Cognitive Linguistics Second Language Acquisition and Foreign La 2004

Cognitive Linguistics, Second Language Acquisition,
and Foreign Language Teaching
≥
Brought to you by | The University of Texas at Dallas (UTDALLAS)
Authenticated
Download Date | 4/24/17 6:41 PM
Studies on Language Acquisition
18
Editor
Peter Jordens
Mouton de Gruyter
Berlin · New York

Authenticated
Cognitive Linguistics,
Second Language Acquisition, and
Foreign Language Teaching
edited by
Michel Achard
Susanne Niemeier
Mouton de Gruyter
Berlin · New York

Authenticated
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.
앪
앝 Printed on acid-free paper which falls within the guidelines
of the ANSI to ensure permanence and durability.
Library of Congress Cataloging-in-Publication Data
Cognitive linguistics, second language acquisition, and foreign

language teaching / edited by Michel Achard and Susanne Nie-
meier.
p. cm. ⫺ (Studies on language acquisition ; 18)
Includes bibliographical references and index.
ISBN 3-11-017357-3 (alk. paper)
1. Language and languages ⫺ Study and teaching ⫺ Psycho-
logical aspects. 2. Second language acquisition. 3. Cogni-
tive grammar. I. Achard, Michel, 1959⫺ II. Niemeier, Su-
sanne, 1960⫺ III. Series.
P53.7.C59 2004
4181.0071⫺dc22
2003027003
ISBN 3 11 017357 3
Bibliographic information published by Die Deutsche Bibliothek
Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data is available in the Internet at ⬍http://dnb.ddb.de⬎.
쑔 Copyright 2004 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin.
All rights reserved, including those of translation into foreign languages. No part of this book
may be reproduced in any form or by any means, electronic or mechanical, including photocopy,
recording, or any information storage and retrieval system, without permission in writing from
the publisher.
Cover design: Sigurd Wendland, Berlin.
Printed in Germany.

Authenticated
Contents
Cognitive Linguistics, Language Acquisition, and Pedagogy

Michel A chard and Susanne Niemeier
Expressing Motion Events in a Second Language:

a Cognitive Typological Perspective 13
Teresa Cadierno
Construal, Convention, and Constructions in L2 Speech 51

Renee Waara
Input Versus Transfer? - The Role of Frequency and Similarity

in the Acquisition of L2 Prepositions 77
Wander Lowie and Marjo/ijn Verspoor
Linguistic and Cultural Relativity - 95

Reconsidered for the Foreign Language Classroom
Susanne Niemeier
The Figure / Ground Gestalt and Language Teaching Methodology 119

Peter Grundy
"Cultural Scripts": a New Medium for Ethnopragmatic Instruction 143

Cliff Goddard
Grammatical Instruction in the Natural Approach:

a Cognitive Grammar View 165
Michel A chard
Teaching Temporal Connectors and their Prototypical

Non-temporal Extensions 195
Angeliki Athanasiadou
Expanding Learners' Vocabulary Through Metaphor Awareness:

What Expansion, What Learners, What Vocabulary? 211
Frank Boers

Authenticated
vi Contents
A cognitive Linguistic View of Polysemy in English

and its Implications for Teaching 233
Szilvia Csabi
Applying Cognitive Linguistics to Pedagogical Grammar:

The Case of Over 257
Andrea Tyler and JYvyan Evans
Index 281

Authenticated
Introduction: Cognitive Linguistics, Language
Acquisition, and Pedagogy
Michel Achard and Susanne Niemeier
The cognitive linguistics movement consists of different theories that share

at least two important conceptions about linguistic organization. The first
one, most thoroughly investigated in Langacker's Cognitive Grammar
(Langacker 1987, 1991) emphasizes the inherent symbolic function oflan-
guage. The second one states that the cognitive processes that enable speak-
ers to understand and produce language represent the acute specialization of
more general cognitive abilities (Tomasello 2000). This volume investigates
the relevance of the cognitive linguistics view of language to the fields of
second language acquisition and pedagogy. In this introduction, we briefly
present some of the most important concepts of cognitive linguistics; the
different papers then apply these concepts to specific issues of second lan-
guage learning and teaching.
1. The Symbolic Function of Language
In the cognitive view, the purpose of linguistic inquiry is to describe its

semiotic function, that is, the symbolic association between a meaning and
a phonological form. The lexicon, morphology, and syntax form a contin-
uum of symbolic structures that cannot easily be separated into discrete
compartments. Consequently, the grammar of a language can be described
as "a structured inventory of conventional symbolic units" (Langacker
1987: 73). Each symbolic unit is composed of a semantic and a phonological
pole. Grammatical constructions (the rules) represent a specific kind of
symbolic structure. They take the form of templates that sanction the use of
actually occurring expressions. These templates are also symbolic, and thus
meaningful, even though their meaning is more abstract than that of actu-
ally occurring expressions.
Symbolic units constitute the totality of the grammar of a language. The
only units that can be invoked to describe a linguistic system are: i) seman-
tic, phonological, or symbolic units which are (part of) occurring expres-

Authenticated
2 Michel Achard and Susanne Niemeier
sions, ii) schematizations of permitted structures, iii) categorizing relations

between permitted structures (Langacker 1987: 488). Let us consider the
word "cat" as a phonological example. It is composed of the occulTing
expression [het]. One schematization over that occulTing expression is of
the type [CVC], and the categorizing relation existing between the occur-
ring expression and the schema is one of instantiation: [CVC] ~[k<et]. The
three units described here constitute the only types of constructs which can
be posited for the representation of any level of linguistic construction. This
"content requirement" (Langacker 1987) rules out purely grammatical con-
structs, i.e., elements which are given no phonological or semantic value
(diacritics, traces, filters, etc.). It also implies that rules cannot be different
in nature from the expressions they describe. They can only be schematiza-
tions of actually occulTing expressions. In a cognitive grammar, the rules
and their instantiations cohabitate without fear of reduplication because
they represent different facets of a speaker's linguistic knowledge.
2. The Nature of Meaning
Because linguistic expressions are inherently symbolic, the investigation of

their meaning (the content of their semantic pole) represents a major field of
studies in cognitive linguistics. Researchers are most specifically interested
in assessing the role of human experience in providing the basic meaning
and intentions coded in natural language. Meaning is equated with concep-
tualization, or, to be more specific, in the human interpretation of the world.
It is subjective, anthropocentric, and reflects dominant cultural concerns
and culture-specific modes of interaction as well as features of the world as
such (Lakoff, 1987, Langacker 1991, Wierzbicka 1988).
In Cognitive Grammar (one of the cognitive linguistics models where
the semantic theory is best articulated), a linguistic expression is character-
ized semantically relative to one or more knowledge structures called "cog-
nitive domains" (Langacker 1987: 147-166). The notion of cognitive
domains shares important similarities with Fillmore's frames (1982),
Lakoff's Idealized Cognitive Models (ICM's) (1987), and Wierzbicka's cul-
tural scripts (see Goddard this volume). Some domains are ineducible: they
pertain to our experience of space, time, the different senses, emotions, etc.
Most linguistic expressions are, however, characterized with reference to
complex domains. Any knowledge system or conceptualization can TIlllC-
tion as a domain for the characterization of a linguistic expression, regard-
less of its possible complexity or abstractness. Presumably such structures

Authenticated
Cognitive Linguistics, Language Acquisition, and Pedagogy 3
are ultimately grounded in the set of cognitively irreducible domains

alluded to above, but this relationship may be quite indirect, and a particular
expression must be described in terms of structures that occupy appropriate
positions in hierarchies of conceptual elaboration. In particular, different
folk models that pertain to our conventional conception of certain concepts
can be used as cognitive domains relative to which the meaning of linguistic
expressions is characterized. To take an example, the domain of time in
general, as well as the way we socially learn to subdivide it into years,
months, and weeks, is necessarily activated when the word "Tuesday" is
used. Similarly, the whole system of kinship relations is the background
against which the meaning of the word "father" gets characterized.
The meaning of an expression represents the conceptualizer's desire to
construe the relevant cognitive domains in a specific way. One particular
(and crucial) dimension of construal concerns her ability to impose a "pro-
file" on a "base", which derives the semantic value of a linguistic expres-
sion. The base consists of those facets of active cognitive domains that are
directly relevant to the expression, hence necessarily accessed when the
expression is used. The profile is a sub-region within the base. It is that sub-
region that the expression designates and thus makes prominent within the
base. It is important to bear in mind that the particular profile imposed on a
base is a consequence of the way in which the conceptualizer construes the
scene, and not an inherent property of that scene. As an example, the con-
ception of a wheel constitutes the base relative to which the meaning of the
word "rim" is characterized. The outer edge of the wheel constitutes the
specific sub-structure "rim" profiles.
Cognitive linguistics embraces a semantic theory based on the ideas of
family resemblance and complex categories (Rosch 1977, Lakoff 1987
among others). A fixed expression is often polysemous. It has a variety of
group of people
operating together
(clandestinely)
'----r---.,-----' mmmmmj ''On, I
..............-
circular circular circular piece

mark object ofjewelery
Figure 1. Ring: semantic network ofa complex category

Authenticated
related senses that form a complex category which can be represented as a

network. For instance, a portion of the semantic network of the noun ring is
given in Figure 1 (based on Langacker 1988a: 52).
A particular meaning can stand in two types of relations to another:
extension and elaboration. In Figure 1, the solid arrows represent elabora-
tion relations, and the dashed arrows represent extension relations. An
entity elaborates another entity when it is fully compatible with it, but is
construed with a greater degree of precision. An entity represents a semantic
extension from another entity if it is not completely compatible with it, but
nonetheless assimilated to it on the basis of perceived resemblances. The
type of representation of a complex category illustrated in Figure 1 is not
only valid for lexical items. Other linguistic expressions are amenable to the
same type of characterization, namely, organized around a central proto-
type, with extension and elaboration relations linking that prototype to other
instances of the category.
Metaphorical extensions represent specific cases of semantic extension.
The study of metaphor represents an important part of cognitive linguistics
research because it directly links language to the speaker's conceptual sys-
tem (Lakoff and Johnson 1980, Kovecses 1986, Lakoff 1987, Gibbs and
Steen 1999 among many others). Because "metaphor is pervasive in every-
day life, not just in language, but in thought and action" (Lakoffand Johnson
1980: 3), language represents a privileged entry point into our conceptual
system. In the cognitive view, metaphorical concepts are systematically
organized into coherent systems, and these systems are reflected in the
metaphorical expressions used in language. The social implications of this
view are important. Since metaphors are largely culture specific, learning
the metaphors used in a given language provides invaluable insights into the
way in which the speakers of that language act and think.
3. A Usage-based Model
The methodology of cognitive linguistics is ''usage-based'' in that it is pri-

marily concerned with the characterization of language as it is spoken and
understood, as well as with the dynamics of its use (Langacker 1988b, 2000,
Barlow and Kemmer 2000, Tomasello 2000). Langacker (1987: 494)
describes usage-based models as follows: "Substantial importance is given
to the actual use of the linguistic system and a speaker's knowledge of its
use; the grammar is held responsible for a speaker's knowledge of the full

Authenticated
range ofthe linguistic conventions, regardless of whether those conventions

can be subsumed under more general statements". The usage-based approach
constitutes a "non-reductive approach to linguistic structure that employs
fully articulated schematic networks and emphasizes the importance of low
level schemas" (Langacker 1987: 494).
3.1. Learning in a UBM
The goal of a usage-based-model is not to achieve mathematical elegance,

but to depict the complexity of language use. Consequently, it is composed
of an eclectic array of expressions at different levels of complexity, abstrac-
tion, and generality. Individual lexical items cohabitate in the system with
idioms, conventionalized idiosyncratic collocations, and fully productive
grammatical constructions (Tomasello 2000). Tomasello argues that usage-
based models constitute strong theoretical frameworks for the description of
child language acquisition because they do not demand that a child's gram-
mar be identical to the adult system. He expresses his proposal for a usage-
based view of Ll acquisition as follows (from Tomasello 2000: 237-238,
emphasis in the original):
The proposal would thus be that the child initially learns individual, item-
based linguistic constructions (e.g., verb island constructions), and ifthere
are patterns to be discerned among these different item-based constructions
in adult usage, she could then make abstractions and create inheritance hier-
archies of constructions. In this view of language and its acquisition, there-
fore, there is continuity not of structures - adults control a more diverse and
abstract set of constructions than do children - but there is continuity of
process in the sense that the processes of learning and abstraction are the
same wherever and whenever they are applicable ... This general approach is
usage-based in the sense that all linguistic knowledge - however abstract it
may ultimately become - derives in the first instance from the comprehension
and production of specific utterances on specific occasions of use.
In a usage-based model, the goal of child language acquisition research is to

characterize the steps by which the child's inventory of conventionalized
units comes to resemble the adult's. It predominantly involves the investiga-
tion of the development of the cognitive abilities (making analogies, com-
bining structures, drawing inferences, to name just a few) that allow chil-
dren to eventually master the adult system.

Authenticated
3.2. Second Language Acquisition and Teaching
In a fundamental way, the task of the second language researcher is quite

similar. L2 learners are also attempting to master the specific array of sym-
bolic units that represents the linguistic conventions of the target language.
The learner's interlanguage resembles a child's grammar in that it is also
composed of an assortment of eclectic constructions at various levels of
systematicity, abstraction, and productivity. The difference between the two
comes from the conditions under which the L2 speaker comes to learn native
conventions. Developmental issues are obviously less critical, at least in the
case of adult learners who already control their native system of linguistic
conventions. Rather, the emphasis of research shifts to the retraining that
needs to take place in order to learn a new set of symbolic units. In a devel-
oping L2 system, the target units are in direct competition with the native
ones because they both represent alternative ways of construing the same
reality. L2 learning can therefore be viewed as a gradual process by which
the target system gains more and more differentiation and autonomy from
the native one. This autonomy is complete when the learner exercises full
control over two separate sets of conventionalized linguistic impressions.
Three papers in this volume investigate the factors that enhance this
differentiation, as well as the specific aspects of the native inventory that
influence the learning of the target language conventions. In their attempt to
isolate the factors that facilitate the learning of English prepositions by
Dutch native speakers, Lowie and Verspoor investigate the importance of the
(semantic and phonological) similarity of the prepositions in both languages
versus the input frequency of the English prepositions learners are exposed
to at different stages of their linguistic development. Waara analyzes the
acquisition by Norwegian speakers of the various kinds of argument struc-
tures associated with the English verb "get". Her use of the concept of a
usage-based model shows that the constructions representative oflearners'
interlanguage include conceptual blends, elements directly transferred from
Norwegian grammar, as well as overgeneralizations of their developing
English system.
The competition between the native and target systems can also be inves-
tigated at a more local level. With respect to specific conceptual domains,
different languages illustrate different systematic construal choices by their
speakers. The coding of motion events represents an interesting example.
Talmy (1985) showed that the components of "motion", "figure", "ground",
"cause", "manner", and "path" compose the conceptual structure of complex
motion events. He further showed that different languages package these

Authenticated
components in different ways. In satellite-framed (S-framed) languages, the

verb conflates "motion" with "manner" or "cause", and "path" is expressed
separately by a satellite (a preposition, for instance). In verb-framed (V-
framed) languages, the verb conflates "motion" and "path", and "manner"
or "cause" are coded in a separate satellite. This typology has constituted
the base for a large number of recent studies of motion constructions in lin-
guistics (Talmy 1991, Slobin 1996, 2003), and first language acquisition
(Berman and Slobin 1994, McNeill 2000). In this volume, Cadierno uses
Talmy's typology as well as Slobin's notion of "thinking for speaking"
(Slobin (1991) to show how speakers of Danish (a S-language) learning
Spanish (a V-language) as a second language learn to express motion events
in a language that differs typologically from theirs. She illustrates the
process by which Danish speakers retrain themselves to attend to specific
details of the events in order to describe them in a way consistent with the
linguistic conventions of the Spanish language.
In addition to being a valid framework for the description of systematic
L2 learning patterns, the cognitive linguistics model also offers important
contributions to second language pedagogy because the kinds of generaliza-
tions it posits to describe linguistic organization can easily be made explicit,
and thus incorporated into classroom practices. Eight papers in this volume
show that instruction structured around cognitive linguistics principles
helps solve some difficult problems in various areas of second language
teaching ranging from curriculum development to the teaching of specific
lexical items.
As mentioned earlier, one of the most important tenets of the model is
that the language faculties constitute highly specialized uses of more general
cognitive faculties. For example, Langacker (1987, 1991) and Talmy (2000)
have demonstrated the linguistic relevance of our capacity to impose a figure/
ground organization on a conceptualized scene. In this volume, Grundy
argues that the figure / ground gestalt undermines the discrete-item-syllabus
assumption upon which second language instruction generally rests.
The recognition of the centrality of meaning to linguistic organization
provides pedagogical insights to second language teachers on several levels.
First, the symbolic character of a linguistic system, and thus the absence of
a strict delineation between the lexicon, morphology, and syntax, provides
interesting methodological possibilities for grammatical instruction in a sec-
ond language. In this volume, Achard argues that the symbolic nature of
grammatical constructions (and therefore their semantic import) affords a
kind of grammatical instruction perfectly congruent with the goals and
practices of the communicative models of language instruction. This is

Authenticated
highly desirable because of the notoriously difficult integration of the teach-

ing of grammar in communicative classrooms.
Secondly, the very constructs relative to which the meaning of linguistic
expressions is characterized constitute potent teaching guides, because they
provide the necessary social context to learn difficult, often culture-specific
concepts. Three papers in this volume explore the benefits of making these
constructs explicit for the teaching of vocabulary. At the curricular level,
Niemeier assesses the relevance of the kind of cultural and linguistic rela-
tivity commonly associated with most cognitive linguistics models to for-
eign language teaching methodology. On a more specific level, Goddard
presents the "structural scripts" approach to the teaching of etlmopragmatic
knowledge. This technique makes use ofWierzbicka's semantic metalan-
guage to describe the lexical semantics of specific concepts, as well as the
communicative norms of a given community in simple, cross-translatable
terms. Goddard illustrates the pedagogical power of these scripts by analyzing
some aspects of the cultural pragmatics of English and Malay. Boers is con-
cerned with expanding learners' vocabulary using metaphor awareness. He
argues that learners retain the meaning of a metaphorical expression better
if they can be made aware of its motivation. He also shows that the optimal
success of metaphor awareness critically depends on the careful selection of
the kind of vocabulary presented, as well as the cognitive style and motiva-
tion of the students.
Finally, the radial category approach to polysemy provides invaluable
help to teach specific lexical items. Because the peripheral senses of an
expression are related to the more central ones by perceived similarities
between them (semantic extension), they are "motivated" by those central
senses. In this volume, three papers investigate how making this motivation
explicit can enhance the retention of the more peripheral senses of a lexical
item. The basic idea is that teachers can guide their students through the
paths of semantic extension and emphasize what the peripheral senses share
with the more central ones in order to facilitate their learning. Athanasiadou
illustrates this methodology by showing how the non-temporal uses of the
expressions "when", "as long as", "since", "as soon as", and "while" are
motivated by their specific temporal senses. She further argues that being
aware of the relation that exists between the temporal and non-temporal
senses helps students to learn the more difficult non-temporal senses. Csabi
shows that the Hungarian students who have explicitly been shown the
semantic connection between the central senses of the English verbs "hold"
and "keep" and their more peripheral and idiomatic counterparts retain the
latter senses better. Finally, Tyler and Evans present an in-depth analysis of

Authenticated
the polysemous "over" where the multiple senses associated with the prepo-
sition are organized in a connected network of related meanings. Tyler and
Evans further argue that their approach enables teachers to rely on the well-
known meanings of the central senses to gradually move on to the most
peripheral ones, thereby strengthening their semantic motivation.
The second language research program inspired by cognitive linguistics
is still in its infancy, and a lot of the results presented here are preliminary.
However, it is our hope that the different chapters of this volume will help
establish the cognitive linguistics model as a valuable framework for the
investigation of second language learning and teaching phenomena, and
provide the methodology to further extend the research.

Authenticated
References
Barlow, Michael and Suzanne Kemmer (eds.)

2000 Usage Based Models ofLanguage. Stanford, CSLI.
Berman, Ruth and Dan Slobin (eds.)
1994 Relating Events in Narrative: A Cross linguistic Developmental Study.
Hillsdale, New Jersey: Lawrence Erlbaum.
Fillmore, Charles
1982 Frame Semantics. In The Linguistic Society of Korea (ed.), Linguistics
in the Morning Calm. Seoul: Hanshin Publishing Co.
Gibbs, Raymond and Gerard Steen (eds.)
1999 Metaphor in Cognitive Linguistics. AmsterdamlPhiladelphia: John
Benjamins.
K6vecses, Zoltan
1986 Metaphors of Anger, Pride and Love: A Lexical Approach to the
Study of Concepts. AmsterdamlPhiladelphia: John Benjamins.
Lakoff, George
1987 Women, Fire and Dangerous Things: What Categories Reveal about
the Mind. Chicago/London: University of Chicago Press.
Lakoff, George and Mark Johnson
1980 Metaphors we Live by. ChicagolLondon: University of Chicago Press.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites.
Stanford: Stanford University Press.
1988a "A view oflinguistic semantics". In B. Rudzka-Ostyn (ed.), Topics in
Cognitive Linguistics. Amsterdam/Philadelphia: John Benjamins:
49-90.
1988b "A usage-based model". In B. Rudzka-Ostyn (ed.), Topics in Cognitive
Linguistics. AmsterdamIPhilade1phia: John Benjamins: 127-161.
1991 Foundations of Cognitive Grammar. Vol. 2: Descriptive Application.
2000 "A dynamic usage-based model". In Michael Barlow and Suzanne
Kemmer (eds.), Usage Based Models ofLanguage. Stanford, CSLI:
1-63.
McNeill, David
2000 "Analogic/Analytic representations and cross-linguistic differences
in thinking for speaking." Cognitive Linguistics 11: 43-60.
Rosch, Eleonor
1977 "Human categorization." In Neill Warren (ed.), Studies in Cross-
Cultural Psychology. Volume 1, 1-49. London: Academic Press.
Slobin, Dan
1991 "Learning to think for speaking: Native language, cognition, and
rhetorical style." Pragmatics 1: 7-25.

Authenticated
1996 "Two ways to travel: Verbs of motion in English and Spanish." In M.

Shibatani and S. Thompson (eds.), Grammatical Constructions: Their
Form and Meaning, Oxford: Oxford University Press: 195-219.
Slobin, Dan
2003 "The many ways to search for a frog: Linguistic typology and the ex-
pression of motion events." In S. Stromqvist and L. Verhoeven (eds.),
Relating Events in Narrative: Typological and Contextual Perspec-
tives. Mahwah, NJ: Lawrence ErlbaumAssociates.
Ta1my, Leonard
1985 "Lexicalization patterns: Semantic structure in lexical forms." In T.
Shopen (ed.), Language Typology and Syntactic Description. Vol. 3:
Grammatical Categories and the Lexicon. Cambridge University
Press, 57-149.
1991 "Path to realization: A typology of event conflation." Proceedings of
the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics
Society: 480-519.
2000 Toward a Cognitive Semantics. 2 Volumes. MIT Press.
Tomasello, Michael
2000 "Do young children have adult syntactic competence?" Cognition 74:
209-253.
Wierzbicka, Anna
1988 The Semantics of Grammm: Amsterdam and Philadelphia: John
Benjanlins.

Authenticated
Authenticated
Expressing Motion Events in a Second Language:
a Cognitive Typological Perspective
Teresa Cadierno
1. Introduction
The question of how adult second language learners come to express spatial
relations in a second language is a rather neglected area within second
language acquisition research (but see Becker and Carrol1997 for an excep-
tion).l Within first language acquisition research, on the other hand, extensive
research has been conducted on how children perceive and express spatial
relationships in their native language (see Bloom et al. 1996 for a recent
overview). While earlier research focused on the influence of the child's
nonlinguistic knowledge on the development of spatial language (e.g.,
Piaget and Inhelder 1956; Levine and Carey 1982), recent cross-linguistic
approaches under the general framework of functional-cognitive linguistics
(e.g., Berman and Slobin 1994; Bowerman 1994; Bowerman 1996a & 6b)
have examined the ways in which children from different languages initially
use spatial language.
This cross-linguistic research, which has compared same-age children
learning languages with different spatial categories (e.g., Choi and Bowerman
1991; Bowerman 1994) has shown that the way in which children initially
classify space for language is the outcome of a complex interplay between
the child's nonlinguistic spatial knowledge and the input of the language
that he/she is learning. The conclusion reached by Bowerman (1996a) is
that the influence of the input language is stronger than it was believed. In a
similar way, cross-linguistic research conducted by Berman and Slobin
(1994) within a cognitive typological framework (Talmy 1985) has shown
that children learning typologically different languages differ with respect
to how much and what kind of information they provide when referring to
particular spatial situations involving motion.
As expressed by Bowerman (1996b), the hypothesis that language can
influence children's construction of semantic categories has been rather
unpopular given its associations with the controversial Whorfian hypothesis,
according to which the way in which human beings view reality is shaped

Authenticated
14 Teresa Cadierno
by the particular language they speak. In order to account for the above-
mentioned findings in cross-linguistic research and at the same time decouple
such issues from the loaded Whorfian hypothesis, Slobin (l996a) has coined
the expression "thinking for speaking" to refer to the influences of language
on the ongoing process of interpreting and expressing verbal messages.
According to Slobin (1996a), the results of the cross-linguistic research in
first language acquisition conducted by Slobin and his colleagues (Belman
and Slobin 1994) support the hypothesis that when acquiring his/her first
language, a child learns specific ways of thinking for speaking.
The purpose of the present investigation is to extend this line of research
conducted within L1 acquisition to the field ofL2 acquisition. More specifi-
cally, the study examines the implications of Slobin's thinking for speaking
hypothesis for L2 learners within a cognitive typological approach, and thus
investigates how L2 learners come to express spatial relationships involving
motion in a language - Spanish - that is typologically different from their
native language - Danish.
2. Conceptualization in Language:
the Thinking for Speaking Hypothesis
The approach to language known as Cognitive Linguistics (CL) belongs to

the functionalist linguistic tradition in that linguistic functions are conceived
as fundamental to the description of linguistic form. However, within func-
tionalism, CL constitutes a framework which focuses on the semiological
function of language and stresses the important role of conceptualization in
social communication (Langacker 1998). The term conceptualization is used
in a broad sense and includes not only abstract concepts but also sensory,
motor, and emotive experience and full apprehension of the physical, social,
cultural, and linguistic context (Langacker 1998).
Central to CL is the adoption of a subjectivist view to linguistic meaning.
As indicated by Langacker (1987: 6-7), "the semantic value of an expression
does not reside solely in the inherent properties of the entity or situation it
describes, but critically involves as well the way we choose to think about
this entity or situation and mentally portray it." In other words, different
semantic meanings can be obtained on the same objective situation by
adopting different perspectives or construals of that situation. Critically,
however, the perspectives that speakers take on given events and situations
are often influenced by the particular language that they speak, i.e., by the
available grammatical resources that can be used for given semantic

Authenticated
Expressing Motion Events in a Second Language 15
domains. Languages thus often favor particular perspectives on a given situ-

ation. As Slobin (1996a) has indicated, our experiences of the world are not
only filtered into verbalized events through the choice of the individual
speaker's perspective, but also through the particular set of options provided
by the particular language we are speaking. This conception of the relation-
ship between language and thought, which has been termed "thinking for
speaking", claims that there is a special kind of thinking that is intimately
tied to language, i.e., the thinking that is carried out on-line in the process of
speaking. In other words, the language that we speak directs our attention,
while speaking, to particular ways of filtering our experiences of the world.
In Slobin's (1996a: 76) own words, thinking for speaking would "involve
picking those characteristics of objects and events that (a) fit some concep-
tualization of the event, and (b) are readily encodable in the language".
Slobin's formulation contrasts with stronger claims made in the past by
Whorf ([1940] 1956) who related language with a particular worldview, as
well as with current claims made by researchers such as Lucy (1992, 1996),
who advocate the influence of language on nonlinguistic cognition, i.e., on
speakers' patterns of attention, categorization, and memory. It must be noted,
however, that Slobin (1998) reports results from initial experiments examin-
ing the effects of language typology on memory, which indicate differences
in the mental images built by native speakers from typologically different
languages in relation to manner of motion. These results seem to suggest
possible typological effects on nonlinguistic cognition, and can therefore be
considered to be in line with stronger claims such as Lucy's.
3. Motion Events in First Language Acquisition
Slobin's thinking for speaking hypothesis has received empirical support in

a cross-linguistic study (Berman and Slobin 1994) where native speakers
from three main age groups (preschool, school-age, and adults) produced
oral narratives elicited by means of a wordless picture book, Frog, where are
you?, i.e., the so-called "frog story" (Mayer 1969). Five different languages
were studied - English, German, Spanish, Hebrew, and Turkish.
One of the aspects that was analyzed in this study was the expression of
motion events. The description of these motion events was based on Talmy
(1985: 60) who defines a motion event as " ... a situation containing move-
ment or the maintenance of a stationary location alike ... " According to this
author, there are six cognitive components which play a role in the conceptual
structure of a motion event, the first four constituting the central elements

Authenticated
16 Teresa Cadierno
while the last two are associated components: (a) Motion: the presence per
se of motion; (b) Figure: the moving entity; (c) Ground: the object with
respect to which the Figure moves, which can potentially include the source,
the medium, and the goal of movement; (d) Path: the course followed by the
Figure with respect to the Ground; (e) Manner: the manner in which the
motion takes place; and (t) Cause: the cause of its occurrence.
Talmy (1985: 61) offers the following two examples to illustrate the
components:
The pencil rolled off the table

Figure Motion Path Ground
Manner
The pencil blew off the table

Figure Motion Path Ground
Cause
In both sentences the pencil functions as the Figure and the table as the
Ground which in this example expresses source of movement, while the
particle off expresses the Path. In addition, the verbs rolled and blew express
Motion while in addition the verb rolled expresses Manner of motion while
blew expresses Cause of motion.
According to Talmy (1985), there are crucial cross-linguistic differences
in the ways in which different languages of the world express motion
events. This author has identified two main typological patterns in which
the universal components of motion events are expressed by different
means in different languages: (a) satellite-framed languages (S-languages),
such as Chinese and all branches of the Indo-European family except
Romance Languages, where the main verb conflates Motion + Manner /
Cause, and Path is expressed separately by means of a satellite, i.e., verb
particles (e.g., down, out, up, etc) and prefixes (e.g., mis- as in misfire).
Some examples in English are: The rock slid / rolled / bounced down the
hill (Motion + Manner); The napkin blew off the table (Motion + Cause);
and (b) verb-framed languages (V-languages), such as Romance, Semitic,
and Polynesian languages, where the main verb conflates Motion + Path,
and Manner and Cause are expressed separately by means of an adverbial or
a gerundive type of constituent. In some cases, the expression of this con-
stituent may result in an awkward style, so that it is often omitted. An exam-

