Beyond the Test: L2 Dynamic

Assessment and the Transcendence

of Mediated Learning
MATTHEW E. POEHNER
304 Sparks Building
The Pennsylvania State University
University Park, PA 16802
Email: [email protected]

A major preoccupation in assessment is connecting examinees’ performance in assessment

and nonassessment contexts. This preoccupation has traditionally been framed in terms of
generalizability. This article reconceptualizes this problem from a qualitatively different per-
spective on human abilities and their development, namely, the Sociocultural Theory of Mind
outlined in the work of Vygotsky (1986, 1998). From this perspective, assessment occurs not
in isolation from instruction but as an inseparable feature of it. Assessment and instruction
are dialectically integrated as a single activity that seeks to understand development by actively
promoting it. This pedagogical approach, known as Dynamic Assessment (DA), challenges the
widespread acceptance of independent performance as the privileged indicator of individuals’
abilities and calls for assessors to abandon their role as observers of learner behavior in favor
of a commitment to joint problem solving aimed at supporting learner development. In DA,
the traditional goal of producing generalizations from a snapshot of performance is replaced
by ongoing intervention in development. Following Vygtosky’s argument that true develop-
ment goes beyond improvement on a given assessment task, DA practitioners have devised a
method known as transcendence (TR), in which they collaborate with learners on increasingly
complex tasks. In this article, transcendence in the second language (L2) domain is illustrated
with examples of advanced learners of French composing oral narratives with support from
a mediator. The article concludes with recommendations for future research on TR in L2
development.

ALL EDUCATIONAL PROGRAMS MUST AT past learning, in reality they are often used to
some point appraise learners’ knowledge and abil- make decisions about individuals’ futures (e.g.,
ities; that is, they must assess them. Tradition- admittance to a program, promotion to a higher
ally, two purposes for assessing are identified: level of study, or conferral of a degree or certi-
summative and formative. Bachman (1990) ex- fication). Bachman went on to explain that for-
plained that summative assessments occur at the mative assessments, in contrast, are administered
end of an instructional period and are intended before completion of a course. Formative assess-
to capture the results of instruction. The most ments have a much closer relationship to instruc-
well-known summative assessments include mas- tion because their results feed back into class-
tery tests and achievement tests used to deter- room teaching. Ellis (2003) pointed out that a
mine the extent to which students have learned wide range of classroom practices can be con-
specific course content. Although summative as- ducted for formative assessment purposes, includ-
sessments are concerned with the products of ing planned tests and quizzes as well as anecdotal
records resulting from teacher observations of stu-
dent work and informal teacher–student interac-
The Modern Language Journal, 91, iii, (2007) tions. The summative–formative distinction per-
0026-7902/07/323–340 $1.50/0 tains to the reasons for conducting the assessment


