- tions about language and language teach-
guage Laboratory Association of Japan. A tenn ing should not be transferred uncritically,
as a Mellon Fellow at the National Foreign but should be examined carefully in order
Language Center, Washington, D. C., during to determine whether and how they might
the spring of 1991, facilitated the reworking of apply to the Japanese situation. Only then
the author's original remarks.] can the American experience provide input
that leads to viable solutions.
Before one can engage in language te~t
ing in general, in testing a learner's speak- LANGUAGE TESTS AND OUR
ing ability specifically, one must first clarify PROFESSIONAL BELIEFS
a number of central issues regarding essen-
tial features oflanguage and its use and how Any testing is inherently a sampling pro-
these are acquired by non-native speakers. cedure. Tests cannot possibly include ev-
erything that was taught prior to the test,
For this reason I will address the follow- nor can they project precisely how learners
ing major points. I will begin by exploring will have to use their language skills after
the connection between testing and our ba- the test. As a consequence the test developer
sic professional beliefs. I will then summa- must choose, ideally selecting those aspects
rize current assumptions in foreign language of language which are important. But "im-
pedagogy that have attained prominence in portance" does not exist of its own. Rather,
the United States and Europe, assumptions it comes about as a result of numerous con-
that are captured by the term "communica- siderations.
tive language teaching." The unique chal-
lenges to established testing practice brought The most important of these consider-
on by this shift will be outlined in the third ations regarding "importance" pertains to
section. And finally, the largest portion of
this paper will be devoted to exploring ap-
proaches to oral language testing that have Heidi Byrnes is a Professor of Linguis-
developed in the United States in the last tics at Georgetown University.
decade or so.
the goals of language teaching and learning acquisition. What steps are likely to be
which apply in a given case. It follows that necessary for a learner to attain the global
if speaking ability is to be tested, then functional ability to which she or he aspires,
speaking ability must have been targeted as or, rephrased in terms of testing, what fea-
a goal of instruction, and instruction must, tures of the language should the learner
in fact, have allowed students to engage in have acquired at what stage with what de-
spoken interaction in order to acquire this gree of certainty to allow us to feel confident
complex ability in a well-articulated and that she or he is progressing satisfactorily?
well-motivated curricular sequence. While
this may sound all too basic to be worthy of Second language acquisition research
repetition, there is every indication that obviously admits to many lacunae regard-
many programs, be they at the institutional, ing how language is learned. Even so, nu-
state, or even national level, have major merous studies have convinced us, particu-
gaps precisely in this area. Numerous mis- larly through the concept of interlanguage,
matches exist. For example, one teaches that such learning is not a simple on/ off
something in a way that really does not procedure, such that students can produce
support the stated long-term goals; one tests a form correctly all the time or they cannot.
abilities that have not been explicitly taught; Instead, learning is a lengthy and, at times,
or one tests something that has been taught, circuitous road, leading from initial aware-
but which bears little resemblance to what ness of forms and their meaning, to their
learners ultimately are expected to do with better understanding, to halting and error-
the language. prone variant use of these forms in restricted
contexts, and, finally, to complete mastery
Thus, only if speaking ability is the ex- in all contexts (for studies in interlanguage
plicit goal of a program and if the instruc- development, see Eisenstein, 1989).
tional approach has provided opportuni-
ties for the learners to develop it, is it appro- If one applies these insights to the devel-
priate and fair to test this multi-faceted skill. opment of speaking ability, one must con-
clude that neither past teaching nor past
In addition to the mismatches between testing have adequately reflected the fact
teaching and testing just mentioned, both that correct speaking in all contexts is the
teaching and testing have tended to focus very last step in a long process. Thus, testing
on short-term goals, usually the more easily will have to recognize intermediate stages
defined, more form-related components of which indicate that learners are progressing
language which, inherently, are more ame- steadily toward this lofty and demanding
nable to prevailing modes of testing. If it is goal.