Authenticated
pie in Spanish would be: La botella entro / salio de la cueva (flotando) (The
bottle entered / exited the cave (floating)).
In their choice oflanguages, Berman and Slobin (1994) together with his
associates used Talmy's typology selecting S-languages (i.e., English and
Gemlan) and V-languages (i.e., Spanish, Hebrew, and Turkish). The results
of the study showed consistent typological differences in the way motion
events were narrated. More specifically, children leaming typologically dif-
ferent languages differed strikingly by age 3 in the way they talked about
motion. These differences, which resulted in distinct rhetorical styles, can
be summarized as followed:
1. Speakers of S-languages used a greater variety of motion verbs as com-

pared to speakers of V-languages. As indicated by Slobin (1996a), this is
greatly due to the higher frequency of motion verbs conflating motion
and manner in S-languages as compared to V-languages.
2. Speakers of S-languages exhibited a higher degree of elaboration in their
description of path of motion as compared to speakers of V-languages.
In other words, there was a richer and more detailed description of paths
(trajectories) in S-languages than in V-languages. This was shown by a
more frequent use of ground-adjuncts, i.e., prepositional phrases refer-
ring to the source, medium, or goal of movement (e.g.,fell into the pond
vs. se cayo 'fell down').
3. Speakers of S-languages evidenced a more frequent use of event confla-
tion, i.e., incorporation ofthe different composites of locative trajectories
(Path, Ground -source, medium, and goal) within the clause as compared
to speakers of V-languages (e.g., The deer threw him off over a cliff into
the water vs. Lo tiro al agua '(It) threw him into the water'). Even
though both S-languages and V-languages can in principle allow for
such phrasal path descriptions (e.g., He fell down from the window onto
the street- Se cayo de la ventana a la calle), S-languages make a larger
use of this option than V- languages.
4. Speakers of S- languages tended to specify the details of trajectories, i.e.,
provide rich path descriptions, thus leaving setting to be inferred,
whereas speakers of V-languages tended to describe aspects of the static
scene in which the movement took place, leaving trajectories to be
inferred. Compare The deer threw them off over a cliff into the water,
where the trajectories described allow one to infer that there is a cliff
above the water, with Lo tiro. Por suerte, abajo, estaba el rio. El niiio
cayo en el agua (Slobin 1996b: 204, age 11) (['The deer] threw him.
Luckily, below, was the river. The boy fell into the water'), where the

Authenticated
18 Teresa Cadierno
locative description 'below was the water' allows one to infer that the
trajectory went from some elevated place to the pond. As shown by the
two examples, the two language types suggest a differential allocation of
attention between description of movement (i.e., details of trajectories)
and description oflocation (i.e., static descriptions).
5. Speakers of S-languages provided more manner information than speakers
of V-languages. This is mainly due to the fact that S-languages have a
larger lexicon of manner of motion verbs, which allows their speakers to
make finer manner of motion distinctions than speakers of V-languages.
The systematic differences found in children learning typologically different

languages have been explained by Slobin (1996a) as reflecting different pat-
terns of thinking for speaking. That is, children from typologically different
languages pay differential online attention to the particular details involved
in a motion event. Based on these results, Bernmn & Slobin (1994: 611)
concluded that
in becoming a native speaker of a given language, the child learns to attend

to particular aspects of experience and to relate them verbally in ways that
are characteristic of that language. Becoming a native speaker thus involves
not only the acquisition of the phonology and morphosyntax of that language;
it also requires attention to the grammaticized semantic distinctions of the
language and to the ways in which grammatical forms are deployed in the
construction of connected discourse.
In other words, each language "trains" its native speakers to pay different
kinds of attention to particular details of events when talking about them.
This is possible given that "the language or languages that we learn in
childhood are not neutral coding systems of an objective reality; rather,
each one is a subjective orientation to the world of human experience, and
this orientation affects the ways in which we think while we are speaking"
(Slobin 1996a: 91). That is, the ways in which native speakers of different
languages express and describe motion are largely determined by the lexi-
calization patterns of their native language.
The implication of this "training" for second language acquisition has
been pointed out by Slobin himself (1996a) who hypothesizes that the train-
ing received while learning our native language may turn out to be rather
resistant to restructuring in adult second language acquisition. The present
study is presented as an attempt to examine the implications of Slobin's
thinking for speaking hypothesis for adult second language learners whose

Authenticated
L 1 and L2 belong to two different typological patterns, and as such extends

previous research on cognitive typology and first language acquisition to
the research field of second language acquisition (see also Cadierno & Lund
in press).
4. The Present Study
The general research question that guided the present study was the follow-
ing: How do L2 learners come to express motion events in an L2 that is
typologically different from their Ll? More specifically, how do adult lan-
guage learners whose Lt can be characterized as an S-language (Danish)
come to express motion events in a V-language (Spanish)? The general
hypothesis is that the learners' Lt-thinking for speaking patterns will be the
point of departure for the interpretation and production of L2 patterns, and
that, consequently, learning a second language will involve learning another
way ofthinking for speaking, that is, learning how the semantic components
of a motion event are mapped into L2 surface forms, and learning which
particular details of a motion event must be attended to in the input and
expressed in the L2.
In the case of the present study, the Danish learners of Spanish must
learn: (l) the characteristic L2 form-meaning mappings, which are different
from the Lt mappings, i.e., Path typically mapped in the verb, Manner typi-
cally mapped in a separate expression; and (2) the (subtle) differences
between the Ll and the L2 with respect to the particular aspects or details of
the motion event that must be focused on: (a) relatively less elaboration in
descriptions of Path of motion, i.e., less use of ground adjuncts; (b) relatively
less frequent use of event conflation; (c) relatively more narrative attention
to location-static descriptions vs. movement-trajectories; and (d) relatively
less elaboration of Manner of motion.
The present study, which focused on the semantic components of Path
and Ground, addressed the following specific research questions:
1. Are there differences in how Danish learners of Spanish and native

speakers of Spanish express motion events in this language? More
specifically, will there be differences with respect to (a) the amount of
motion verb types; (b) the degree in which path of motion is elaborated;
(c) the use of event conflation; and (d) the relative allocation of attention
to movement and setting?

Authenticated
20 Teresa Cadierno
2. If differences are found, does the learners' performance in the L2 parallel

that of the Lt? In other words, is there a congruity between the learners'
Ll and L2 performance, which could reflect a possible transfer of Lt
typological patterns?
Based on Slobin's thinking for speaking hypotheses and on prevIOUS

research on Ll acquisition, it was hypothesized that:
1. There will be differences in how Danish learners of Spanish and native

speakers of Spanish express motion events in this language. It is thus
hypothesized that the L2 learner group will (a) use fewer motion verb
types in Spanish than the native speaker group; (b) exhibit a higher
degree of complexity and elaboration of the path of motion in their nar-
ratives than the native speakers. This elaboration could be formalized by
a "satellization" of Spanish locative constructions, i.e., the use of path
particles which can be redundant and/or ungrammatical in native
Spanish, and by a more frequent use of ground adjuncts than that exhib-
ited by native speakers; (c) make use of event conflation, a construction
not easily allowed in the Spanish language; and (d) tend to provide
descriptions of trajectories as opposed to static descriptions.
2. There will be a parallelism between the way in which motion events are
expressed in the learners' Ll and L2. That is, the expected L2 patterns
mentioned above will also be observed in the learners' L 1. A transfer of
(at least) some ofthe Ll typological patterns is thus hypothesized.
4.1. Description of the Languages Involved in the Study
Applying the typology suggested by Talmy (1985), Danish and Spanish rep-
resent two different typological patterns. Danish is a Germanic language
charactet1zed as an S-language, where Path is typically expressed by an elab-
orate system of satellites, i.e., verb particles which indicate an opposition
between translocative and locative function (e.g., ud-ude 'out (translocative)-
outside', ind-inde 'in (translocative)-inside'; Han gik ud 'He went out' vs.
Han var ude 'He was outside'). Manner and Cause of motion are typically
encoded in the verb (e.g., kravle 'crawl', snige sig 'creep', smide 'throw').
Tn fact, Danish could be characterized as a more prototypical example of an
S-language than English in that it lacks Latinate verbs (e.g., enter, ascend,
descend), thus preventing the conflation of Motion and Path in the verb
root.

Authenticated
Spanish, on the other hand, is a Romance Language characterized as a V-

language, where Path tends to be encoded in the verb (e.g., entrar 'enter',
salir 'leave', bajar 'go down " subir 'go up '), whereas Manner and Cause of
motion tend to be encoded separately, e.g., an adverbial or prepositional
phrase (Entra a la casa corriendo 'He entered the house running'). How-
ever, since typologies are not rigidly fixed, Spanish also has verbs that con-
flate Motion and Manner (e.g., caminar 'walk', correr 'run', volar 'fly') as
well as Motion and Cause (e.g., El/a 10 empuja al agua 'She pushed him into
the water'), although their occurrence is less frequent than in S-languages.
4.2. Method
4.2.1. Participants
A total of thirty-two subjects participated in the study: sixteen adult learners

of Spanish and sixteen native speakers (NS) of Spanish. The learner group
was composed of first year university students, all of whom were native
speakers of Danish. Since no placement exam is given to Danish university
students when they start their studies in a foreign language, background
information about the informants' prior contact with Spanish was obtained
by means of a questionnaire. The questionnaire inquired about the learners'
former exposure to Spanish both in Denmark and in a Spanish-speaking
country, and both in formal and informal settings. In addition, information
about knowledge of other foreign languages was also obtained.
With respect to previous contact with Spanish, the analysis of the ques-
tionnaires revealed that fourteen learners had studied Spanish in high school
for a period of three years in Denmark (average 105 teaching hours in the
first two years and 127 teaching hours in the third year). The other two had
received no formal instruction in Spanish in Denmark, but had received
instruction in a Spanish-speaking country and had been living there for long
periods of time. In addition, most of them had lived and studied Spanish in
a Spanish-speaking country for a period ranging from four months to one
year. When questioned about knowledge of other languages, all subjects
reported knowledge of English and German, while four informants reported
a knowledge of a third foreign language (Gnmlandish or Swedish).
Due to the different types of prior exposure to Spanish before data col-
lection, the informants' proficiency level was not completely uniform. Two
sub-levels of proficiency were apparent on the basis of their biographical
data as well as on the data collected: (1) a subgroup of learners with a

Authenticated
22 Teresa Cadierno
higher proficiency level consisting of the two learners who had studied and
lived in a Spanish-speaking country, plus six learners who had studied
Spanish in Danish high schools as well as in Spanish-speaking countries for
long periods of time (between seven months and one year). A total of eight
learners fell into this group. The proficiency level of these learners would
be comparable to the advanced proficiency level described in ACTFL
guidelines on writing; and (2) a subgroup of learners with a lower profi-
ciency level, consisting of the remaining eight learners who had studied
Spanish in Danish high schools but who had either not been in a Spanish-
speaking country, or had only been there for very short periods of time
(under two months). The proficiency level of these learners would be com-
parable to the intermediate-mid proficiency level described in the ACTFL
guidelines on writing.
Finally, the native speaker group consisted of university students from
Spain who had just arrived in Denmark for one semester's study as part of
the European Socrates interchange program. At the time of data collection,
they all reported no knowledge of Danish.
4.2.2. Experimental Procedures
Following previous research in this area, narrative data were elicited by

means of the "frog story". This picture book was chosen because (I) com-
parable data to that obtained by Slobin and his colleagues on L1 acquisition
could be obtained; and (2) it portrays a rich array of locative trajectories,
which involve the search of a boy and a dog for a frog in a forest.
With respect to the learner group, the story was elicited twice by each
subject: first in the L2-Spanish- and approximately one week later in the
LI-Danish. This order was chosen in order to minimize as much as possible
the potential effects of the subjects' Lt on the L2. Subjects were told that
they were going to look at a wordless picture book where there were three
main characters, a boy, a dog, and a frog. They were further instructed to look
at the 24 pictures first in order to get an idea of what the story was about,
and then to write a narration describing what they saw in each picture. 2
Subjects were likewise presented with a bilingual list consisting of Spanish-
Danish pairs referring to the key nouns (not verbs) that appeared in the pictures
(e.g., window, deer, beehive) in order to provide them with the necessary
vocabulary to describe each picture. In order to ensure that subjects described
as many pictures as possible, they were instructed to use these nouns in
their narratives. A total of forty-five minutes was allotted for the task.

Authenticated
The experimental procedures for the native speaker group were identical to
those of the learner group with two exceptions: (a) they only performed the
task in Spanish; and (b) they were not presented with a bilingual list of
nouns.
4.3. DataAnalysis
The written data were first divided into clauses which, following Berman &
Slobin (1994: 660), were defined as "any unit that contains a unified predi-
cate". By "unified" was meant a predicate that expressed a single situation,
and it included both finite and nonfinite verbs. In general, clauses consisted
of a single verbal element, but infinitives and participles functioning as
complements of modal or aspectual verbs were included with the matrix
verb within a single clause (e.g., querer ir 'want to go', empezar a caminar
'start walking').
Once the written texts had been divided into clauses, the clauses
expressing motion events were identified, i.e., motion clauses. These were
the clauses that formed the basis for further analyses. Following Slobin
(1996b: 206), a motion event was defined as " ... the description of the
movement of a protagonist from one place to another. ..." The criterion was
thus that the protagonist had to end up in a different place from where he
started. Thus, for example, in the sentence Om natten da drengen sover i sin
seng,jlygter freen ('In the night when the boy sleeps in his bed, escapes the
frog'), two clauses were first identified - Om natten da drengen sover i sin
seng and jlygter jreen - and only the second clause was subsequently iden-
tified as involving a motion event.
The following analyses were conducted on the data: (1) a comparison of
the length of the texts produced by the two subject groups, i.e., the Spanish
native speaker group and the L2 learner group. This analysis was conducted
to examine whether the narratives written by the two groups were compara-
ble in this respect; (2) a type/token analysis of the motion verbs found in the
Danish and Spanish data; (3) an analysis of the degree of elaboration of
path of motion. This analysis comprised: (a) an examination of the satellites
found in the Danish and the Spanish data; and (b) an examination of the
extent to which the Danish and the Spanish data included minus-ground vs.
plus-ground clauses; (4) an analysis of the extent to which the Danish and
the Spanish data included event conflation; and (5) an analysis of the rela-
tive allocation of attention to movement (trajectories) vs. location (static
descriptions) in both languages.

Authenticated
24 Teresa Cadierno
In the analyses the following comparisons were made: Ca) Spanish narratives
written by the Spanish native speaker group, i.e., Ll Spanish narratives; Cb)
Danish narratives written by the learner group, i.e., Ll Danish narratives;
and Cc) Spanish narratives written by the learner group, i.e., L2 Spanish nar-
ratives.
4.4. Results and Discussion
4.4.1. Length of the Spanish and Danish Texts
As indicated above, this first analysis entailed a comparison of the total

number of clauses included in the frog story narratives, i.e., their total
length. The results of this analysis revealed the following pattern: the L 1
Spanish narratives (Mean: 64.6) were longer than the Ll Danish narratives
(Mean: 50.4) and these two were longer than the L2 Spanish narratives
(44.2). A non-parametric Wilcoxon signed-rank test performed on the data
revealed that the texts produced by the native speakers of Spanish were sig-
nificantly longer than the ones produced by the learners in their Ll-Danish
as well as in their L2-Spanish (p = 0.0053 and p = 0.0003, respectively).
However, no significant difference was found between the Danish and the
Spanish versions of the learner group (p = 0.15).
A qualitative analysis of the narratives showed that the Spanish native
speakers tended to provide certain types of information not strictly related
to the chain of events depicted in the "frog story". For example, reference to
the emotional states as well as the personal traits of the main characters -
the boy, the dog, and the frog - were much more frequent in the Spanish Ll
narratives (e.g., the initial happiness of the boy with the frog in his posses-
sion and the subsequent sadness and anxiety at its loss; the good-natured
character of the boy and the naughtiness of the dog). This finding parallels
the one obtained by previous studies such as McClure et al. (1993) where
narratives written by Spanish monolingual children - Mexicans - also con-
tained a higher proportion of information about emotional states and per-
sonal traits than narratives produced by English monolingual children -
Americans. This difference was attributed to cultural differences between
Americans and Mexicans, in the sense that Mexicans are "being more con-
cerned with psychological attributes, emotions, and subjective reality than
with pragmatic outcomes or objective reality" (p. 218). Whether the same
type of cultural difference exists between the two language groups included
in the present study is an empirical question, but the parallelism in the

Authenticated
results seems revealing. In addition, it was observed that the Spanish native
group included numerous instances of dialogues and conversations between
the main characters, either in direct or indirect speech. Such instances were
rare in the narratives produced by the Danish group.
Given the significant differences in total length mentioned above, a sec-
ond analysis was conducted which examined the relative frequency of
motion clauses used, i.e., clauses encoding motion events. That is, this
analysis examined the amount of motion clauses used in relation to the total
number of clauses produced by both subject groups. A Wilcoxon signed-
rank: test performed on the three means (Ll Spanish: 30.8; Ll Danish: 34.9;
L2 Spanish: 28.5) revealed no significant differences in the performance of
the two subject groups in their Ll, i.e., the Ll Spanish narratives and the Ll
Danish narratives (p = 0.1134). Additionally, no significant difference was
found between the Ll and the L2 Spanish narratives (p = 0.4509). In other
words, a comparable relative frequency of motion clauses was obtained by
(a) the Spanish NS group and the Danish learner group performing in the
Ll; and (b) the Spanish NS group and the Danish learner group performing
in the L2.
To summarize, the narratives written by the Spanish speakers were sig-
nificantly longer than the narratives written by the Danish speakers in both
Danish and Spanish. This difference, which was attributed to the inclusion
of additional information not strictly related to the chain of motion events
depicted in the "frog story" picture book, did not, however, have an effect
on the relative frequency of motion clauses used by the two subject groups.
Thus we can be confident as to the validity of the following analyses, which
are based solely on information contained in the motion clauses.
4.4.2. Type and Token Analysis
This analysis examined the total number of tokens and types found in the
Danish and the Spanish data. The motion verbs used by the learner group in
both Danish and Spanish are presented in Appendix A and the motion verbs
used by the Spanish NS group are presented in Appendix B. The lists
include both verbs of self-motion (e.g., caminar 'to walk') and of caused-
motion (e.g., tirar 'to throw'). Furthermore, following Slobin (1996b), a
plus sign following a verb indicates that it has been used with one or more
satellites.
The learner group used a total of 63 verb types in Danish and a total of
37 verb types in Spanish, whereas the Spanish NS group used a total of 67

Authenticated
26 Teresa Cadierno
verb types in Spanish. Table 1 shows the mean values for the tokens and
types used by the two subject groups along with the token-type ratios.
Table 1. Tokens, types, and token-type ratios in Danish and Spanish: Mean values
Learner group NS group
Danish Spanish Spanish
Tokens 17.4 12.8 19.7
Types 12.8 8.4 14.3
Ratio 0.75 0.67 0.74
Note: NS = Spanish native speaker
A Wilcoxon signed-rank test conducted on the mean scores for the verb types
in isolation revealed the following: (a) there were no significant differences
in the amount of verb types produced by the Danish and the Spanish speak-
ers in their Lt (p = 0.1659); (b) the learner group produced significantly
more motion verb types in the Lt-Danish than in the L2-Spanish (p =
0.0012); and (c) the learner group used significantly fewer motion verb
types in Spanish than the native speakers of this language (p = 0.0001).
Another Wilcoxon signed-rank test was conducted on the mean scores
obtained for the token/type ratios. This analysis differs from the previous
one in that it examines the verb types used by the subject groups in relation
to the total number of verb tokens produced, and it thus provides a more
balanced description of the data. The results of this analysis parallel those of
the previous analysis with respect to (a) (p = 0.9099) and (b) (p = 0.0410).
With respect to (c), however, the difference in performance between the
learner group and NS group did not reach significance (p = O. 0568).
However, the low p value obtained here in combination with the significant
result from the previous analysis seems to point to a difference in the amount
of Spanish motion verb types used by the learner and the NS groups.
In sum, the results obtained in the present analysis support the first
hypothesis posited at the outset of the study, namely, that the L2 learner
group would use fewer motion verb types in Spanish than the NS group.
Second language acquisition entails a process of progressive vocabulary
learning, and it was expected that these learners, although belonging to the
intermediate and advanced stages of language acquisition, would still not be
able to use as wide a variety of motion verbs as native speakers of this lan-

Authenticated
guage. It was noted that, for example, learners would use only one verb
(e.g., ir 'go') whereas Spanish native speakers would use additional verbs
with similar semantic import, such as dirigirse 'go' and encaminarse 'be on
one's way'. Furthermore, the difference in the Ll-L2 performance by the
learner group was also to be expected. That is, it was expected that the
learners would have at their disposal a greater variety of verb types in their
LI-Danish than in the L2-Spanish.
What was unexpected was the result concerning the non-significant dif-
ference between the two subject groups in their L1. This result seems to
contradict previous research by Berman & Slobin (1994) and Slobin (1998)
which has shown that native speakers of S-languages produced more verb
types than native speakers of V-languages. The results of the present inves-
tigation indicate that even though the two language types may differ with
respect to the total amount of motion verb types contained in their lexicons,
the subject groups investigated here showed no difference as to their use. A
possible explanation for the difference in results between Slobin's study and
the present study may be that whereas the former used oral narratives as
elicitation method, the latter used written narratives. Stylistic differences
between the written and the oral mode of communication, which are well
documented both at the grammatical and the lexical level (e.g., Schiffrin
1987; Vigara 1992), may explain the different results of the two investiga-
tions. Thus, at the lexical level, it is to be expected that the native speakers
will use a greater variety of motion verbs in the written mode than in the
oral mode, and that consequently the differences between native speakers of
S- and V-languages found in oral language will be leveled out in the case of
written language. A detailed analysis of the Spanish native data showed that
each individual tended to use several motion verbs with similar meanings in
their narratives (e.g., andar and caminar 'walk'; tirar and arrojar 'throw';
huir and escapar 'flee'). If the narratives had been oral, the speakers might
not have produced such a wider variety of lexical units encoding motion.
This is a matter that deserves future investigation.
4.4.3. Degree of Elaboration of Path of Motion
The present analysis examined whether the Danish learners of Spanish

exhibited a higher degree of complexity and elaboration of the path of
motion in their Spanish narratives than the Spanish native speakers. As indi-
cated in the analysis section of the paper, this analysis consisted of two dif-
ferent analyses which are presented below.

Authenticated
28 Teresa Cadierno
Satellites in the Danish and the Spanish data. Under (1 a) is the complete
list of Danish verbs + satellites contained in the Danish data:
(la)
Danish verbs + satellites:

begive sig + ud lrene sig + ind, op, ad, ud
bevrege sig + ud 10fte + op
bremse+ op 10be + afsted, efier, hen, om, over, op, vrek
brere + hen myldre+ud
drage + ud rende + rundt
dukke + frem, op ryge+ned
falder + derned, ned, ud, udover samle + op
fare + afsted skubbe + ned, ud
flyve + efier, ud slippe + ud
forsvinde + ud smide + ned, ud over
f01ger + efier, med snige sig + ind, om
fa + object + af, frit, ind, ned, ud snuble +ud
fa + participle + fast, ned stige + op
gallopere + afsted stikke + af, frem, ind, op, ned, ned i, ud
ga + hen, imod, over, pa, til, ud, udenfor stille sig + op
hoppe + op, ud stoppe + op
kaste + ned, ud st0tte sig + til
klatre + op, over sm +ud
komme + frem, op, til, tilbage, ud srette (sig) + ned, op
kravle + hen over, op, ud tage + afsted, med, ud
lande + i vrelte + ned
liste sig + afsted, forbi, op, over
As shown in (la), Danish has a great diversity of verb + satellite combina-

tions. Of the total 64 verb types used in Danish (as shown in Appendix A),
43 verbs were used with one or more satellites. Satellites were not only used
with general-purpose motion verbs such as bewege sig 'move' and sa!Ue
'set' but also with manner verbs such myldre 'swarm' and kravle 'crawl',
which parallels the findings reported in Berman & Slobin (1994) for
English adult speakers who, unlike children, used satellites with these two
types of motion verbs.
Satellites, however, were not only found in the Danish data. In the Spanish
L2 data, i.e., the narratives produced by the L2 learners, five directional
locative adverbs were used accompanying some verbs of motion: abajo
'downwards', (a)fuera 'outwards', arriba 'upwards', dentro 'inwards', and
encima 'topwards' (onto an upper surface). Under (lb) is the complete list
of the Spanish verbs + satellites that were used by the L2 learners:

Authenticated
(lb)
Spanish verbs + satellites:

caer + fuera, abajo 'fall + outwards, downwards'
ir(se) + abajo, encima, hacia aniba 'go + downwards, topwards, upwards'
hacer caer + abajo 'make fall + downwards'
meter + dentro 'insert + inwards'
mover + abajo 'move + downwards'
perseguir + dentro 'chase + inwards'
saltar + (a)fuera 'jump + outwards'
volar + aniba 'fly + upwards'
A total of 15 instances of verb + directional adverb were found in our data.

These instances, which were not present in the Spanish native data, were
produced by the learners belonging to the intermediate level of language
proficiency. Most of the adverbs were used redundantly, thus reinforcing the
meaning of the accompanying verb of motion (e.g., El perro ha metido la
cabeza dentro del tarr0 3 'The dog has inserted his head inside the jar', Al
final consiguio hacerla caer aba;o la colmena 'At the end he managed to
make it fall down the beehive'). This type of construction has also been
documented in first language acquisition by Sebastian & Slobin (1994) who
found that half of the preschooler children they investigated, especially the
4 and 5 year olds, produced instances such as the ones described above. Its
occurrence was explained by Berman & Slobin (1994) as reflecting a U-
shaped developmental curve, where some children seem to go through a
stage of redundantly encoding path of motion. 4 Consequently, the presence
of this type of construction in our data seems to indicate a developmental
process, one found in both first and second language acquisition, where
learners attempt to provide more information about path of motion than is
generally found in the language. In the case of the Danish second language
learners investigated here, this natural tendency would seem to be rein-
forced by the frequent presence of satellites in the learners' L 1, i.e., Danish.
Future research using cross-sectional designs with L2 learners from various
proficiency levels, and with L2 learners from typologically different LIs,
i.e., S- vs. V-languages, is needed to examine this issue in greater detail;
that is, to examine whether in the case of second language acquisition, the
presence of this type of redundant construction reflects a universal phenom-
enon or the influence of the Ll lexicalization patterns.
The learner Spanish data, however, also contains a few instances of non
redundant uses, as in Entonces el perro salto afilera de la ventana 'Then the
dog jumped out of the window' and El perro se cayo fuera de la ventana

Authenticated
30 Teresa Cadierno
'The dog fell out of the window', where the adverb (a)fuera specifies the path
of motion followed by the Figure - the dog. This type of construction, which
involves boundary-crossing situations was not found in the data from the
Spanish native speakers who, instead, favored constructions such as El perro
se cayo de la ventana 'The dog fell off the window' where path of motion is
omitted. The constructions found in the learner data thus seem to indicate that
the learners are adding some extra semantic information, namely information
on the path of motion, by means of a satellite. This semantic information,
on the other hand, must be obligatorily expressed in the learners' Ll-Danish
by a satellite, as in Sd hoppede hunden 1!:.d of vinduet 'So jumped the dog out
of the window' / Hunden faldt 1!:.d af vinduet The dog fell out of the window'.
Consequently, the presence of this construction in the Spanish L2 learner
data seems to indicate a process ofL! transfer at work. More specifically, in
accordance with the hypothesis posited at the outset of this study, some
learners, i.e., those who are at the intermediate stages of language acquisi-
tion, do exhibit a higher degree of complexity and elaboration of the path of
motion in Spanish than the native speakers of this language. This degree of
elaboration is, in turn, parallel to that found in the learners' Ll-Danish.
In addition to the anomalous constructions discussed above, the L2
learner data included another type of anomalous construction which is
exemplified by sentences such as El ciervo mueve al niiio y a su perro abaiQ
en un precipicio 'The deer moves the boy and his dog down in the abyss'
and Cuando el chico intenta irse arriba de algunos arboles 'When the boy
tries to go up of some trees'. In these examples, the learners have used a
non-directional verb of movement (i.e., mover 'move', ir 'go') followed by
a satellite encoding vertical path (i.e., abajo 'down', arriba 'up'). In addi-
tion, in the first example, the learner has omitted the semantic component of
cause which is present in the native Spanish rendering of the sentence, as in
El ciervo tiro / arrojo al niiio y a su perro a un precipicio 'The deer threw
the boy and his dog in the abyss' .
There are two possible explanations for all these types of anomalous
sentences, which are in agreement with some of the views proposed in the
literature on transfer: 5 (a) the learners are inadvertently transferring an L 1
pattern to the L2, possibly as the result of shared conceptual systems under-
lying both their Ll and their L2 structures (as discussed in Slobin 1993 and
Jarvis 2000, among others); or (b) the learners are evidencing a type of
communication strategy; that is, when the learners are confronted with a
communicative situation where they lack the knowledge of the appropriate
Spanish verb of movement, they purposely use an Ll structure to fill in the
gaps in their knowledge of the L2 (as suggested by Corder 1983, among

Authenticated
others). R. Ellis (1994) has likewise distinguished between learning transfer,

(i.e., role ofLl in interlanguage hypothesis construction) and communication
transfer (i.e., role of L 1 as a perfonnance phenomenon or strategy utilized to
compensate for deficiencies in the learners' interlanguage).
As indicated by R. Ellis (1994), even though it is possible to make a
clear theoretical distinction between learning transfer and communication
transfer, it is less clear how the two can be distinguished in empirical
research, given that we are normally dealing with production data. A possi-
ble solution is to collect both production and introspective data from the
learners as in Poulisse (1990), where learners were asked to retrospectively
comment on the types of strategies they had used in a previous task. Since no
retrospective data were collected in the present study, no hard conclusions
can be reached as to the nature of our learners' non-targetlike uses. However,
it seems plausible to hypothesize that in the case of the first type of anomaly
(i.e., addition of a semantic component - path of motion), a learning transfer
seems to be at work. The learners seem be inadvertently transferring the
lexicalization pattern of their Ll into the L2, thus explicitly encoding
semantic infonnation that is explicitly encoded in their Ll but omitted in the
target language. With respect to the second type of anomaly (i.e., creation of a
construction consisting of a non-direction motion verb + directional particle),
the fact that such uses were restricted to very few verb types (such as move
and go) combined with very few satellites (such as abajo and an·iba),
seems to pinpoint a communication strategy where the learners' apparent
lack of knowledge of the appropriate Spanish verbs of motion resulted in a
non-target-like expression: mover abajo instead of tirar and irse arriba
instead of subir. These examples seem to suggest that when the learner does
not know the exact Spanish verb of motion required for a specific context,
his/her communicative strategy is to use an Ll-type structure, i.e., a verb of
motion followed by a pm1icle expressing path. Future research, however,
needs to address this issue in a more systematic way.
4.4.4. Minus-ground vs. Plus-ground Clauses
The present analysis examined the extent to which the narratives produced
by the L2 learners and the Spanish native speakers included clauses incor-
porating a ground specification. As indicated by Slobin (1996a), descrip-
tions of movement can be equally enriched in both S- and V- languages by
means of prepositional phrases which add locative details to the verbs, and
which can express the ground of movement, i.e., source, medium, or goal:

Authenticated
32 Teresa Cadierno
Drengen og hunden gemte sig bag en trrestamme
El niiio y el perro se eseondieron detnls de un troneo
'The boy and the dog hid themselves behind a tree trunk. '
Slobin (1996 a & b) explored whether native speakers of typologically dif-

ferent languages - English and Spanish - showed the same tendency in
exploiting this possibility, i.e., he examined the extent to which they added
prepositional phrases to the verbs of motion in these two languages. His
results showed an overall tendency for English narratives to use more
ground adjuncts with verbs of motion than Spanish narratives, a tendency
that was more marked for mature speakers but also noticeable at younger
ages. Slobin (1996b) thus concluded that English native speakers seem to
pay more attention to path details than Spanish speakers.
Of interest to the present study was whether the Danish language learners
of Spanish investigated here exhibited a more frequent use of ground
adjuncts in the Spanish narratives than the native speakers of this language.
In this analysis a comparison was therefore made between the occurrence of
clauses that, on the one hand, consisted of verb + prepositional phrases
referring to the ground, i.e., plus-ground clauses (e.g., De faldt ned i vandet
'They fell down in the water'), and clauses consisting of bare verbs or verbs
+ satellite, i.e., minus-ground clauses (e.g., Defaldt ned 'They fell down').
This analysis, whose raw data is presented in Table 2, revealed the following
pattern: the L 1 Danish narratives ( Mean: 81.7) contained more ground
adjuncts than the L2 Spanish narratives (Mean: 67.1) and these contained
more than the narratives produced by the Spanish native speakers (Mean:
53.7). A Wilcoxon signed-rank test performed on these data showed a sig-
nificant difference between the L 1 and the L2 Spanish narratives (p =
0.0050), a significant difference between the Lt Spanish and the Lt Danish
narratives (p = 0.0000) and a significant difference between the Lt Danish
and the L2 Spanish narratives (p = 0.0048). In other words, the Spanish
native speaker group produced significantly lower number of plus-ground
clauses than the learner group independently of the language used, i.e., Lt-
Danish or L2-Spanish.
These results again support the hypothesis posited at the outset of the
study, i.e., that Danish learners of Spanish would exhibit a more frequent
use of ground adjuncts in their Spanish narratives than the native speakers of
this language. This overproduction of ground adjuncts, i.e., overproduction
in relation to the Spanish native data, seems to indicate that the L2 learners
are again transferring some of the typological characteristics of the first

Authenticated
Table 2. Percentages of plus-ground clauses in Danish and Spanish
S Danish Spanish Spanish
1 75.0 83.3 36.8

2 69.0 247.0 55.6
3 57.1 71.4 57.9
4 30.4 60.0 57.1
5 69.2 71.4 55.6
6 30.0 90.0 56.3
7 43.7 33.3 59.3
8 46.2 50.0 47.6
9 42.1 64.3 53.8
10 61.1 77.8 52.0
11 52.4 54.5 54.2
12 68.2 71.4 52.0
13 66.7 75.0 54.5
14 50.0 66.7 56.5
15 55.6 83.0 53.3
16 61.5 77.8 56.0
Note: S= Subjects; NS = Spanish native speaker
language, thus resulting in Spanish narratives being richer in locative trajec-

tories than the ones produced by Spanish native speakers. The fact that
transfer can manifest itself as overproduction has been repeatedly pointed
out in the literature (Odlin 1989; R. EIlis 1994). What is important to
emphasize here is that (a) the overproduction of locative details in Spanish
by our Danish language learners does not result in any kind of error, but just
in a different kind of rhetorical style than the one found for native speakers
of Spanish; and (b) such overproduction is allowed by the structure ofthe
L2. That is, it is possible in Spanish to add such locative details to verbs of
motion by means of prepositional phrases. As expressed by Andersen (1990)
in his Transfer to somewhere principle, a grammatical structure will be likely
to be transferred from the learner's Ll if the potential for generalization to
produce the same structure already exists within the L2 input.
4.4.5. Event Conflation
The aim of this analysis was to examine whether the Danish learners of
Spanish tended to make use of event conflation, i.e., to investigate whether

Authenticated
34 Teresa Cadierno
different composites of locative trajectories (path, ground-source, medium,

and goal) were or were not incorporated within a single clause in the two
languages. This analysis was conducted on the "cliff scene", which com-
prises two pictures (see Appendix C): Picture 17, where the deer is standing
at the edge of the cliff and the boy and the dog are in mid-air, and Picture
18, where the boy and the dog fall down into the water below.
This analysis was of interest given that research within L1 acquisition
again has shown clear differences among speakers of typologically different
languages (Berman & Slobin 1994). Thus, while English native speakers
(especially older children-9 years old-and adults) tended to make reference
to both source and goal either (a) in two different clauses, with some chrono-
logical sequencing overt or implied, or (b) by means of event conflation,
where both the cliff-source and the water-goal were combined within a single
clause (e.g., The deer dropped the boy off a cliff into the water), Spanish
native speakers of all ages very rarely specified both source and goal within
a single clause.
The absence of event conflation in Spanish has been explained by the fact
that its grammar does not easily allow for the expression of such compact
expressions. According to Slob in (1997), this is due to the presence of a
constraint in V-languages which prevents the accumulation of more than
one ground per clause when a boundary-crossing within a given path is
involved. In Slobin's own words, "It appears to be a universal characteristic
of V-languages that crossing a spatial boundary is conceived of as a change
of state, and that state changes require an independent predicate in such lan-
guages" (Slobin 1997: 441). In other words, while English allows for bound-
ary-crossing expressions such as She went downstairs and out of the house
(example cited in Slobin 1997: 441), Spanish does not: *Ellafue abajo y
fuera de la casa. The appropriate Spanish expression - Ella bajo la escalera
y salio de la casa 'She descended the stairs and went out of the house' -
would require breaking the extended path into two separate predicates.
Furthermore, even though V-languages allow for such compact expres-
sions when a boundary-crossing is not involved as in Subi los anchos
escalones hasta el/a 'I went up the great stairs towards her' (example cited
in Slobin 1997: 441), such expressions are extremely rare in these languages.
According to Slobin (1997), this is due to the fact that the combined effect
of a lexicon rich in path-verbs and the boundary-crossing constraint creates
a characteristic narrative style in V-languages that would go beyond its core
constraints.
The results of the analysis on event conflation, which is shown in Table 3,
indicate that the descriptions pertaining to the "cliff scene" in our data differ

Authenticated
from those described in the work of Slobin and his colleagues in that not
only two locations (the cliff-source and the water-goal) were mentioned, but
also a third one, the so called afgrund 'abyss'. This inclusion was very
interesting since its use seemed to invoke different conceptualizations of the
scene, depending on the grammatical construction in which it appeared, an
issue that will be discussed below.
Table 3. Reference to different locations in Danish and Spanish (cliff scene)
Narratives I lac. 2 lac. 3 lac. E.C.
Ll Danish 2 7 2 5
Ll Spanish 3 13 o o
L2 Spanish 2 11 3 o
Notes: I loc.= Reference to one location; 2 lac. = Reference to two locations;
3 loco = Reference to three locations; E. C. = Event Conflation.
The following categories were used in this analysis:
a) Reference to one location alone (always the water-goal)
Hjorten smider bade drengen og hans hund ned i en lille so.

The deer throws both the boy and his dog down in a little pond.
En el bosque se encuentran con un topo, un buho y un ciervo / y con todos
tienen problemas / Al ultimo se caen en una clzarca.
In the forest they meet a mole, an owl and a deer / and with all of them
(they) have problems / At the end (they) fall into a pond.
(b) Reference to two locations - either the cliff, the abyss or the water - in
separate clauses:
Sa kaster dellne hjort drengen ned i en afgrund / Begge falder pa hovedet i

vandet.
So throws this deer the boy down in an abyss / Both fall on the head in the
water
Hjorten lober mod en afgrund / hvor den stopper hrat / og drengen faldel' ud
over og /led i vandet
The deer runs towards an abyss / where it stops suddenly / and the boy falls
out over and down in the water

Authenticated
36 Teresa Cadiemo
These two Danish examples are interesting to compare because they show
that the same word, i.e., the afgrund 'abyss' seems to invoke two different
conceptualizations when used in different grammatical constructions. Thus, in
the first example, the word afgrund 'abyss' seems to profile6 the undefined
endpoint of the downward movement; that is, the boy is profiled as moving
on a path downwards from an implicit unstated source - the cliff - towards
an abyss which is pictured as being below that source. And in the next
clause, the endpoint of that abyss is defined as the water. In the second
example, on the other hand, the word afgrund 'abyss' seems to profile the
starting point or origin of the downward movement; here the boy is profiled
as moving on a path from the beginning of an abyss, from which he falls
down into the goal-ground, the water. In other words, in the first example,
the word afgrund focuses on the lower part or the depth of the abyss, and as
such it is conceptualized as an undefined goal, while in the second example
it focuses on the top part of the abyss, and as such it is conceptualized as a
sort of origin or beginning of the downward movement.
The same patterns in the conceptualizations are found in the Spanish data:
El ciervo lleva al niiio colgado encima de su cabeza. I Lo tira a un vreciDicio I

y el perro cae tambiim I Alfondo hay una charca
The deer carries the boy hanging on top of its head / (It) throws him into an
abyss / and the dog falls as well / At the bottom there is a pond.
Un ciervo 10 coge con sus cuemos I y empieza a con'er! El ciervo se acerca

a un precipicio I donde para duramente I y el niiio cae en ~a.
The deer takes him with its antlers / and (it) starts running/ The deer gets
close to an abyss / where (it) stops suddenly / and the boy falls into the
water.
In the first example, the abyss is conceptualized as an undefined endpoint,

and the pond as the specified goal; in the second example, on the other hand,
the abyss is conceptualized as the origin of the downward movement and
the water as the goal of movement.
(c) Reference to three locations: the cliff, the abyss and the water, in separate
clauses:
/fiorten anbringer drengen i sit gevir I og gar imod en steil skrt:ent I hvor den
lader drengenfalde ned i gfgrunden I Hunden og drengen lander samtidigt i
vandet.

Authenticated
The deer places the boy in his antler / and goes towards an abrupt cliff /
where he lets the boy fall down in the abyss !The dog and the boy land at the
same time in the water.
El ciervo coge alniiio con sus cuemos / y le lleva hasta un acantilado / y le

deja caer en el vreciDicio / El perro ladra / y se cae con el niiio al agJHl.
The deer takes the boy with its antlers / and (it) takes him to a cliff / and (it)
lets him fall into the abyss / The dog barks / and (it) falls with the boy into
the water.
In the examples in both languages, the cliff is conceptualized as the source,

the abyss as the undefined goal, and the water as the defined goal. This
same conceptualization is found in all the examples where reference is
made to the three locations.
(d) Event conflation: inclusion of at least two of the locations within a

clause:
Hjorten lager ham med / og smider ham ud over en afg,rund ned i en brfJnd
The deer takes him with / and throws him over an abyss down in a pond.
In this example, the second clause specifies two locations, the abyss and the
pond.
As shown in Table 3, the predominant patterns of expression for the Lt
Danish narratives were (a) the reference to two locations in separate clauses
(7 subjects); and (b) event conflation (5 subjects). The remaining subjects
chose either to make reference to the three locations or only one of the loca-
tions (the goal). In contrast, the predominant pattern of expression for the
Lt Spanish narratives was the reference to two locations in separate clauses
(13 subjects). The remaining 3 Spanish native speakers made reference to
only one location (the goal). With respect to the L2 Spanish narratives, the
predominant pattern of expression was the reference to two locations in two
separate clauses (11 learners). The remaining learners either made reference
to three locations or only one location.
A Fisher's exact test performed on these data revealed the following: (a)
a significant difference was found between the patterns followed by the two
groups in their Lt (p = 0.020). This result mirrors the one found by Berman
& Slobin (1994) concerning the different rhetorical styles followed by adult
speakers of S- vs. V- languages: whereas the former tend to favor event con-
flation (i.e., the most compact construction) and event serialization, i.e., the
expression of two or three locations in separate clauses, the latter tend not to
use event conflation, favoring instead event serialization; and (b) no signifi-

Authenticated
38 Teresa Cadierno
cant differences were found between the Spanish native speakers' patterns
and the Spanish L2 learners' patterns (p = 0.258). This result does not support
the hypothesis stated at the outset of the study, namely that there would be
differences in the way in which Danish learners of Spanish and native
speakers of this language would use event conflation in their nalTatives. As
shown in Table 3, the L2 learners investigated here did not employ event
conflation and consequently, did not incorporate the typological pattern
characteristic of their mother tongue into their second language.
A possible explanation for this finding may be what Kellerman (1978,
1986) has termed "psychotypology", which refers to the learners' perceived
distance between the constructions in the two languages. Kellerman demon-
strated that the L2 learners have some perceptions about a) what structures
from their native language are or are not prototypical and therefore poten-
tially transferable or non-transferable, and b) whether the native and target
languages are relatively similar or different. These perceptions influence
what they actually transfer into the L2. In the case of the present study, it
could be the case that the Danish learners perceive constructions involving
event conflation as characteristic of Danish and uncharacteristic of Spanish,
thus constraining or preventing transier. This is a matter that deserves further
investigation.
4.4.6. Movement and Setting.
Given the finding from L1 acquisition research that typological differences

between languages not only have implications for the description of motion
per se but for the description of location as well, the aim of the present
analysis was to examine the L2 learners' relative allocation of attention to
elements of movement and setting. More specifically, this analysis exam-
ined the extent to which the L2 learners and the Spanish native speakers
provided static descriptions or descriptions of trajectories in relation to the
cliff scene which comprises picture 17 and picture 18 (see Appendix C).
Table 4 below shows the pattern of use of the learners and the Spanish
native speakers both at the individual and the group level. The Table speci-
fies whether or not the informants provided static scene descriptions. If no
static descriptions were provided, i.e., if a minus value appears on the Table,
this means that descriptions oftrajectories were given by the informants.
As shown in Table 4, the predominant pattern of expression for the Ll
Danish nalTatives was to provide explicit descriptions of trajectories (15
subjects). Only one subject provided a static description. With respect to the

Authenticated
L2 Spanish narratives, the distribution looks different: 9 subjects provided

static descriptions whereas 7 provided descriptions of trajectories. Finally,
the pattern for the L1 Spanish narratives looks very similar to that of the L2
Spanish narratives: half of the subjects provided static descriptions, half
provided descriptions of trajectories.
A Fisher's exact test performed on these data revealed the following: (a)
a significant difference was found between the patterns followed by the two
groups in their L1 (p = 0.003). This result agrees with previous findings in
first language acquisition (Berman & Slobin 1994) and provides further
support for the claim concerning the different rhetorical styles used by
native speakers of S- and V- languages; it is worth noting, however, that the
L 1 Danish group showed a more consistent pattern of use than the Spanish
Lt group. All subjects except one in the Lt Danish group followed the
expected S-language pattern, i.e., allocation of attention to trajectories,
whereas subjects in the Spanish L1 group varied with respect to the rhetorical
style used; (b) a non-significant difference was found between the Spanish
Table 4. Elaboration of setting-description in Danish and Spanish (cliff scene)
S Danish Spanish Spanish
1 + +
2 + +
3 + +
4 + +
5 + +
6 + +
7 + +
8 + +
9 +
la
11
12
13
14
15
16 +
Total 9 8
Note: S= Subjects; NS = Spanish native speaker

Authenticated
40 Teresa Cadierno
L 1 patterns and the Spanish L2 patterns (p = 0.500). This result does not
support the hypothesis posited at the beginning of this investigation, namely,
that there would be differences in the relative allocation of attention to
movement and location by Danish learners of Spanish and Spanish native
speakers. It is important to note, however, that this result is due to the mixed
pattern of use followed by the Spanish native speakers; and Cc) a significant
difference was found between the learners' performance in the L 1 and in the
L2 (p = 0.0156). That is, whereas the learners followed the typical pattern of
S-languages, i.e., allocation of attention to trajectories, in their LI-Danish,
their pattern of use in the L2-Spanish was mixed.
Table 4 further indicates different patterns with respect to the learners'
performance in the Ll and the L2. Ten learners allocated differential
attention to movement and setting in the Danish and the Spanish versions of
the frog story. Nine of them (No. 1-9) provided descriptions of trajectories
in the Ll-Danish and descriptions of settings in the L2-Spanish, whereas
the other learner (No. 16) followed the reversed pattern, i.e., she provided a
setting description in the Ll-Danish and a description of the trajectory
involved in the L2-Spanish. The following example illustrates the former
pattern, which reflects the characteristic rhetorical style of the two typologi-
cally different languages involved in this study:
Ved hjrelp af sit stor gevir bar hjorten drengen hen til en skrrent. / hvorefter
den smed ham ned i afgrunden. / Hunden ... rag samme vej / Heldigt for
dem var det / at de faldt ned i en dyb sa.
With the help of his big antler carried the deer the boy to a cliff / where after
he threw him down in the abyss / The dog ... went the same way / Luckily
for them was / that they fell down in a deep pond.
Un ciervo con sus cuernos 10 lleva / y luego 10 avienta hacia un precipicio /

El perro queriendo ayudarlo / tambien cae al precipicio / Como habia un
klgo mas aba;o. / sobrevivieron tanto el nino como el perro.
A deer with its antlers carries him / and afterwards (it) throws him towards
an abyss / The dog wanting to help him / also falls into the abyss / As there
was a lake further down / survived both the boy and the dog.
In the Danish version, the segments of the path are explicitly encoded by
means of a satellite - ned - and the setting thus needs to be inferred: i.e., the
fact that the cliff is located above the lake. In the Spanish version, on the
other hand, the setting is explicitly stated. That is, it is explicitly stated that
there is a lake below at the bottom of the abyss, a description which allows
the path to be inferred.

Authenticated
The remaining six learners (No. 10-15) followed the same narrative style in
both languages, i.e., they specified the segments of the path, thus leaving
the setting to be inferred by the reader, as in the following example:
Men /ige for skrcellten naede ~jorten at stoppe /og bade Peter og Frede
faldt ned i afgrunden, men heldigvis - lige i en skovso.
But just before the cliff, managed the deer to stop / and both Peter and Frede
fell down in the abyss - but luckily in a forest pond.
El ciervo empezo a carrer / y al parar a un precipicio / se cayo otra vez el

nino nuis el pe/TO bajo en una charca.
The deer started to run / and when (it) stopped at an abyss / fell down again
the boy plus the dog down into a pond.
The fact that one informant (No. 16) provided a static description in the
Danish version together with the mixed pattern of use exhibited by the
Spanish native speakers included in this study suggests that the rhetorical
styles characteristic of the two language types are a result of strong tendencies
in the languages, thus allowing for possible overlapping across typologies:
5. Conclusions
The purpose of this article was to report an investigation of how learners

whose L1 can be characterized as an S-language (Danish) come to express
motion events in a L2 V-language (Spanish), two languages involving dif-
ferent lexicalization patterns with respect to the expression of the various
motion components.
The results of the study partially support the hypotheses initially stated.
Thus, as hypothesized, differences were found between Danish learners of
Spanish and native speakers of this language in terms of the total amount of
verb types used, i.e., the learners used fewer motion verb types than the
native speakers. In addition, some ofthe learners - those in the intermediate
stages of language acquisition - exhibited a "satellization" of the Spanish
locative constructions, i.e., the use of redundant and anomalous path parti-
cles not found in the adult native speaker data. Finally, the learners added
more ground adjuncts to the motion verbs than the Spanish native speakers.
The latter two results indicate that the learners exhibited a relatively higher
degree of complexity and elaboration of the semantic component of path of
motion than the one exhibited by the Spanish native speakers. This higher

Authenticated
42 Teresa Cadierno
elaboration seems to be attributable to the influence of the learners' L 1-

Danish, given that the learners have been trained to provide such elaboration
in their native language. This conclusion, however, which is in agreement
with Slobin's thinking for speaking hypothesis, must be taken with care since
the present investigation did not include a comparison group of learners
with a verb-framed Ll.
On the other hand, the hypotheses concerning the differences between the
L2 learners and the Spanish native speakers on the use of event conflation
and the allocation of attention to movement and setting were not supported.
The learners - like their native speakers counterparts - did not use event
conflation in Spanish, a construction not easily allowed in this language but
common in Danish. In other words, the learners did not transfer the charac-
teristic typological pattern of the Lt into the L2. It was argued that this
could be due to the learners' psychotypology, i.e., the perceived distance
between the characteristic constructions of the two languages (Kellerman
1978, 1986). Finally, with respect to the allocation of attention to movement
and setting, both the L2 learners and the Spanish native speakers exhibited a
mixed pattern of use in the Spanish narratives, with roughly half of the sub-
jects providing static descriptions and the other half providing descriptions
of trajectories. This result was unexpected in the light of previous research
(Berman & Slobin 1994) which had shown a stronger tendency among
native speakers of Spanish to provide static descriptions. This finding seems
to indicate the possibility of a strong variability among native speakers of
V- languages, and evidences the existence of overlapping patterns of use
across typologies. This conclusion is further supported by previous research
conducted on other V-languages such as Basque (Ibarretxe-Antufiano 2001)
which has shown that the native speakers of this language tend to devote
more narrative attention to the dynamics of movement than to setting-
descriptions, and their rhetorical style in this respect thus resembles more
that of S-languages than V-languages.
The results of this investigation, therefore, do not show a consistent
picture with respect to the role of the learners' Lt in the L2 expression of
motion events. 7 The influence of the learners' L 1 seems to be present in
some aspects of their L2 use, such as the high degree of complexity and
elaboration of path of motion, but not in others, such as the absence of event
conflation constructions. These contradictory results seem to support the
view ofLI crosslinguistic influence as a complex phenomenon which mani-
fests itself in multiple ways (e.g., errors, avoidance, and overuse of certain
linguistic structures), and which interacts or is constrained by a number of
factors such as the learners' general level of development and the learners'

Authenticated
perceptions about similarities and differences between the Lt and the L2

(Ellis 1994). More specifically, the learners' level ofL2 proficiency seemed
to interact with the phenomenon of cross-linguistic influence given the dif-
ferent patterns of use exhibited by intermediate and more advanced learners
in terms of the "satellization" of Spanish locative construction. Likewise, the
learners' perceived distance between characteristic Ll and L2 constructions
could explain the non-production of event conflation construction in L2
Spanish by our Danish informants. Given the limitations of the present
paper, which have been mentioned above, future research on cognitive
typology and SLA could benefit from designs which would include (a)
learners from various language proficiencies, especially beginners and well-
advanced or near natives; (b) two groups oflearners, i.e., one whose Ll and
L2 share the same typological pattern, and one whose Ll and L2 do not
share it; and (c) retrospective data as well as both oral and written production
data in order to investigate the difference between learning transfer and com-
municative transfer more thoroughly, as well as the issue of the learners'
psycho typology. Research designs such as these will help to shed more light
on the different factors that can interact with the phenomenon of cross-
linguistic influence in the expression of motion events in an L2.
To conclude, as indicated by Odlin (1989), research based on linguistic
typologies can be useful in the study of SLA phenomena such as transfer
since it allows for the investigation of systemic influences, and for a clearer
understanding of the complex interplay between transfer and developmental
sequences or the natural principles ofL2 acquisition. It is the main claim of
this paper that this type of research in SLA can benefit from the framework
of cognitive linguistics, and more specifically cognitive typologies such as
Talmy's (1985), which have already been fruitfully used in cross-linguistic
research in first language acquisition.

Authenticated
44 Teresa Cadierno
AppendixA.
List of motion verbs produced by the learner group.
Danish verbs:
anbringe 'place', baske 'flap', begive sig+ 'move', bremse+ 'stop', bringe
'bring', brere+ 'carry', drage+ 'pull', dukke+ 'emerge', falde+ 'fall', fange
'catch', fare+ 'rush', flygte 'escape', flyve+ 'fly', forf£llge 'chase', forlade
'leave', forsvinde+ 'disappear', f£llge+ 'follow', fH 'get', fa + past partici-
ple+, galopere+ 'gallop', gemme sig 'hide', gH 'walk', havne 'land', hente
'fetch', hoppe+ 'jump', jagte 'chase', kaste+ 'throw', klatre+ 'climb',
komme+ 'come', kravle+ 'crawl', lande+ 'land', liste sig+ 'creep', lregge
(sig) 'lay', lrene sig+ 'lean', l£lfte+ 'lift', l£lbe+ 'run', medbringe 'bring
along', medtage 'bring with', myldre+ 'swarm', nrerme sig 'approach', na
'reach', ramme 'fall on', rende+ 'run', ryge+ 'fly off', ryste 'shake', samle+
'assemble', skubbe+ 'push', slippe+ 'release, let fall', srnide+ 'throw', snige
sig+ 'creep', snuble+ 'stumble', stige+ 'rise', stikke+ 'put, thrust', stille sig+
'place', stoppe+ 'stop', st£ltte sig+ 'lean', stii+ 'stand', srette (sig)+ 'place',
tage+ 'take', undslippe 'escape', vandre 'stroll', vrelte+ 'knock over', vrere
pa vej 'go'.
Spanish verbs:
acercarse 'approach', acostarse go to 'bed', aparecer 'turn up', asomar
'begin to appear', aventar 'throw', caer(se) 'fall', capturar 'capture', coger
'take', correr 'run', dejar 'leave behind', dejar caer 'let drop', desaparecer
'disappear', encerrar 'shut in', escalar 'climb', escapar(se) 'escape', escon-
der(se) 'hide', hacer caer 'make fall', huir 'flee', ir(se) 'go', levantarse 'get
up', llegar 'arrive', llevar(se) 'carry', meter 'insert', mover 'move', parar
'stop', perseguir 'chase', poner 'put', resbalar 'slide', sacar 'take out', salir
'leave', saltar 'jump', seguir 'follow', subir(se) 'ascend', tirar 'throw', venir
'come', volar 'fly', volver 'return'.

Authenticated
AppendixB.
List of motion verbs produced by the Spanish native speaker group
Spanish verbs:
abandonar 'leave', acercarse 'approach', acostarse 'go to bed', adentrarse
'go into', alejarse 'go away', andar 'walk', aparecer 'appear', apoyarse
'lean', arrastrar 'drag along', arrojar(se) 'throw', asomarse 'lean oneself
over', atravesar 'cross', caer(se) 'fall', caminar 'walk', capturar 'seize',
coger 'take', colocar 'put', correr 'run', dejar 'leave', depositar 'put', deten-
erse 'stop', derrapar 'skid', desaparecer 'disppear', dirigirse 'go', empujar
'push', encaminarse 'be on one's way', enganchar(se) 'hook up', entrar
'enter', escapar(se) 'escape', esconderse 'hide', flotar 'float', hacer caer
'make fall', huir 'flee', introducir 'insert', ir 'go', ir de vuelta 'go back', lan-
zar 'throw', levantarse 'rise', llevar(se) 'transport', llegar 'arrive', mar-
charse 'leave', merodear 'maraud', meter 'insert', moverse 'move', parar
'stop', pasar 'go by', perseguir 'pursue', poner pies en polvorosa 'flee',
echarse para atras 'back out', menear 'waggle', precipitarse 'precipitate',
recoger 'pick up', regresar 'return', refugiarse 'take refuge', retirarse 'go
away', sacar 'take out', salir 'leave', saltar 'jump', seguir 'follow', subir(se)
'ascend', tirar 'throw', transportar 'transport', trepar 'climb', venir 'come',
volar 'fly', volcar 'capsize', volver 'return'.
Appendix C.
Pictures 17 and 18 of the "frog story".

Authenticated
46 Teresa Cadierno
Notes
1. This research was funded by Aarhus University Research Foundation and The
Faculty of Humanities at the University of Southern Denmark. Some of the
issues discussed in this paper were presented in a talk by Teresa Cadierno and
Anne Jensen at the AAAL conference in Vancouver (2000). I would like to
thank Werner Vach for his valuable statistical help, as well as Eva Dam Jensen
and Lucas Ruiz for comments on earlier versions of this article. Thanks also to
Sharon Millar for linguistic revisions of the paper.
2. Given that previous research on interlanguage variability (e.g., Tarone 1983) has
shown L2 learners' performance to be more target-like in written tasks than in
oral tasks - the reason being that in written tasks they can attend more closely to
the language code - written data were collected in this study. Differences in
results between previous research and the present one, which could be attributed
to differences in elicitation mode, are discussed later in the paper.
3. This sentence was also produced by a few adult native speakers of Spanish in
Sebastian and Slobin's (1994) study. Its use was accounted for by the fact that
the situation describes a kind of non-canonical change of location, which makes
the native speaker emphasize the fact that the dog had actually put his head
inside the jar.
4. Another possible explanation offered is that the two groups of presehoolers, i.e.,
the ones that used the directional adverbs and the ones that did not, simply re-
present two different types of individual styles. Further research is needed to
shed some light on this issue.
5. I would like to thank Nick Ellis for bringing this issue to my attention.
6. The term "profile" has been used by Langaeker (1987, 1998) to refer to a kind of
prominence involving the focusing of attention: "Within the full conceptualization
it evokes, an expression directs attention to some particular substructure - its
profile - as being the entity that it designates or refers to. An expression's profile
is thus its 'referent', in a psychological ... sense of that term." (Langacker 1998: 9).
7. I would like to thank an anonymous reviewer of this paper for his/ her comments
on this regard.

Authenticated
References
Andersen, R.
1990 Models, processes, principles and strategies: Second language acqui-
sition inside and outside of the classroom. In B. VanPatten & J. F. Lee
(Eds.), Second language acquisition - Foreign language learning.
Clevedon, Avon: Multilingual Matters.
Becker, A. & M. Carrol
1997 The acquisition of spatial relationships in a second language
Amsterdam / Philadelphia: John Benjamins.
Berman, R. A. & D. I. Slobin
1994 Development of linguistic fonns: English. In R. A. Berman & D. I.
Slobin (eds.). (1994). Relating events in narrative: A crosslinguistic
developmental study. Hillsdale, New Jersey: Lawrence ErIbaum.
Bloom, P., M. A. Peterson, L. Nadel, & M. F. Garret
1996 Language and space. Cambridge, Mass: The MIT Press.
Bowerman, M.
1994 From universal to language-specific in early grammatical develop-
ment. Philosophical Transactions of the Royal Society. London,
B34637-45.
1996a The origin of children's spatial semantic categories: Cognitive versus
linguistic determinism. In J. J. Gumperz & S. C. Levinson (eds.),
Rethinking linguistic relativity (pp. 145-176). Cambridge: Cambridge
University Press.
1996b Learning how to structure space for language: A crosslinguistic
perspective. In P. Bloom, N. A. Peterson, L. Nadel & M. F. Garrett
(eds.), Language and space (pp. 387-436). Cambridge, Mass: The
MITPress.
Cadierno, T. & K. Lund
In press. Cognitive linguistics and second language acquisition:
Motion events in a typological framework. In B. VanPatten, J.
Williams & S. Rott (eds.), Form-meaning connections in second
language acquisition. Hillsdale, New Jersey: Lawrence ErIbaum.
Choi, S. & M. Bowerman
1991 Learning to express motion events in English and Korean: The influ-
ence of language-specific lexicalization patterns. Cognition 41:
83-121.
Corder, P.
1983 A role for the mother tongue. In S. Gass & L. Selinker (eds.),
Language transfer in language learning (pp. 85-97). Rowley, MA:
Newbury House.
Ellis, R.
1994 The study ofsecond language acquisition. Oxford: Oxford University
Press.