C 2007 The Modern Language Journal
rather than to the instruments or tasks employed.
324 The Modern Language Journal 91 (2007)
Traditional assessments, including multiple- & Grigorenko, 2002). Furthermore, DA is not a
choice exams and open-response essays, as well standalone activity carried out in isolation from
as so-called alternative assessments such as port- other pedagogical activities. It is instead an on-
folios, projects, and presentations, can all serve going, development-oriented process of collab-
a summative or formative function depending orative engagement that reveals the underlying
upon how their results are used. causes of learners’ performance problems and
This article considers a radically different con- helps learners overcome those problems. In other
ceptualization of assessment and its relation to words, DA does not differentiate instructional ac-
instruction, according to which these activities tivities from assessment activities because every
are not separate or even complementary under- mediator–learner interaction encompasses both
takings but rather a single activity that seeks to types of activities. Instead, DA sessions vary ac-
understand (i.e., assess) abilities by promoting cording to learner development so that over time
their development. That is, a comprehensive as- learners engage in increasingly complex tasks with
sessment of abilities requires instructional inter- less mediation.
vention, which in turn leads to the development of The dialectic unification of assessment and in-
those abilities. Conversely, for instruction to guide struction that DA represents has profound im-
development optimally, it must take account of plications for classroom practice, which second
the full range of individuals’ abilities. Assessment language (L2) researchers are only beginning to
and instruction, therefore, are dialectically inte- explore (Antón, 2003; Lantolf & Poehner, 2004;
grated into the same development-oriented activ- Lantolf & Thorne, 2006; Poehner & Lantolf, 2005;
ity. This pedagogical approach has come to be for a full review of this literature, see Poehner,
known as Dynamic Assessment (DA), although as in press). This article is intended to familiarize
I argued elsewhere (Poehner, in press) this ap- the reader with basic concepts and principles in
pellation is to some degree a misnomer because DA and to illustrate their potential to enhance
the term assessment in DA is qualitatively different our understanding of L2 development. It is be-
from how it is typically understood in psychology yond the scope of a single article to consider fully
and education. In lieu of understanding assess- the relevance of DA to L2 assessment and instruc-
ment as the observation and recording of indi- tion. In this article, I attempt only to open the
viduals’ behaviors for the purpose of inferring discussion. I focus my remarks on the issue of the
underlying abilities, assessment in the dynamic generalizability of assessment outcomes and how
sense involves transformation of those abilities this construct may be reconceptualized within a
through dialogic collaboration between learners dynamic framework.
and assessor–teachers, or mediators. Moreover, the According to Messick (1989), validation in as-
drive toward standardization and measurement sessment refers to the process of establishing the
that characterizes many forms of assessment is legitimacy of a particular interpretation of as-
usually replaced in DA with descriptive profiles sessment outcomes, and the generalizability of
of learner development, although, as we will see, an interpretation depends upon various factors
some dynamic approaches provide quantitative thought to affect the abilities in question. Cron-
reports of mediator–learner interactions, includ- bach (1990) offered a helpful illustration of gen-
ing test scores. eralizability involving a test of typing speed. He
DA is grounded in Vygotsky’s Sociocultural explained that the purpose of such a test is not
Theory of Mind (henceforth, SCT), which differs to determine how quickly an individual can type
both ontologically and epistemologically from a given passage at a given time of day under spec-
the mainstream psychological perspectives on ified conditions. Rather, the test is intended to
mental abilities that inform other approaches reveal the speed at which we might expect the
to assessment. According to SCT, individuals’ individual to type other texts in other circum-
responsiveness to support, or mediation to use stances. Cronbach went on to suggest that many
Vygotsky’s (1978) term, that is sensitive to their variables may impact an individual’s performance
current level of ability reveals cognitive functions on a typing speed test, such as the kind of day the
that have not yet fully developed. Moreover, ap- examinee is having, the length and complexity of
propriate mediation enables individuals to ex- the sentences in the passage, whether the passage
ceed their independent performance, and this in contains words that are difficult to spell, and so
turn stimulates further development (Vygotsky, on. It is important to note that these factors are
1986, 1998). Thus, DA targets what individu- considered sources of variance because changing
als are able to do in cooperation with others any of them will likely change an individual’s score
rather than what they can do alone (Sternberg on the test (Cronbach, 1990).
Matthew E. Poehner 325
In theory, if one were to observe every instance ment allow for interaction and feedback in order
of an individual’s typing, it would be possible to impact learning (e.g., Donato & McCormick,
to discern his or her true abilities from these 1994; Spence-Brown, 2003).
confounding variables. In reality, however, assess- I also wish to address Ellis’s (2003) description
ments usually involve very few observations, or of what he refers to as incidental informal formative
even a single observation of performance, serving assessment, in which learners are assessed as they
as the basis for generalizing individuals’ proba- carry out daily classroom tasks. In their review
ble functioning in other contexts. For this rea- of formative assessment practices, Poehner and
son, assessors go to great lengths to control for Lantolf (2005) argued that this approach also dif-
sources of variance in their procedures. Indeed, fers from DA because it is oriented toward provid-
most assessments require individuals to perform ing affective support to learners and helping them
in isolation, forbidding the use of computers, cal- get through the task at hand rather than support-
culators, reference materials, and other aids. In ing their development. As I explain in this article,
addition, many assessment procedures favor stan- DA—and especially the concept of TR—involve
dardization to reduce the risk that their results far more than improving learners’ performance
will be contaminated by factors external to an in- of a given task. For the purposes of the present
dividual examinee. discussion, I limit the use of the term DA to those
In contrast, DA links all assessment–instruction approaches that follow Vygotsky’s (1998) meth-
interactions through the provision of appropriate ods of providing mediation in order to reveal the
mediation to understand and intervene in devel- depth of learners’ abilities and simultaneously act
opment. Transcendence (TR), also referred to as as a catalyst for their further development.
transfer,1 has to do with individuals’ ability to re- In a classroom organized according to DA, ob-
contextualize their learning and apply it to new, servers would not be able to discern whether they
more demanding problems (Feuerstein, Rand, & are witnessing an assessment or an instructional
Hoffmann, 1979). Relying on examples of TR lesson because these are one and the same. Every
with L2 learners, I argue that in DA concerns interaction performs both an instructional and
over generalizability are obviated because the pur- evaluative function. During all sessions, learners
pose of the procedures is not to infer true abili- are encouraged to take on as much responsibility
ties from independent performance but rather for task completion as possible, and the media-
to promote learner development in a continuous tor remains ready to catch them when they “slip
fashion through appropriate interaction. I sug- over the edge” of their abilities (Newman, Griffin,
gest that the challenge for DA practitioners is to & Cole, 1989, p. 87). Mediators may document
fine-tune their support as learners become more their interactions with learners in different ways,
agentive and to introduce increasingly complex depending upon their need to generate results
tasks. In order to help the reader appreciate these (i.e., grades, scores, profiles, etc.) for administra-
distinctions, I now turn to a more general discus- tors, parents, and others. However, one must not
sion of DA and its theoretical origins in Vygotsky’s lose sight of the fact that in DA, all activities seek
work. to understand learners’ abilities by supporting
their development. This is true even when exter-
DYNAMIC ASSESSMENT nal pressures might require that learning be con-
ceptualized as a product that can be described and
The term dynamic assessment was coined by displayed for accountability purposes. In some DA
Vygotsky’s colleague Luria (1961) and popular- programs, the initial DA session is used as a diag-
ized by Israeli researcher and special educa- nostic of learners’ abilities and later repeated in
tor Reuven Feuerstein. Feuerstein contrasted his order to track developmental changes over time.
methods with other forms of assessment, which However, even these diagnostics are not set apart
he labeled static (Feuerstein et al., 1979). Given from the rest of the program; instead, they serve
the obvious negative connotations associated with as a point of departure for subsequent interven-
the latter term, I will use the terms DA and non- tions (see Feuerstein, Rand, Hoffmann, & Miller,
Dynamic Assessment (NDA) to capture Feuerstein’s 1980).
distinction between assessments that require the There are several dynamic approaches to psy-
provision of mediation to promote learner devel- choeducational assessment currently being pur-
opment and assessments that do not. I also recog- sued (for a review, see Lantolf & Poehner, 2004;
nize that nondynamic practices are not uniform Lidz & Elliott, 2000; Sternberg & Grigorenko,
with regard to not supporting learners. Indeed, 2002). Lantolf and Poehner (2004) explained
some forms of portfolio and project-based assess- that these models differ in how they approach
326 The Modern Language Journal 91 (2007)
mediation, with some models adhering to scripted potential development they are sharply different”
prompts and hints, and other models encourag- (Vygotsky, 1956, pp. 447–448).
ing open-ended dialogue between mediators and In the four decades since the ZPD was intro-
learners. They referred to the former models as duced to Western researchers, more and more
interventionist DA, and the latter models as interac- psychologists and educators have recognized the
tionist DA. Both are implementations of Vygotsky’s potential relevance of the concept for assess-
(1986) well-known proposal of the Zone of Proxi- ment, and consequently an increasing number
mal Development (ZPD). Because the ZPD is the of systematic procedures integrating mediation
theoretical foundation of DA procedures, I will into assessment have emerged (see Sternberg &
discuss it in detail before moving on to examples Grigorenko, 2002, for a review). In Feuerstein
of DA in L2 instruction. et al. (1979), Feuerstein explained the appeal of
DA by pointing out the approach’s inherently op-
timistic view of individuals. He argued that NDA
Theoretical Basis of Dynamic Assessment reveals only “the manifest repertoire of the indi-
vidual,” the abilities that have developed up to the
The ZPD is Vygotsky’s approach to understand- present (p. 95). Following Vygotsky’s argument
ing and supporting cognitive development. It rests that learners’ past achievements are not adequate
on two important, interrelated constructs: media- for predicting their future possibilities, Feuerstein
tion and internalization. According to SCT, indi- maintained that the principal goal in DA is “mod-
viduals are always mediated by cultural artifacts, ifying this repertoire [of abilities] by appropriate
social practices, and activities. They are mediated strategies of intervention” in order to help learn-
even when they are working alone, in which case ers construct a better future (p. 95).
their cognitive functioning is mediated by their
history of interactions with the world (Vygotsky, Illustration of Dynamic Assessment
1986). In other words, those abilities originally
residing in an individual’s social interactions be- To understand more clearly how a dynamic
come internalized and reemerge as new cognitive procedure might unfold, consider the following
functions. The individual no longer relies on the implementation of the Graduated Prompt Ap-
external environment for mediation but is able to proach to DA, as reported by Palinscar, Brown,
self-mediate, or self-regulate to use Vygotsky’s term. and Campione (1991). In contrast to many educa-
Observing a person’s independent performance tional applications of the ZPD in which mediation
reveals those functions that have been fully inter- unfolds dialogically (e. g., Aljaafreh & Lantolf,
nalized. In contrast, the level of performance an 1994), the Graduated Prompt Approach follows a
individual cannot reach independently, but only standardized approach to interaction: Mediation
through external forms of mediation, indicates is scripted prior to the assessment and standard-
those abilities that are still forming—the next or ized as a series of hints arranged from most im-
proximal level of development.2 plicit to most explicit. When learners encounter
One of Vygotsky’s favorite illustrations of the problems, the mediator first offers the most im-
ZPD involves two children who are both able to plicit hint in the repertoire. If this hint is suffi-
solve problems independently at a standard level cient for the learner to overcome the problem,
for 7-year-olds (Vygotsky, 1956, pp. 447–448, cited they move on. Otherwise the mediator provides
in Wertsch, 1985). Vygotsky used this example to the next hint, and so on until, if necessary, the
caution against making assumptions about indi- mediator actually reveals the solution to the prob-
viduals’ abilities solely on the basis of their inde- lem and offers an explanation of the principles
pendent performance. He explained that when involved.
these two children were offered assistance in the The procedure described by Palinscar et al.
form of hints, leading questions, and demonstra- (1991) was intended to assess school-aged chil-
tions, they responded differently. One child ben- dren’s pattern recognition abilities. The chil-
efited by improving his performance to that of dren were given a series of eight letters and
a 7.5-year-old while the other progressed to the asked to complete the pattern by determining
level of a 9-year-old. Vygotsky concluded that the the next four letters in the sequence. The pat-
similarity of the children’s independent perfor- tern GWHWIWJW_ _ _ _, for instance, required
mance masked important differences in their abil- the learners to identify that the letters proceeded
ities. As he put it, “From the point of view of alphabetically but were interspersed with a W. Ac-
their independent activity they are equivalent, cording to the authors, this sequence was formal-
but from the point of view of their immediate ized according to two “relations” — N (next) and
Matthew E. Poehner 327
I (identity, a repeating relation). In this case, the should not be understood as “training. . . oriented
correct solution was KWLW. Palinscar et al. (1991) toward a specific content” but rather as a series of
reported that not all the children were able to procedures designed to help learners continually
complete the pattern independently. The authors move beyond their current abilities and the here-
mediated the children’s engagement in the task and-now demands of a given problem (Feuerstein
by first asking if the problem was similar to any et al., 1979, p. 105). TR sessions, then, are part of
they had previously encountered. Children who a DA program and occur after learners have been
recalled other problems that followed the same brought to a point where they can perform with
pattern were asked to try again to complete the little or no reliance on the mediator. According to
task, whereas children who still failed to recog- Feuerstein, their independent performance sug-
nize the pattern were told to read each of the gests that they have internalized the support of-
letters aloud. If this reading also failed to help the fered to them through intervention, but the de-
child, the mediator began to ask relatively explicit gree to which they have developed is revealed only
leading questions such as “Are there any letters when they recontextualize their learning. TR tasks
written more than once in the problem? Which are more complex than those tasks previously en-
ones? Does this give you any ideas about how to countered by learners and require them to apply
continue?” (Palinscar et al., 1991, pp. 75–76). In principles they have already learned, as well as to
the event that even these leading questions did work out new ones with support from mediators
not help the child to answer correctly, the media- as needed. However, it is important to stress that
tor revealed the answer and explained the pattern to an observer a TR session would appear like any
before moving on to the next item. other DA interaction, with mediator and learner
Palinscar et al. (1991) reported that although jointly carrying out tasks.
some children needed minimal intervention to Let us return to the earlier DA example in
complete the tasks, others were able to do so which the learners searched for patterns in strings
only with extensive support from the mediator. of letters. Campione et al. (1984) explained that
The authors concluded that learners who would they formalized the notion of TR with a series
have received similar scores in a nondynamic pro- of transfer tasks that deviated to varying degrees
cedure actually varied in their ability level and from those previously introduced to learners.
that it was only by observing their response to Near transfer problems involve the same under-
mediation that these differences emerged. More- lying principles as original tasks but are often
over, this group of researchers contended that af- used in new combinations. For example, the sim-
ter learners have become proficient at complet- ple N I (next-identity) pattern for sequencing
ing the tasks used during the intervention, they the original string of letters presented to learn-
can be further differentiated according to how ers would be replaced by an NNNN relation, as in
well they are able to sustain their performance VHDPWIEQ _ _ _ _. This replacement would re-
when variations are introduced and tasks become quire learners to use the already learned “next”
more complex (Brown & Ferrara, 1985; Campi- relation to ascertain that the new pattern in-
one, Brown, Ferrara, & Bryant, 1984). That is, volved an alphabetic progression of every fourth
they are interested in learners’ capacity to tran- letter. The correct solution in this case would be
scend their abilities beyond a given DA session. XJFR. Far transfer problems are those problems in
The following section addresses this topic in more which learners use familiar principles in conjunc-
detail. tion with new ones. Here the learners received
UCTDSERF _ _ _ _. The solution to this problem
entailed the “next” relation but also identification
Development and Transcendence in Dynamic
Assessment of a “backward” (B) relation in order to complete
the pattern with QGPH. The final stage in this
To date, Feuerstein has articulated the most in- model, very far transfer, involved even more com-
depth formulation of TR (Feuerstein et al., 1979; plex problems. In the letter-sequencing exam-
Feuerstein, Rand, & Rynders, 1988). Feuerstein’s ple, children had to search for relations between
clinical work with underperforming learners in Is- strings of letters rather than within a string. They
rael revealed that many of the children struggled were given the phrase “six ships gone” and the se-
to establish connections among tasks and prob- quence THY RIHQR _ _ _ _. It was explained that
lems that were similar but not identical. He rea- the letters were a coded version of the message,
soned that while such learners may have gained and the children’s task was to encode the final
efficiency in performing a very specific task, they word (“gone”) following the same pattern. The
had not developed the ability to transcend their solution, HNOD, was reached by determining a
learning to new contexts. He explained that TR backward-next relation between the decoded and
328 The Modern Language Journal 91 (2007)
coded versions of the message (Campione et al., The prompt for the DA sessions was a scene
1984). from the Hugh Grant comedy Nine Months. The
Research on TR indicates that learners who clip was in English and contained both dialogue
perform in similar ways when completing famil- and action. During the first TR activity, clips from
iar tasks sometimes differ sharply when extending the film The Pianist were used. These were far
their abilities to new problems. In a particularly from comedic, and in fact, contained no dialogue
compelling example, a DA approach was used but were filled with intense war images and re-
with normal and learning-disabled school-aged quired learners to use fairly specialized vocab-
children engaged in reading tasks (Brown & ulary. The second TR task abandoned video in
Ferrara, 1985). During their initial DA session, the favor of written text—Voltaire’s Candide. During
two groups of learners diverged greatly in their both the DA and the TR sessions, the learners
ability to complete basic reading comprehension composed their narratives in French, but medi-
tasks. After regular intervention sessions where ation in English was offered to them in order
learners received mediation from tutors, they to be sure that they understood and could re-
were asked to reattempt the initial DA tasks. The spond to help. The examples below illustrate how
differences between the two groups all but disap- mediator–learner interactions during the DA and
peared, although in many cases they reemerged TR sessions promoted development and provided
as the children engaged in transfer tasks, with insights into the learners’ functioning. It must also
the greatest disparities occurring during very far be noted that, although the discussions we will
transfer. These results argued strongly in favor consider address grammatical features of the L2,
of DA’s potential to guide development, but they dynamic approaches can be developed for all as-
also pointed to the considerable variability among pects of language learning. In each of the follow-
learners that would be missed by a snapshot (or ing cases, the participants’ real names have been
static, to use Feuerstein’s term) approach to cap- replaced by pseudonyms.
turing learners’ abilities. In the next section, I
present examples of insights into L2 development Transcendence of a Learner’s Understanding
gained through TR. of Verbal Aspect