true that instruction rarely" gets to" or "has
time for" enhancing the learners' acquisi- While the concern with intermediate
tion of the long-term, comprehensive func- stages and goals of learning extends to all
tional goals, then testing practice is even aspects of use, the form side of language is
more remiss in addressing these aspects likely to be particularly prominent in in-
which are much more descriptive of what structed language learning. The question
using a language is all about. then becomes: how do we test learners'
command of specific language forms while
While reference to overall goals, includ- recognizing that these forms are a means to
ing linguistic, pragmatic, discourse, and an end, namely successful communication,
sociolinguistic competence, is critical for rather than the end itself?
language testing (see Canale and Swain,
1980}, "importance" comes about also by It is impossible in this context to provide
considering aspects of second language details, if for no other reason than that they
are likely to be language-specific. But one Given this primacy for goals articulation
overall observation is appropriate and criti- it is all the more surprising that this area
cal: the formal components of a language frequently receives only perfunctory atten-
carry different weight at different points tioninforeignlanguage pedagogy. To avoid
in the process of learning, and this possible misunderstanding, by no means is
differentiation must be incorporated into it the case that all programs should aspire to
testing practice. the same goals in lock-step fashion, whether
that is communicative ability or language
For example, without a doubt learners learning for the sake of enhancing one's
must learn to produce proper past tense research access. By the same token, it is also
forms for English irregular verbs, such as inadmissible to simply drift along without
"to go-went-gone", and, therefore, one may having clarified objectives and ways of
wish to test them on this knowledge. How- reaching them. The following parameters
ever, in isolation these forms and the degree would seem to apply in this critical deci-
of control a learner has over them hold no sion-making process:
inherent significance. Only in terms of their
place and role in language learning and • Language teaching and learning, and by
language use do they become important, an implication language testing, is embedded
observation which implies that the "impor- in social contexts. What social contexts might
tance" or role of a given form shifts over impact on the setting of goals? For example,
time. To return to the earlier example, once knowing another language may be consid-
students possess basic familiarity with En- ered to be a characteristic of the educated
glish irregular verbs, it makes little sense to elites. If these educated elites have little
test them as decontextualized forms in terms occasion or little incentive for personal con-
of mastery. Their real use and usefulness, tact with speakers of other languages, and if
and thus their real importance, lies in en- the overall cultural climate is toward inter-
abling learners to provide extended narra- nal self-sufficiency rather than reaching out
tives. Such narration has certain textual re- to others, then the capability to read foreign
quirements, among them devices of coher- language texts, literary or non-literary, is a
ence and cohesion typical for the English valid goal. A commonly used indicator to
language, alongside proper verb morphol- gauge comprehension is translation of texts
ogy. Thus, in terms of the learners' ability to into the native language. Intermediate goals
narrate,.it is less critical that they can pro- toward attaining this ability would include
vide the form "went" correctly in a list of an extensive vocabulary that deals with is-
irregular verbs and more informative of sues in the target language culture, aug-
their progress if they can create utterances mented by facility with dictionaries, exten-
such as Mterwards we went to a restau-
11
sive familiarity with the literate norms of
rant where we enjoyed a pleasant meal and the language, and the ability to analyze
wonderful conversation. However, unfor- texts for their literary value. By contrast,
tunately, my sister could not come along good pronunciation habits, fluency in pro-
but went home because of a terrible head- duction, ready access to the vocabulary of
ache.11 daily life, or familiarity with the interactive
norms pertaining to speaking, would hardly
I have argued that we can assign impor- be of interest, neither in teaching nor in
tance to specific features of learner perfor- testing.
mance, and thus devise appropriate testing
modes and procedures only on the basis of • Aside from social expectations, goals
stated goals and by considering pervasive reflect a network of professional expecta-
aspects of language learning. tions and convictions. Not infrequently these
two clash. For instance, in the United States
the senior university professorate in lan- beliefs, let us focus on the relationship be-
guage deparbnents continues to consider tween testing and curricular goals. Under
the study of literature to be of paramount ideal circumstances one first determines
importance and makes demands of students overall goals, spells these out as a sequence
similar to the case just described. By of intermediate and specific goals, and then
comparison, the dominant pedagogical dis- expresses them in terms of learner perfor-
cussion takes language as a means for inter- mance statements. Only then can and should
active communication, as a way of perform- testing procedures be devised. But we may
ing socially derived tasks, of give and take have, in foreign language pedagogy, a situ-
in speaking and, increasingly more, also in ation that is the exact reverse of what things
reading and writing. Obviously, such di- ought to be. Tests often are the way they are
vergent attitudes and goals for language because it is easiest to test in a certain fash-
learning surface not only in teaching, they ion, not because these testing procedures
also affect testing practice. and the resulting test items reflect our goals.