Authenticated
48 Teresa Cadierno
Ibarretxe-Antuiiano, I.
2001 Motion events in Basque narratives. Linguistic LAUD Agency. Series
A. General and Theoretical Papers N. 528. Universitiit Essen.
Jarvis, S.
2000 Methodological rigor in the study of transfer: Identifying L1 influence
in the interlanguage lexicon. Language Learning 50 (2): 245-309.
Kellerrnan, E.
1978 Transfer and non-transfer: where are we now? Studies in Second
Language Acquisition 2: 37-57.
1986 An eye for an eye: crosslinguistic constraints on the development of the
L2lexicon. In E. Kellerman & M. Sharwood Smith (eds.), Cross-lin-
guistic influence in second language acquisition. Oxford: Pergamon.
Langacker, R. W.
1987 Foundations of cognitive grammar, Vol 1: Theoretical perspectives.
Stanford, California: Stanford University Press.
1998 Conceptualization, symbolization, and grammar. In M. Tomasello
(ed.), The new psychology of language: Cognitive and functional
approaches to language structure. London: Lawrence Erlbaum.
Levine, S. C. & S. Carey
1982 Up front: The acquisition of a concept and a word. Journal of Child
Language 9: 645-657.
Lucy, J. A.
1992 Grammatical categories and cognition: A case study of the linguistic
relativity hypothesis. Cambridge: Cambridge University Press.
1996 The scope of linguistic relativity: An analysis and review of empiri-
cal research. In J. J. Gumperz & S. C. Levinson (eds.), Rethinking
linguistic relativity (pp. 37-69). Cambridge: Cambridge University
Press.
Mayer,M.
1969 Frog, where are you? New York: Dial Press.
McClure, E., M. Mir, & T. Cadierno.
1993 What do you include in a narrative? A comparison of the written nar-
ratives of Mexican and American fourth and ninth graders.
Pragmatics and Language Learning Monograph Series, Volume 4:
209-224.
Odlin, T.
1989 Language transfer: Cross-linguistic influence in language learning.
Cambridge: Cambridge University Press.
Piaget, J. & B. Inhelder
1956 The child's conception of space. London: Routledge.
Poulisse, N.
1990 The use of compensatory strategies by Dutch learners of English.
Enschede: Sneldruk.
Schiffrin, D.
1987 Discourse markers. Cambridge: Cambridge University Press.

Authenticated
Sebastian, E. & D. I. Slobin.

1994 Development of Spanish forms: Spanish. In R. A. Berman & D. I.
Slobin (eds.), Relating events in narrative: A cross linguistic
developmental study (pp. 239-284). Hillsdale, New Jersey: Lawrence
Er1baum.
Slobin, D. I.
1993 Adult language acquisition: A view from child language study. In C.
Perdue (ed.), Adult language acquisition: Cross-linguistic perspectives
(pp. 239-252). Cambridge: Cambridge University Press.
1996a From "thought and language" to "thinking for speaking". In J.
Gumperz, J. John & S. C. Levinson (eds.), Rethinking linguistic rela-
tivity. Cambridge: Cambridge University Press.
1996b Two ways to travel: Verb of motion in English and Spanish. In M.
Shibatani & S. A. Thompson (eds.), Grammatical constructions:
Their foml and meaning. Oxford: Clarendon Press.
1997 Mind, code, and text. In J. Bybee, J. Haiman & S. A. Thompson
(eds.), Essays on language jimction and language type: Dedicated to
T Givon (pp. 438-467). Amsterdam: John Benjamins.
1998 Verbalized events: A dynamic approach to linguistic relativity and
determinism. Working Paper for the LAUD Symposium. Essen:
Linguistic Agency University-GH Essen.
Ta1my, L.
1985 Lexicalization patterns: Semantic structure in lexical forms. In T.
Shopen (ed.), Language typology and syntactic description. Vol. 3:
Grammatical categories and the lexicon. Cambridge, UK: Cambridge
University Press.
Tarone, E.
1983 On the variability of interlanguage systems. Applied Linguistics 4:
142-163.
Vigara, Ana Maria
1992 Mo rfos in taxis del espaiiol coloquial. Esbozo estilistico. Madrid:
Gredos.
Whorf, B. L.
1940 Linguistics as an exact science. Technology Review 43: 61-63.
Reprinted in 1956, Language, thought and reality: Selected writings
of Benjamin Lee Whorfby J. B. Carroll (pp. 220-232). Cambridge,
MA: MIT Press.

Authenticated
Authenticated
Construal, Convention, and Constructions
in L2 Speech
Renee Waara
1. Introduction
A principled account of how learners produce "slightly off' utterances is

difficult for several reasons related to theoretical orientation to the nature of
language and language acquisition. Language usage and speaker construal
are important considerations in accounting for language acquisition and
these aspects are not a direct part of a purely syntactic approach, which
excludes the nature of the conceptualization process. Consequently, a cogni-
tive linguistics approach that does take speaker construal and usage into
account could shed some light on aspects of second language acquisition
otherwise left undiscovered. Achard (1997) argues for Cognitive Grammar
as a plausible and preferred alternative to traditional generative grammar
accounts of second language acquisition, partially because an utterance can
be syntactically correct and yet be pragmatically incorrect. This mismatch
appears to be the case in a group of constructions identified here as learner
constructions, and will be taken a step further to include not only pragmati-
cally odd expressions, but also form-meaning reconstructions or the use of a
construction in which the meaning is retrievable but not in the conventional
sense. Even though learner constructions deviate from conventional usage,
communication does not break down. The emergence of learner constructions
in L2 speech will be presented using notions central to a usage-based
approach that relies on the notions of construal, convention, and construc-
tions.
A central concern in second language acquisition theory is the way in
which a learner reorganizes and utilizes input from the target language. The
intermediate state of second language acquisition has commonly been
referred to as "interlanguage". In a usage-based model there are certain
basic assumptions regarding the nature of interlanguage. First, knowledge
of L 1 consists of grammatical structures that capture generalizations of the
inventory of specific expressions. Therefore, we would expect some degree

Authenticated
52 Renee Waara
of transfer-related phenomena in L2. Second, the developing L2 language

system contains a specific list of expressions from which grammatical struc-
tures emerge due to the processes of categorization and schema extraction.
Third, learners have access to their general knowledge of the world, which
allows them to construe events. Furthermore, the existing structures of Ll
will naturally be reflected in the L2 because of accessibility and degree of
automaticity attained in L 1. However, linguistic items in L2 are put in focus
while the speaker is speaking L2 and therefore we might anticipate some
blending effects.
Following Fauconnier (1997), blending is the combination of two inputs
in mental space that yield a third mental space that is called the "blend".
The blend contains characteristics from each of the inputs, yet it has its own
structure. In some learner constructions, e.g., Can I get a dance?, it appears
that two otherwise conventional uses, e.g., Can I get a coffee? and May I
have this dance? are the sources for the learner construction, Can I get a
dance? The notion of blending will be evoked to discuss learner constructions
in section 4.2.
Frequency plays a role in language production in terms of accessibility
as well as facilitating comprehension. Although frequency is not the only
factor, it will be used in this paper as a relative indication of verb choice.
One of the characteristics of frequent verbs is that they tend to be used in a
variety of different ways, i.e., combine with a variety of argument structures.
F or example, get can combine in structures that range from expressing
'acquire something', 'move', 'become', and 'undergo'. Given the various
combinations for get, and other high frequent verbs, it has been claimed that
these combinations have to be learned. Hence, "light verbs" such as get tend
to express numerous senses. The argument structure is therefore difficult to
learn because the input must contain numerous examples in order to experi-
ence and be able to produce the structures correctly.
Input is vital in Ll acquisition and considered equally important for L2
acquisition. However, in a classroom learning situation, input is substan-
tially less than in a natural language learning setting. Therefore, it would
seem natural to assume that L2 learners struggle with sorting out the various
senses of highly frequent words, which will result in the production of a
variety of non-accurate structures. Nevertheless, the data indicate that, in
the case of get, structures are primarily used correctly. There are exceptions,
however, with respect to what I have called learner constructions, and these
will be the focus of this paper.

Authenticated
Construal, Convention, and Constrnctions in L2 Speech 53
Learner constructions are defined as:
A learner construction is a construction, i.e., a meaning and syntax corre-

spondence, but which is used in a slightly unconventional manner. Although
usage does not result in a communication breakdown between participants,
it deviates in some way.
In other words, somehow the utterance makes sense, but it sounds odd in a
way that may be attributed to the conceptualization process. Learner con-
structions provide insight into the conceptualization process, which reflects
the state of the learner's language system.
In cases where learners have chosen to use get, they use it in comprehen-
sible ways, and prefer certain constructions to others, but most importantly
they produce slightly off utterances. However, the nature of the task and the
learners' approach to the task may be responsible for the different sense
preferences as well as the choice of verb. In other words, the fact that the
non-native speakers (NNS) use get less frequently than native speakers
(NS) may be related to either avoidance or simply the lack of accessibility
to other linguistic forms. Therefore, what I can say about constructions and
language acquisition is restricted to an analysis of the learner constructions
and how these relate to the actual occurrences of get constructions, the dis-
tribution of these get forms in terms of what is more frequent and which uses
are less frequent, as well as a comparison to the frequency of forms chosen
by the native speakers. The construction senses, e.g., obtain, move, undergo,
that occur in the learner constructions correspond to the most frequent
senses that were used in conventional ways by the learners. The availability
of a construction is an important factor in L2 production because it allows
speakers to take what they have as far as they can, and often this entails that
the extension of a construction overrides the conventional usage. This may
be because conventional usage is less accessible, whereas constructions are
more accessible in that learners are not dependent on knowing or remem-
bering specific linguistic items. In cases where communication is the goal,
constructions allow L2 speakers to be understood simply because construc-
tions capture experience in a language-specific way and already exist in the
Ll system.
1.1. Abstract Rules, Formulaic Language, and Constructions
We seek patterns. We have a physical need for patterns. It is a general human

trait that we rely on patterns to create order out of a world of many choices.

Authenticated
54 Renee Waara
Patterns are essential and make our lives easier and livable. The learning
process, in general, is often a matter of seeing a pattern. L2 language learn-
ing may be similar in that in order to grasp an L2 we have to recognize,
remember, and recall the patterns of lexemes, and the combination of these
lexemes, and the inherent semantic connection. Seeing patterns in units that
consist of more than one word has spawned a series of studies on the phe-
nomenon often referred to as "formulaic language". However, describing
patterns and the role of patterns is dependent on the theoretical assumptions
regarding the nature oflanguage.
In the generative linguistics approach (cfChomsky 1965; 1981; 1986;
1995), the focus has been on the relationship between chunks of language
for the purpose of abstract rule formulation to accommodate an approach to
language in which streamlining and high-order generalizations are essential.
Crucially, context and use are viewed as independent of the language system.
Researchers have been particularly interested in the production of "gram-
matically advanced" expressions that appear to be beyond the capacity of a
language user (cfClark 1974; Wong Fillmore 1976; Huang & Hatch 1978;
Wagner-Gough 1978; Hatch, Peck & Wagner-Gough 1979; Vihman 1982;
Karnio11990; Willett 1995; Henry 1996; Myles, Hooper & Mitchelll998;
Myles, Mitchell & Hooper 1999). Many researchers have suggested that a
language learner performs some form of reanalysis of chunk language, result-
ing in the formation of productive rules. However, reanalysis is an assumption
held within the framework of universal grammar, and the problem becomes
making a concrete connection between the use of larger chunks of language
and universal grammar. This has been referred to as the linking problem (cf.
Tomasello 2000: 232ft). If a theory of language stipulates a core grammar
(universal grammar) that is accessible to the language user's a priori lan-
guage input, then the theory must also account for the link between universal
grammar and the particular language learned by the language learner.
Tomasello (2000: 232ft) proposes that this is an inherent problem for a theory
that stipulates a universal grammar. A linking problem would apply equally
to L2 language learners. Consequently, studies on L2 acquisition within the
generative linguistics framework have a linking problem between the param-
eters to be set in the universal grammar and the particular language used by
the language learner.
In a usage-based approach to language, there is no linking problem
because rule formation can only derive from specific, concrete expressions
that are already in the language user's possession. Working within a cognitive
linguistics framework, Dabrowska (2000) has shown how language chunks
are utilized in the formation of schema extraction through a process moving

Authenticated
Construal. Convention. and Constructions in L2 Speech 55
from lower level schemas to more abstract schemas. The acquisitional process
of emerging schemas at various levels of abstractness is reconcilable with
patterns found in learner language as well as the notion of categorization. If
verbs are learned on a verb-by-verb basis, then repetition of these patterns
will be found in language production in the form of chunks of language or
formulaic language. Language acquisition is then perceived as a process of
categorization of patterns rather than a reanalysis of chunks of language for
abstract rule formation. These chunks are related to constructions in that they
are part of the inventory of language knowledge, and from them constructions
emerge.
1.2. Speaker Construal
If a usage-based approach is assumed, then the role of the speaker is altered

from being subject to distant rules governing language to taking an active
role in putting language together based on the events surrounding the usage
event. The speaker's perspective, language use, and context are primary
considerations in Langacker's notion of usage event. A usage event is an
actual instance of language use; the language user, not a linguistic system,
is responsible for the linguistic expression and is in command of not only
linguistic resources but such factors as memory, planning, problem-solving
ability, general knowledge, as well as "a full apprehension of the physical,
social, cultural, and linguistic context" (Langacker 2000: 9). In short, the
focus is on language users and everything available to them.
Accordingly, the notion of speaker construal must play a central role in
an account of L2 learner speech production. Only by looking at patterns of
language in terms of usage and L2 speaker construal can we say something
about a mutually beneficial relationship between a usage-based model and
second language acquisition theory. It seems intuitively correct to assume
that speakers choose language for a particular purpose based on numerous
factors surrounding the act of speaking, and that they use what they have to
take them as far as possible and that this involves both Lt knowledge and
L2 knowledge. An L2 speaker's choice of expression is not always conven-
tional in the sense that the utterance directly corresponds to conventional
usage, and most often this is not the case. For the purposes of this paper,
conventional usage is confined to combinations that are socially anticipated,
culturally generated, and in some sense entrenched in the community. Con-
ventional usage entails specific or concrete expressions that have attained
some degree of unit status through use within a language community.

Authenticated
56 Renee Waara
So far the following issues have been addressed. It is a general cognitive

condition that we seek and utilize patterns, and that we attempt to account for
language in terms of patterns. A usage-based approach provides a plausible
theoretical framework for accounting for language acquisition with respect
to these patterns and the developing L2 involves transfer and blending. The
notion of speaker construal is a central aspect and must be considered as a
factor in the language acquisition process.
2. Constructions and Construction Grammar
Goldberg (1995: 1) proposes that basic sentences of English are instances of

constructions that consist of a form-meaning pair that are independent of
particular verbs, and that "constructions themselves carry meaning, inde-
pendent of the words in the sentence".
The data used in this study suggest that L2 speakers rely on constructions
to carry meaning. In order to illustrate the notion of construction as outlined
by Goldberg, the Caused Motion Construction will be used here. As with all
constructions, the Caused Motion Construction consists of a semantic part
and syntactic part.
(l) Caused Motion Construction

Meaning: X causes Y to move Z
Form: Subj V Obj Obl
Example: Pat sneezed the napkin off the table.
If we attribute caused motion to a construction, and not a particular property

of a verb, then there is no need to specify in the lexical entry that sneeze can
cause motion. Sneeze can occur in several other constructions as well
(Goldberg 1998: 204):
(2) sneeze
a. Pat sneezed.
b. Pat sneezed the napkin off the table. (caused motion construction)
c. She sneezed a terrible sneeze. (cognate object construction)
d. She sneezed her nose red. (the resultative construction)
e. She sneezed her way to the emergency room. (the way construction)
Sneeze is a "standard" intransitive verb, but we see from (2b) that it can
combine with three other syntactic components. Goldberg argues that it is

Authenticated
unnatural to assume that sneeze is responsible for the fact that someone
caused something to move somewhere. If we attribute the caused motion
meaning in (2b) to the caused motion construction, then there is no need to
specify a special sense for sneeze. Moreover, the additional senses found in
examples (2c-e) are attributed to their respective constructions. These exam-
ples can be accounted for by the way in which the verb sneeze is construed in
a particular construction. It is not the construction that has changed meaning,
but rather that a situation is construed and expressed in words. The notion
of construction as presented here is also extremely useful for verbs that are
less specific than sneeze, the so called "light verbs", e.g., put, do, make. and
get, in that the form-meaning correspondence is neatly conveyed in con-
structional meanings, as well as the variety and diversity of possible argu-
ment structures. In the next section the corpora from which the data are
extracted are described.
3. Method
The corpora and the speaking test
The data used in this study are taken from two corpora that were collected
for a speaking test, including a non-native speaker corpus and a native
speaker corpus. The non-native speakers were Norwegian 14--15 year olds,
while the native speakers were 14-15 year olds from schools in the
Newcastle area. The non-native speaker corpus contains approximately
33,000 words (tokens), and the native speaker corpus contains some 20,000
words. The non-native speaker corpus has more individual subjects than the
native speaker corpus, 58 compared with 26. A tester administered every
test in pairs, so there are 29 pairs in the non-native speaker corpus and 13
pairs in the native speaker corpus. Each test took an average of 30 minutes,
of which approximately half is tester talk, so there are approximately 435
minutes of non-native pupil speech and 200 minutes of native pupil speech.
Both the natives and non-natives pupils took the same set of tests. The tasks
are designed to elicit evidence of the various components of communicative
language ability, and presented in detail in Hasselgren (1998).
There are two different versions of the speaking test in the corpus. The
structure ofthe test is the same, with different themes. The theme of version
one is a disco and that of version two a football match. In the non-native
speaker corpus there were 17 pairs doing version one and 12 pairs doing
version two. In the native speaker corpus there were six pairs doing version

Authenticated
58 Renee Waara
one and seven pairs doing version two. The division between the two different
versions has no effect on the analysis because while the themes are different,
the test design ensures similar conditions and comparable contexts for lan-
guage production.
3.1. A Brief Description of the Four Tasks
The tasks taken from the corpora represent four different types.
- Task One: Telling a story

- Task Two: Two minutes of self-determined topic
Task Three: Giving instructions
- Task Four: Providing directions
In task one, Pupil A tells the story using the picture series. These are generally
longer stretches of speech, with only minor interruptions from the tester,
almost never from the other pupil. In task two, pupil B is asked to talk for
two minutes on something slhe did recently. This is something unique to the
pupil, no pictures. In task three, one of the pupils explains in detail to the
other pupil how to do something in the pupil's home. There is not usually
very much interaction between the pupils. In task four, one of the pupils
gives directions to the other pupil.
In task one, pupil A is asked to tell a story based on the series of pictures
she has in her test booklet. In version one the theme is a disco. A girl has a
curfew that she has broken, and when she gets home there is a man breaking
into her family's car. In version two, there are two teenagers who are on
their way to a football match, and they miss their train because they left
their bag back in a cafe. They set out to hitchhike and get a lift with the
director, winding up in the director's box. In short, there is enough action
for good storytelling. Pupil A's speech in this section is generally longer
stretches with only acknowledgements andlor encouragement from the
tester. Rarely does the other pupil intercede with a comment. The tester
tends to be most active towards the end of the storytelling, while trying to
tease out the action in each story, with statements from version two such as
where are they sitting? In the picture there is a thought bubble that says
"thief get dad". The speakers tend to put this in rather strange contexts, such
as she saw a thief get dad (the thief didn't get her dad) or she thought/thinks
thief get dad (in which the pupil is telling us what the girl is thinking),
which indicates that they are reading the text. There are more of these in the

Authenticated
Construal, Convention, and Constrnctions in L2 Speech 59
NNS than in the NS corpus, and the former seem to be motivated to use the
phrase because they know it is correct language and want to get it in the
conversation.
In task two, pupil B is asked to talk about something she did recently. In
version one, the time frame is set specifically to last weekend or a time she
went out with her friends, and in version two, pupil B is supposed to talk
about a journey or event she attended. There is often a rather lengthy negoti-
ating section before the pupil actually starts her two-minute talk. Each pupil
talks about something that is unique to her; i.e., there are no pictures, and
the listeners are not familiar with what the other pupil is talking about. The
difference between this task and the other tasks is that there are no pictures,
or predetermined guidelines for the pupils to follow.
In task three, one pupil is to instruct the other pupil on how to do a chore
in her/his own home. In version one, the chores include cook potatoes, wash
clothes, make coffee or tea, and record on the video-player. In version two,
one of the pupils has to instruct the other about how to take care of the other
pupil's puppy for the afternoon. There is a detailed picture showing the
kitchen and backyard that is to be used in instruction giving.
Giving directions is in task four. In version one, the task is to retrieve a
CD from the other pupil's house. One pupil directs the other inside herlhis
house to where the CD is located. In version two, one pupil is supposed to
be living in Slode, and is expecting a visit from the other pupil. The pupil in
Slode has to provide explicit directions on how to get to Slode via Newcastle
and Coventry. Detailed timetables and platform numbers are provided.
Language produced in this section occurs in longer stretches, and although
lexical searches are common, they do not tend to break down the communi-
cation.
The speech that is produced by these speakers is not natural conversation,
with the possible exception of the two-minute talk, but even that is guided
through topic choice. Instead of free conversation, we have pupils who are
talking within specific contexts under the duress of the speaking test frame-
work.
3.2. The Data
A concordance program, Wordsmith, is used to identify the most frequent

verbs. A preliminary word frequency count of verbs reveals that the native
speaker corpus (NS) and the non-native speaker corpus (NNS) have, sur-
prisingly, very similar lists:

Authenticated
60 Renee Waara
Table 1. Frequency counts of verbs in corpora (most frequent verbs)
Raw numbers % of corpus Raw numbers % of corpus

NNS NNS NS NS
BE 1656 3,37 804 3,49

GO 426 0,87 246 1,07
HAVE 335 0,68 187 0,81
DO 265 0,54 123 0,53
THINK 207 0,42 66 0,29
CAN 175 0,36 108 0,47
COME 150 0,31 63 0,27
TAKE 150 0,31 42 0,18
GET 123 0,25 254 1,10
KNOW 112 0,23 48 0,21
SAY 101 0,21 80 0,35
SEE 99 0,20 39 0,17
PUT 56 0,11 67 0,29
Table 2. Ten most frequent verbs in sub corpora
Task 1 Task 2 Task 3 Task 4
Be * *
Go * *
Have
Take * * * *
Do * *
Come *
Get * * * *
Think
See
Put *
Significant difference p < .05 indicated with *

Authenticated
Construal, Convention, and Constmctions in L2 Speech 61
Among the most frequent verbs found in both the corpora, light verbs such
as go, do, get, put are highly represented, thus coinciding with verbs that are
learned and produced in L1 at high frequency levels (cf Clark 1978).
Moreover, light verbs appear as the most frequent verbs in the two corpora.
Within the four tasks selected for this study, the following verbs were the
most frequent. The ten most frequent verbs in order of frequency in the
NNS corpus are listed in table 2.
The verbs that are significantly different are marked with *. Get was chosen
for this paper because it is one of the verbs that differs significantly with
respect to frequency of occurrence in each task. Every occurrence of get
was extracted from the corpora, i.e., get, got, getting, gotten, and to get. For
each sub corpora each occurrence of get was classified by its form-meaning
construction. Table 3 lists the total number of occurrences in each task:
Table 3. Number of get occurrences in corpora
Task 1 Task 2 Task 3 Task 4 total
NNSs 34 16 18 11 79
NSs 31 22 40 48 141
The NS have significantly more occurrences than the NNS.
3.3. Get and Argument Structures
Following have and be, get is the most frequent verb form in conversation.
(cf. Biber et al. 1999; Krug 2001) Get is a prime example of a verb with no
clear argument structure because it appears in so many different construc-
tions. This is not a property solely attributed to get; according to Biber et al.
(1999) the next eight most frequent verbs are just as difficult to assign a
clear argument structure, i.e., say, go, know, think, see, come, want and mean.
Hence the more frequent a verb is, the more combinations it tends to occur
in. Thompson & Hopper (2001: 50) take the lack of argument structure a step
further and claim that "The data show very clearly that the most frequent
verbs in the language have no 'argument structures', but occur in a wide
range of lexicalized expressions that must be learned." Thus it would seem

Authenticated
62 Renee Waara
that get combines with many different constructions and therefore it is diffi-
cult to know what kinds of argument structures are permissible. However,
the data in this study at hand indicate a higher degree of regularity.
Get occurs in a variety of combinations from idiomatic instances to sys-
tematic patterning. Idiomatic in the sense that the sum of the parts does not
provide meaning, but rather refers to something completely different; for
example, get it 'understand', get it over with 'suffer through', get into it
'become interested', get the phone 'answer', get down 'party'. Idiomatic
occurrences were extremely rare in the corpora. There were only two
instances in the data: one in NNS in the instruction task, I didn ~ get it,
meaning lack of understanding of what was said; and one in NS, in the two
minute free speech task, as we got into it, meaning 'as we became more
involved' and not into a car, for example. The systematic patterning of get,
in terms of constructions, is described in Table 4.
Table 4 is divided into three columns: construction meaning, basic sense,
and construction form. The form-meaning correspondence is represented in
each row, such that "X OBTAIN Y" will always correspond to a transitive
construction with the basic sense OBTAIN. Table 4 gives the range of
senses of get. However, other verbs are also possible in these patterns, to
some degree. For example, bought can express OBTAIN, as in I bought it at
the store, instead of I got it at the store. In other words, these constructions
bear the meaning of "obtain", "cause obtain", "cause someone to receive
something" so even though other verbs combine with constructions render-
ing a slightly different reading, bought versus got, they do not occur in such
a diversity of constructions. Whereas a verb such as buy can combine in
OBTAIN and CAUSE OBTAIN, get nms the gambit from OBTAIN to DO
as illustrated in Table 4. So in a manner of speaking, get is a "watered-
down-pretty-much-do-anything" kind of verb, which appears to get meaning
through specific constructions, i.e., the combination of the meaning in the
left-hand column of the table and the form in the right-hand side of the table.
For each construction in Table 4 there is a non-causative reading and a
causative reading, with the exception of the basic sense "obligation". Each
syntactic form has a corresponding semantic sense; if there is a change in
syntactic form, then there is a change in meaning. Each occurrence of get in
the corpora has been categorized according to the constructions described in
Table 4.
Given that get can integrate with a variety of constructions, we might
expect that NNS speakers would use it incorrectly. If there are no discernible,
reliable argument structure patterns, then each individual use must be learned
and can only be achieved through contact with many examples. It does,

Authenticated
Construal, Convention, and COl1stnlctions in L2 Speech 63
Table 4. Constructions with get
Construction meaning Basic sense Construction form
X obtains/possess Y OBTAINIPOSSESS Transitive / Possessive

Ex. I got it at the store. Subj V Obj
We've got a taxi.
X causes Y to receive Z CAUSE OBTAIN Double object

Ex. He got me some stamps. Subj V Obj Obj2
X moves Y MOVE Intransitive /

Ex. When will we get there Simple motion
We got home late. Get out Subj VObl
of my house. Get on the bus.
X causes Y to move CAUSE MOVE Caused motion

Ex. She got me into the show. Subj V Obj Obl
We got the bus home.
X becomes Y BECOME Inchoative

Ex. I get tired/sad Subj V Adj
X causes Y to become Z CAUSE BECOME Resultative

Ex. I got him sad/drunk Subj V Obj Adj
X undergoes Y UNDERGO Passive

Ex. I got trapped/caught/lost Subj V(passive)
X causes Y to undergo Z CAUSE UNDERGO Causative

Ex. She got me trapped Subj V Obj V(passive)
X acts on Y DO Subj V to-verbal inf.