TRANSCENDENCE IN L2 DYNAMIC The first illustration of L2 transcendence comes

ASSESSMENT from Donna, a learner who struggled to select the
verbal aspect appropriate to her intended mean-
The following examples of L2 TR form part of a ings. During the initial DA session, it became
DA program for advanced (seventh semester) un- clear that Donna had been taught various rules-
dergraduate learners of L2 French (see Poehner, of-thumb that described instances when perfec-
2005). Participants were recruited from a tive and imperfective aspects (in French, the passé
4th-year French language course centered on composé and the imparfait) should be used, but
speaking and listening. The DA program supple- that she was not always able to apply the rules
mented regular classroom activities, with students correctly. When asked to verbalize why she had
meeting with a mediator (who was not otherwise selected one aspect over the other, her reasons
known to them) outside of class to help them were not always appropriate for the situation.
develop their oral abilities in French. The task The mediator therefore tailored Donna’s enrich-
put before the participants was to compose an ment sessions to focus on the concept of aspect
oral past-tense narrative recounting events from so that she could overcome her concern to se-
a video clip they were shown. This approach was lect the “right answer” and could instead select
selected because the video clip provided a com- aspect according to how she wished to portray
mon frame of reference for mediator and learner certain events in her narrative (i.e., emphasizing
to discuss and also required a high level of control their completed or ongoing aspect). Although
over the L2. The initial DA session identified prob- this presentation sometimes conflicted with the
lematic areas for individual learners, which were rule-based approach Donna had learned in her
then addressed through six weekly DA sessions. previous classes, she quickly came to appreciate
Much of the intervention focused on control of that a conceptual understanding allowed her to
verbal tense and aspect, which was especially diffi- make choices to reflect the meanings she wanted
cult for all the learners, but the intervention was to express. At the end of the enrichment period,
tailored to learners’ individual needs. At the end Donna was generally able to use the passé composé
of the program, the initial DA was repeated and and imparfait appropriately and could offer rea-
then two TR activities were completed. soned explanations for her selection of each
Matthew E. Poehner 329
respective aspect. Her new abilities were appar- the way she wished to portray the event in her nar-
ent when she repeated the initial DA session and rative. Donna’s hesitation may be due in part to
required only minimal support from the media- the use of être ‘to be’, which in this case would be
tor. Moreover, Donna’s understanding of aspect equivalent to was in English. Typically, French im-
also manifested itself in her performance during perfect constructions are translated into English
the transcendence sessions. using the to be verb in this way (e. g., elle parlait
The following exchange between Donna and ‘she was talking’), and so the use of was here may
the mediator is representative of her performance suggest to her the imparfait.
when she repeated the initial DA. At this point in In this instance, Donna receives minimal help
her narrative, Donna is trying to describe a char- from M. Continuing to use English to mediate
acter’s anger. While considering lexical alterna- herself, as she and M had done frequently during
tives, she becomes confused as to which aspectual enrichment, Donna considers both a perfective
form best fits the context she wishes to create, and and imperfective version of the statement, ulti-
she turns to the mediator (M) for help. Donna mately settling on the perfective version. From
(D) initially uses the verb devenir ‘to become’ but this choice, it is clear that Donna is aware of the
then switches to the expression être en colère ‘to be consequences of both linguistic forms for how the
angry’: event is portrayed. This conclusion is further cor-
roborated in lines 7 and 8 when Donna explains
1 D: . . . elle devenait uh elle avait elle deve-
her decision, referencing the change in the char-
nait fâchée elle devenait elle a été
acter’s state of being that is emphasized by the
. . . she was becoming uh she was having she
perfective aspect.
was becoming she was
In this exchange, M’s most obvious contribu-
2 elle était en colère quelle était la mieux?
tion was to suggest a lexical alternative, se fâcher
she was angry which was the better one?
(to be angry), in line 9. Donna responded by mak-
3 M: well uh—
ing a mental note (line 11) connecting the adjec-
4 D: she became angry
tive fâchée, which she had used earlier, with this
5 M: she well uh do you want to use imparfait
verb. The responsibility for performance rested
or passé composé how do you
largely with Donna. It was she who constructed
6 want to do it?
the meaning and selected perfective aspect. Of
7 D: she became angry she was being angry
course, it is not possible to know how Donna
she became angry that’s what I
would have performed in this case had M not
8 want to say
been present to serve as a sounding board for
9 M: right well um you could use the verb
her ideas. In other words, even though she was
se fâcher [but would it change
able to resolve the problem independently, simply
to be angry
knowing that M was present to offer pedagogical
10 sort of how you
support may have served a mediational function.
11 D: (to self) it’s a verb]
Aljaafreh and Lantolf (1994), in their study of
12 M: you know what you’re emphasizing if
tutor–tutee interactions in the ZPD, argued that
you’re using imparfait or passé
there is a qualitative difference between a learner
13 composé like um if you were saying just
performing a task in isolation and that same indi-
here a second ago she got angry
vidual performing the task when in the presence
14 D: there was a definite point where she be-
of a tutor or mediator. For Aljaafreh and Lantolf,
came angry so that would be
even if the tutor or mediator does not engage in
15 passé composé
overt interaction with the learner, his or her pres-
16 M: yeah
ence creates a “collaborative posture” whereby the
17 D: elle s’est fâchée? Elle s’est fâchée et uh
learner expects that such an interaction is possi-
juste après ça . . .
ble (p. 471). In the previous example, M was not
she got angry? She got angry and uh just
needed to guide Donna or to provide her with
after that. . .
information that would help her overcome the
In lines 1 and 2, Donna considers both the per- problem, but his presence may have prompted
fect and imperfect and asks M which aspect is bet- her to talk through the problem and thereby me-
ter, suggesting that she is aware that her choice diate her own performance. Of course, in most
of aspect will impact the meaning. Then, before forms of NDA, assessors are permitted only to
M can respond, she switches to English (line 4) observe learners’ behaviors rather than support
in order to clarify for M (and possibly for herself) their striving toward autonomy, in which case the
330 The Modern Language Journal 91 (2007)
assessor’s presence not only fails to provide a col- She produces both a perfective and imperfective
laborative posturing but may even have a debili- construction and, as before, looks to M for ad-
tating effect on the learner. vice. Once again, M does not provide an answer
As mentioned, Donna was able to sustain her but redirects her to consider which form best cap-
high level of control over verbal aspect even dur- tures her meaning. In lines 29 to 32, Donna ex-
ing the more challenging TRs. In the following ex- plains how the scene would be framed differently
ample from TR1, Donna is narrating a scene from by both forms, revealing again that she is very
The Pianist in which the title character avoids cap- much aware of how to use the passé composé and
ture by German soldiers. Once again, she strug- the imparfait. In line 33, we see that, as in the
gles with a verbal construction but resolves the earlier example, M still has a role to play, albeit
difficulty largely on her own: a limited one. To some degree, he continues to
evaluate Donna’s performance, either accepting
18 D: il savait bien qu’il y a quelqu’un qu’il y
her choices and explanations or prompting her
avait quelqu’un qu’il y avait
to reconsider them.
he knew well that there is someone that there
During both the DA and TR, it is not clear what
was someone that there was
impact, if any, the mediator’s positive evaluation
19 quelqu’un dans l’atelier mais le soldat ne
had on Donna’s performance. What must be em-
peut∗ trouver donc tout à fait—
phasized is that in both instances Donna was able
someone in the attic but the soldier can’t find
primarily to mediate herself. She independently
therefore completely
determined how to discuss various past events, se-
20 M:il savait bien qu’il y avait quelqu’un dans
lected the verbal aspects appropriate to the mean-
l’atelier mais il?
ings she wished to convey, and correctly produced
he knew well that there was someone in the
the necessary forms. The fact that she was able to
attic but he?
maintain her high level of performance during
21 D: il ne pouvait pas trouver il ne pouvait pas
the TR, when she had to circumvent lexical chal-
le trouver, c’est mieux que il
lenges, further supports the view of her abilities
he couldn’t find he couldn’t find him, that’s
that emerged from her second DA session. Thus,
better than he
the TR provided additional evidence of Donna’s
22 n’a pas pu le trouver?
level of functioning. The next case considered il-
couldn’t find him?
lustrates a decidedly different function of TR.
23 M: I guess it depends on the meaning right?
il ne pouvait pas trouver or il
24 n’a pas pu trouvé, either is grammatical. . .
25 D: je peux faire l’imparfait je crois Transcendence and the Reemergence of Problems
I can do the imperfect I believe
Not all learners were able to transcend their
26 M:all right
level of performance from DA to the TRs as easily
27 D: il ne pouvait pas trouver—
as Donna. One participant, Jess, had developed a
he couldn’t find—
sophisticated conceptual understanding of verbal
28 M: you see the difference in meaning be-
aspect whereby she, like Donna, had learned to se-
tween the two?
lect the passé composé and the imparfait according
29 D: well he couldn’t find him and then he
to how she wished to bring events into discourse.
stopped looking for him would
However, while engaging in the more demanding
30 be the passé composé, l’imparfait would be
TRs, she reverted to a less functional, rule-based
he couldn’t find him but
approach for selecting verbal forms, revealing that
31 there’s no it doesn’t imply a time when
her new basis for selecting aspect was coexisting
the soldier stopped looking for
with the rules-of-thumb she had learned in her
32 him
courses.
33 M:right so it kind of like depends I think on
Negueruela (2003) reported a similar finding
what you follow it up with
in his study of concept-based instruction among
This exchange begins when M realizes that Spanish L2 learners. In his study, after learners
Donna inappropriately slipped into the present had worked extensively with linguistic concepts,
tense in line 19 while using the verb pouvoir ‘to including tense, aspect, and mood, their per-
be able to’. Rather than point out the error, M formance revealed that for some learners their
simply asks that she repeat what she had said, at history of learning rules occasionally resurfaced,
which point Donna correctly puts pouvoir in the overlapping with and even overriding their con-
past tense but is uncertain which aspect is better. ceptual understanding.
Matthew E. Poehner 331