• Finally, and certainly not least impor- However, problems do not stop here.
tantly, the individual learner comes to the Any language teacher has the desire to ap-
task of language learning with certain ex- pear as having achieved what she or he set
plicit or implicit goals. As recent research out to achieve. We want to be successful and
into motivation, attitudes, and the role of we want our students to be successful. One
anxiety in language learning has brought to of the easiest ways for creating at least the
the forefront again (Horwitz and Young, appearance of success is to direct our teach-
1991), it is the individual who learns or does ing toward our testing. In other words, tests
not learn a language because she or he feels thatwereoriginallydevised because of con-
that instruction does or does not address siderations of testing expediency or for psy-
what she or he wants to accomplish. An chometric reasons, all of a sudden drive our
emphasis on the individual learner does not teaching practice and ultimately our cur-
negate the faCt that instructional systems riculum.
inherently group together people with a
range of goals, where some of these may not One all too· obvious example shall suf-
even agree with a particular institution's fice. We know that tests that require stu-
mission. But it is best to uncover that infor- dents to perform all kinds of manipulative
mation at the outset, to share it with learn- tasks, such as switching nouns from singu-
ers, and to adjust the instructional approach lar to plural, altering the subject, or chang-
accordingly. Among the alternatives are the ing the tense of a sentence, can be created
attempt to have learners modify their ex- relatively easily and can also be checked
pectations in the direction of those underly- and scored withouttoo much difficulty. We
ing a given program, or, in reverse, of alter- also know that, necessary though these skills
ing institutional goals so that they can re- may be, their relationship to functional lan-
flect what may well have become general guage use in speaking is quite tenuous.
social trends regarding the purposes of lan- How many students with commendable
guage learning. For example, if trends in scores on tests made up of such items
student interest in the United States bear utterly lack an ability to communicate any-
any resemblance at all to developments in thing in speaking? Yet, despite this repeated
Japan, then the demands for communica- experience, disturbingly little· has changed
tive language teaching that reaches out into in our approach to testing language perfor-
the professions must be taken seriously. mance, even when speaking ability is
explicitly stated to be the instructional goal.
To conclude this section on the connec- On the contrary, the practice of discrete-
tion between testing and our professional point, decontextualized testing has created
The Relationship of Norm and Variation All of this amounts to a central require-
ment for the new communicatively oriented
Perhaps thegreatestchange results from foreign language class: the individual must
our expanded understand ing of the be given g rea ter freedoms and,
commensurately, must accept greater re- of langu11ge skills has provided a much-
sponsibilities in the process ofleaming. This needed balance (Swain, 1985). At the very
leads directly to the following point. least, apE dagogy which recognizes the spe-
cial char• 1cteristics of input processing as
Language Replication and Language well as output processing will differentiate
Creation: Second Language Norm and between language use that primarily relies
Interlanguage on the lea mer's background knowledge and
is seman Cically based (reading and listen-
One can say that, up until now, the learn- ing), and language use that must also focus
ers' central task was to replicate the model on specif .c language forms, is thus syntacti-
of the instructor. Imitation to the best of cally bas1 ~d (speaking and writing). It goes
their ability would summarize their level of without 1;aying that, for most second lan-
engagement. Now, however, they are asked guage lea mers, comprehension tasks can be
to create language and to make choices, at a significantly higher level than produc-
something that has complicated their task tion task:;.