Ex. I got to kiss her.
She gets to go home.
X causes Y to act on Z CAUSE DO Subj V Obj to-verbal

Ex. He got me to kiss her. inf. Obj2
X mustY OBLIGATION indicated by Subj HAVE V

has or have + got. to-verbal inf.
Ex. He's got to go to the
hospital

Authenticated
64 Renee Waara
however, appear that, despite the surfeit of examples, the L2 learners in this
corpus manage quite well. Their success is reflected in the fact that commu-
nication rarely breaks down, and that restarts, which indicate hesitation and
uncertainty, are not overly common. However, the existence of a set of
learner constructions suggests that get (and perhaps light verbs in general)
presents some challenges for learners in terms of knowing precisely which
arguments are permissible or acceptable in particular situations. This uncer-
tainty is realized by the emergence of conceptual blends, elements of transfer,
and overgeneralizations.
3.4. The Norwegian Get
In the discussion of the learner constructions, some cases appear to be influ-

enced by the Norwegian use of get. This is anticipated and warrants some
comments on the Norwegian a fa, which is a highly frequent verb in
Norwegian. As a main verb afa 'to get' has the primary sense of RECEIVE,
as in hanfar hjelp 'he gets help'. The Norwegianfa also occurs in several
idiomatic expressions, such as fa i stand meaning' fix'. However, it is the
auxiliary uses that are perhaps the most common, such as fikk du snakket
med henne? 'got you talk with her' and kan jeg fa prate med deg litt? 'can I
get talk with you a little'. Many of the senses of the English get are not
directly transferred from the Norwegian fa because they simply do not
exist, e.g., MOVE. The data, however, indicate that the NNS speakers man-
age to successfully produce senses of get, e.g., MOVE, that do not occur in
the Norwegian cluster of get senses. This may suggest that constructions
acquired in L2 play an equally important role as L1 transfer. Constructions
are relied on for successful communication in settings where knowledge of
specific lexical items is not available.
4. Results
The frequency of occurrences differs significantly between the NNS and the
NS speakers. In Table 5, all of the occurrences of get are categorized in
terms of basic sense or other combinations that emerged from the data.!
The emergent combinations are four non-sense categories that are
included to highlight other aspects of the uses of get. For example, when get
occurs in the infinitive form as in trying to get the car, even though the basic
sense of OBTAIN is obvious, there is another verbal element that flavors its

Authenticated
use and represents a form difference. This particular combination also

appears to be favored by the NNS more than by the NS. Restarts are also
considered relevant as their own category to show that restarts occur in both
corpora and are therefore not predominantly an NNS feature. Although the
number of occurrences is relatively small, they serve to indicate tendencies
and distributional insights within specific types of tasks between the two
specific groups of speakers. Even though get is purported to be notorious
for its wide idiomatic flare, these types of cases do not abound in the data,
and are represented by 1.3 percent among the NNS and less than one percent
among the NS. The C stands for CAUSE, i.e., something causes something
to do something.
The causative readings, e.g.,m CAUSE OBTAIN, CAUSE MOVE, occur
far less often than non-causative readings, e.g., OBTAIN, MOVE, BECOME,
such that 3.9 percent of the NNS examples are causative, compared to 16.9
percent in the NS examples. More widely used are non-causative construc-
Table 5. Percentage distribution of get in NNS and NS corpora
Distribution Task 1 Task 2 Task 3 Task 4 SUM SUM

by% NNS NS NNS NS NNS NS NNS NS NNS NS
OBTAIN 29.4 19.4 12.5 22.7 16.7 65.0 9.1 18.8 20.3 32.6
C-OBTAIN 2.9 0 0 4.5 0 2.5 0 0 1.3 1.4
MOVE 14.7 45.2 37.5 18.2 0 7.5 81.8 41.7 25.3 29.1
C-MOVE 0 6.5 0 13.6 5.6 0 0 22.9 1.3 11.3
BECOME 5.9 0 6.3 13.6 5.6 0 0 0 5.1 2.1
C-BECOME 2.9 0 0 0 0 7.5 0 2.1 1.3 2.8
UNDERGO 0 3.2 0 4.5 0 0 0 2.1 0 2.1
C-UNDERGO 0 0 0 0 0 2.5 0 2.1 0 1.4
DO 0 0 0 0 5.6 0 0 0 1.3 0
C-DO 0 0 0 0 0 0 0 0 0 0
OBLIGATION 0 9.7 0 0 5.6 10 0 4.2 1.3 6.4
INF+OBJ/OBL 26.5 16.1 12.5 13.6 0 2.5 9.1 2.1 15.2 7.1
RESTART 5.9 0 6.3 4.5 5.6 2.5 0 4.2 5.1 2.8
IDIOMATIC 0 0 0 4.5 5.6 0 0 0 1.3 0.7
LEARNER 11.8 0 25 0 50.0 0 0 0 21.5 0
% of total sum 43 22 20.3 15.6 22.8 28.4 13.9 34

Authenticated
66 Renee Waara
tions, which represent 52 percent in the NNS and 70.2 percent in the NS.
The NNS and NS are both relatively low for the causative and relatively high
for the non-causative, showing similar tendencies. In order to illustrate how
the uses of get are realized in the corpora, below is an example of each con-
struction:
OBTAINIPOSSESS: you get the dogs dish/ she 50 got the bag
- C- OBTAIN: you'll have to get it a drink
- MOVE: they get to the railway station
C-MOVE: we'll be getting the next train
BECOME: it 50 got dirty
C-BECOME: they may not get them right
UNDERGO: you'll get killed
C-UNDERGO: its got "dog" written on it
DO: so he'll get to know you
OBLIGATION: you've got to go upstairs
INF + OBJ/OBL: she went to get her dad
RESTART: when we got when here we come to the
- IDIOMATIC: I didn't get that
LEARNER: she get to know that she must be home
If the focus of study is solely on the combination of words with a strict top-
down abstract set of rules or principles of language, then learner constructions
are quite difficult to account for because syntactically they are generally
correct. However, from a constructional perspective we can assume that
underlying knowledge that verbs have different senses influenced the
speaker's choice. Activation of possibilities must play an important role, a
certain degree of automaticity with the pressure of the online processing.
Learner constructions sound slightly odd because they rely more on the
application of a constructional sense than a conventional expression. In a
general way, constructions dominate at the expense of conventionality.
Specifically, with the notions of construction as the point of departure,
learners blend conventional uses within the target language, transfer aspects
ofLI constructions into the L2 system, and overgeneralize existing L2 con-
structions in their developing L2 system.
The overall results will be discussed first. Then, each sub corpora which
correspond to the four individual tasks described in section 3.1 will be dis-
cussed in terms of group tendencies and preferences for certain senses of
get. Finally, learner constructions will be discussed in terms of how they

Authenticated
bring construction meaning to bear on the intended meaning and manage to

communicate successfully.
The overall results in Table 5 suggest that general similarities between
the two groups exist. There are strong similarities with respect to the senses
OBTAIN and MOVE. The distribution between the senses is also fairly
similar. Even though the overall numbers reveal a similar distribution, a
break down of individual tasks suggests that NNS speakers seldom resort to
using the same sense as the NS speakers within the same task. NNS and NS
use different constructions to talk about the same even. This difference
occurs somehow in the conceptualization process and may be attributed to
accessibility or simply knowledge of which constructions are more com-
monly used. Moreover, this discrepancy brings to mind the puzzles of
native-like fluency and native-like selection proposed by Pawley & Syder
(1983). They suggest that even though there are vast amounts of ways of
expressing a particular event, some ways are perceived as more native-like
than others. Perhaps the choice of construction type reveals elements of
native-like fluency and native-like selection.
4.1. Distribution of Constructions: NNS Compared with NS
Task One: Telling a story
This task involves telling a story based on a series of pictures and was dis-
cussed in more detail in section 3.1. In terms of construction distribution,
when they are telling a story the NS clearly prefer the MOVE sense, as in
they get to the disco, and this represents 45 percent of the total number of
occurrences in task one. In contrast, the NNS do not utilize the MOVE
sense as frequently as the NS, and prefer the categories OBTAIN and the
INF+XP. In addition, the NNS produce the most instances of get in task
one, i.e., 43 percent of all the instances are found in task one. The distribu-
tion suggests a different approach to the task, such that instead of telling a
story the NNS speakers describe the pictures, such as she s got the bag,
whereas the NS speakers utilize the MOVE sense, such as they get to the
railway station, in accordance to what we would anticipate in telling a story,
i.e., the movement of a narrative.

Authenticated
68 Renee Waara
Task Two: Two-minute talk
In task two, pupils are asked to talk about a recent event or journey. The dif-
ference between this task and the other tasks is that there are no pictures or
predetermined guidelines for the pupils to follow. The pupils introduce a
variety of experiences from babysitting, football games, and shopping. Thus
the topics have no commonality apart from the fact that they are not imposed
as a specific task. We might expect this choice to allow the pupils freedom
to avoid language they are uncertain about. This is also the task in which
both the NS and the NNS produce the least number of occurrences of get. In
contrast to task one, the usage differs. Whereas the NNS strongly favor the
MOVE sense, e.g., I got home and We got in, the NS favor OBTAIN, e.g.,
I've got money. The use of the MOVE sense is an indication that this con-
struction is accessible to the NNS speakers when they see reason to express
movement.
Task Three: Giving instructions
In task three, one pupil instructs the other pupil on how to do something.
Thus we would expect to find a sequence of actions, most likely connected
with and then. This is what we find. The NS produce most OBTAIN con-
structions, e.g., you get the dog s dish, and it s got a plug. There is a skewing
effect towards possessive construction caused by the fact that the native
speakers were told to pretend they did not know certain words. The project
leader responsible for the project was uying to simulate L2 conditions for
the search for lexical items. Consequently there are several instances of
descriptive strings, e.g., its got holes/ dials/ circles, which complicates the
data interpretation. However, there is another rather interesting result from
this task, in that there appears to be a structural difference that may be related
to a cultural difference. The UK pupils have a slightly different approach to
the matter of telling somebody what to do. Following the and then signal,
the UK pupils prefer get + object, as in and then you get the tin/the towel/
the dogs dish/dirty clothes, whereas the Ll Norwegian speakers prefer
combinations such as and then you got to put/search/looklfill. The UK
pupils focus on things, whereas the Norwegian pupils focus on actions in a
way that breaks with a convention of politeness when instructing someone
to do something.

Authenticated
Task Four: Providing directions
In task four, one pupil gives directions to a house or place to another pupil.
Of the four tasks discussed here, there are the most occurrences of get in the
NS and the least in the NNS, 34 and 13.9 percent respectively. There is a
strong preference or tendency for MOVE in the NNS corpus, e.g., get off.
There is a clear preference in the NS corpus towards the MOVE and the
CAUSE MOVE constructions, e.g., we 'U be getting the next train to
Coventry. The NS speakers take the MOVE construction a step beyond with
the causative reading, and specify both movement and destination. 2
4.2. Learner Constructions
In this section the emergence and usage of the learner construction will be
discussed in terms of how they reflect elements of transfer, blending, and
overgeneralizations of the developing L2 system. The intentions of the
speakers in these corpora are identifiable insofar as they are related to specific
tasks, which provide a starting point to distinguish how the speakers construe
a given expression. In other words, viewing the utterance in light of the sur-
rounding context allows us a clearer picture of the intentions of the speaker.
The learner constructions are divided by tasks, i.e., the examples in (a-d)
are from task one, the examples in (e,f) are from task two, and the examples
in (g-j) are from task three. No learner constructions occurred in task four.
Learner constructions in task one:
a. Can I get a dance with you?

b. She sees a man get to the car.
c. The thief is getting shook.
d. She get to know that she must be home.
In (a), the form is subject-verb-object, dance is an object and suits the transi-
tive form, but a dance requires two people, albeit under traditional conditions.
Can I get has the conventional use of offering to get something for someone,
as in can 1 get you a coffee? The conventional expression used to obtain a
dance with someone is May I have this dance? The example in (a) gives us
a blend of two conventional usages instantiated by can I OBTAIN a dance
with you?
In (b), the MOVE construction captures the sense that the man is moving
towards and arriving at the car. The speaker expresses or describes what is

Authenticated
70 Renee Waara
in a picture, a masked man with his hand on the car handle. Based on the
series of pictures we can tell that he has just moved over to the car. The
speaker has used the intransitive motion construction to describe this event.
The man moves along some path to the goal, which is the car. When get is
used in the intransitive motion construction it has restricted conventional
usage. These uses include expressing future events, e.g., when we get to the
cabin, or past-completed action, as in we got home late. The example in (b),
a "man get to the car", is used in an unconventional manner to express "X
moves Y" because it overgeneralizes the use of get within the speaker's sys-
tem of L2 language.
In (c), the thief is undergoing a process. In the picture, the thief is literally
being shaken by a man, so this construction is the passive construction, "X
undergoes Y". It would seem that the meaning of getting is being extended
to the meaning of being, and the construction contributes most of the mean-
ing. Moreover, in the two-minute task there is a NS example of, we were
gettingfollowed. In other words, the form is acceptable and it is not only the
NNS who can stretch a construction, i.e., constructions have the ability to
be stretched.
In (d), the parents are telling the girl that she has to be home by midnight.
The meaning is retrievable from the context; she OBTAINS, in a loose sense,
a message from her parents, but the form is confusing because get + to-
infinitive verb form denotes some action in "X acts on Y", and yet the
intended meaning of the expression is passive, in the sense that the girl is
told something. However, there is a transfer effect from the Norwegian, afa
vile noe 'to get to know something', which has the basic sense 'receive' or
'obtain'. But we also 'get to know someone' which conventionally refers to
'be acquainted with someone'. This learner construction is a transfer from
L 1 combined with an over generalization with the existing L2 system.
There are three very different learner constructions found in the NNS
task two.
e. One of my friend get trouble with another.

f. You get better known to somebody.
The example in (t) is a transitive construction, i.e., with the sense of

OBTAIN, the oddity in this case stems from the fact that trouble is usually
something that happens, not something that we obtain, acquire, or receive.
Instead, normally, we get into trouble, which may be a lexicalized phrase.
However, based on the context, it appears that the speaker conceptualizes
the construction "X obtains Y" in a way that is very close to the Norwegian

Authenticated
Construal, Convention, and Constmctions in L2 Speech 71
rendition or use ofla 'get', i.e.,flkk problem med en annen 'got problem
with one other'. This is a case of overgeneralization of the basic sense of
OBTAIN and L1 transfer.
There are not many direct transfers from Norwegian; however, the
example in (t) is a clear exception, i.e., du blir bedre kjent med noen 'you
become better known with someone'. The construction is very descriptive
in that "X becomes Y". Get relates the subject to the state of 'getting to
know somebody'. Although this is an unconventional use, communication
is successful. The Norwegian use of get in (e) and in (t) instantiates Ll
transfer, which may be related to the strong relational sense between
become and the Norwegian verb blir.
In task three, the following learner constructions are found:
g. You got to putlsearch/looklfill. ..

h. The aroma gets sooner away.
1. You have to get upstairs.
J. He'll get to you because he's hungry.
The examples in (g) correspond to the construction meaning 'X acts on Y'
and is, on its own, an acceptable expression. However, in the course of giv-
ing instructions, it is generally not conventional to imply that someone must
do something, because of the edge of impoliteness it implies and because it
diverges from a general convention of politeness in the given context.
In (h), the aroma is attributed with a property of motion; the aroma moves
away to someplace else or is "dispersed". The pupil has described how to
boil the water and put the coffee granules in the water and let it stand. The
pupil says that for filter coffee the powder is ground more finely than for
boiled coffee, and because "the aroma gets sooner away", he prefers the
boiled coffee. In other words, the flavor of boiled coffee is better because it
does not lose the aroma by being ground as finely as the filter coffee. The
verb disappear is suggested by the tester in this exchange, and even used by
the pupil earlier in the turn, but for some reason it is later rejected and a
conceptualization process involving "X moves Y" is preferred by the pupil.
The pupil feels more expressive or more accurate by describing the aroma
in a MOVE construction, based on a basic human experience, "move some-
thing". This example is an overgeneralization within the existing L2 system.
The example in (i) is conventional in the sense that you can say, Get
upstairs!, with the reading 'go or move upstairs'. But in this context it is
slightly odd because it is a part of a series of directions about moving from
one point to another point, i.e. first you go there and then you go there. In

Authenticated
72 Renee Waara
other words it is not grammatically wrong, but it is pragmatically odd in the

context it is used.
In 0), he, referring to a dog, is the point of reference and is perceived as
moving towards you. The dog is construed as the object that moves. The basic
concept of motion is intact, but just simply reversed with the choice of the
verb. Normally, the dog comes to you. The fonn, subject-verb-oblique, coin-
cides with the constructional sense of MOVE, which is perhaps the reason
that communication is successful. The speaker overgeneralizes the intransi-
tive motion construction that already exists in the speaker's L2 language
system.
In the majority of the learner constructions, the form-meaning corre-
spondence is retrievable from the construction through a conceptual blend,
transfer effects or by an overgeneralization. It is the combination of the way
the speaker construes the event and the use of a construction that allows for
successful communication. Moreover, the constructions that are represented
in the learner constructions correspond closely to the remaining examples
that are most frequently used in the corpora (cf Table 6), such that the
OBTAIN and MOVE senses are most frequent both in the corpora as well
as in the set of learner constructions.
5. Conclusion
What is striking is that it looks like the learners have mastered the construc-
tion level better than specific lexical items, such that the construction is
supplied with semantic components that are either 100% compatible or
slightly unconventional, hence giving us a learner construction. The problem
or point of conflict is the semantic compatibility of the components in the
construction. Although many of the get constructions produced by the learn-
ers are acceptable and in many cases quite similar to the native speaker
choices, there are some differences found in the learner constructions, dif-
ferences that suggest that constructional meaning guides production at the
expense of complete conventional usage. Perhaps these speakers do not
have the lexicalized expressions in their language inventory. Evidence for
this is found in the overgeneralization of constructions and descriptive
power of the construction.
In the majority of the learner constructions presented here, the construc-
tion itself is a composite whole. It is the interaction between construction
and its components in a specific context that produces the unconventional
uses. That is, the manner in which the speaker construes the words, i.e.,

Authenticated
components, results in a slightly unconventional use. However, and most

importantly, there is no breakdown of communication. The L2 speakers
blend, transfer, and overgeneralize constructions, thus using what they have
to facilitate communication, at the expense of conventional usage. In the
case of light verbs, such as get, the semantic contribution is highly, if not
solely, dependent on the construction and can be an asset for the L2 speaker,
and they utilize this somewhat neutral aspect in the learner constructions.
Notes
1. Cases of 'thief get dad' were not included because it is assumed to be influenced
by the language bubble in the picture that the subjects used to tell the story.
2. It is worth noting that both English and Norwegian are satellite-framed languages
in Talmy's (1991) typological framework of motion events.

Authenticated
74 Renee Waara
References
Achard, Michel
1997 Cognitive Grammar and SLA Investigation. Journal of Intensive
English Studies 1112: 157-176.
Biber, Douglas, Stig Johansson, Geoffrey Leech, Edward Finegan & Susan Conrad
1999 Longman Grammar of Spoken and Written English. London:
Longman.
Chomsky, Noam
1965 Aspects of the theory ofsyntax. Cambridge, MA: MIT Press.
1981 Lectures on government and binding. Dordrecht: Foris.
1986 Knowledge oflanguage: lts nature, origin, and use. New York: Praeger
Publishers.
1995 The Minimalist Program. Cambridge,MA.: MIT Press.
Clark, Eve V.
1978 Discovering what words can doPapers from the Parasession on the
Lexicon, Chicago Linguistic Society, 34-57.
Clark, Ruth
1974 Performing without competence. Journal of Child Language 111: 1-10.
Dabrowska, Ewa
2000 From formula to schema: The acquisition of English questions.
Cognitive Linguistics 11/112: 83-102.
Fauconnier, Gilles
1997 Mappings in thought and language. Cambridge: Cambridge University
Press.
Goldberg, A. E.
1995 Constructions: A Construction Grammar Approach to Argument
Structure. Chicago: University of Chicago Press.
Goldberg, Adele.
1998 Patterns of Experience in Patterns of Language. In M. Tomasello
(ed). The New Psychology of Language. New Jersey: Lawrence
Erlbaum Associates, 203-219.
Hasselgren, Angela
1998 Smallwords and Valid Testing. Ph.d, University ofBergen.
Hatch, Evelyn, Sabrina Peck & Judy Wagner-Gough
1979 A Look at Process in Child Second-Language Acquisition. In E. Ochs
& B.B. Schieffelin (eds). Developmental Pragmatics. New York:
Academic Press, 269-278.
Henry, Alex.
1996 Natural Chunks of Language: Teaching Speech Through Speech.
Englishfor Specific Purposes 1511: 295-309.
Huang, H & Evelyn Hatch
1978 A Chinese Child's Acquisition of English. In E. Hatch (ed). Second
Language Acquisition, A Book of Readings. Rowely, MA: Newbury
House, 118-131.

Authenticated
Karniol, Rachel
1990 Second-language acquisition via immersion in daycare. Journal of
Child Language 17: 147-170.
Krug, Manfred G.
2001 Frequency, iconicity, categOl;zation: Evidence from emerging modals.
In J. Bybee & PJ. Hopper (eds). Frequency and the emergence oflin-
guistic stntcture. Philadelphia: John Benjamins, 309-336.
2000 A Dynamic Usage-Based Model. In M. Barlow & S. Kemmer (eds).
Usage Based Models of Language. Stanford, California: Center for
the Study of Language and Information (CSLI).
Myles, F., 1 Hooper & R. Mitchell
1998 Rote or rule? Exploring the role of formulaic language in classroom
foreign language learning. Language Learning 48/3: 323-364.
Myles, Florence, Rosamond Mitchell & Janet Hooper
1999 Intenogative Chunks in French L2: A Basis for Creative Construction?
Studies in Second Language Acquisition 21/1: 49-80.
Pawley, Andrew & Frances Hodgetts Syder
1983 Two puzzles for linguistic theory: nativelike selection and nativelike
fluency. In le. Richards & R.W. Schmidt (eds). Language and Com-
munication. London: Longman, 191-223.
Talmy, Leonard
1991 Path to realization: A typology of event conflation. In Teh Seventh
Annual Meeting of the Berkeley Linguistics Socie~v: 480-519.
Berkeley: Berkeley Linguistic Society.
Thompson, Sandra & Paul 1 Hopper
2001 Transitivity, Clause, and Argument Structure. In J. Bybee & P.J.
Hopper (eds). Frequency and the emergence of linguistic stntcture.
Amsterdam: John Benjamins Publishing Company, 27-60.
Tomasello, Michael
2000 Do young children have adult syntactic competence? Cognition:
internationaljournal of cognitive psychology 74/3: 209-253.
Vihman, Marilyn M.
1982 Formulas in First and Second Language Acquisition. In L. Menn &
L. Obler (eds). Exceptional Language and Linguistics. New York:
New York Academic Press, 261-284.
Wagner-Gough, Judy
1978 Excerpts fonn comparative studies in second language learning. In
E.M. Hatch (ed). Second Language Acquisition: a book o{readings.
Rowely,MA: Newbury House, 155-174.
Willett, Jeni
1995 Becoming First Graders in an L2: An Ethnographic Study of L2
Socialization. TESOL Quarterly 29/3: 473-503.
Wong Fillmore, Lily
1976 The Second Time Around: Cognitive and Social Strategies in Second
Language Acquisition. Ph.D., Stanford University.

Authenticated
Authenticated
Input Versus Transfer? -
The Role of Frequency and Similarity in the
Acquisition of L2 Prepositions
Wander Lowie and Marjolijn Verspoor
1. Introduction
In cognitive linguistic theory, a usage-based approach to language, the notion

of "entrenchment" is well known. "Entrenchment pertains to how frequently
a structure has been invoked and thus to the thoroughness of its mastery and
the ease of its subsequent activation" (Langacker 1991: 45). Entrenchment
is interrelated with input. Tomasello (2000: 70) points out that an important
aspect of first language (L 1) learning is some form of imitative learning and
that ,,[i]t is also important that children seem to have special difficulties in
going beyond what they have heard when they have heard it multiple times,
that is, when it is entrenched." Also, in second language (L2) acquisition, the
role of entrenchment (operationalised as input or frequency of occurrence)
has been accepted by many as one of the most decisive factors in acquiring
a second language (cf. Ellis 1994: Chapter 7). Of course, in L2 acquisition
the first language also plays a role, and the more related the L1 and L2 are,
the easier the learner finds it to learn the L2 (cf. Ellis 1994: Chapter 8). For
example, Dutch students learning English have the advantage of many cog-
nates such as Dutch breken and English break.] In this paper, we will examine
the role of these two variables, similarity between Ll and L2 versus fre-
quency of L2 input ("entrenchment"), in the acquisition of L2 over time. To
do so, we will make use of an interactive activation model of the mental
lexicon - a model very much in line with the cognitive notion and function
of entrenchment - that predicts that a frequently occurring word will have a
higher level of activation than a word that is scarcely used. For students
beginning to learn a second language, only the Ll is activated to a high
degree, and therefore similarity would play a major role, but the more a stu-
dent is exposed to the L2, the more the frequency of occurrence in L2 will
play a role.

Authenticated
78 Wander Lowie and Marjolijn Verspoor
2. Similarity and Frequency in the Bilingual Mental Lexicon
Recent (psycholinguistic) models of the mental lexicon tend to agree on the

general contents of an item in the lexicon. Each item will refer to at least
three separate units of information, semantics, syntax, and
phonology/orthography, which are divided between lemmas and lexemes.
Following Levelt (1989), the lemmas are referred to as abstract units com-
prising the syntactic and semantic information, whereas the lexemes refer to
the orthographic and phonological information associated with a lexical
item, as illustrated in Figure 1.
Figure 1. The basic components ofa lexical item.
In interactive activation models of the mental lexicon (cf. Lowie 1998,

2000; Schreuder & Baayen 1995; Dijkstra, van Heuven & Grainger 1998) the
lexical item is represented in a similar manner. Here each lexical representa-
tion comprises a lemma node, which is the central node linking the semantic-
pragmatic information, the syntactic properties, and the orthographic-
phonological information (the lexeme). The model takes a compositional
view on the relation between the semantic contents of the lexical item and
the conceptual representations associated with it, in which the latter must be
seen as the different aspects of semantic content of a word. Through a
mechanism of activation and inhibition, the level of "resting" activation is
primarily determined by the frequency of a lexical item. A frequently occur-
ring word will have a higher level of activation than a word that is scarcely
used. Figure 2 schematic ally represents a simplified representation of an
item in the mental lexicon. Although different lemmas may share conceptual
representations, no two lemmas in the mental lexicon can refer to a fully
identical set of conceptual representations. In other words, this model allows
partially overlapping word meanings, but it will not allow pure synonyms,
as these would entail fully redundant items in the lexicon.
The main purpose of this study is to determine the role of the learner's
language and how it develops over time in L2 acquisition.

Authenticated
Input Versus Transfer? 79
• 'E
Lemma Seman:ic / ~
Lexeme node
FO~. ~
·~~I if
• U§
Syntactic
properties
Figure 2. A simplified representation of a lexical item in the mental lexicon.
This image of lexical item in the mental lexicon can be adjusted to the bilin-
guallexicon by assuming an additional source of infonnation linked to the
lemma node referring to the language a lexical item is associated with. The
items associated with a particular language can be regarded as a subset of
the lexicon (cf. Lowie 1998; Woutersen 1997).
A question that is relevant to the bilingual mental lexicon is whether and
to what extent L2 learners make use of the lexical knowledge from their
first language in the acquisition and use of the second language. As the
adult L2 learner has a fully developed lexicon, it makes sense to assume that
an L2 learner will make use of the knowledge already required. A question
that is central in current debates on the bilingual mental lexicon is whether
L2 words have direct links with conceptual memory or are accessed through
Ll lemmas present in the lexicon. Recently, Nan Jiang (2000) argued that
the role of the first language differs in three stages of development. In the
first stage, L2 fonns are mapped onto existing (Ll) meanings. At this stage,
an "empty" L2 lemma is created that is linked to a Ll lemma: the L2 lexical
item only has the fonnal characteristics and full equivalence to an L 1 lexical
item is assumed and there is no direct link from the L2 lemma to conceptual
content. At the second stage, the infonnation of an existing Ll lemma is
copied onto the L2 lemma: this is the situation where the Ll lemma mediates
L2 word processing; now there is a link from the L2 lemma to both concep-
tual content and the Ll lemma. Only at the third stage will the Ll lemma no
longer be accessed and a direct link has been created between the conceptual
representation and the L2 lemma. Jiang's model sketches a picture similar
to the one proposed by, for instance, Kroll (1993), in which lexical items in
L 1 and L2 are connected:

Authenticated
Figure 3. Nan Jiang's model
Although this model conveniently explains what Jiang calls "lexical fos-
silization", it cannot account for the fact that lexical items in Ll and L2
hardly ever fully overlap in meaning. In a model that takes compositional
meanings as a starting point, this can be accounted for much more easily.
By referring to the activation metaphor, it is no longer necessary to distin-
guish between different ways of lexical organization; activation models
hypothesize that all individual lexical entries are stored identically, but that
major differences between the entries can be expected based on their fre-
quency, expressed by their relative level of activation. Ll entries are never
directly linked to L2 entries, but information shared between the languages
will result in activation feedback flowing to the lemma nodes concerned. In
other words, Ll and L2 entries can never be lexically mediated, but are
always conceptually mediated to a degree dependent on the relative activa-
tion of the conceptual representations, the lemma nodes, and the lexemes.
Similar to how partially overlapping meanings in the monolingual mental
lexicon can be accounted for, the activation model can also account for over-
lapping meanings between Ll and L2. Figure 4 exemplifies the partial over-
lap between a Dutch and an English item in the bilingual mental lexicon.
The same framework can also be used to account for the development of
the bilingual mental lexicon. At initial stages of L2 acquisition, a full overlap
may be assumed between the conceptual representations of the Ll lemma
and the L2 lemma. Gradually, the differences between the L1 and the L2
lemma will be acquired, which may eventually lead to a "native-like" lexical
representation. This process can be entirely based on positive evidence and
is guided by the same principle of contrast that is at work in L1 acquisition
(cf. Clark 1993). When the learner encounters a new L2 word, this may lead
to the partial restructuring of the semantic form of existing concepts by
adding or deleting the match with some of the conceptual representations.
This process is exemplified in Figure 5. 2 At some early stage of acquisition
(t 1), the Dutch learner of English will assume full overlap between between
and among, since Dutch does not make this conceptual distinction. 3
Subsequently, the principle of contrast will ensure that the learner will not

Authenticated
accept two fully identical lemmas, leading to the discovery of the semantic
differences between between and among. This will then lead to restructuring
of the semantic form of between and the creation of a new lexical item
among. The ultimate result of the acquisition process can be a "balanced"
bilingual lexicon in which all semantic forms of all lemmas have been fully
specified. However, cases where this happens for all lexical entries in both
languages will be highly exceptional, as most bilinguals will not be fully
"balanced". The additional advantage of this approach is that it is no longer
necessary to assume the same stage of development for entire language sub-
sets. While some L2 lexical items may be fully developed, including all
semantic and syntactic regularities and restrictions, other items may be found
in different stages of acquisition.
0
0 0
eerlijk o O~o DUTCH
D----o~g
HONEST
FAIR
o 0 ADJ
o 0 +ABSTR
0 - - - ~O
fair
o 0 ENGLISH
0 0
0
0 0
Figure 4. An example of partial overlap between lexical entries in Ll and L2. In

this simplified representation, the different units of information associated
with a lemma (semantic, syntactic, language) have been collapsed.
After this elaboration of the interactive activation model of the bilingual

mental lexicon, let us now return to our main question: what is the role of the
first language in the acquisition oflexical items in L2? In the interactive acti-
vation model, cross-linguistic similarity can be expected to affect the acqui-
sition ofL2 lexical items at two levels. First, it can be expected to occur on
the left-hand side of the model, at the end of the phonological/orthographic
representations. Orthographic and phonological similarity to Lt lexical items
may affect the acquisition of L2 lexical items. This effect may be facilitating
in the case of cognates, but can be confusing when orthographic and phono-
logical similarity does not coincide with semantic similarity (Lowie 1991;

Authenticated
Dijkstra, Grainger & van Heuven 1999). Second, cross-linguistic influence

will be related to the degree of semantic overlap between lexical items in
different languages. This effect, Translation Equivalence (Lowie 1998) will
facilitate the acquisition of an L2 lemma through interactive activation: not
only will the differences between languages be gradually acquired, but also
the similarities. Translation equivalents in L1 and L2 lead to the coactivation
of semantic forms. It can therefore be assumed that if translation equiva-
lence is combined with orthographic and phonological similarity, i.e., if
translation equivalents are cognates, the equivalence will be noticed sooner.
In the study described below, the orthographic and phonological similarity
between lexical items in L1 and the L2 is one of the main variables included.
LX SF CR LX SF CR
o o
000
00 0 among
tussen o
0-1,40 o~~ 0 tussen
00 0
_-0
between 0-1,40 0 between
o 0 o
o
Figure 5. Simplified representation of two time slices (tl and t2) in the process of
acquiring the new L2 concept "among". In this figure, the lemma nodes
have been left out. LX= lexeme; SF = semantic form; CR=conceptual
representations.
The other variable factor, frequency, is independent of the mother tongue and
expresses the extent to which the L2 lemma is used. The role offrequency
in activation models is obvious, as processing in the lexicon is driven by
frequency-induced activation; all elements in the lexicon can attain variable
degrees of activation, which increases each time a node is used, and decreases
over time. Activated nodes spread activation to nodes with which they are
connected. Frequency is thus the major drive behind lexical acquisition.
The two factors described here, input frequency representing the inde-
pendent effect of L2 on the acquisition process, and formal similarity repre-
senting cross-linguistic influence, were also the main variables in a recent

Authenticated
study on the acquisition of derivational morphology in the bilingual mental

lexicon (Lowie 2000). The L2 factor in this study was productivity, which is
strongly related to frequency. The study showed that translation equivalence
plays a major role in written production at all levels of acquisition, indicating
that the learners in this experiment rather strongly relied on their morpho-
logical experience in L 1. This study also showed that learners, especially at
lower levels of proficiency, have not (yet) acquired the productivity of L2
morphological types. Only at the highest level of proficiency did productiv-
ity positively contribute to the scores in the test, and then only in the context
where L 1 was not explicitly activated. This finding is in agreement with the
expected lexical development: at early stages ofL2 acquisition, the learner's
main source of information is his or her mother tongue. At these stages, a full
conceptual overlap is assumed between lexical entries in L 1 and L2. At later
stages, after prolonged exposure to the second language, the restructuring of
the semantic form, as exemplified in figure 5, will take place.
3. Prepositions in English and Dutch
To answer our main research question - to what extent is L2 development

related to input (relative frequency of occurrence) or language transfer - we
set up an experiment in which the comparative effect of the Ll-related
variable "similarity" and the L2-related variable "frequency" was deter-
mined for Dutch learners of English at four different levels of language pro-
ficiency.
To keep variables to a minimum, we limited the word classes in the
experiment to prepositions. The choice of prepositions was based on the
fact that they occur relatively frequently, so even beginners are familiar
with some of them. In addition, they are easily controlled for meaning in
context, so they can easily be elicited. To avoid any problems with figurative
or idiomatic uses, the prepositions were used in their central senses, referring
to place (e.g., he is in his room), time (e.g., he has been ill for three
months), direction (the ashtray fell off the table), possession (the legs of the
table), and beneficiary (I bought the presentfor my friend) (see Appendix 1
for a full list of the test items used).
A relative frequency list of prepositions was obtained through CELEX
(Cobuild Corpus; Baayen, Piepenbrock & van Rijn 1993). All occurrences
of the prepositions disregarding different senses were taken into account.
Prepositions for the test were selected according to the relative frequency,
those with high frequency (75,000 or more occurrences in the CELEX/

Authenticated
COBUILD corpus, which contains 18,636,970 English lemmas) and those

with low frequency (20,000 or less in the CELEXlCOBUILD corpus). From
both lists, items were selected that were highly similar in orthography and
meaning to their Dutch counterpart when used in a literal sense and those that
were not, resulting in the item selection represented in Table 1. These preposi-
tions were elicited in a cloze test consisting of 25 rather simple English sen-
tences, with the blank to be filled with a targeted preposition (see Appendix
1). To avoid positively affecting "transfer", we did not provide the Dutch
equivalent. The test, which took about 15 minutes to complete, was admin-
istered during regular class times. All prepositions that fit the English con-
text were considered correct, even if they were not the targeted preposition.
Table 1. Overview of the English prepositions and their Dutch translation equiva-
lents in the four categories selected.
High frequency in Low frequency in

English / Dutch English / Dutch
equivalent equivalent
High similarity to BY/BIJ SINCE / SINDS

Dutch equivalent IN/IN UNDER/ONDER
FOR/VOOR ABOVE/BOVEN
ON/AAN
Low similarity to AT/AAN NEAR/BIJ

Dutch equivalent TO/AAN OVER/BOVEN
BY/DOOR BELOW / ONDER
FOR/GEDURENDE AMONG / ONDER
TO/NAAR AS FARAS / TOT
AT/SINDS BETWEEN/ TUSSEN
FOR/VAN AMONG / TUSSEN
OF/VAN OFF/VAN
FROM IN FRONT OF / VOOR
Seventy-five Dutch learners of English participated in this experiment. These

participants were taken from four naturally occurring groups at the begin-
ning level (a 7th grade junior high-school class), at the low-intermediate
level (a 9th grade high-school class of a college preparatory school), the
high-intermediate level (first year non-English majors at the University of
Groningen), and the advanced level (third year English majors at the Uni-
versity of Groningen). Based on previous research and on the literature

Authenticated
described above, our expectations were that, especially at the lower levels
of acquisition, a high degree of similarity between the Dutch and English
prepositions would positively affect the correctness of the answers. The
effect of frequency, which was previously shown to occur only after pro-
longed exposure to the second language, was expected to be strongest at
higher levels of acquisition.
4. Results
The scores of the four groups in this experiment were analyzed using a
MANOVA, with group as the between-subjects variable (four levels) and
similarity and frequency as within-subject variables (two levels each).
The main effect of the between-subjects variable, group, was significant
(F[3,71]= 45.2; p<O.Ol); the highest number of correct scores was found in
the group with the highest proficiency (see Figure 6). Also the main effects
of the within-subjects variables similarity (F[1,71]=71.0; p<O.Ol) and fre-
quency (F[1,71]=82.S; p<O.Ol) were both significant.
100
90T-----------------------------------r=====~--
.. -:-:..-:-.71-,------1
80------·----------·----------r..:-:-.-:-:
70 --------------------------4
60 - j - - - - - - - - - - - - - - - - - j . . . . . . . . . t - - - - - l ..... -
. . . . .. .. .
50 +----------------i t - - - - - l ......... r------j ....... - -
40+----------------i ...... -
30 r-----: .... -
20 r-----.' .,-
10 1------ ••. ..-
O+-~----~--~~----~--~~----~--~~-----J__.
group 1 group 2 group 3 group 4
Figure 6. Mean percentage of overall correct scores per group,
The interactions between group and frequency (F[3,71]=6.7; p<O.Ol) was

significant (see Figure 7), where the smallest frequency effect was found at
the highest level of proficiency.