In the example below, taken from Jess’s sec- described a Jewish uprising depicted in The Pi-
ond DA session following enrichment, her expla- anist, and included the detail that German sol-
nation of aspectual choice makes it clear that her diers were killed. We pick up the exchange as she
decisions are grounded in the meaning conveyed is considering how to produce this passive voice
by the forms. In particular, Jess (J) employs both construction:
the imparfait and the passé composé to describe
two female characters, Christine and Rebecca, as 45 J: [to self] ils étaient how do you say killed?
praying mantises: they were
46 M: killed? Tué
34 J: . . . et Sean il il explique il a expliqué uh, 47 J: tués par des coups de fusil aussi
cette cette chose à Samuel et il il a killed by gunshots also
. . . and Sean he he explains he explained uh, 48 M: using the plus-que-parfait?
this this thing to Samuel and he he the pluperfect?
35 déclaré qu’elle était comme une insecte 49 J: uh what is it? La voix passive?
declared that she [Christine] was like an the passive voice?
insect 50 M: oh okay right you can use the voix passive
(M and J laugh) but you’re using which verb
36 elle était une mantis et Samuel il avait peur 51 tense though?
de Rebecca que et il parce 52 J: um imparfait
she was a mantis and Samuel he was afraid
of Rebecca that and he M intervenes in line 48 to determine which
37 qu’elle était comme une mantis main- form Jess is trying to produce. At first, he mistak-
tenant enly assumes that she has incorrectly formed the
because she was like a mantis now plus-que-parfait or past perfect (they had killed).
(M and J laugh) Upon hearing that she wishes to use the passive
38 il a regardé à Rebecca et elle est devenue voice, M shifts his focus to helping her select the
une insecte appropriate aspect to convey her intended mean-
he looked at Rebecca and she became an insect ing. Initially, she states that she wants to use the
imperfect and so M pursues this statement to un-
In lines 35 to 37, Jess uses the imperfect of derstand why:
the verb être ‘to be’ but then switches in line 38
to the perfect of the verb devenir ‘to become’. 53 M: instead of passé composé
These choices are indeed appropriate, but M sub- 54 J: . . . yeah
sequently intervenes to determine Jess’s level of 55 M: to say that they were they were killed
awareness of her choices. She explains her rea- 56 J: uh huh
soning as follows: 57 M: okay and how come?
58 J: I don’t know actually should be saying ils
39 J: . . . it’s a description of Christine who’s
(XXX). . . because I have to say was
like in this situation so we don’t
59 otherwise they would be killing someone
40 ever meet her we just get a description
else
but Rebecca all of a sudden becomes
60 M: ils ont tué would be they killed but you
41 this same thing so it’s like it’s an actual
want to say they were killed
event in the movie she becomes a
61 J: right par quelqu’un
42 praying mantis
by someone
43 M: okay okay
62 M: so you need another verb in there
44 J: so that would have been passé composé
63 J: ils étaient tués ils avaient tué
The fact that Jess is able to verbalize precisely they were killed ∗ they had killed
how the same phenomenon—a woman being like 64 M: well then using plus-que-parfait they had
a praying mantis—is framed differently depend- killed?
ing upon the aspect used suggests that she has 65 J: no how would I say they were killed? Ils
become sensitive to the meaning potentials of the étaient tués
passé composé and the imparfait. This conclusion is 66 M: étaient? So you’re using être but être can
further supported by her overall high level of per- be used in the imparfait or the
formance during the remainder of that session. 67 passé composé right so you could use the
During the TR, however, her interactions with passé composé—
the mediator revealed that some confusion over 68 J: ils ont été tués
the use of aspect lingered on. At one point, Jess they were killed
332 The Modern Language Journal 91 (2007)
69 M: ils ont été tués so it’s the passé composé of The difference in mediational support for these
être and tués as an adjective two learners during TR suggests that Donna had
they were killed more fully internalized her previous interactions
70 J: oui ça marche with the mediator and was consequently closer to
yes that works full independent performance. Generalizations
71 M: okay makes sense? about Donna’s and Jess’s abilities based solely on
72 J: oui, ils ont été tués par des coups des fusil their DA performance would have overlooked the
yes, they were killed by gunshots differences that became apparent only when the
learners moved beyond the assessments to more
Extensive mediation was required to uncover complex tasks. We now turn to a final contribution
the source of the problem and to help the learner of transcendence that is worth considering and
produce an appropriate response. In lines 58 and that involves the extension of mediated learning
59, Jess explains that her use of être is motivated from one linguistic feature to another.
by her need to include was, in order for the con-
struction to be rendered in the passive rather than
Transcendence to Other Linguistic Features
active voice. As noted in the example involving
Donna, learners sometimes assume that using was The concept of transcendence, as outlined by
in English denotes the imperfect in French. In this Feuerstein and his colleagues (Feuerstein et al.,
case, the meaning Jess constructs in English is not 1988), involves not only extending performance
matched by her French utterance. M endeavors to other tasks but also creating new connections
to help her realize that either aspect can be used within a domain of study. The following interac-
with être and that the perfect is in fact a better tion between the mediator and Amanda captures
choice. M, at first, proceeds implicitly by drawing this aspect of transcendence. In this case, the me-
Jess’s attention to the perfect construction ils ont diator and the learner are collaboratively moving
tué ‘they killed’ in line 60 and then pointing out beyond their discussion of the passé composé and
that another verb is needed for the passive. When imparfait and considering employing the plus-que-
this assistance fails to help, M explicitly reminds parfait, or past perfect, to reframe events in the
her in lines 66 and 67 that être can be marked for narrative. In this exchange, M prompts Amanda
either aspect. Only then is Jess able to produce (A) to rethink her use of the passé composé with
the correct form, and M follows this up with an the verb finir (to finish). Although he does not
explanation in line 69. explicitly name the form he has in mind, it is clear
This example highlights the contribution of the that he believes the past perfect is the appropri-
transcendence model. Both Donna and Jess ex- ate form to express the meaning Amanda is after.
hibited high levels of performance during DA Use of the past perfect requires a shift in orien-
and required minimal support from the media- tation, with the reference point occurring in the
tor. However, differences between the two learn- past and a second event taking place at an earlier
ers emerged as they engaged in TR tasks. Whereas time than the reference point. Amanda attempts
Donna was able to sustain her functioning more to use her understanding of the imparfait and the
or less independently, Jess was not. Both learn- passé composé as a basis for considering how she
ers needed some support from the mediator as might employ the past perfect:
the tasks became more complex, but the nature
of this support was different. Donna was able to 73 A: les deux se sont en colère∗ , parce qu’ils
make aspectual choices and produce the correct n’ont pas fini leur discu argument
linguistic forms, but she remained uncertain of the two are angry, because they didn’t finish
their appropriateness. In effect, this is an example their discu argument
of what Appel and Lantolf (1994), following the 74 et uh
work of von Kleist, described as trying out one’s 75 M: actually they hadn’t finished their discus-
meanings on another; in this case, Donna tested sion right because it kind of
her choices by bouncing them off the mediator. 76 takes place before (. . .) so they arrived in
Approval of performance, then, still resided with the past but they hadn’t finished
the mediator during Donna’s TR, but in Jess’s case 77 their discussion at an earlier time in the
the mediator had to do much more. Jess struggled past
to orient properly to the task, oscillating between 78 A: ils ne finissions∗ ? Or uh finissent?
a conceptual and a rule-based approach to select- they were finishing? Or uh finish?
ing aspect, and needed the mediator’s help to 79 M: actually it’s more with the auxiliary verb
carry out the activity. that you see the change
Matthew E. Poehner 333
80 A: oh se (. . .) s’était? s’était under assessment conditions and drawing infer-
81 M: almost almost right (. . .) except you’re ences about their underlying abilities in order
using the other verb the other to make statements regarding their probable per-
82 auxiliary formance in nonassessment situations. Bachman
83 A: uh, right (. . .) (looks confused) (1990) added that assessors must take great care
84 M: you were using avoir with uh uh finir to model abilities such as language proficiency ap-
right? propriately so that one can have confidence that
85 A: oui uh ils n’avaient pas fini assessments actually reveal something about the
yes uh they hadn’t finished abilities they purport to measure. Interestingly,
86 M: right exactly McNamara (2004) referred to this mode of oper-
ating as “a testing version of Labov’s Observer’s
In line 78, Amanda is unable to incorporate M’s Paradox,” which states that an object under ob-
comment and searches for various forms of finir, servation changes as a result of being observed
and so M becomes more explicit, directing her (p. 765). In NDA, this mode is highly problem-
attention to the auxiliary verb. Although Amanda atic because NDA relies on strict procedures for
requires a good deal of support, it is ultimately she interpreting assessment performances—many of
(not M) who produces the correct form in line 85. which involve statistical modeling—that are pred-
To be sure, this exchange cannot be taken as icated on the assumption that abilities are not
evidence that Amanda had full understanding of changing during the assessment (see Glutting &
and control over the plus-que-parfait. In fact, she McDermott, 1990). In this way, assessors are able
did not employ any other past perfect structures to use numbers (e. g., scores, percentages, and
in that session. Nevertheless, she was able to use percentile rankings) to describe the amount of
the appropriate morphological marking (line 80), ability an individual possesses.4
and after she was prompted to switch from être to As the above cases of L2 transcendence illus-
avoir, she managed to combine the elements—the trate, in DA one continually alters both tasks and
subject pronoun, the auxiliary, the past participle, mediation in order to work successfully within a
and the negation—to produce the correct struc- learner’s ZPD because individuals’ abilities and
ture ils n’avaient pas fini. Moreover, her search corresponding developmental needs are always
for an alternative form after hearing M’s trans- emergent. DA follows Vygotsky’s reasoning that
lation of the preferred structure in line 76, they the principal responsibility of education is not to
hadn’t finished, suggests that she was aware that her document whether learner performance is prob-
earlier passé composé construction did not carry lematic but to discover the underlying causes for
that precise meaning. With appropriate media- that performance in order to help learners set
tion, Amanda extended her knowledge of present new developmental trajectories. Vygotsky (1998)
perfective and imperfective aspects to a related likened the results of conventional assessment