tremendously (Swaffar, 1989). Small won-
der that, instead of producing correct forms If one compares such an approach with
in a mastery mode, they are likely to offer much of •:urrent classroom practice one ob-
more or less successful approximations of serves a curious mismatch. In the begin-
the target language system, a phenomenon ning, stu :Ients often get simplistic compre-
that is captured with the term hension ,tasks in reading and in listening,
11
interlanguage." while thEy are expected to perform tasks in
speaking that go far beyond their capabili-
This term is not simply a new way of ties. By c )ntrast, in more advanced classes,
referring to the old troublesome occurrence students are frequently asked to read any-
of errors. Far different, the concept of thing printed in the foreign language. But
interlanguage recognizes that language their spc ~aking involves little more than
learning, as a process of approximations single SEntence answers, or even just the
toward the multiple norms of a language, is completion of the teacher's sentence with a
inherently error-prone. Testing with a phrase a r a word. After all, the instructor
communicative orientation must recognize knows all too well that an appropriate dis-
this fact and attempt to incorporate it in cussion <•f a given text would go far beyond
defensible ways into testing practice. the learr .ers' abilities. The result is a high
level of f~ustration due to a total reversal of
Comprehensible Input- the natu:~al relationship between receptive
Comprehensible Output and pro iuctive skills and their develop-
ment. In: ;tead of initially incorporating stu-
In contrast with wide-spread misper- dents' bc1ckground knowledge to compen-
ceptions, communicative language teach- sate for their limited knowledge of the lan-
ing is by no means the equivalent of a near- guage, we present them with rather
exclusionary emphasis on speaking. On the unenticing texts that often come close to
contrary, as previously mentioned, it recog- insultin~; their intelligence. At the same time
nizes the critical relationship between com- beginnir .g language classes often demand a
prehension and production. 11Comprehen- tremend ousamountofprocessinginspeech
sible input", a term popularized by Krashen, rightfro:n the start. Later on, when learners
thus became one of the key terms in the have be~ unto automatize certain aspects of
initial stages of a move toward communica- languag• ~ in their speech, instruction often
tive language teaching (Krashen and Terrell, does not allow them to grow. Essentially we
1983). However, gradually, the notion of lack a pEdagogy which supports the devel-
output as being critical for the development opment of discourse competence and,
The previous discussion has laid out In sum, if one contrasts an approach
some key characteristics of teaching toward which includes the world around us with
functional language use. We can now inves- the limited engagement of students' knowl-
tigate their implications and special chal- edge of the world in language teaching and
lenges for appropriate testing. testing, one cannot help but conclude that
teaching, as well as testing, frequently tar-
The Purposefulness of Speaking gets a use of language that does not exist
anywhere else, but is totally artificial.
As contrasted with much classroom lan-
guage, language in the real world is pur- The Audience-relatedness of Speaking
poseful. It is intended to fulfill a communi-
cative function, such as persuading, inform- Language in communication obtains its
ing, reprimanding, congratulating, or the motivation, derives its purposes, and takes
less obvious function of establishing, main- its form from the communicative partners
taining, or clarifying role relationships be- that are involved. Closer analysis indicates
tween the communicative partners. The that everything, from the content of a
message, to the level of its explicitness, its restricted settings. But within these settings
degreeofformality or informality, the choice the Ieamer can strive to be as native-like as
of words and structures, the degree of di- possible right from the start.
rectness or indirectness, the level of
involvement of the respective communica- Let nte further illustrate this shift by
tive partners, relates to the audience to whom examining three terms that have become
a particular communication is directed. promine::tt in communicative testing in the
Again, if one contrasts the non-descriptness American context, proficiency testing,
and nonspecificness of audience in most achieventent testing, and prochievement
testing tasks with real life one finds another testing.
important area which must be attended to if
communicative language testing is to be- Proficiency testing refers to curriculum-
come a reality. independent testing that assesses the
Ieamer's ability to function in the target
To give a simple example: it makes all languagE!, irrespective of how he or she
the difference in the world, even in a com- acquired that language, or over what length
municative task that is as routine and for- of time, ·~tc. Learners are said to possess a
malized as exchanging greetings, to know certain I·~vel of proficiency depending on
how old the partners are, what social status whatconununicativetaskstheycanhandle,
they have, how long they have known each everythbtg from the simple interactive tasks
other, what time of the day it is, how long it of daily :life to the highly complex formal-
has been since they last saw each other, ized tas~s that educated speakers must be
before one can judge the correctness or, able to h&andle if they wish to use the foreign
perhaps better, the appropriateness of a languagE~ in their professional environments.
particular language form. Thus, proficiency testing is open-ended and
a learneJ· really does not obtain a perfect
Proficiency Testing - Achievement score.