Authenticated
Also, the interaction between group and similarity turned out to be signifi-
cant ([F3,71]=S.O; p<O.Ol). The analysis showed that for similarity also the
smallest effect was found at the highest level of proficiency (see Figure 8).
The interaction between frequency and similarity (F[1,71]=3S.3; p<O.Ol)
showed that the largest similarity effect was found with the low-frequency
prepositions (see Figure 9).
The three-way interaction between group, similarity, and frequency was
significant as well (F[3,71]=5.1; p<O.Ol). The analysis showed that the inter-
action between similarity (as represented in Figure 7) was only significant for
the low and the intermediate level of proficiency, but not for the highest level.
A qualitative items analysis revealed that the lowest overall scores were
found with English prepositions that indicate a conceptual distinction that
does not occur in Dutch. Examples are the English prepositions over and
above, both represented by boven in Dutch, and among and between, both
represented by tussen in Dutch. The same effect, but less strong, was found
100.---------------------------------------.
__ ---i
80 _------ ---- --- --------

---- ---- ;.~
60 '~.~~.~~.-
•••••••••••••• - •••••••••• - ••••••••• - •••••• - ••••••- _ . . . .•••••
group
40
_._._._+
Imavo
.~
.....................
~ 20 3vwo
en
eo
1;l
------.
Juni
~o 3uni •
u O+-------------------------------------~
low high
FREQUENCY
Figure 7. Frequency effect for the four groups in the experiment

Authenticated
Table 2. Representation of correct scores (%) for some individual items in the three
groups.
Group boven / over tussen / among van/ of bij / by
1 30 lO 40 50
2 12 4 65 65
3 80 29 80 86
for from and of, both represented by van in Dutch. The number of correct
scores for these words was particularly low at the lower levels of English
proficiency. The leamers typically opted for the more frequent altemative in
these cases. Another striking result was that some items that are very similar
in both form and meaning in English and Dutch (e.g., by - bij, as in 'sit by
the fire') had comparatively low scores at the lower proficiency levels.
100,-----------------------------------------,
--~
-- -- -- -- -- -- ---
80
--- --- ---
4 ---
--- --- .......................
...............................................•
..........
..........
60 " .................... .
group
_._._.-+
Imavo
40 .....................
3vwo
------.
luni
20+---------------------------------------~ 3uni •
low high
SIMILARITY
Figure 8. Effect of similarity for the four groups in the experiment.

Authenticated
100,-----------------------------------------,
90
80
70 '11""
60
SIM
50
..................•
1
u
~
8 40+---------------------------------------~ 2
•
low high
FREQUENCY
Figure 9. The interaction between frequency and similarity. High and low frequency
are represented horizontally; similarity is represented by the different lines:
the dotted line for low similarity and the continuous line for high similarity.
5. Discussion
The experiment showed a clear effect for both similarity and frequency for
the low and intermediate levels of proficiency, but hardly any effect for the
highest level of proficiency. Considering the high scores of the high profi-
ciency group (see Figure 6), this observation could be attributed to a ceiling
effect for the highest level of proficiency. Apparently the subjects in this
group had acquired full lexical representations for all the prepositions in the
experiment, regardless of degree of similarity to Dutch prepositions or the
frequency in the input.
At beginning and intermediate stages, both frequency and similarity tend
to affect the score. The effect of similarity was in agreement with our expec-
tations and cOlToborates results from previous studies. Apparently, the formal

Authenticated
similarity to L 1 prepositions facilitates the acquisition and use of preposi-

tions in L2. The finding that frequency also affected scores for these learn-
ers was not in agreement with expectations. A possible explanation for this
finding is that Dutch children are extensively exposed to English. Contrary
to the morphological types investigated in an earlier study (Lowie, 2000),
prepositions are rather frequent. Apparently, subjects at the lower and inter-
mediate levels of proficiency are sufficiently exposed to frequently occurring
prepositions to show a difference in scores related to frequency levels.
The interaction between frequency and similarity showed that the degree
to which L2 prepositions are similar to prepositions in L 1 only affected the
scores if these prepositions were not very frequent: for the frequently occur-
ring prepositions no effect of similarity was found. The explanation for this
finding would be that subjects tend to rely on their first language only for
the more unfamiliar prepositions. This explanation is supported by the
observation that this interaction did not occur at the highest proficiency
level: these students had been sufficiently exposed to the L2 to develop full
representations for all prepositions.
The results of the qualitative data analysis clearly indicates the develop-
mental process predicted by the model (and exemplified in Figure 6). At the
lower proficiency levels most subjects show a presumed conceptual overlap
between Ll and L2 lexical items. At the highest proficiency level almost all
subjects seem to have acquired the additional conceptual category in
English. Only the difference between the prepositions among and between
has not (yet) been acquired by the majority of subjects at the highest profi-
ciency level. The striking finding that in some cases low scores were found
while both formal similarity and conceptual overlap were high seems to point
to the learners' reluctance to use formally similar words for the translation
of lexical items not used in a frequently used sense (cf. Kellerman 1987).
6. Conclusion
In this paper we have summarized an interactive activation model of the

bilingual mental lexicon and presented an empirical study investigating one
aspect of this model. The model summarized here claims that all overlap
between items in the mental lexicon is mediated by conceptual representa-
tions. In the bilingual lexicon, sets of items that share the same language
property (e.g., English or Dutch) constitute different language subsets.
Through a mechanism of interactive activation, both lexical subsets and
individual lexical items can have different levels of resting activation. The

Authenticated
process of acquisition of lexical items is a matter of gradual development

from assumed full conceptual overlap between L1 and L2 items at beginning
stages of acquisition to native-like L2 representations at advanced levels of
acquisition.
The main purpose of this study was to determine the role of the learners'
first language in the acquisition of lexical items in a second language and to
investigate how this role develops over time. We argued that the first lan-
guage plays a role at two levels of the model described: at the level of seman-
tic overlap between lexical items in the two languages, mediated by the
conceptual representations, and at the formal (phonological/orthographic)
level. The focus of the current paper was on the latter type of cross-linguistic
influence. The amount of formal overlap between the two languages was
represented as the different levels of formal similarity between Dutch and
English prepositions. The experiment shows that formal similarity plays a
role at the beginning and at the intermediate level of proficiency, but not at
the highest level of proficiency. This finding is in agreement with what we
expected to find. We hypothesized that beginning learners will presume a full
overlap between lexical items in the two languages and that the learner's
assumptions are reinforced by formal similarity of L1 and L2 lexical items.
This shows that at early stages of L2 acquisition learners rely on their first
language in learning and using L2 lexical items. Only at the highest level of
acquisition has the learner developed complete L2 entries and no longer has
to rely on hislher first language.
On the basis of previous research on the acquisition of L2 derivational
morphology, we also expected that frequency would play a role only at the
highest levels of L2 acquisition. The reason for this was that the effect of
frequency was expected to start affecting L2 performance only after pro-
longed exposure to the second language. The experiment, however, showed
that frequency did play a role at the lower levels of L2 acquisition, but not
at the highest levels. We accounted for this observation by arguing that,
unlike morphological types in earlier experiments, the average frequency of
prepositions is comparatiVely high. This results in high frequency preposi-
tions that are abundant even in the beginning learner's input. The absence of
an effect for the highest level of acquisition could be explained by a ceiling
effect: the scores for this group on prepositions were already very high for
the low frequency prepositions, leaving no room for a further frequency
effect.
Although the findings in the study reported here are clearly significant,
the results must be interpreted with some care. The subjects in this study
were taken from four intact classes and their number was limited. However,

Authenticated
for the main effects of significance and frequency, this is no serious draw-
back as these were within-subjects variables. And although the cross-
sectional design with these groups cannot be considered identical to real
development as measured in a longitudinal design, the clear difference
between the cross-sections taken gives a strong indication in the direction of
the development observed.
This study suggests that up to the intermediate level of acquisition learners
rely on formal similarities with their L1 in using L2 prepositions. Further
research will be conducted to determine whether the translation equivalence
of lexical items, i.e., the amount of conceptual overlap between L1 and L2
lexical items, is noticed if it does not coincide with formal similarity. It can be
assumed that the greatest difficulty for learners can be found in cases where
formal similarity does not coincide with translation equivalence. This cate-
gory (of "false friends") needs further investigation with learners at different
stages of L2 acquisition. Another interesting observation that merits further
investigation is that qualtitative items analysis of the current experiment
revealed that English prepositions that indicate a distinction that does not
occur in Dutch get the lowest scores (e.g., tussen = between / among; boven
= over / above). The difference between the groups for the scores on these
items seems to support the gradual restructuring of L2 lemmas from assumed
conceptual overlap with similar Ll items to independent (though partially
overlapping) L2 lemmas.
The experiment described here is by no means sufficient to support (or
falsity) the entire model of the bilingual mental lexicon described here. Only
one aspect of the model has been tested and many other questions are yet to
be answered. However, with this experiment we hope to have contributed to
the gradually emerging picture of the bilingual mental lexicon that can only
be established by the constant interaction between empirical studies and
theoretical modeling.

Authenticated
Appendix - full representation of test items used in the experiment
The quiz contained a blank for the underlined preposition. The Dutch equiv-
alent was not provided (but is given here to show level of similarity).
1. Let's sit by (BIJ) the fire.

2. He is in (IN) her room, I suppose.
3. I bought this present jar (VOOR) my friend.
4. There is a picture on (AAN) the wall.
5. T have lived here since (SINDS) 1995.
6. The cat is lying under (ONDER) the table.
7. The plane was flying above (BOVEN) the clouds.
8. He is at (A AN het) work
9. I will never lend my car to (AAN) my son.
10. She was killed by (DOOR) a bullet.
11. He has been illIor (GEDURENDE) three months.
12. He went to (NAAR) London for a week.
13. They threw tomatoes at (NAAR) him.
14. He has been absent jar (SINDS) three days.
15. The legs of (VAN) the table are wobbly.
16. Tknew I could trust himfram (VAN) the moment I saw him.
17. He lives Ileal' (BIJ) Cambridge
18. He was leaning over (BOVEN) the body when he was struck on the head.
19. His apartment is below (ONDER) mine.
20. Among (TUSSEN) my friends, there is no one who can help me.
21. To get to the station, walk as far as (TOT) the traffic lights, and then turn right.
22. His house stands between (TUSSEN) his sister's and mine.
23. T couldn't find my friend among (TUSSEN) all those students.
24. The ashtray fell ojJ(VAN) the table.
25. I saw somebody standing inFont of (VOOR) your window, trying to look inside.
Notes
1. However, Kellerman (1987) has pointed out that Dutch learners resist transferring
non-prototypical uses of such words.
2. Analogous to an example worked out in Schreuder & Baayen (1995).
3. Even though the distinction between henveen and among is one that is not
observed by all native speakers much anymore, it still exists. Benveen refers to
two entities; amollg to three or more.

Authenticated
References
Baayen, H., Piepenbrock, R., & van Rijn, H.

1993 The eELEX Lexical database (CD-ROM). Philadelphia: Linguistic
Data Consortium, University of Pennsylvania.
Clark, E.v.
1993 The lexicon in acquisition. Cambridge: CUP.
Dijkstra, T., van Heuven, W. 1. B., & Grainger, 1.
1998 Simulating cross-language competition with the Bilingual Interactive
Activation model. Psychologica Belgica, 38, 177-197.
Dijkstra, T., Grainger, J., & van Heuven, W. 1. B.
1999 Recognition of cognates and interlingual homographs: The neglected
role of phonology. Journal ofMemory and Language, 41, 496-518.
Ellis, R.
1994 The study ofsecond language acquisition. Oxford: Oxford University
Press.
Jiang, N.
2000 Lexical representation and development in a second language. Applied
Linguistics 21 (1),47-77.
Kellerman, E.
1987 Aspects oftransferability in second language acquisition. Unpublished
Ph.D. thesis, University ofNijmegen.
Kroll,1.F.
1993 Assessing conceptual representations for words in a second language.
In R. Schreuder & B. Weltens (eds.), The bilingual lexicon (pp. 52-81).
Amsterdam/Philadelphia: Benjamins.
Langacker, R. W.
1991 Foundations of Cognitive Grammar: Descriptive Application.
Levelt, W 1. M.
1989 Speaking: from intention to articulation. Cambridge, Mass.lLondon:
Bradford.
Lowie, W.M.
1991 Derivationele morfologie en tweede-taalverwerving. Toegepaste
Taalwetenschap in Artikelen 39, 83-92.
1998 AIM: The Acquisition of Interlanguage Morphology. Ph.D. thesis,
Rijksuniversiteit Groningen.
2000 Cross-linguistic influence on morphology in the bilingual mental
lexicon. In Studia Linguistica 54,2,175-185.
Schreuder, R. and H. Baayen
1995 Modelling morphological processing. Morphological aspects of
language processing. ed. L. B. F eldman, l31-156. Hillsdale: Lawrence
Erlbaum.

Authenticated
Tomasello, M.
First steps toward a usage-based theory of language acquisition.
Cognitive Linguistics 11, 112,61-82.
Woutersen, M.
1997 Bilingual word perception. Ph.D. Thesis, University of Nijmegen.
Nijmegen: Katholieke Universiteit Nijmegen.

Authenticated
Linguistic and Cultural Relativity -
Reconsidered for the Foreign Language Classroom
Susanne Niemeier
1. Introduction
The purpose of this contribution is not so much to provide evidence for the
Whorf theory complex of Linguistic Relativity itself, but rather to discuss
how the renewed attention on relativity in language and culture may be rele-
vant to current tendencies in foreign language teaching methodology.
However, let us state right from the beginning that relativism as such is
independent from the concept of a first or second language, but applies lan-
guage-internally, and is not used here as a cover term for "cultural stereo-
types".
Over the last few years we have witnessed a very noticeable rise of interest
in Whorf's thoughts and ideas (cf. Lucy 1992 and 1997, Gumperz and
Levinson (eds.) 1996, Lee 1997, Niemeier and Dirven (eds.) 2000, Piitz and
Verspoor (eds.) 2000). This reappraisal is not primarily due to Whorf's cen-
tenary in 1997, but is also based on more solid grounds such as a strength-
ened acceptance as well as a re-interpretation and re-evaluation of Whorf's
scientific research. This renewed interest stems from various sources: First of
all, many scientific domains slowly abandon the idea that knowledge can be
compartmentalized but argue that we have to take a more holistic perspec-
tive. Second, in linguistics, there is no longer any agreement that language
is to be seen as an autonomous faculty - as the generative paradigm had it -
but there is also the view that it is directly interacting with other mental fac-
ulties. For example, the slow but steady rise of Cognitive Linguistics from
the 1980s onwards has from its very beginning emphasized the fact that lan-
guage, culture, and thought are inextricably intertwined, and that therefore
it is not possible to analyze anyone of them without taking the others into
consideration. Much of our linguistic behavior can be shown to be based on
extralinguistic experiences and cultural knowledge. Even if today Cognitive
Linguistics is not as enthusiastic about Whorf's ideas as it seemed to be in its
beginnings - because his ideas do not seem to allow for universalist tenden-
cies - the interrelation of both research directions is nevertheless certainly a
given.

Authenticated
96 Susanne Niemeier
2. Current Trends in Foreign Language Teaching Methodology
In foreign language teaching methodology, several new areas of reflection

have been developing over the last fifteen years. From a German perspec-
tive, I the most promising of these are the concern with awareness raising,
the learning objective of intercultural competence, the targets of auton-
omous learning, multi-channel learning, holistic learning and teaching, as
well as the action-oriented approach to learning. Furthermore, we should
also take into account the rather recent focus-on-form approach where we
are currently witnessing a movement back from a concentration on mere
communicative competence, aiming at reintegrating grammar into the foreign
language classroom by combining formal instruction and communicative
language use (cf. Doughty & Williams 1998). These approaches and their
connection to relativity in language and culture as well as to Cognitive
Linguistics will be discussed in more detail below.
Suffice it to mention at this point that the above-mentioned trends in
foreign language teaching methodology, which are to a certain degree inter-
related, also seem to have a relation to Whorf's hypotheses concerning the
interconnectedness of language, culture, and thought; and that it is thus a
truism to claim that Whorfian ideas should be very much present in today's
foreign language classroom. Put in a nutshell, Whorf proposed the idea that
language has an influence on our way of thinking and thus on our perspec-
tive of reality - or in more "cognitive" terms, on conceptualization. I will
briefly highlight the connections between Whorf's ideas and the new EFL
trends mentioned in the above paragraph and then move on to an analysis of
how to profit from these connections in the foreign language classroom.
The concept of language awareness was developed in the 1980s (for
more recent publications, see James and Garrett (eds.) 1995, van Lier 1996,
Gnutzmann 1997) and is geared at a different kind of learning than tradi-
tional, mainly grammar-oriented language learning. Being aware of the
idiosyncrasies and thus of the underlying systems of values and attitudes of
a foreign culture became a learning target in itself, and was no longer
regarded as merely an instrument to advance the learners' language compe-
tence. In current language awareness approaches, language is seen as being
much more than a medium for communication. This kind of awareness rais-
ing became accepted as a cognitive and affective asset in itself, widening
the learners' world knowledge and their (inter-)cultural competence. Being
aware of how language works, how meaning is created, and how language
is connected to culture and to our conceptualizations provides a different
and deeper understanding of a foreign language and culture as well as of

Authenticated
Linguistic and Cultural Relativity 97
one's own language and culture, and thus potentially generates a different
worldview. All in all, the insights connected to the concept of language
awareness seem to fit well into Whorf's thoughts on the interconnectedness
of language, thought, and culture.
Quite a similar connection can be discovered when reflecting on the
learning target of intercultural competence (cf. Buttjes and Byram, eds.
1991), a possible means to remedy the learners' lack of familiarity with or
even lack of knowledge about the target culture, which until today is mainly
being taught by texts or, less frequently, by audio and video materials. In a
cognitive-linguistic view, culture is present in the language itself, on all its
levels, and can be detected by language work and language analysis.
"Culture" does not reside somewhere outside of language, but is implicit in
its every word and every grammatical construction and is thus constantly
communicated by the language. Detecting and analyzing intercultural
differences via the foreign language can even provide learners with a differ-
ent, more distanced view of their own culture and thus initiate very helpful
processes of a change in their self-image and of more empathy toward the
foreign culture or foreign cultures in general.
As to the learning objective of autonomous learning (cf. Wolff 1994,
Little 1994), this is rather close to the theory of constructivism in language
learning (Wendt 1996). Learners do not necessarily learn what they are
taught, but they construct their own knowledge little by little from the input
that they are offered. They form hypotheses and try them out, either veri-
fying or falsifying them, and in the latter case they form new hypotheses
(this idea is also quite close to the concept of "interlanguage", cf. Selinker
1992). In this way, learners learn rather independently at their own pace and
the teacher's role becomes that of a monitor and of an input provider.
Autonomous learning can thrive best when the learners are offered authen-
tic learning situations (which involve insights into foreign thinking patterns
and into the foreign culture) in a classroom environment where the above-
mentioned principles of language awareness and intercultural competence
are present as well. Of course, most of the language input still has to come
from the teacher, who at the same time shows the learners how to learn and
how to reflect on their own learning strategies. In this way, learners may
construct their own image of the foreign language, the foreign way of think-
ing, and the foreign culture. The connection to Whorfs ideas is not as obvi-
ous here as it was concerning the before-mentioned trends, but it is present
nevertheless as the hypotheses that learners are able to form are necessarily
influenced by the learners' own linguistic and cultural experiences, be it in
their native language or in a foreign language.

Authenticated
98 Susanne Niemeier
Learners are not a homogeneous group; everybody has their own ways and
strategies of how to construct knowledge. This is why teachers should be
aware of the fact that they have to offer different kinds of information to
their learners. Concentrating on multi-channel learning (cf. Ellis 1994)
means that teachers have to cater to various types of learners by, if possible,
offering for example visual, auditory, audiovisual, tactile, verbal-abstract,
and action-oriented approaches to the information they want to transmit. In
this way, knowledge gets stored in various parts of the brain and different
brain links are constructed, which again means an easier retrieval of the
stored knowledge (see also Lamb 2000). A related approach is that of holistic
learning and teaching (cf. Timm 1995, Loffier 1996) which claims that
cognitive and affective aspects should go hand in hand (involving both
hemispheres of the brain, where traditional teaching was mainly centered on
the right hemisphere) and where all the learners' senses should be activated.
This is supposed to heighten the learners' motivation as well as their atten-
tion and creativity, and to have them enjoy learning (about) the foreign lan-
guage and the foreign culture. Whorf' s ideas can be called holistic as well,
as they highlight the connections between language, thought, and culture
and thus do not see them as separate modules.
Although none of these trends in EFL teaching methodology can be
explicitly connected to Whorf's ideas, they are still in line with the idea that
language is only one facet of a person or a culture, and that it is in a very
close relationship with other mental capacities. Activating these other capac-
ities helps the process of learning and of discovering and understanding
connections, for example those between language and culture, because
culture is not only expressed language-internally, but also, as cultural anthro-
pology continues to point out (cf. Palmer 1996), in other spheres oflife, be it
housing, dressing, agriculture, kinship systems, and the like. These aspects
have to be taken together, and this is not possible by using only language
but needs other channels and other ways of making connections; for example
by raising the learners' interest and emotional involvement, which again
forms part of a holistic way of teaching.
To give another example, the action-oriented approach to learning (cf. Bach
and Timm 1996) fits nicely into the pattern outlined above as well. According
to this approach, learners are expected to develop a competence for acting in
the foreign language, first in the classroom but ultimately also with foreigners
in their own country or within the foreign language culture. The teaching
relies heavily on the methodological approach of "learning by doing", which
involves offering authentic learning environments (as, for example, project
work) and pursues the goal that the learners relate affectively as well as

Authenticated
cognitively to the learning tasks. The learners are expected to use the lan-
guage as much as possible and not just to passively take it in, and by using the
language they learn about its culture-relatedness at the same time. The teacher
has to tolerate a higher rate of mistakes because the general motto is "message
before accuracy", giving the learners the possibility of taking communicative
risks (as we all do in everyday discourse), of building their own hypotheses
about language in general and the foreign language in particular, and of en-
hancing their linguistic as well as their social skills in communication.
Last but not least we should mention the rather recent focus-on-form
approach. Although communicative language teaching (CLT) is probably still
the most widespread view on foreign language teaching, we are currently
witnessing a movement back from a concentration on mere communicative
competence to a so-called "focus-on-form", aiming at reintegrating grammar
into the foreign language classroom by combining formal instruction and
communicative language use (cf. Doughty & Williams 1998). SLA re-
searchers such as Pica see "the need for direct instruction and corrective
feedback" (2000: 11). Another recent development is discourse-based ap-
proaches to grammar instruction, which focus on authentic language uses and
structures and their meanings in discourse and text, i.e. an application of
corpus-based research (cf. McCarthy 1991), so that learners may be con-
fronted with "real" language and not with specially constructed texts. Again,
if we take a cognitive-linguistic perspective, we can claim that grammar is
as closely connected to culture as every other aspect of the language, and
that therefore a renewed focus on grammatical structures, combined with
other approaches, helps raise (inter-) cultural awareness.
It is striking that in all these newer approaches within foreign language
teaching methodology, culture is seen as an inherent part of language and
that every single approach, as different as they may each be, aims at inte-
grating learning the language with learning about the language as well as
with learning about the culture. The main idea, then, is that foreign language
learning or learning a language different from one's mother tongue also
opens up the possibility of seeing the relativity of one's own language and
culture. Language awareness raising combined with foreign language
awareness raising creates the potential of freeing oneself from one's own
prejudiced Ll views. This is not merely a linguistic aim but also a peda-
gogical one. Taking this perspective also aims at developing the learners'
personalities by offering them the possibility of analyzing and evaluating
differing cultural values - as they show up in the language - and thus
helping them to get rid of their own ethnocentricity, which is in line with
Whorf's claims concerning the impact of culture on language.

Authenticated
100 Susanne Niemeier
Furthermore, a certain commitment to "Whorfianism" would not only prove

helpful for the foreign language learners, but also in the development of new
foreign language curricula for teacher trainees, or for on-the-job training for
teachers, as well as for the development of new textbooks. So far, teaching
about the foreign culture is left to a very large extent to the teachers them-
selves, who - at least in Germany - do not necessarily have any personal
experience with the foreign culture and are thus in no position whatsoever
to teach about the culture apart from what they may have read or heard or
what the textbooks provide. However, even if in the newer generation of
English textbooks for German learners the keyword of "intercultural learn-
ing/competence"2 does appear quite frequently, often enough this is just a
cover term for introducing single items as diverse as British school uniforms
or the relevance of Thanksgiving Day in the USA. Such textbooks neither
try to transmit an insight into the diverse interlinked systems of values and
judgments in the foreign culture nor into the "habitual behavior patterns"
connected to these systems. Therefore what the textbooks do is certainly not
what is meant by raising the learners' "intercultural competence" or "inter-
cultural awareness". What should be aimed at, however, is to show or to let
the learners experience that language and culture are interrelated and that -
speaking in terms of two of the most relevant items when learning a foreign
language - not only do grammar and vocabulary acquisition go hand in
hand but they are also inseparably connected to the cultural background
and, therefore, the learning objective of intercultural competence should be
considered as equally important to that of learning the grammar and the
vocabulary. A problem that many teachers are confronted with in this
respect is that intercultural competence does not appear to be a testable item
at first sight, but even here it would not be too difficult to find an alternative
way of judging the outcome of the learning processes, such as, for example,
a portfolio assessment over a longer period of time or various pragmatic
tests (cf. Doye 1996).
The ultimate goal, then, is not only to be aware of or even get rid of pre-
judices which might be present in the learners' own cultures and minds, but
in the first place this means to realize and to understand that different cul-
tures see and experience the world in different ways than their respective
languages and value systems have conceptualized them. Different thought
and value systems, crystallized in each culture, claim the right to be consid-
ered to be as mature as any others. Acquiring intercultural competence
means that the learners realize that each culture is as valuable as their own
and that they get sufficiently involved with that other culture to get the feel
of at least some of its values. If at all possible, learners should be taught not

Authenticated
to evaluate or judge the differences they perceive but merely to state them
and to reflect upon their repercussions on their own self-image and their
own culture. The foreign language classroom is a perfect environment to
pursue such aims, because learning a foreign language necessarily includes
learning about aspects of a foreign culture. This is also where teaching
methods based on Cognitive Linguistics may come in, as this theoretical
paradigm does not separate language and culture but is based on the insight
that language is always embedded in and part of its cultural environment
and that these two factors are thus inseparably intertwined.
Ifwe take Whorf's insights seriously, we might even want to ask ourselves
how it is feasible at all to teach a foreign language - which is necessarily
rooted in a culture different from that of the learners - without taking the
learners' culture into account. The logical conclusion that imposes itself is
that, first of all, raising the learners' intercultural awareness is a prerequisite
for all successful foreign language teaching, in the sense of making learners
aware of different "isolates of experience", as Lee (1997) would suggest
along with Whorf.3 Thus, for example, even if Germans physiologically
experience much the same natural phenomena when they talk about
"Nebel" and "Dunst" as native English speakers would when they talk about
"haze", "mist", and "fog", the German language has cut up the continuum
at different points than the English language, and German speakers thus do
not talk about exactly the same concepts as English speakers do. The same
applies to the famous example of the multitude of Inuit words for snow:
members of other cultures may well be able to experience the same kinds of
snow but as those are not vital to their everyday lives, their languages have
not lexicalized them. This does not mean that these phenomena do not exist,
just that the language communities chose not to focus on these "isolates of
experience". Second, we have to relativize the dogma of translatability - of
words or categories, of grammatical concepts as well as of elements of cul-
ture - because in translation, words, categories, and concepts, linguistic as
well as otherwise, would simply be transferred to a different conceptual
world where they may be completely out of place (for a Whorfian view on
translation, see also House 2000). Adhering to such principles, foreign lan-
guage teaching should be rethought so as to stress the aspect of the foreign
culture at least equally strongly as that of the lexical and grammatical struc-
tures themselves, and the new trends in foreign language teaching method-
ology as outlined above all seem to point to that direction. Taking Whorf seri-
ously in the foreign language classroom definitely means taking a much more
holistic view on language than the view that is currently present in Germany's
English textbooks and, accordingly, English language classrooms.