linguistic feature that she had not previously dis- procedures to an “empty” medical diagnosis in
cussed with the mediator. which a doctor merely restates in scientific terms
what is already apparent to the patient (p. 205).
For Vygotsky, this emptiness is unacceptable and
DISCUSSION: RECONSIDERING
GENERALIZABILITY FROM A DYNAMIC must be replaced with a “true diagnosis” that in-
ASSESSMENT PERSPECTIVE cludes “an explanation, prediction, and scientific
basis for practical prescription” (p. 205). DA en-
Thus far, we have considered how DA differs deavors to do just that, and, as a result, many of
theoretically and methodologically from NDA, the traditional principles in NDA for interpreting
with L2 transcendence examples highlighting in- assessment performance do not apply. I take, as
sights into learner development afforded by a a case in point, the concept of generalizability,
monistic approach to instruction and assessment. which contrasts sharply with transcendence.
In the remainder of this article, I address the con- Generalizability has been a major preoccupa-
sequences of a dynamic perspective for the in- tion for assessors since Campbell and Stanley
terpretation of assessment outcomes. As should (1963) first expressed concern over the authority
be clear from the preceding examples, DA and granted to findings obtained under experimental
NDA are fundamentally different endeavors.3 Ac- conditions, which they referred to as the “exter-
cording to McNamara (2004), NDA is “a process nal validity” of experimental procedures (Messick,
of gathering information about test-takers from 1989, p. 56). Since that time, several aspects
observed performance” (p. 765). He explained of generalizability have emerged in the assess-
that this process involves observing individuals ment literature. Messick (1989) grouped these as
334 The Modern Language Journal 91 (2007)
follows: population, ecological, temporal, and charged that this omission has led to “an erasure
task generalizabilities. Population generalizability of the social” in which performance is assumed
refers to generalizations across groups of individ- to be a direct outcome of an individual’s compe-
uals. This form of generalizability, characteristic tence rather than the result of interactions among
of experimental research, concerns how well re- individuals and their available resources (p. 337).
sults of procedures carried out with one set of Van Lier (2004) similarly railed against the treat-
individuals can be said to pertain to other sets ment of context in much L2 research, arguing that
of individuals who are thought to share common contextual information is not something that can
relevant characteristics. Because my interest here simply be “added on to whatever is investigated, in
is in the assessment of individuals’ abilities, I will a supplementary sort of way” (p. 5). He observed
not address the problems involved in generalizing that such practices raise questions regarding how
from one population to another.5 The discussion much contextual information is important as well
will instead focus on the other facets of generaliz- as how it is important.
ability described by Messick, and in particular on Context gains new relevance in DA as the fo-
how a dynamic approach to assessment compels cus broadens from individuals working in isola-
us to rethink these constructs. Where appropriate, tion to interactions among individuals and cul-
I will reference the preceding L2 transcendence tural artifacts. As explained earlier, NDA captures
interactions to illustrate the implications of fol- independent performance, during which indi-
lowing a dynamic model. viduals mediate their own functioning, but DA
encourages mediator–learner cooperation and
Contextual Variables and the Individual the use of artifacts when learners are unable to
self-regulate. An important consequence of this
Ecological generalizability has to do with gener- approach for ecological generalizability is that as-
alizations from one context to another. In ex- sessment is brought more in line with other ac-
perimental research, ecological generalizability is tivities in life where individuals cooperate with
usually discussed in terms of differences between one another and employ various tools to achieve
the laboratory and other contexts, whereas in as- their intended outcomes (Luria, 1973; Wertsch,
sessment it leads to statements about individuals’ 1998). Put another way, DA removes learners from
likely performance in various situations on the the “curious kind of isolation” (McNamara, 1997,
basis of their assessment performance. The pref- p. 449) that is characteristic of NDA but unchar-
erence of NDA for quantification and measure- acteristic of most other activities in daily life.
ment proceeds from methods developed in the In the case of formal education, DA fully in-
natural sciences, where it is often assumed that tegrates assessment with instructional activities,
certain variables can be held constant while oth- since both involve providing appropriate medi-
ers are manipulated (see Ratner, 2006, for discus- ation to support learner development. General-
sion). Contextual factors framing human activities izing assessment performance to classroom ac-
can then presumably be selected and controlled, tivities becomes a nonissue because instruction
and so assessors can create contexts into which is part of assessment just as assessment is ongo-
individuals—and their relatively stable repertoires ing during instruction. In this way, it is possible
of abilities—can be deposited. To have confidence to track learner development during and across
in their generalizations, assessors devise contexts sessions. As I have argued elsewhere (Poehner,
thought to parallel closely those to which they in press), tracking development involves care-
wish to generalize. ful attention to changes regarding the types of
The relationship between the individual and problems learners encounter, the mediation they
context posited by ecological generalizability is require to overcome these problems, their respon-
problematic in DA. Whereas NDA views perfor- siveness to mediation, and their efforts to gain
mance as the provenance of the individual and greater autonomy.
context as a backdrop to that performance, DA Transcendence is fundamental to tracking de-
understands performance to emerge from the in- velopment because it involves going beyond the
terplay among individuals, the activities in which test as learners endeavor to recontextualize their
they engage, and the culturally constructed arti- abilities while engaging in new tasks. To return
facts they employ. McNamara (2001) explained to the L2 French examples described above,
this perspective in his critique of language as- Donna and Jess performed in similar ways upon
sessment practices. He contended that assessors repetition of the initial DA task. It was only
have not taken account of findings from research when the task was made more complex that
into the social nature of human cognition, and he differences between these two learners became
Matthew E. Poehner 335
manifest. Had their second attempt at the DA task Temporal generalizability makes little sense in
been regarded as a standalone assessment and DA because, as explained, change rather than sta-
used as the basis for generalizations about their bility is a defining feature of this approach. Each
likely performance in other situations, one would mediator–learner interaction is designed not only
have expected them to function at a similar level. to capture learners’ current level of ability but to
However, Donna’s successful use of verbal aspect help them develop beyond it. If a learner’s perfor-
during transcendence, compared with the prob- mance remained constant over time, this might
lems encountered by Jess, reveals that such an in- suggest a lack of development, in which case the
ference would have been incorrect. Furthermore, mediator would need to reevaluate carefully his
analysis of mediator–learner interactions during or her approach to interacting with the learner
transcendence illuminates the type of support in order to provide support that lies within the
both learners needed to meet the demands posed learner’s ZPD. As we saw in the example of tran-
by the transcendence tasks. The insights gained scendence involving Jess, the mediator had to pro-
by jointly engaging with learners in more difficult vide different kinds of support at various points
tasks thus clarify the degree to which they can self- in time to remain attuned to her changing needs.
regulate and the forms of mediation needed for During her initial DA session, Jess required exten-
their further development. sive support from the mediator as she selected the
appropriate verbal forms for her narrative. The
From Predictability to a Preferred Future learner’s second attempt at this task, after she had
received explicit instruction in verbal aspect and
The third aspect of generalizability that Messick had been given several opportunities to practice
(1989) described, temporal generalizability, involves using perfective and imperfective aspects when
predictions about how individuals will perform in recounting past events, was markedly different.
the future. Sometimes referred to as predictive va- The mediator’s role this time was merely to con-
lidity, this concept is especially relevant to aptitude firm Jess’s aspectual choices. Later, however, when
testing, where examinees’ present performance presented with a more challenging narrative to ex-
is taken to be a strong indicator of their future plain, Jess’s earlier confusion reemerged, and the
performance. For example, an assessment is of- mediator played a more active part in helping her
ten used to predict how successful individuals are frame the events in the story.
likely to be in a rigorous foreign language pro- An additional point worth noting is that DA’s
gram. Candidates who perform well are believed goal of promoting learner development is rooted
to have a good chance at success. Of course, the in the idea that development itself does not have
obvious flip side of this argument is that candi- an endpoint (such as earning a high score on a
dates with poorer assessment outcomes cannot be test) but is instead about moving beyond one’s cur-
expected to do well. rent level of ability, whatever it might be. Indeed,
As Lantolf and Poehner (2004) observed, the Feuerstein et al.’s (1988) book indicates in its title
reasoning behind such an approach is that the the goal of DA: Don’t accept me as I am: Helping “re-
future is assumed to be a smooth continuation tarded” people to excel . The L2 example presented
of the present. As such, the cognitive processes earlier, in which the mediator and Amanda dis-
that have brought about an individual’s abilities cussed the pluperfect, illustrates the significance
at the present time are unproblematically used as of this perspective. With appropriate mediation,
the basis for predictions about the person’s fu- Amanda extended her understanding of verbal
ture abilities. These authors argued that, from aspect beyond the passé compose–imparfait distinc-
the perspective of DA, the relationship between tion she had studied and explored new ways of
the present and the future is not so easily de- using the L2 grammar to create meanings. Thus,
fined because the goal of DA is to make visible the while Amanda gained confidence with one fea-
cognitive processes underlying a learner’s perfor- ture of the language, she was also preparing to
mance, which is achieved through mediated in- tackle new problems that might arise.
teraction. This approach allows DA practitioners
to address the source of an individual’s difficul- A Task by Any Other Name
ties rather than simply concluding that there is a