Testing - Prochievement Testing
By co::ttrast, achievement testing is based
Therequirementsofcommunicativetests on what a Ieamer was actually taught, the
spelled out thus far appear to be asking material:; covered within different periods
entirely too much of learners, seemingly of time, ·,e that the time spent on an indi-
treating them as though they already pos- vidual unit in a textbook or the content of a
sessed complete mastery of the second lan- semester or an entire program of study, for
guage. But the real issue is the following: instance in a high school. Ideally, test items
instead of having students invest years into would be~ taken from the syllabi on the basis
leamingdecontextualized language forms- of their presumed importance, as previ-
something that experience tells us is essen- ously duocussed. While there is no inherent
tially impossible-, and then asking them to and unalterable need to assess students'
apply their knowledge in communicative achieve111ent in a particular unit of instruc-
settings-a transfer that, in most cases, is tion by nteans of discrete-point testing, that
highly problematic-, communicative lan- has esser ltially been the kind of testing char-
guage teaching and testing assumes a con- acteristi<· of the foreign language classroom.
text of language use right from the start.
Thus, communicative pedagogy echoes first This brings us to the third term,
language learning by reuniting learning and prochiev ement testing, an approach which
use, aspects which experience, corroborated attempt:: to combine the two. It incorpo-
by research, tell us should never have been rates the communicative thrust of current
separated in the first place. Clearly, for the pedagogy and, by taking the communica-
learner such use is initially possible only in tive task; students are required to perform
from a particular instructional unit, it avoids where each level presupposes the ability to
the open-endedness of proficiency testing. perform all the previous tasks:
Prochievement testing aims to capture the
interactive, purposeful audience driven, Global Tasks
creative use of language which is the hall-
Superior Can discuss extensively
mark of natural language use. It does so by
identifying limited tasks which can chal- by abstracting,
lenge our learners' ability to use the lan- supporting opinions and
guage in a valid context but which do not hypothesizing
presuppose total command of the language. Advanced Can describe and narrate
in major time/ aspect
TESTING SPEAKING ABILITY IN A
COMMUNICATIVE LANGUAGE frames
TEACHING APPROACH Intermediate Can maintain simple
face-to-face
We have now set the stage for a closer
conversations by asking
look at how speaking ability, in a communi-
and responding to simple
cative mode, might be tested in a sensible
way (ACTFL, 1989). questions
These seemingly simplistic facts are re- • What is the setting where this communi-
flected in the hierarchy of global tasks that cation takes place, the home, the work
has been established in the so-called ACTFL place, the public arena?
oral proficiency scale which underlies much
of oral proficiency testing in the United • How frequently does this occur?
States. It shows the following progression,
their own language. These are crucially Here another matter deserves clarifica-
important skills to be developed by tion. It may well be true that many of our
beginners. They are not signs of deficiency, advanced learners would rather not deal
but signs of competence. Therefore they are with matters pertaining to daily life, prefer-
worthy of being tested. ring to handle content in their own area of
professional expertise, - such as engineer-
By extension, it stands to reason that the ing, or the natural sciences. But closer ex-
kinds of monologic situations that we some- amination of their language reveals the
times ask our students to handle should following pattern: they are likely to speak
really be best reserved for advanced stu- individual sentences, with little or no con-
dents. Only they possess the requisite range nectors between them. Though they will
of grammar and vocabulary to handle these show basic awareness of the forms of the
contexts effectively, and even they can only language, perhaps even possess some of the
be expected to possess that ability if our required vocabulary, success in the conver-
instruction has provided ample opportu- sation would depend very much on a sym-
nity for its development, something that is pathetic conversational partner that shares
not always the case. their professional background knowledge.
In other words, their ability to converse is
The Content more the result of shared previous informa-
tion than it is the result of their ability to
The next level of consideration as we test handle the language competently. This is an
for communicative ability is what content important distinction to make, even if we
our learners are likely to be able to deal acknowledge, as we should, that any suc-
with. The following progression seems rea- cessful communication depends crucially
sonable: on shared background knowledge.
So how might one go about doing this You would like to meet with a classmate
kind of testing? or friend sometime over the weekend.