Authenticated
Although we should keep in mind, as Lucy (1997: 295) correctly reminds us,
that "linguistic relativity is not the same as cultural relativity", there is still a
very strong interrelation to be found, or as Bickel (2000: 185) formulates it:
"Correlations between linguistic and cultural patterns ... suggest mutual
influence ... , since both speaking and social behavior are publicly shared
activities that are transmitted across generations". And especially in foreign
language teaching, language and culture seem to be very narrowly inter-
twined because the learners' access to the other culture and their means to
discover it are always mediated by the language they are learning, although
one should be aware of the fact that the learners will always be influenced by
their own cultural and linguistic background/s as well. Thus we are dealing
here with a kind of double responsibility, because on the one hand foreign
language teachers are supposed to transmit the language correctly, and on
the other hand they are also supposed to be trying to "translate" and/or to
explain aspects of the target culture without letting the learners' home culture
interfere too much.
How could these claims be satisfied in a foreign language classroom,
apart from reflecting on and possibly acting according to the current trends
in foreign language teaching methodology? One suggestion would be to
turn to some major issues within the framework of Cognitive Linguistics (in
the Langacker-Lakofftradition) and see whether they can help reach at least
some of these aims. As the contributions in this volume show, this seems to
be quite a promising venue. Cognitive Linguistics, with its underlying
humanistic approach, takes a holistic view on language and culture and thus
shares a first common denominator with some of the above-mentioned cur-
rent trends in foreign language teaching methodology.
It is a truism that foreign language learning should not take place in bits
and pieces but aim at integrating all mental capacities of the learner at once
and from the very beginning onwards. Furthermore, a cognitive-linguistic
approach to language teaching and learning also allows for multi-channel
learning insofar as Cognitive Linguistics claims that language is but one of
our mental activities and is directly related to other mental activities such
as, for example, cognition, memory, vision, or senso-motor strategies - the
last also hinting at a connection to the action-oriented approach. Raising the
learners' awareness would then be the major target in this kind of teaching
and learning, and this not only entails raising their language awareness but
also raising their cultural as well as their intercultural awareness so that they
are successfully endowed with the capacity to realize the relativity of con-
cepts within their own culture, as well as those within the culture and the
language they are getting acquainted with. As a consequence, learners may

Authenticated
be willing to shed their cultural egocentrism and to accept the existence of

concepts other than the familiar ones.
3. Cognitive-linguistic Insights Relevant for the Foreign Language

Classroom
In the following paragraphs I will elaborate on two of the major tenets of

Cognitive Linguistics, namely on categorization and prototypicality as well
as on metaphor and metonymy, and point out their connections to the con-
cepts of linguistic and cultural relativity on the one hand and to the above-
mentioned current trends and targets within foreign language teaching
methodology on the other hand, and finally discuss some possibilities of
how to introduce these issues into a foreign language classroom. As this
topic has been at the mere periphery of interest, little research has been
done so far. However, the interest of the cognitive-linguistic community
seems to be slowly but steadily awakening. 4
3.1. Categorization and Prototypicality
So far, we have been asking the question of what the notion of linguistic
and cultural relativity could contribute to a strengthening of the learners'
recognition of the learning objective of intercultural competence (as tenta-
tively defined above), or at least to an increased awareness of the fact that
there are indeed major intercultural differences. I will try to answer this
question by drawing on an example from Cognitive Linguistic theory, i.e.,
how can we use the concepts of categorization and prototypicality in order
to achieve the aim outlined above.
First of all, if we consider that Gumperz and Levinson at the very be-
ginning of the introduction to their 1996 volume claim that "culture, through
language, affects the way we think, especially perhaps our classification of
the experienced world" (1996: 1), the association with categorization as it is
discussed in Cognitive Linguistics comes up immediately. The claim is that
different cultures categorize given concepts in different ways - for example,
an avocado is conceptualized as a vegetable in Germany and as a fruit in
other languages, which just goes to show that the boundaries between these
categories are fuzzy - and that such examples are illustrative enough to be
used even with ab initio learners. Of course, the outcome would be even
clearer if we chose cultures farther apart from each other than the German

Authenticated
and the British or American one - but nevertheless, even between these cul-
tures there are many differences to be shown, which furthermore do not
necessarily refer to concrete objects. Thus, for instance, in English we find
the expressions joy and happiness to describe very positive states of mind,
whereas in German we find the expressions gliicklich and zuJrieden (cf.
Wierzbicka in Dirven and Verspoor 1998: 146-148). Although these ex-
pressions do not refer to exactly the same states of mind (as can be proven
by an analysis along the lines ofWierzbicka's Natural Semantic Metalan-
guage approach), they still belong to the same meaning continuum; however,
the two languages in question have cut up this continuum in different ways
and have thus established different categories. Examples of this kind are not
hard to find and are useful to awaken the learners' interest in and openness
for linguistic as well as (inter-) cultural differences.
A central phenomenon within categorization is that of saliency or proto-
typicality. Thus, according to experiments by Rosch and others (cf. Rosch
1978), there is always a "best example" within a category, the so-called pro-
totype. This prototype is extremely culture-dependent - in a culture like
Germany, the prototypical bird would be a sparrow but in a culture like the
USA it is a robin because this bird is much more frequent there than, for
example, a sparrow. Admittedly, this is a rather obvious example; others may
be more subtle. But as prototypes depend on experiences within a culture
and are also often transmitted inside a culture, they have to be seen against
this cultural background.
In this context, research on colors is also of interest. In 1969, Berlin &
Kay already claimed the existence of "focal colors", i.e., best examples or
prototypical colors within a color continuum. These focal colors are culture-
related and also language-related because there are cultures such as the
Papuan Dani culture (see, for example, Heider 1972) - the extreme case that
is often quoted - which have in their language only two color terms and
therefore split up the color continuum in a different way than, for example,
the English language or the German language do, which possess eleven
focal color terms each. Thus, the linguistic color continuum is not a univer-
sally given fact but - as Lucy (1997) also claims - depends on the cultural
background and influences language and culture at the same time. For in-
stance, in languages with fewer basic color terms, people's color memory
seems to be less accurate than in languages with more basic color terms
(Lucy & Shweder 1979, 1988). As the focal color terms are usually mono-
morphemic and are among the first lexemes to be learnt when learning a
first or foreign language, the differences between two cultures can be pointed
out by focussing on the different referents for those terms. However, at first

Authenticated
glance, this does not seem to work when focus sing on the English vs. the
German language as they seem to have the same focal colors.5 However, if
we take color expressions such as "blue-eyed boy" or "red tape" into ac-
count (see Niemeier 1998) as well as the non-existence of possible German
counterparts, we are able to explain the culture-boundedness and thus the
non-translatability of these expressions.
The question remains of what the issues of categorization and prototypi-
cality have to offer for second or foreign language learning. First of all, let us
stress again that their specific strength is to help raise language awareness.
Monolingual people tend to think - in a rather ethnocentric way - that the
terms and prototypes found in their mother tongue describe universal issues,
and only via the detour of learning a foreign language may learners become
aware of how differently languages can be structured 6 and how different the
cultural concepts behind those linguistic structures are. The above mentioned
and similar examples may all be used even in beginners' classes and it is
decisive to use them exactly there, in order to show the learners from the very
beginning that they should not expect one-to-one meaning relationships or
structural similarities between their mother tongue and the foreign language.
This is an important issue as well when it comes to translation, because such
cultural differences in categorization and prototypicality as named above
implicitly negate the existence of a one-to-one meaning transfer, not to
mention the different associations with cultural norms and values that most
lexical items entail. This is a crucial issue for the learners to internalize be-
cause it will help them deal with more intricate concepts later on, and because
it allows them to construct their knowledge of the foreign language and the
foreign culture more independently from their own culture and language.
Thus, learning to realize and even to make sense of the cultural concepts
behind the words of the foreign language may also make learners more con-
scious of their own language, and therefore this way of learning (about)
intercultural differences often induces a process of reflecting on one's own
culture and cultural image for the first time. Intercultural understanding
does not necessarily mean stating what is different between the cultures, but
it focuses on showing that there ARE such differences, that one should be
aware of them, and that one should use the foreign language accordingly,
i.e., not referring to one's own cultural background but being aware of the
foreign cultural background. In a nutshell, learners should especially learn
to see their own language and culture as only one among many sensible
experiences of the environmental world. As a very trivial and often quoted
example, we can take the famous "How do you do?" utterance, which does
not call for a detailed report on one's health problems, but as an act of

Authenticated
phatic communication only warrants an answer such as "Fine. And you?",

whereas it is much more acceptable in Germany to give more information
on one's state of health as an answer to "Wie geht's?".
To come back to the issues in question, namely categorization and pro to-
typicality, I would like to suggest that they are valuable tools to be used
even at the very onset of foreign language learning. New words that are pre-
sented within a category are much easier learnt and remembered than isolat-
ed words, because in that way the learners can establish new neural links
and thus organize their knowledge much more efficiently. This effect is
even enhanced when multi-channel learning is involved as more neural
links can be established in that way. And if the categories of the mother
tongue and the target language are compared, differences may become evi-
dent and beg for an explanation.
Ideally, the learners discover such differences on their own and suggest
explanations of their own, which is a highly motivating endeavor. Even if
the explanations offered are not completely correct, finding or constructing
them entails focusing on the foreign language and "playing" with it, and in
that way also entails an activity on the learners' side. The action-oriented
approach maintains that learning via being motivated to do things with the
language - as searching for explanations or as realizing the need to use cer-
tain grammatical structures because of the need to express something or to
get the correct meaning across - facilitates learning and also facilitates
recall. Therefore, teachers should welcome such hypotheses from the learn-
ers' side, as those are proof of the fact that the learners construct their own
hypotheses regarding the foreign language and thus, "learn". Of course,
quite often teachers will have to present the explanations themselves as they
are the mediators between the two cultures in question. Also, in general,
working with explanations is a way of teaching that is very cognitively ori-
ented which does not cater to all learning styles. But explanations pave the
way for understanding, and learning via understanding is so much more
efficient than merely learning by heart. If learners are able to integrate and
extend their knowledge of and about the foreign language and culture in a
holistic way, then they can start being creative in the foreign language and
fulfill their own communicative needs. Therefore, such an approach may
very well be the second-best way of letting the learners experience cultural
differences when there is no possibility for them to spend some time within
the target culture itself.
Putting the concept of prototypicality to use is rather similar: by showing
or leading the learners to discover that other cultures have different or at least
differently organized prototypes - and thus different or at least differently

Authenticated
organized concepts of categories based on different cultural concepts -

teachers may be able to induce a heightened linguistic as well as cultural
awareness and probably a raised level oftolerance for the other culture (and
for other cultures in general). This can be done on a very simple level with
beginning learners and on a more sophisticated one for more advanced
learners. For example, beginners might prepare interviews with an English-
speaking partner class asking them to mention "typical" aspects of their cul-
ture (concerning animals, food, clothes, sports, etc.) and then compare those
aspects with their own answers. Alternatively, one might have the partner
class draw "a dog/a cat/a tree", in accordance with the multi-channel ap-
proach, and then compare these drawings with the drawings done inside the
German classroom - and one will quite probably end up with the various
prototypes because that is what we draw (or try to draw) when there is no
further specification of the drawing task. For more advanced learners, one
might want to use some of Rosch's research on categories such as that on
the category of furniture (Rosch 1975) and have the learners compare these
categories and their prototypes with research conducted by themselves on
the same categories in their home culture (this task would fit the action-
oriented approach), and then have the group of learners discuss the results
and try to draw their own conclusions. Of course, they need a teacher to
monitor the task and to give extra information when needed, but on the
whole the learners should try to approach the task in their own autonomous
way (even design it, if possible) and use their own special learning strate-
gies and approaches.
This is not only valid for lexical items, as prototypicality is also a factor
when it comes to grammar (see Niemeier 2001). Ifa grammatical structure,
such as, for example, the progressive form in English is introduced via the
prototype, then it is but one step more to also have the learners discover the
reasons for so-called exceptions to the rule, which in a cognitive-linguistic
approach are no longer seen as exceptions but as marginal members of the
category (see Niemeier forthcoming) and which thus become explainable
and understandable. Using the categorizationlprototypicality approach also
in the realm of grammar fits in well with the current focus-on-form approach
in SLA, which aims at reintegrating grammar into the foreign language
classroom by raising the learners' awareness to certain grammatical features.
Cognitive Linguistics, however, goes one step further as it is also able to
explain the structures as such in a way that the learners can follow and com-
prehend.

Authenticated
3.2. Metaphor and Metonymy
Two other, interrelated areas within Cognitive Linguistics that can be made
fruitful for foreign language teaching - especially on a more advanced level
- are conceptual metaphors and metonymies. Metaphors as well as
metonymies are so heavily entrenched in the foreign culture that they can
not be properly interpreted or explained without referring to their cultural
background. Therefore they seem to be a good means to induce knowledge
about the culture in question or at least to have the learners baffied by their
non-translatability and thus raise their interest. As to metaphors, Cognitive
Linguistics is not primarily interested in innovative, "creative metaphors"
(see also Lakoff and Turner 1989) as they are dealt with in traditional litera-
ture studies and that exploit the same conceptual metaphors as folk
metaphors do, but focuses on the strategies that underlie both these creative
metaphors and the so-called conventional metaphors, which are not used
consciously as metaphors anymore. These fossilized metaphors are never-
theless very efficient, as they may even be able to influence our thought
processes. Not only do we use them subconsciously but we also think in
terms of these metaphors without being aware of it, as they are so much a
part of our language that we do not pay any special attention to them; but
there is indeed a structuring behind the widespread usage of linguistic or
everyday metaphors.
Lakoff coined the term "conceptual metaphor" to highlight the fact that
within our conceptual systems we have a metaphorical structure which allows
or even constrains us to follow certain patterns of thought. A frequently
quoted example of a conceptual metaphor is TIME IS MONEY and the meaning
transfer (or mapping) from "time" to "money" is not a conscious one but a
cognitive predisposition within our culture. The concept of TIME IS MONEY
comes to the fore in many linguistic instantiations, such as I spent an hour
with my grandmother yesterday, This flat tire cost me an hour, I lost a lot of
time in the traffic jam and many others, as there is an unlimited number of
expressions that we can use. However, the mental structuring behind all of
these expressions is the TIME IS MONEY concept in our minds. We talk
about time in money terms because time is such an abstract concept that it is
difficult to express directly, so we turn towards a more concrete domain of
knowledge - such as money - and map it onto time.
From a Cognitive Linguistic perspective, metaphors function by mapping
the meaning structure of a more concrete source domain onto the as yet
dimly seen conceptual structure of a more abstract target domain in order to
facilitate understanding of this second domain in some way. Thus, we see a

Authenticated
similarity between two different domains (in our example, money being the
concrete source domain and time being the abstract target domain) and map
characteristics from the source domain onto the target domain. Generally this
does not happen in a conscious way, it is part of our linguistic and cultural
socialization. In a culture where people are not paid by the hour, the month,
or the year, such a mapping would presumably not come into existence.
However, in this respect, there is no big cultural difference between Anglo-
American cultures and the German one, so we find this conceptual meta-
phor in German as well, although the linguistic instantiations are not exactly
the same. Thus we do not "spend" time in German, but we can "Zeit inves-
tieren", "Zeit gewinnen", "Zeit verlieren", "Zeit sparen", "Zeit verschwen-
den" and "ein biBchen Geduld kann sich auch auszahlen".
The TIME IS MONEY metaphor is, of course, not the only way to talk
about time in English, as there are other conceptual metaphors connected to
the target domain of time, depending on what we want to express. We may
also use the conceptual metaphors TIME IS A MOVING OBJECT (in examples
such as The years flew by, or It is getting close to bedtime), or TIME IS A
THIEF (in examples such as The passing of the years has robbed me of my
beauty, or when Milton in Sonnet 7 calls time the subtle thief ofyouth), or
TIME IS A CHANGER (in examples such as Over the last years, I have
become a lot more patient, or Time heals all wounds) or as many other
things (cf. Lakoff/Turner 1989). This quantity of conceptual metaphors is
due, on the one hand, to the fact that time is such an important and salient
concept for us that we need various ways to talk about it and, on the other
hand, to the fact that no culture is homogeneous and we need a choice of
how we want to conceive of time, depending on our social, religious, or
individual backgrounds. However, this choice is not an unlimited one be-
cause we are not free to choose our conceptual metaphors as they are part of
our linguistic and cultural heritage, and we can only choose among those
that exist within the target culture.
If we take the example of "life", a concept that is hard to define but
which is nevertheless very central for us humans, we find many different
conceptualizations. One of Lakoff's best-known examples is LIFE IS A
JOURNEY (Lakoff1980). A journey is a very concrete event which all of us
have some experience with, whereas the more abstract concept of "life" is
hard to describe, apart from a purely biological definition. But once we try
to grasp its meaning by mapping our everyday knowledge of "journeys"
onto the concept of life, it becomes much less obtuse and we may see life as
having a beginning, as pursuing a certain route, as possibly involving side-
tracks and obstacles, and as having a final aim or an end. In this way, we are

Authenticated
able to talk about "life" and make ourselves understood, at least inside the
culture. Even if we only rarely think of life explicitly in terms of a journey,
this conceptual structuring nevertheless exists in our minds, since it comes
to the surface in a number of concrete metaphors.
Even the Bible makes abundant use of this conceptual metaphor, for
instance when it talks about the "good way" and the "evil way" in the Old
Testament. However, different belief systems have chosen to see "life" from
other perspectives, such as the Hindu culture where life is seen as a transi-
tory stage to a better existence (also present in Christian culture in the belief
in a better "life" after death). Lakoff and Turner (1989) quote 17 different
basic metaphors for "life" that are present in the Western culture, which,
according to them, is an astonishingly small number for such an important
concept. The different conceptualizations of "life" mainly rely on the every-
day experiences made in a certain cultural context, which are handed down
from generation to generation. Also, new creative metaphors ultimately rely
on these conceptual metaphors because they make use of the schemata
underlying the conceptual metaphors and extend them in innovative ways
(cf. Lakoffand Turner 1989).
How can this approach be made fruitful for the foreign language class-
room? From a functional point of view, we can state that metaphors are
often used for manipUlative purposes, be it in advertising or in political
propaganda of all kinds. If news coverage on a war or on political unrest
metaphorically frames one party as "the bad guys" and the other one as "the
good guys" and does so by exploiting common conceptual metaphors, we
are normally not even aware of this kind of manipulation, so that we do not
have many possibilities of seeing the events in any other light. There are
examples of this concerning the news coverage on the Gulf War or also the
September 11 event. The manipulative potential of metaphors, then, is one
of the reasons why learners should at least be made aware of the notion of
conceptual metaphors, so that they may have a chance to see as well as react
to the manipulative potential of language, and that they may realize what a
powerful and flexible tool language really is. This is, of course, not only
true for a foreign language but also for the native language, since, via the
alienating effect of realizing how the foreign language works, one may also
arrive at a more realistic view of the native language. Working with
metaphors in the foreign language classroom therefore helps raise the learn-
ers' awareness for structuring principles within language and thought and
for the cultural differences within this structuring.
Metaphors always involve cultural knowledge, which is another reason
to focus on metaphors in the foreign language classroom because they can

Authenticated
raise the learners' intercultural awareness and may be used for a contrastive
approach. We can see this, for example, when learners actively work with
metaphors in comparing conceptual metaphors, as well as their linguistic in-
stantiations, in English and in German texts of any kind and of their own
choice - because metaphors are everywhere! - and compare and discuss
their results. It may also be connected to the analysis of literature or used as
a conscious strategy in creative writing. Working with metaphors also
entails an affective element, as such an approach may raise the learners'
motivation because it is fun to uncover manipulation and may induce learn-
ers to look more closely at language and to decipher "hidden meanings". In
this way, they can also monitor their own language output in a more
meaningful way. The main point to stress is that working with metaphors is
a way to understand how language works, how it changes and how meanings
are created and extended. In this way, learners may develop a new perspec-
tive on language in general.
Conceptual metonymies are even more pervasive than conceptual meta-
phors and, as with metaphors, more often than not people are not even aware
of using them. Metonymies come in a great variety of shapes, each one of
them highlighting another relation between the referent as such and the
metonymy denoting it. In each case, the most salient aspect of the context is
verbalized to evoke the whole concept or scene. To give an example, in the
WHOLE FOR PART metonymy as in "The Rolls-Royce left the gas station
without paying", where it is of course the less salient driver of the more
salient Rolls-Royce who did not pay, we immediately recognize the whole
scenario of events.
As with metaphors, the coding and decoding processes connected to
metonymy are largely determined by extralinguistic experiences and conno-
tations. Therefore, intercultural differences will show up in the existence or
the "congealing" of such metonymies, because different aspects are seen as
the most salient ones, and thus different metonymies are conceptualized.
These different aspects have much to do with the cultural background in
which the language in question is spoken. Thus, when we find expressions
with "green" in English (or closely related cultures) which refer to ecologi-
cal preservation (green party, Greenpeace) via the meaning link of "green"
as a metonymy for nature (green belt, green thumb). This is due to the pre-
dominance of naturally occurring green entities like leaves, plants, or grass
(cf. also Niemeier 1998). However, such metonymies would probably not
develop in cultures where there is not much green in the daily environment.
Making the learners aware of such differences might be a useful inroad to
raising their intercultural awareness. Another reason why the concept of

Authenticated
metonymy may be introduced into the foreign language classroom is to use

it in order to restructure the learners' vocabulary by leading them to dis-
cover new, "foreign" connections between words and concepts, and there-
fore help them to internalize conceptualizations belonging to the foreign
language and culture as well as to oppose those to the conceptualizations
belonging to their own language/s and culture/so
The learning objective of awareness raising in foreign language teaching
may therefore also refer to making learners aware of the culture-relatedness
of metaphorical and metonymic concepts, which may entail a completely new
view of the foreign language for the learners, and thus broaden their horizons
in various ways. With reference to metonymy, we see that different cultures
tend to highlight different aspects of a concept, and these highlighted aspects
then come to stand for the complete concept and are thus able to influence
and guide even our thought processes. This notion may be related to Whorf's
ideas concerning the connection between culture and language. Whorf con-
sidered a language to be inseparably intertwined with a specific world-view.
These different world-views, or cultures, may then influence the way we
think and speak and categorize the world around us. Following up this
thought may lead to the conclusion that metonymies in different languages
concentrate on different salient aspects of the world and thus contain dif-
ferent culturally induced concepts. This means that analyzing metonymies
may present an inroad to culturally induced conceptualizations as such.
Although Whorf mainly concentrated on patterns of grammar as guiding
our thoughts, I would like to make the claim that it is not only patterns of
grammar but also patterns of conceptual and linguistic metaphor and
metonymy, the schemata built up in language as well as many other related
entities that are able to influence our habitual thinking. If we claim that
grammar patterns reflect the underlying culture in an iconic way, we also
have to see that by doing so, they are just one example of iconicity within a
given cultural frame and that other language- and thought-related issues like
those mentioned above behave in much the same way, mainly because they
are all closely interrelated - in Cognitive Linguistics, there is no difference
between, e.g., grammar and the conceptual constructs within lexis, as they
underlie the same structuring principles.
4. Conclusion
It cannot be stressed enough that the methods outlined in the text are
designed - and in my opinion well able - to help learners realize the close

Authenticated
connection between a language and its culture, and to see how the meaning
networks in the foreign language are constructed and how a language adds
new concepts and meanings almost every day. This way of looking at the
foreign language does not just entice the learners to learn concepts by heart,
but it offers them an insight into the ways the language works. Especially
metaphors and metonymies are packed with cultural knowledge. Their
meanings are not ad hoc meanings at all, but they are all motivated by dif-
ferent aspects of the culture in question, and thus able to transfer information
about this culture.
Therefore, using some of the insights and principles of the cognitive
linguistic approach in the foreign language classroom or even - for a start -
inserting cognitive linguistic elements into the "normal" teaching methods
seems to be particularly relevant for vocabulary training and for intercul-
turallearning for advanced learners. Furthermore, the examples mentioned
above are also relevant for grammar teaching, because in grammar we also
find categorization and prototypicality and even metaphors, such as, for
example, metaphorized prepositions (see Niemeier 2001). Furthermore, as I
have tried to show, the methods discussed are compatible with most of the
current didactic theories such as the concern with awareness raising, the
learning objective of intercultural competence, the learner-centered approach
(Le., autonomous learning), multi-channel learning, the tendency towards a
more holistic kind of teaching and learning, the action-oriented approach to
learning and teaching, as well as the focus-on-form approach. What all of
these approaches have in common is that they do not compartmentalize dif-
ferent aspects of the language (as, for example, grammar teaching as
opposed to or at least as more or less unrelated to vocabulary training) but
that they are based upon a concept of language that regards language in a
holistic way as inseparably connected with culture and with people's mental
as well as bodily experiences.
Such an approach to teaching and learning foreign languages seems to be
in line with Whorf's ideas insofar as it helps learners to not try to look for
equations between their native language and the foreign language, but that
they are open for other conceptualizations within the foreign language and
its culture. As pointed out above, a metonymy may arise when a salient
detail in a given domain is taken to represent the whole scenario or domain.
However, the saliency of an object, event, or relation is construed by the
language user and is not objectively or automatically given. The language
user, on the other hand, does not live in a culture-free environment and is
thus influenced by the surrounding culture, if only subconsciously. There-
fore, the meaning extensions are bound to mirror the underlying culture to a

Authenticated
certain degree. For the foreign language learner this means that these under-
lying and possibly contrasting factors - closely bound to the language -
have to be brought to mind, and that possible discrepancies between the
learners' home culture and the foreign target culture have to be dealt with
much more consciously.
Acknowledgment
I would like to thank Rene Dirven (UniversiHit Duisburg) and Robert

MacLaury (Pennsylvania State University) as well as my two reviewers for
their most valuable comments on a first draft of this paper. All remaining
flaws are, of course, entirely my own responsibility.

Authenticated
Notes
1. This text can only account for the German point of view, as educational systems
vary markedly in international comparison.
2. Currently a generally accepted defmition of "intercultural competence" does not
seem to be available, therefore I will not try to give one here (for further discus-
sion, see, for example, Byram 1997, House 1997, House and Edmondson 1998).
My own working definition of "intercultural competence" is an understanding
of the major general differences between the target culture and the home culture,
and thus a prevention of prejudices and a neutral re-evaluation of both cultures.
This working definition is relatively close to Kramsch's "third domain" (see
Kramsch 1993), which, however, seems to me too static a concept.
3. Lee (1997) defines these "isolates of experience" as following: "An isolate of
experience can be figure or ground. An isolate is any phenomenon which we
can abstract away from the matrix of its occurrence as something separately
perceivable or noticeable. Isolates can occur in any sensory domain."
4. The 6th ICLC conference in Stockholm, July 1999, featured three plenaries on
language acquisition and a one-day workshop on "Cognitive linguistics and lan-
guage acquisition" organized by AchardlNiemeierlSinha. At the 7th ICLC con-
ference in Santa Barbara, July 2001, several papers on the topic in question were
presented. Furthermore, the 28th LAUD Symposium (organized by Piitzl
Niemeier) in Landau/Germany, March 2000, was completely devoted to the topic
of "Cognitive Linguistics: Second Language Acquisition, Language Pedagogy,
and Linguistic Theory", and GURT 2003 (Georgetown University, Washington)
has as its general topic "Cognitive and Discourse Perspectives on Language and
Language Learning".
5. However, this is by no means clearly determined. Currently there is research in
progress which tries to fmd out the subtle meaning differences between focal
color terms in a couple of European languages, among them German and English
(personal communication by Robert MacLaury).
6. Again, although this may work even better with languages that are culturally
further apart from each other than English and German are, between these two
languages we can also find salient structural differences, which influences the
language users' conceptualization processes and linguistic output, such as, for
example, the progressive aspect in English verbs, which is not directly marked
on German verbs - or, to use cognitive linguistic terminology, the notion of
boundedness vs. unboundedness in verbs (see Niemeier 2001 and forthcoming).

Authenticated
References
Bach, Gerhard and Johannes-Peter Timm

1996 Englischunterricht. Grundlagen und Methoden einer handlungs-
orientierten Unterrichtspraxis. 2nd ed. TiibingenlBasel: Franck.
Barcelona, Antonio (ed.)
2000 Metaphor and Metonymy at the Crossroads. Berlin: Mouton de
Gruyter.
Berlin, Brent and Paul Kay
1969 Basic Color Terms: Their Universality and Evolution. Berkeley:
University of California Press.
Bickel, Balthasar
2000 "Grammar and social practice: On the role of 'culture' in linguistic
relativity". In: Susanne Niemeier and Rene Dirven (eds.), 161-191.
Buttjes, Dieter and Michael Byram (eds.)
1991 Mediating Languages and Cultures. Towards an Intercultural Theory
ofForeign Language Education. Clevedon: Multilingual Matters.
Byram, Michael
1997 Teaching and Assessing Intercultural Communicative Competence.
Clevedon: Multilingual Matters.
Dirven, Rene
1993 "Metonymy and metaphor: different mental strategies of conceptu-
alization". Leuvense Bijdragen 82, 1-28.
Dirven, Rene and Marjolijn Verspoor
1998 Cognitive Exploration of Language and Linguistics. Amsterdam:
Benjamins.
Doughty, Catherine and Jessica Williams (eds.)
1998 Focus on Form in Classroom Second Language Acquisition.
Cambridge, UK: Cambridge University Press.
Doye, Peter
1996 "Priifung der Handlungskompetenz durch pragmatische Tests". In:
Gerhard Bach and Johannes-Peter Timm (eds.), 192-209.
Ellis, Nick C. (ed.)
1994 Implicit and Explicit Learning of Languages. London: Academic
Press.
Gnutzmann, Claus
1997 "Language awareness". Praxis des neusprachlichen Unterrichts
3/97,227-236.
Gumperz, John J. and Stephen C. Levinson (eds.)
1996 Rethinking Linguistic Relativity. Cambridge: Cambridge University
Press.
Heider, Eleanor
1972 "Probabilities, sampling and the ethnographic method: The case of
Dani colour names". Man 7 (3),448-466.