problem. That is, the desired outcome of assess- Task generalizability refers to generalizations
ment procedures should be to understand indi- from assessment tasks to other tasks assumed to
viduals’ potential for future development and to require the same abilities. As with ecological gen-
help them realize that future. eralizability, task generalizability presupposes a
336 The Modern Language Journal 91 (2007)
bifurcation between assessment and teaching. In for every kind of learning task about which one
other words, one need only worry about general- might make predictions? Moreover, what if one
izing from assessment tasks to instructional tasks wishes (or needs) to make predictions about how
if these are separate activities. Researchers typi- individuals will perform in nonacademic settings?
cally seek to design assessment tasks that require Are specific tasks needed to approximate every
the same abilities and knowledge as those tasks nonpedagogical situation that might occur out-
that are used for pedagogical purposes because side of school?
they can then generalize with greater confidence. As should be clear from this discussion, DA
The work of Coughlan and Duff (1994) and differs from task-based approaches in that the
Spence-Brown (2003) suggested that even when purpose of the assessment is to support learner
assessment and learning activities are carefully development, and this purpose goes beyond im-
coordinated, learners are very much aware when proved performance on any particular task. For
they are being assessed, and this awareness affects this reason, transcendence involves altering tasks
how they orient to tasks. For example, Spence- to make them more challenging, rather than de-
Brown reported on an authentic performance- signing them to mirror those tasks learners have
based assessment that required learners of L2 already encountered. As we saw in the contrasting
Japanese to interview native speakers of Japanese L2 examples of Donna and Jess, this approach
in the local community. Some of the students fol- has the potential both to distinguish learners
lowed the teacher’s preference for open-ended who would otherwise erroneously be assumed to
and spontaneous interactions with their intervie- have achieved a similar level of functioning and
wees, but others admitted to scripting and rehears- to help learners develop beyond their current
ing their dialogues carefully. She argued that the abilities.
former group of students benefited the most from
the encounters but did not earn grades as high as
those learners who were better at playing the as- CONCLUSION
sessment game.
The growing interest in task-based learning and As noted, the widely accepted approach to un-
assessment further attests to the desire to align derstanding abilities in NDA involves theorizing
assessment and instruction practices. Task-based constructs and making inferences based upon ob-
practitioners design parallel tasks, some of which servations of independent performance, a process
are used for assessment purposes and some for in- that has been likened to Labov’s Observer’s Para-
struction. These assessments therefore represent dox (McNamara, 2004). Labov (1972) noted that
an important departure from traditional assess- informants often changed their manner of speak-
ments because generalizations are not based ex- ing in response to his presence, leading to his for-
clusively on theorized constructs but are instead mulation of the well-known paradox: The object
grounded in the actual instructional activities in we wish to study (especially in the case of human
which learners take part. Indeed, Candlin (2001) phenomena, such as language performance) is
observed that assessment and learning tasks can transformed as a result of our efforts to under-
be distinguished primarily by the presence of stand it. For researchers who model their meth-
teacher support during the latter.6 Of course, ods after those developed for the natural sciences,
given the importance in task-based approaches Labov’s paradox may be seen as a challenge to the
of bringing instruction and assessment together, entire enterprise.
one is left to wonder why teacher support is con- In DA, the reality that assessment procedures
sidered relevant only for learning but not for as- themselves affect the abilities being assessed is
sessment. It seems, therefore, that even when the not cause for concern, and fits quite well with
assessment and instructional tasks are parallel, the Vygotsky’s central argument that higher cogni-
conditions under which learners carry them out tive functions are social in origin. It makes lit-
are not, and the examinee-in-isolation model of tle sense to assess them by studying solo perfor-
assessment persists. mance because, as Vygotsky (1998) pointed out,
At first glance, task-based assessments might this solo performance reveals only part of the pic-
seem to offer a way around generalizing to other ture. Independent functioning focuses on indi-
contexts because assessment and instructional ac- viduals’ ability to self-regulate, and therefore re-
tivities are so similar. However, at some point, the veals only those abilities that have already fully
issue of underlying abilities must reenter the pic- developed. Of course, in daily life most human
ture. For instance, are specific assessments needed activities are not performed independently but
Matthew E. Poehner 337
collaboratively, and involve various types of cul- In DA, generalizations to hypothetical situa-
tural resources. “Showing off” one’s ability to func- tions are replaced with concrete TR activities
tion independently is indeed an artifact of con- in which mediators and learners collaboratively
temporary testing practices. In this regard, DA’s carry out new tasks, with prior interactions serving
interactive nature is in itself important for asses- as a point of departure. Rather than aligning as-
sors seeking to generalize about individuals’ abil- sessment tasks with pedagogical ones, TR tasks are
ities because it reduces the extent to which assess- made increasingly complex in order to challenge
ment is set apart from other contexts. learners, and the forms of mediation offered are
As we have seen, offering learners mediation, continually attuned to learners’ needs (i.e., their
especially through dialogic interaction, brings to growing autonomy but also the emergence of new
light abilities that are still developing. Analysis of difficulties). TR activities are yet another step in
mediator–learner interactions makes it possible an ongoing series of mediator–learner interac-
to understand how close learners are to gaining tions designed to bring about new levels of de-
control over these emerging functions. Moreover, velopment. From a Vygotskian perspective, there
the instructional orientation of mediated inter- is no endpoint to development because there are
actions means that this form of assessment also always new problems to solve.
promotes learner development. This is the signif- The examples of L2 DA discussed in this ar-
icance of the ZPD: It integrates assessment and ticle powerfully illustrate the benefits of moving
instruction according to the principle that me- interaction, rather than standardization, to the
diation reveals the full range of an individual’s fore of the procedure. For instance, differences
abilities while guiding development. in Donna’s and Jess’s ability to control verbal as-
Labov’s paradox is not a source of concern for pect come to light when one considers their dia-
DA practitioners because they embrace the op- loguing with the mediator, explanations they gave
portunity to intervene in learners’ development. for their linguistic choices, and their responsive-
Like NDA, DA seeks to understand cognitive abil- ness to the support the mediator offered. Whereas
ities, but it argues that this understanding can Donna primarily needed the presence and ap-
be realized only through actively helping learners proval of another person in order to sustain her
develop. Thus, while NDA proceeds by produc- high level of performance, Jess required more
ing snapshots of performance and inferring from extensive interaction with the mediator, includ-
these snapshots the state of an individual’s abilities ing translations of contrasting structures into En-
at a certain moment in time, DA is concerned with glish. It would therefore be a mistake to assume
helping individuals. To be sure, DA practition- that both these learners have the same abilities, as
ers recognize the accountability demands faced would have likely been concluded in an NDA pro-
by most classroom instructors, and for this rea- cedure. In fact, even in the DA session following
son, have devised various means of describing DA enrichment, both learners performed well; their
interactions, such as scores and counts of mediat- differences emerged only during the cooperative
ing moves (in interventionist DA), as well as qual- dialoguing that took place during transcendence.
itative profiles of the processes of learners’ de- Because DA unifies assessment and instruction
velopment (in interactionist DA). However, DA’s according to Vygotsky’s and Feuerstein’s concep-
primary goal is to help learners become more tualization of mediated mind, every mediator–
than they currently are. As Feuerstein has re- learner interaction—whether it is an initial DA,
peatedly argued, the purpose of DA is to “undo an enrichment activity, or focused more on tran-
the predictive value of the initial assessment by scendence to new problems—forms a link in an
modifying functioning through the mediational ongoing series of ZPD interactions in which me-
process” (Feuerstein et al., 1988, p. 199, empha- diation is continually attuned to learners. An
sis in original). Any interest in comparing an in- interesting area for future research would be to in-
dividual’s performance on a given occasion to a vestigate transcendence across modalities. In the
set of standards or to the performance of oth- present study, different modalities were used for
ers in a population is superseded by the primary stimulation (e. g., video clips and written texts),
goal of promoting learner development. Famil- but learners’ production was oral during both the
iar terms, such as criterion-referenced and norm- DA and TR tasks. Future work might include TR
referenced assessment, imply priorities that are tasks that require learners to apply their knowl-
not shared by DA. It may therefore be more ap- edge during listening, writing, or reading activi-
propriate to describe DA as development-referenced ties. Indeed, both Vygotsky and Feuerstein under-
or development-centered. stood that development can manifest itself in a
338 The Modern Language Journal 91 (2007)
variety of ways, and this development is after all at
the heart of transcendence. ACKNOWLEDGMENTS
The work presented in this article is a first step
toward understanding the contributions of a DA I would like to thank Jim Lantolf for his insightful
approach to L2 assessment. If this framework is suggestions and feedback on an earlier version of this
to be more fully developed within the L2 field, paper.
a good deal of additional research is needed.
For instance, a major challenge to implement-
ing DA concerns the feasibility of dialogic interac- NOTES
tion when classroom teachers may see dozens or
even hundreds of learners. Vygotsky (1998) him- 1 Transcendence and transfer , as will be explained later
self seemed to have recognized this problem and in this article, are terms developed in two different DA
proposed the possibility of constructing a ZPD traditions—Feuerstein’s Mediated Learning Experience
with groups of learners rather than with individu- (Feuerstein, Rand, & Rynders, 1988) and Brown’s Grad-
als. This approach entails mediating the group as uated Prompt Approach (Brown & Ferrara, 1985), re-
its members cooperatively engage in activities. In- spectively. In the DA literature, the concept of transcen-