Call up the friend on the phone. Find out
. After our lengthy look at the intimate when it will be convenient to get to-
relationship between the goals of language gether. Make plans for what you would
learning and teaching practice it should like to do and how and where you will
come as no surprise if we now extend this meet.
relationship to include testing.
This seemingly simple task requires a
As mentioned earlier, communicative tremendous amount of interactive,
testing is possible and fair only if it was negotiative work between the partners. In
preceded by communicate teaching and that, all likelihood it involves using greeting ritu-
in turn, is possible only if such use of lan- als over the telephone, questioning about
guage is, in fact, the goal of language learn- time, about places, about things one might
ing. Only then is it possible to establish the do, perhaps suggesting some possibilities
kind of learner outcomes which can then or rejecting them as not very practicable,
become the focus of teaching as well as finally deciding on a convenient meeting
testing. time and place and, in the end, concluding
the conversation with the appropriate clos-
This allows another look at the relation- ing ritual. With relatively little change in the
ship between teaching and testing. Not in- parameters, such that one or the other stu-
frequently teachers subtly subscribe to the dent may not have too much time, or may
notion that tests are most useful if they not have much money to spend, the entire
indicate a learner's ability to perform under conversation would turn out quite differ-
the worst conditions. The hidden agenda ently. Thus, it could easily be used on a test:
seems to be to prove to the students what it is the same kind of task that students have
they cannot do, presumably to motivate practiced previously, yet, it also requires
new creativity which proves their commu- what they might do to improve, rather than
nicative ability. just hearing that they were not good enough
in some ill-defined way.
Role plays whose parameters have been
specified with respect to global task, setting, In this fashion instructional goals that
content, and communicative intention can may initially have been set in an abstract
be typed up on cards ahead of time and can way turn into outcomes that are definable,
be used both during class as well as in for teachers and students alike. In turn,
testing. Over a period of time teachers can these defined and tangible outcomes can
develop a repertoire of tasks which makes inform the setting of curricular goals : how
testing communicative ability no longer the realistic were the goals, can they be achieved
dreaded extra task, butanextensionofwhat in the time frame that was originally set, are
has already taken place in class. they possible with some students, but not
with others, is more time required, are dif-
Obviously, this close connection between ferent methods necessary, might different
teaching and testing can become a tremen- materials be incorporated?
dous motivational force for the students.
The teacher's expectations will be very clear, Only if we perform this loop back be-
and students will see thatit is to their advan- tween testing and curriculum and instruc-
tage to participate in these activities in class, tion will these new trends in testing as well
even if they may have been reluctant to do as in teaching fulfill the promise toward
so initially. improved language learning that they seem
to hold.
Teaching and testing will be more closely
connectedinanotherimportantway,namely
the diagnosing of learner progress. Here the
audio tape can be an invaluable aid in help- REFERENCES
ing students develop their abilities. By hav-
ing small tape recorders available in class The ACTFL Oral Proficiency Interview Tester
two students can record their role play tasks Training Manual. 1989. Yonkers, NY:
without the teacher being present. The ACTFL.
teacher can later collect these tapes and
listen to them and provide feedback. Like- Canale,M. andM.Swain. 1980. Theoretical
wise, the students will have an opportunity Bases of Communicative Approaches to
to listen to themselves, perhaps even collect Second Language Teaching and Testing.
something like a portfolio of their progress Applied Linguistics 1: 1-47.
with tapes that were produced over an ex-
tended period. In this way, they will get a Eisenstein, M. R., ed. 1989. The Dynamic
much better sense of what they sound like Interlanguage: Empirical Studies in Second
and, ultimately, how they might improve Language Variation. New York: Plenum
their language use. Press.
By informing students of the criteria for Horwitz, E. and D. Young. 1991. Language
assessment as they were discussed one of Anxiety: From Theory and Research to Class-
the greatest oppositions to communicative room Implications. Englewood Cliffs, NJ:
testing, namely its alleged subjectivity, is Prentice Hall.
also addressed. Once they have been in-
formed about criteria for evaluation, learn- Krashen, S. D. and T. Terrell. 1983. The
ers are remarkably adept at assessing their Natural Approach: Language Acquisition
own abilities and feel good about knowing in the Classroom. Oxford: Pergamon Press.
28 IALL Journal of Language Learning Technologies
Heidi Byrnes
Advertisement