dividual learners no doubt experience their own dence has been more robustly developed and is theo-
difficulties and can be expected to contribute dif- retically in line with Vygotsky’s (1986) research. In this
ferently, but mediating the group not only moves article, the term transcendence will be used throughout,
the group’s ZPD forward but also promotes the de- and transfer will be reserved for discussion of research
on the Graduated Prompt Approach to DA.
velopment of individuals (see Poehner & Lantolf, 2 The ZPD itself has been the subject of much debate
2005, for further discussion). Group DA repre- among Vygotskian researchers and has been interpreted
sents perhaps an even greater break from the and applied in various ways. A full discussion of the his-
preoccupation with the autonomous individual tory of the concept is beyond the scope of this article.
that underlies most accepted assessment practices The interested reader is referred to Chaiklin (2003) for
because comparisons among individuals cease to a general review and to Lantolf and Thorne (2006) for
make sense and the dynamics of learners’ cooper- a review that relates to L2 learning.
3 As explained at the outset of this article, DA dif-
ative activity becomes more relevant.
DA clearly posits a new educational ideology fers from both formative and summative forms of assess-
oriented toward learner development, and as such ment, although it is more in line with the former. In
the following discussion, my remarks about generaliz-
it adopts an inherently optimistic view of human
ability pertain most obviously to large-scale NDA when
beings and their development. Following Feuer- it adopts a summative character. However, as Cronbach
stein, DA chooses not to accept the view of in- (1990) observed, most of the concepts used to interpret
dividuals that emerges from a given assessment the results of standardized tests (e.g., generalizability, re-
procedure but to go beyond the test to help them liability, and validity) apply to other forms of assessment
initiate a new developmental trajectory. In this as well.
4 Many assessors are no doubt aware that when they
way, DA can be conceived as a liberatory peda-
gogy that supports learners while they construct speak of quantifying cognitive abilities, they are oper-
futures that would not be predicted on the basis of ating metaphorically. However, as Lakoff and Johnson
the past. Indeed, DA originates in Vygotsky’s and (1980) pointed out, the power of metaphor resides in
its ability to highlight certain aspects of a concept and
Feuerstein’s work in special education, which of-
hide others. In this case, the largely unspoken metaphor
ten produced life-altering results. The principles that abilities exist as quantities individuals possess in
of DA have subsequently been applied with simi- different amounts has become the normalized way of
larly impressive outcomes with such diverse pop- understanding human cognition, and assessment per-
ulations as immigrants, minorities, penitentiary formance is consequently taken to be a representative
inmates, gifted learners, and patients suffering sampling of the abilities in question.
5 It is worth noting, however, the considerable crit-
from dementia (see Tzuriel, 2001). The common
thread running through all this work is perhaps icism that has been leveled at population generaliz-
best captured by a remark made by Bruner (1980, ability. In linguistics, for example, approaches such as
cited in Feuerstein et al., 1988) regarding Feuer- Chomsky’s (1985), which make normative claims about
grammar on the basis of an educated native speaker’s
stein’s Mediated Learning Experience approach:
judgments (sometimes those of the linguist himself),
“MLE is not only for the handicapped, it is for have been increasingly called into question (see Hanks,
all of us since it’s MLE which makes us human!” 1996, for discussion). As a consequence, linguists are
(p. 58). It is my hope that DA will contribute to now turning their attention to corpus-based approaches
the developmental trajectory of L2 education in that capture what individuals actually do with language
the years ahead. as they go about the activity of living.
Matthew E. Poehner 339
6 It is important that Candlin (2001) did not provide Cronbach, L. J. (1990). Essentials of Psychological Testing
a description or examples of the support learners were (5th ed.). New York: Harper & Row.
offered during task-based instruction, and it is therefore Donato, R., & McCormick, D. (1994). A sociocultural
not possible to determine whether this support can be perspective on language learning strategies: The
considered mediation and whether it promotes learn- role of mediation. The Modern Language Journal,
ers’ ability to self-regulate. 78, 453–464.
Ellis, R. (2003). Task-based language learning and teach-
ing . Oxford: Oxford University Press.
Feuerstein, R., Rand, Y., & Hoffman, M. B. (1979). The
REFERENCES
dynamic assessment of retarded performers: The learn-
ing potential assessment device, theory, instruments,
Aljaafreh, A., & Lantolf, J. P. (1994). Negative feedback and techniques. Baltimore: University Park Press.
as regulation and second language learning in the Feuerstein, R., Rand, Y., Hoffman, M. B., & Miller, R.
zone of proximal development. The Modern Lan- (1980). Instrumental enrichment. Baltimore: Uni-
guage Journal, 78, 465–483. versity Park Press.
Antón, M. (2003). Dynamic assessment of advanced for- Feuerstein, R., Rand, Y., & Rynders, J. E. (1988). Don’t
eign language learners. Paper presented at the accept me as I am. Helping retarded performers excel.
American Association of Applied Linguistics, New York: Plenum.
Washington, DC. Glutting, J. J., & McDermott, P. A. (1990). Principles and
Appel, G., & Lantolf, J. P. (1994). Speaking as mediation: problems in learning potential. In C. R. Reynolds
A study of L1 and L2 text recall tasks. The Modern & R. W. Kamphaus (Eds.), Handbook of psycholog-
Language Journal, 78, 437–452. ical and educational assessment of children: Intelli-
Bachman, L. F. (1990). Fundamental considerations in lan- gence and achievement (pp. 296–347). New York:
guage testing . Oxford: Oxford University Press. Guilford.
Brown, A., & Ferrara, R. A. (1985). Diagnosing zones Hanks, W. F. (1996). Language and communicative prac-
of proximal development. In J. V. Wertsch (Ed.), tices. Chicago: University of Chicago Press.
Culture, communication and cognition: Vygotskian Labov, W. (1972). Language in the inner city. Philadel-
perspectives (pp. 273–305). Cambridge: Cambridge phia: University of Pennsylvania Press.
University Press. Lakoff, G., & Johnson, M. (1980). Metaphors we live by.
Bruner, J. (1980). Meeting of the National Institute of Chicago: University of Chicago Press.
Education, Pittsburgh, PA. Lantolf, J. P., & Poehner, M. E. (2004). Dynamic assess-
Campbell, D. T., & Stanley, J. C. (1963). Experimen- ment: Bringing the past into the future. Journal of
tal and quasi-experimental designs for research Applied Linguistics, 1, 49–74.
on teaching. In N. L. Gage (Ed.), Handbook of re- Lantolf, J. P., & Thorne, S. L. (2006). The sociogenesis of
search on teaching (pp. 471–535). Chicago: Rand second language development. Oxford: Oxford Uni-
McNally. versity Press.
Campione, J. C., Brown, A. L., Ferrera, R. A., & Bryant, Lidz, C. S., & Elliott, J. G. (2000). Dynamic assessment:
N. R. (1984). The zone of proximal development: Prevailing models and applications. Amsterdam:
Implications for individual differences and learn- Elsevier.
ing. In B. Rogoff & J. V. Wertsch (Eds.), Children’s Luria, A. R. (1961). Study of the abnormal child. Ameri-
learning in the “zone of proximal development” (pp. can Journal of Orthopsychiatry: A Journal of Human
77–91). San Francisco: Jossey-Bass. Behavior, 31, 1–16.
Candlin, C. N. (2001). Afterward: Taking the curriculum Luria, A. R. (1973). The working brain: An introduction to
to task. In M. Bygate, P. Skehan, & M. Swain (Eds.), neuropsychology. New York: Basic Books.
Researching pedagogic tasks: Second language learn- McNamara, T. (1997). “Interaction” in second language
ing, teaching, and testing (pp. 229–243). London: performance assessment: Whose performance?
Longman. Applied Linguistics, 18, 446–466.
Chaiklin, S. (2003). The zone of proximal development McNamara, T. (2001). Language assessment as social
in Vygotsky’s analysis of learning and instruction. practice: Challenges for research. Language Test-
In A. Kozulin, B. Gindis, V. S. Ageyev, & S. M. ing, 18, 333–351.
Miller (Eds.), Vygotsky’s educational theory in cul- McNamara, T. (2004). Language testing. In A. Davies &
tural context (pp. 39–64). Cambridge: Cambridge C. Elder (Eds.), The handbook of applied linguistics
University Press. (pp. 763–783). Malden, MA: Blackwell.
Chomsky, N. (1985). Methodological preliminaries. In Messick, S. A. (1989). Validity. In R. L. Linn (Ed.), Ed-
J. Katz (Ed.), The philosophy of linguistics (pp. 80– ucational measurement: Third edition (pp. 13–96).
125). Oxford: Oxford University Press. New York: American Council on Education.
Coughlan, P., & Duff, P. (1994). Same task, different ac- Negueruela, E. (2003). A sociocultural approach to teach-
tivity: Analysis of a SLA task from an activity theory ing and researching second languages: Systemic-
perspective. In J. P. Lantolf & G. Appel (Eds.), Vy- theoretical instruction and second language develop-
gotskyian approaches to second language research (pp. ment. Unpublished doctoral dissertation, Pennsyl-
173–194). Norwood, NJ: Ablex Press. vania State University, University Park.
340 The Modern Language Journal 91 (2007)
Newman, D., Griffin, P., & Cole, M. (1989). The con- namic testing: The nature and measurement of learn-
struction zone: Working for cognitive change in school . ing potential . Cambridge: Cambridge University
Cambridge: Cambridge University Press. Press.
Palinscar, A. S., Brown, A. L., & Campione, J. C. (1991). Tzuriel, D. (2001). Dynamic assessment of young children.
Dynamic assessment. In H. L. Swanson (Ed.), New York: Plenum Publishers.
Handbook on the assessment of learning disabilities van Lier, L. (2004). The ecology and semiotics of language
(pp. 75–94). Austin, TX: Pro-Ed. learning: A sociocultural perspective. Boston: Kluwer
Poehner, M. E. (2005). Dynamic assessment of oral profi- Academic.
ciency among advanced L2 learners of French. Unpub- Vygotsky, L. S. (1956). Isbrannye psikhologicheskie issle-
lished doctoral dissertation, Pennsylvania State dovaniya [Selected psychological investigations].
University, University Park. Moscow: Izdatel’stvo Akademii Pedagogischeskikh
Poehner, M. E. (in press). Dynamic assessment: A Vygot- Nauk SSSR.
skyan approach to understanding and promoting L2 Vygotsky, L. S. (1978). Mind in society: The development
development. Norwell, MA: Springer. of higher psychological processes. Cambridge, MA:
Poehner, M. E., & Lantolf, J. P. (2005). Dynamic assess- Harvard University Press.
ment in the language classroom. Language Teach- Vygotsky, L. S. (1986). Thought and language. (A.
ing Research, 9, 1–33. Kozulin, trans.). Cambridge, MA: MIT Press.
Ratner, C. (2006). Cultural psychology: A perspective on Vygotsky, L. S. (1998). The problem of age (M. J. Hall,
psychological functioning and social reform. Mahwah, trans.). In R. W. Rieber (Ed.), The collected works
NJ: Lawrence Erlbaum. of L. S. Vygotsky: Vol. 5. Child psychology (pp. 187–
Spence-Brown, R. (2003). Authentic assessment? The im- 205). New York: Plenum.
plementation of an ‘authentic’ teaching and assess- Wertsch, J. V. (1985). Vygotsky and the social formation of
ment task. Unpublished doctoral dissertation, Uni- mind. Cambridge, MA: Harvard University Press.
versity of Melbourne, Australia. Wertsch, J. V. (1998). Mind as action. Oxford: Oxford
Sternberg, R. J., & Grigorenko, E. L. (2002). Dy- University Press.

Forthcoming in The Modern Language Journal

Tribute to the Editor
Heidi Byrnes & Judith Liskin-Gasparro

Thank You to the Profession: Editor’s Farewell

Sally Sieloff Magnan
Joan Kelly Hall. “Redressing the Roles of Correction and Repair in Research on Second and Foreign
Language Learning”
Response: Paul Seedhouse. “On Ethnomethodological CA and ‘Linguistic CA’: A Reply to Hall”

Response: Joan Kelly Hall. “The Devil’s in the Details: A Response to Seedhouse”
Monika Chavez. “Students’ and Teachers’ Assessments of the Need for Accuracy in the Oral Production
of German as a Foreign Language”
Peter D. MacIntyre. “Willingness to Communicate in the Second Language: Understanding the Decision
to Speak as a Volitional Process”

Vincenza Tudini. “Negotiation and Intercultural Learning in Italian Native Speaker Chat Rooms”
Miyuki Sasaki. “Effects of Study-Abroad Experiences on EFL Writers: A Multiple-Data Analysis”

Caroline H. Vickers. “Second Language Socialization Through Team Interaction Among Electrical and
Computer Engineering Students”

Perspectives: ISSUE: The Common European Framework of Reference for Languages: Perspectives on
the Making of Supranational Language Education Policy