Authenticated
House, Juliane
1997 "A contrastive approach to developing intercultural competence".
Sprak og Marked, 15,3-10.
2000 "Linguistic relativity and translation". In: Martin Piitz and Marjolijn
Verspoor (eds.), 69-88.
House, Juliane and Willis Edmondson
1998 "Interkulturelles Lemen - ein iiberfliissiger Begriff". Zeitschrijt for
Fremdsprachenforschung 9: 2, 161-188.
James, Carl and Peter Garrett (eds.)
1995 Language Awareness in the Classroom. London: Longman.
Kramsch, Claire
1993 Context and Culture in Language Teaching. Oxford: Oxford Univer-
sity Press.
1980 Metaphors We Live By. Chicago: University of Chicago Press.
Lakoff, George and Mark Turner
1989 More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago:
University of Chicago Press.
Lamb, Sydney M.
2000 "Neuro-cognitive structures in the interplay of language and thought".
In: Martin Piitz and Marjolijn Verspoor (eds.), 173-196
Langacker, Ronald
1991 Concept, Image and Symbol. The Cognitive Basis of Grammar.
Berlin: Mouton de Gruyter.
Lee,Penny
1997 "Language in thinking and learning: Pedagogy and the New Whorfian
framework". Harvard Educational Review 67: 3,430-471.
van Lier, Leo
1996 Interaction in the Language Curriculum. Awareness, Autonomy and
Authenticity. London: Longman.
Little, David
1994 "Learner autonomy: A theoretical construct and its practical appli-
cation". Die Neueren Sprachen 93, 430-442.
L6ffier, Renate
1996 "Ganzheitliches Lemen: Grundlagen undArbeitsformen". In Gerhard
Bach and Johannes-Peter Timm (eds.), 42-68.
Lucy, John A.
1992 Language Diversity and Thought. A Reformulation of the Linguistic
Relativity Hypothesis. Cambridge: Cambridge University Press.
1997 "Linguistic relativity". Annual Review ofAnthropology 26,291-312.
Lucy, John A. and R.A. Shweder
1979 "Whorf and his critics: Linguistic and noniinguistic influences on
color memory". American Anthropologist 90, 923-931.
1988 "The effect of incidental conversation of memory for focal colors".
American Anthropologist 33, 137-146.

Authenticated
McCarthy, Michael
1991 Discourse Analysis jor Language Teachers. Cambridge, UK:
Cambridge University Press.
Niemeier, Susanne
1998 "Colourless green ideas metonymise furiously". In: Friedrich Ungerer
(ed.), Kognitive Lexikologie und Syntax. Rostocker Beitriige zur
Sprachwissenschaft 5, Rostock: University ofRostock, 119-146.
2001 "Zur Beziehung zwischen Lexik und Grammatik". In: Dagmar
Abendroth-Timmer and Gerhard Bach (eds.). Mehrsprachiges
Europa. Tiibingen: Narr, 305-326.
Forthcoming
"BoundednesslUnboundedness - Blick durchs Schliisselloch. Ange-
wandte Kognitive Linguistik fur den Englischunterricht".
Niemeier, Susanne and Rene Dirven (eds.)
2000 Evidencefor Linguistic Relativity. Amsterdam: Benjamins.
Palmer, Gary.
1996 Toward a Theory of Cultural Linguistics. Austin: University of Texas
Press.
Panther, Klaus-Uwe and Giinter Radden (eds.)
1999 Metonymy in Thought and Language. Amsterdam: Benjamins.
Pica, Teresa
2000 "Tradition and transition in English language teaching methodology".
System 28: 1-18.
Piitz, Martin and Marjolijn Verspoor (eds.)
2000 Explorations in Linguistic Relativity. Amsterdam: Benjamins.
Rosch, Eleanor
1975 "Cognitive representations of semantic categories". Journal ofEx-
perimental Psychology: General 104, 192-233.
1978 "Principles of categorization". In Eleanor Rosch and Barbara B.
Lloyd (eds.), Cognition and Categorization. Hillsdale: Lawrence
Erlbaum, 27-38.
Selinker, Larry
1992 Rediscovering Interlanguage. London: Longman.
Timm, Johannes-Peter (ed.)
1995 Ganzheitlicher Fremdsprachenunterricht. Weinheim: Deutscher Stu-
dien Verlag.
Wendt, Michael
1996 Konstruktivistische Fremdsprachendidaktik. Lerner- und handlungs-
orientierter Fremdsprachenunterricht aus neuer Sicht. Tiibingen: Narr.
Whorf, Benjamin L.
1956 "The relation of habitual thought and behavior to language". In John
(1941) B. Carroll (ed.), Language, Thought, and Reality: Selected Writings
by Benjamin Lee Whoif. Cambridge, Mass.: MIT Press, 134-159.
Wolff, Dieter
1994 Lernerautonomie. Themenheft: Die Neueren Sprachen 93(5).

Authenticated
The Figure / Ground Gestalt and Language
Teaching Methodology
Peter Grundy
1. Introduction
This chapter reviews the ways in which the figure / ground gestalt has been
applied in linguistics and considers its relevance to second language teaching
and learning. In particular, it is argued that the figure / ground gestalt under-
mines the discrete-item-syllabus assumption on which instructed second
language acquisition is typically based and that it is precisely its figure /
ground properties, especially as revealed in its metapragmatic dimension,
that make language comprehensible, and thus learnable.
2. Language Teaching Methodology in the Pre-cognitive Paradigm
Until relatively recently, conventional wisdom had it that a foreign language

needs to be taught (the teachability assumption), that we learn a language in
order to use it (the learn-to-use assumption), and that what has to be taught
and learnt can be described in a syllabus consisting of paradigm lists of dis-
crete phonemes, morphemes, syntactic structures, and lexical items (the dis-
crete-item-syllabus assumption). This conventional wisdom underlies not
only grammar / translation but also later methods, including Reform Move-
ment Direct Method, Situational Language Teaching, Audiolingualism, and
even much Communicative Language Teaching.
Although in recent years methodologists have increasingly questioned the
teachability-of-Ianguage and the learn-to-use assumptions, the discrete-item-
syllabus assumption is still largely unchallenged. As I shall show, however,
this assumption is difficult to sustain in the face of the figure/ground
gestalt.

Authenticated
120 Peter Grundy
2.1. The Teachability Assumption: Generative and Cognitive Positions
The commonsensica1 view that language, like any other "subject", is teach-
able may have some credibility in pedagogy, but has been largely rejected
by both generative and non-generative applied linguists who have come to
argue over the last ten or fifteen years that language is a learnable behaviour
rather than a teachable subject. Writing from a generative perspective,
Wong-Fillmore puts the position clearly: "In current theories of language
'Iearnability', a principle of 'teach-ability' is unnecessary" (1989: 311).
For generativists such as Wong-Fillmore, this is a natural fallout of the
innatist view that language acquisition is biologically determined and thus a
system-internal phenomenon. This position is supported by the good results
obtained by generativists working with learner data which demonstrate that
learners appear to know more than they have been taught (Bley-Vroman,
Felix and loup 1988).
For non-generative, pedagogically oriented applied linguists, the privileg-
ing of learner discovery over teacher instruction and of learner self-expression
over the memorization of model structures has also promoted the learner as
the significant determinant of acquisition. As Crandall writes, "Since the
mid 80s, discussions of effective language instruction have shifted from an
emphasis on teacher-centred to an emphasis on learner-centred classrooms
and from transmission-oriented to participatory or constructivist knowledge
development" (1999: 226).
Rejecting teachability in favour of learnability on the grounds that lan-
guage acquisition is system-internally determined, a reflex of a language
dedicated module in the mind, enables us to speculate on the extent to which
universal grammar is available to post-puberty language learners and on the
influence of prior knowledge on second language acquisition. But substituting
a learnability for a teachability hypothesis does not tell us how the acquisition
process can be assisted in the classroom, beyond vague claims about the
need for comprehensible input and a low affective filter setting (Krashen
1985). Yet we need to know not just that language acquisition processes are
determined by cognitive structures and likely to produce output data accord-
ingly, but, more to the point, how the language input data needs to be pre-
sented to these cognitive structures for successful acquisition to occur.
If the structure of the brain is uniquely adapted to understanding the
nature of the phenomena it encounters in the world, including linguistic phe-
nomena, there is a sense in which it is not just the learner, but rather language
itself, which teaches the learner language. This view obliges us to look at
language not as a counterpart to innate biological structures, but as one of a

Authenticated
set of phenomena whose construction is uniquely processible by the human

mind because of the way in which these phenomena are perceived.
One hypothesis as to the internal structure of observable phenomena
which has influenced cognitive linguists, most notably Langacker (1987,
1991a), is the figure / ground gestalt. This gestalt, which was originally
applied to visual perception, holds that phenomena exhibit figure / ground
properties that make them uniquely comprehensible to us. As we shall see
in the following sections, this insight has significant implications for second
language acquisition.
2.2. The Learn-to-use Assumption: Communicative Competence

and Cognitive Linguistics
In A History of English Language Teaching, Tony Howatt makes the neat

point that in "strong" versions of communicative language teaching, learners
use language in order to learn it, whereas conventionally methodologists
have taken it for granted that we learn language first in order to use it later
(1984: 279).
Although the learn-to-use assumption has lost much of its credibility
among methodologists, we should note, however, that both use-to-Iearn and
more traditional learn-to-use teaching methods presuppose a distinction
between the formal properties of language, or what has to be learnt, and their
function, or use.
The case of contemporary communicative language teaching provides an
illustration of the way in which form and function are divorced. Communi-
cative language teachers typically take as the rationale for their methodology
Hymes's argument that Chomsky's account of what is formally possible
needs to be enriched by a second kind of competence which accounts for our
knowledge of how and when to use language (1971). Hence the dual formal /
functional competence model that underlies contemporary language teaching
methodology.
Others have suggested that communicative language teaching derives its
rationale from the kind of functionalism represented by Jakobson (1960) or
Halliday (1976). (See Richards and Rodgers 2001, for a typical example of
this argument.) However, such interactionalist accounts of language do not
claim that form and function are one and the same thing. Rather, they main-
tain the dualism by arguing that the form of a text is sensitive to its commu-
nicative function.

Authenticated
122 Peter Grundy
The importance of cognitive linguistics for language teaching methodology

is that it provides a usage based model (Langacker, 1987: 46) which does not
distinguish the abstract or symbolic nature of language from the occurrence
oflanguage. In a cognitively based functional account of human language, a
construction is a complex category with semantic, phonological, and sym-
bolic properties. Thus it makes much less sense in the cognitive than in the
generative paradigm to discuss whether we learn language in order to use it
or use language in order to learn it, since in the cognitive paradigm our
knowledge of constructions is a unified knowledge of their semantic, phono-
logical, and symbolic properties.
2.3. The Discrete-item-syllabus Assumption
While second language acquisition research has made the teachability

assumption untenable, and "strong" versions of communicative language
teaching have increasingly supplanted the learn-to-use assumption, the third
conventional wisdom of language teaching methodology, the discrete-item-
syllabus assumption, still underlies virtually every contemporary course-
book, whether in structural, notional, or functional guise. Thus it is virtually
de rigueur for contemporary "international" coursebooks published in UK
to begin with what is often called "a map of the book". This map lists the
"units" which constitute the book, each of which typically contains a broadly
described functional category (e.g., meeting people, finding out, etc.) accom-
panied by sets of structures (e.g., present simple, questions with "do" and
"can") and notions (likes and dislikes, direction, and location) which will be
encountered in the unit. Although coursebook writers intend the functions,
structures, and notions presented in each unit to match so that, for example,
expressing likes and desires will involve using the present simple, each
coursebook is essentially a compilation of discretely identified structures
and functions linked to one another with varying degrees of subtlety or arbi-
trariness. One might argue that the atomistic view of language which
informed those traditional methods that began language instruction with
minimal units, i.e. phonetic segments, and gradually built up to lexical
items and sentences is still present in contemporary international course-
books, but that decontextualised functional primitives have replaced the
phonetic primitives of the earlier approach.
While the discrete-item-syllabus may enable learners to produce and
understand a range of structures, it seems to not work at any deep level,
seemingly because it presents language without a real context, and therefore

Authenticated
without the background in relation to which it is meaningful. Without ground

it is very difficult to distinguish figure. Since discrete-item-syllabuses present
language as a series of "types" or "instances", i.e., figures without ground,
rather than "grounded instances" (Langacker 1991a: 33), they are likely to
fail for most learners.
What kind of methodological alternative to the discrete-item-syllabus
does the figure / ground gestalt suggest, then?
3. The Figure / Ground Gestalt
The insight that figures are associated with discreteness, shape, and singular-
ity, whereas diffuseness is the principal characteristic associated with ground,
is owed to the pioneering experimental work of Rubin (1915) in the field of
visual perception. What we "see" is the figure. We assume that the contour
marking the boundary of figure and ground belongs to the figure and not to
the ground against which the figure is, consequently, profiled. Figures give
an impression of solidity, closeness, and density of colour relative to ground.
The ground appears to continue uninterrupted behind the figure. It is the
figure and not the ground which is remembered.
In relation to the "actual" world, these perceptions are, strictly speaking,
illusions. They are, therefore, direct reflections of cognitive processes which
impose, even on a flat surface such as the page you are now reading, the
impression that some parts of what you see, in this case the graphemic sym-
bols, are closer to you than the background (I cannot avoid the metaphor) on
which they are printed. The salient shapes to which your attention is drawn
are those of the symbols and not of the background page. Moreover, you can
easily reproduce the symbols, but would have immense difficulty reproducing
the shape which shares a common contour with them.
The issue then is whether linguistic phenomena too have salience in a
figure / ground manner and whether the figure / ground gestalt has implica-
tions for language learning and teaching.
3.1. Figure / Ground Accounts of Linguistic Phenomena
The figure / ground gestalt has been appealed to at a number of linguistic

levels. One of the earliest applications was in the Prague School's structuralist
characterisation of poetic language. This is especially clear in the work of
Mukarovsky. Although he does not refer to Rubin or specifically to the

Authenticated
124 Peter Grundy
figure / ground gestalt, Mukarovsky's work and that of his colleagues never-
theless clearly presupposes such a gestalt. For Mukarovsky, poetic language
consists of the fore grounding of phonological, syntactic, or even semantic
features, whose resulting prominence tends to push meaning into the back-
ground. Thus the formal means of expression rather than what is expressed
is the salient figure. What we notice in the case of poetic writing is not what
is conveyed but how it is conveyed.
Working in a different tradition and in a sense anticipating the work of
cognitive linguists, Wall ace (1982) attempts to draw together and expand on
the considerable body of existing descriptive work on linguistic categories
which, he argues, demonstrate the figure / ground polarity. He observes that
languages have a range of contrasting grammatical forms (perfective / im-
perfective, eventive / non-eventive, definite / indefinite, etc.) which are pre-
dominantly oriented to expressing figure / ground relations. Thus, Wallace's
use of figure / ground is oriented to the explanation of discoursal phe-
nomena: a speaker has the means at each point in a discourse to foreground
some element of propositional meaning in relation to some other element.
The most influential application of the figure / ground gestalt in linguis-
tics, however, is found in Langacker's work where it underlies the profile /
base relation, one of the fundamental principles on which the cognitive,
usage based account oflanguage rests:
A predication always has a certain scope, and within that scope it selects a
particular substructure for designation. To suggest the special prominence of
the designated element, I refer to the scope of the predication and its desig-
natum as base and profile [Langacker's emphasis] respectively. Perceived
intuitively, the profile (in the words of Susan Lindner) "stands out in bas-
relief" against the base. The semantic value of an expression resides in neither
the base nor the profile alone, but only in their combination; it derives from
the designation of a specific entity identified and characterized by its position
within a larger configuration (Langacker, 1987: 183).
The profile / base relation is associated in Langacker's work with a single

morphological unit, or predicate. In a grammatical relation, there is also a
fore grounded entity, or trajector, which is profiled in relation to a landmark
(1991b: 301). As the term trajector suggests, the profiled figure may be
dynamic rather than static.
It is not the purpose of this section to survey the considerable range of
work on the application of the figure / ground gestalt to language understand-
ing. However, it is striking that, in both the cognitive and in other paradigms,

Authenticated
the figure / ground properties of pragmatic phenomena have often been noted.
In Talmy's account of complex sentences, for example, figure is seen as a
variable (1978: 628ft) in relation to a ground provided in the subordinate
clause. Thus clauses introduced by after and before, and phrases introduced
by during provide the ground, or presupposition, in relation to which the rest
of the sentence is seen as a figure. This assertion-as-figure / presupposition-
as-ground position is also taken up by Levinson (1983: 180).
Even more strikingly perhaps, at the level of the word, Hanks (1992) argues
that deictics among pragmatic phenomena uniquely capture the relation offig-
ure to ground in a single linguistic expression. Thus what a demonstrative
points to as a figure in an expression like you or this year is related to the
indexical ground or deictic anchoring point of the speaker (and sometimes
of the hearer) at the time when and in the place where the utterance occurs.
Each of these accounts, whether seeing ground as co-text (Langacker,
Talmy) or context (Hanks) or as a linguistic option which is not chosen
(Mukarovsky, Wallace), demonstrates the linguistic reflexes of a fundamen-
tal processing strategy in which salience is perceived as a relation of figure
to ground. And in so doing, the accounts generalize a perceptual theory to
the understanding of language. However, what they do not do is explain
how the grounding process itself occurs. It is to this problem that we turn in
the following section.
3.2. Intrinsically Grounded and Metapragmatically Grounded Instances
In Langacker (1991a), grounding is defined as
A semantic function that constitutes the final step in the formation of a nomi-
nal or a finite clause. With respect to the fundamental "epistemic" notions
(e.g., definiteness for nominals, tense/modality for clauses), it establishes the
location vis-a-vis the ground of the thing or process instance designated by
the nominal or the clause overall (1991 a: 549).
Thus we are able to distinguish as intrinsically grounded full nominal con-

structions or finite clauses in which types or process types become instances
distinguishable from others and related through definiteness (for nominals)
or tense / modality (for finite clauses) to the speaker / hearer's epistemic
perspective.
The question that then arises is how the notion of grounding as a seman-
tic / pragmatic function that applies in the evolution of a construction can be

Authenticated
126 Peter Grundy
extended to include the sense in which a sentential assertion, to take one

construction, is grounded by the presupposition provided by the temporal
clause to which it is linked. Or, more broadly, how it is that the "speech
event, its participants, and its immediate circumstances (such as the time
and place of speaking)" constitute the ground against which speech acts are
constl1led (Langacker 1991 b: 318).
In a footnote accompanying his discussion of deictics, Langacker some-
what obscurely writes: "The convenience of ground as a shorthand reference
to the speech situation leads me to employ the term in this sense despite its
possible confusion with the sense opposed to figure (the two are not unre-
lated)" (1987: 126).
How then are instances of language use grounded in the background
speech situation when this clearly does not occur as the final step in the for-
mation of a constl1lction?
I want to suggest that this is the role of metapragmatic phenomena which
specifically draw attention to the ways in which occurrences of languages
are related to earlier occurrences and to pragmatically defined context
generally.
Metapragmatic phenomena have been very widely discussed over more
than twenty years in the linguistics literature. One of the earliest approaches
in the pragmatics literature which attempts a unified theory is that of
Blakemore (1987). Working in the relevance theoretic model, Blakemore
draws a distinction between what are termed conceptual and procedural
encodings. Thus, in Relevance Theory, discourse particles are treated as con-
straints on interpretation which limit the contexts in relation to which utter-
ances are interpreted. As Wilson and Sperber put it, utterances contain
"information about the representations to be manipulated [conceptual
encodings], and information about how to manipulate them [procedural
encodings]" (1993: 2). This (latter) computational category is thus held to
constrain the interpretation of conceptual meaning by limiting the available
contextual ground in relation to which it is interpreted. An illustration may
help.
To say "Presidents have private lives" is to encode a conceptual meaning.
But when Bill Clinton said "Even presidents have private lives" (18 August
1998), he was encoding not only a conceptual meaning but through the use
of "even" was constraining the background context against which the con-
ceptual meaning was to be interpreted by implying that the (explicated)
proposition that presidents are entitled to private lives would be low on a
scale of expectability. The point is that it is much more difficult to know in
what way the would-be figure "Presidents have private lives" could be rele-

Authenticated
The Figure I Ground Gestalt and Language Teaching Methodology 127
vant (i.e., what might be meant by uttering it) when recovery of the set of
grounding propositions (aI, a2, a3... a n have private lives), is not triggered by
the procedural "even". Thus procedural encodings provide the hearer with an
indication of how to provide contexts in relation to which the implicatures
conveyed by conceptual meanings may be inferred. Put simply, they enable
a hearer to recover a context or ground in relation to which a sentence or an
utterance can be a figure.
The discussion in the previous paragraphs of the way grounding is handled
in Relevance Theory provides a particularly clear account of the role of just
one category of metapragmatic grounding phenomena, discourse particles.
In a survey paper, Verschueren (2000) lists many other metapragmatic means
of grounding, including metapragmatic descriptions in the form of speech
act and performative verbs, self-referential expressions, sentence adverbs,
hedges, deictics, evidentials, aspect, modality, prosody, code switching, and
implicit voices. Verschueren argues that metapragmatic phenomena should
not be regarded merely as linguistic objects (as the list above might suggest)
but rather as a dimension of language, so that the way any proposition is
conveyed provides evidence of the metapragmatic awareness of the speaker
and of the manner in which the proposition is grounded in the context in
which it occurs.
4. The Figure I Ground Gestalt, the Discrete-item-syllabus and Meta-

pragmatic Grounding in Real Talk
In this section, the function ofmetapragmatic grounding phenomena in talk-

in-interaction will be discussed in relation to the use of dialogue in pedagogic
materials.
Despite the fact that the need for a discrete-item-syllabus never occurs to
the parents of children at the language acquisition stage, belief in the efficacy
of such syllabuses is very enduring in instructed second language acquisition.
Thus it is a widely held teacher belief that, because we can describe some of
the regular patterns of a language, we know how to teach that language and
know therefore how languages are learnt. And when we see how little this
belief leads to when operationalised, we tend to put it down to the nature of
the description - thus structural syllabuses are currently passe while the lex-
ical bandwagon with its roots in corpus linguistics is very much of the
moment.
For learning to be possible, we need figure and ground and not merely
the decontextualized items provided by the discrete-item-syllabus. To know

Authenticated
128 Peter Grundy
what is denoted, we need not only a demonstratum but also an origo. Without
presupposition, whether semantic or pragmatic, assertions and other speech
act types are just that, types and not grounded instances, and thus not readily
interpretable. Perfectives need an imperfective discourse for their effect.
Speech acts require their containing speech events. And conceptual mean-
ings frequently require the grounding procedural encodings which enable
the recovery of the base in relation to which their relevance is understood.
There is a sense in which it might be said that without ground, there is no
figure. While this does not seem a very controversial claim to a cognitive
linguist, it seems to not have been widely accepted in language teaching
methodology.
All too often language teaching materials consist of "bare" instances of
language types or "bare", ungrounded dialogues. Although Text A below is
not, as will be seen, an authentic coursebook dialogue, it will serve as
broadly representative of the type. You need to know that the first speaker,
D, is the warden of a student hall of residence and that the anaphor "it" in
his first turn refers to the theft of some butter, while the second speaker, N,
is a student and that the same anaphor in her first turn refers to the butter
that was stolen:
Text A
D Do you know when it happened?

N It was there on Friday and when I went to get it yesterday it wasn't
there.
D I can have a word with people if you want. Do you want to have a
word with everyone, or do you want me to put a sign up, or. .. ?
N A sign'll do.
It is instmctive to compare this exchange with the exchange that actually

occurred when the warden, D, meets two students, Sand N, by chance one
evening in the hall of residence:
Text B
D how's it doing
S yeah no erm Nicole wants to no we need a word with you
D oh Jight_
N _no _ someone's been nicking stuff out the fridge (2.0) so
D which one

Authenticated
The Figure I Ground Gestalt and Language Teaching Methodology 129
N ours (.) well Susie's butter's gone and my cheese has

gone as well
D right _yeah_
N _like _ I opened new packets of stuff so=
D =right the whole lot
N yeah
D well do you know when it happened
N erm (.) right it was there on Friday and when I went to
get it yesterday it wasn't there so
S yeah
D right erm (2.0) there not a lot (.) can do about it I
can sort of like hh (2.0) have a word with people if
you want (1.5) erm hh (3.0) I mean (.) do you do you
want to SOlt oflike (1.0) have a word with everyone or
just (.) do you want me to put a sign up or
(1.5)
N yeah a sign'll do won't it
D's opening inquiry "how's it doing" is answered with an acknowledgement,

"yeah", followed by metasequential "no" whose function is to indicate that
the upcoming utterance is not a conventional small-talk second-pair part; in
fact, it will have the status of official trouble-reporting. S's first attempt at
shifting method (short for "member's method", as in Conversation Analysis
and intended to indicate an expectable conversational routine) to trouble-
reporting ("Nicole wants to") is immediately repaired to the official self-
referential register for trouble-reporting ("we need a word with you"), with
the repair marked by metapragmatic "no". Without these grounding proce-
durals, it would be much more difficult to understand the conceptual content
because this content would no longer have the figure properties provided by
the ground which the procedurals make accessible.
Moving on to a comparison of Text A with the counterpart exchange in
Text B (lines 13-23), we notice first that D's turn at line 13 opens with "well".
Rather than respond to N's invitation at line 10 to formulate a trouble-
response, D asks a potentially dispreferred question. This question is dispre-
ferred as a method at this stage in the talk because it is less procedurally
consequential or outcome oriented than the expectable formulation would
have been, and dispreferred as a question because it lacks felicity since D
knows that N will only be able to provide a very general answer. Thus
"well" at line 13 has an obvious grounding function and makes the speaker's
turn more easily interpretable as D signals that the upcoming question is

Authenticated
130 Peter Grundy
dispreferred, i.e., not fully appropriate to its sequential context, thereby

enabling the other participants in the talk to see what D means by asking it.
When D does eventually propose a set of possible outcomes for Sand N
to choose from, they are set against a further ground which is inferred from
his use of the procedurals "sort oflike", "I mean" and "just". These proce-
durals support the reflexive comment with which his turn opens (line 17) to
the effect that there is not a lot he can do. And when N picks the option that
D should put a sign up (although only after allowing D further time to sug-
gest a possibly more appropriate outcome), she prefaces her choice with
metasequential "yeah" to indicate her preparedness to make a choice from
among the set offered to her. And she invites her friend, S, to agree to her
choice with the tag "won't it" (line 23).
Real talk such as Text B contains regulatory (Chafe 1994), metalinguistic,
metapragmatic, metasequential, and generally reflexive comments which
enable a series of grounds to be constructed against which the propositional
content of the various utterances can be interpreted. These meta-functions
constrain the contexts which are invoked as ground. Without them, it would
be very difficult to find a sequential ground and therefore the propositional
comments themselves would have no background in relation to which their
relevance or salience would be apparent.
Although more economical in terms of language used and therefore peda-
gogically tempting to a materials writer, the doctored Text A is notably more
difficult to interpret than the original Text B from which it is derived. This is
because it lacks the metapragmatic cues which enable the participants to
construct the background in relation to which the propositional content is to
be understood.
The purpose of this comparison of text types is to show how a pragmati-
cally grounded talk exchange is intrinsically more learnable than the kind of
decontextualized dialogues typically found in coursebooks.
5. Grounded Language and Learnability: Explicitly Encoded and

Implicitly Conveyed Ground in Instructed Second Language
Acquisition
The previous section focused on a materials development issue and suggested

that preferred language teaching materials are those which exhibit a base/
profile relation. In this section, we turn to the ways in which instructed
language learning might acknowledge the role of the figure/ground gestalt
in language understanding.

Authenticated
A first question to consider is the level at which the figure/ground relation is

especially salient. Sometimes this occurs even with discrete lexical items,
typically in cases where an item encodes a relation between the lexicalized
figure and the assumed or implicit ground which remains unlexicalized, as
in the case of items such as limb, island, fence, handle, and bridge which
we might hypothesise to be inherently easier to learn than semantically cog-
nate items such as body, land, post, key, and road, where the figure/ground
relation is arguably less salient. As soon as we consider constructions of any
complexity, however, we expect to find both trajector and landmark (i.e.,
profile and base) explicitly lexicalized, as in the end of term. the London
eye, or at this point in time.
That being said, it will inevitably be the case that it is not so much in these
items and constructions as in continuous discourse that the relationship of
each succeeding item to its ground is most clear. Thus constructions rather
than discrete lexical items, sentences rather than constructions, text rather
than sentences, and, pragmatically speaking, speech events rather than
speech acts will be the preferred units of comprehension and production.
In the remainder of this section I will first briefly elaborate some of the
kinds of structure, usually characterized pragmatically, which a skilled
teacher can use to make salient, usually in the form of lexical realization
rather than inference, the ground in relation to which the figure is to be
understood. I will then focus in some detail on the relevance of indexicality
to language teaching and learning. The stress on pragmatics is relevant
since pragmatic meanings are either contextualizing (as when a presupposi-
tion provides the ground in relation to which an assertion is profiled), con-
textualized (as when implicatures result from understanding what is implied
by an utterance in relation to the context or ground judged appropriate), or,
as in the case of deictics, simultaneously conveyed context (or ground) and
denotatum (or figure).
5.1. Contextualizing Pragmatic Phenomena: Presupposition
It seems likely that Talmy's characterization of temporal clauses as grounding

presuppositions, as discussed in Section 2, can be extended to all construc-
tions that are held to give rise to presuppositions, since potential presuppo-
sition triggers could be viewed as a means of indicating a ground in relation
to which a figure is profiled. Thus iteratives ground the present assertion in
relation to previous instances of the phenomenon asserted, while change-of-
state predicates ground the present assertion in relation to previous instances

Authenticated
132 Peter Grundy
or non-instances of the phenomenon asserted. Implicative verbs ground

what they assert to have occurred in relation to the context which the verb
typically presupposes, so that in the case of the implicative verb manage,
the ground in relation to which the asserted action occurred is understood to
have been problematic or challenging. In the case of factive predicates, the
states of affairs which is presupposed as subject or complement sentences is
the ground in relation to which the factive predicate is profiled. Similarly,
pseudo-cleft constructions indicate a ground in relation to which a figure
can be profiled.
Because such constructions encode context explicitly, they provide a
ground which enables comprehension of the accompanying profiled asser-
tion. This is because the assertion is more salient in relation to the context or
ground provided. From a language learning point of view, this ground is a
vital clue to the meaning of the asserted proposition. Thus from a compre-
hension perspective, the contrast between an as yet unlearnt factive and an
as yet unlearnt non-factive predicate followed by an identical complement
sentence is precisely that the learner can take a good guess at the likely
meaning of the factive predicate because its meaning is predictable in relation
to the ground provided by the complement sentence. However, in the case
of the non-factive predicate, the predicate and the complement sentence are
together taken as a complex assertion. Much the same could be said of itera-
tives which, by creating a context of a previous instance, make it easier to
understand a proposition precisely because it is rendered characteristic or
repeatable by virtue of becoming the ground for a repeated instance.
5.2. Contextualized Pragmatic Phenomena: Implicature
While implicitly conceptualized ground may be associated with a small

number of lexical items such as limb, as noted above, very often the most
salient understandings of utterances are pragmatically conditioned implicit
meanings, or implicatures. Implicatures are inferences which a speaker cal-
culates an addressee will recover and which differ in propositional form
from what is actually stated. Such inferences can only be recovered if suffi-
cient context or ground is supplied by the addressee from encyclopedic
knowledge to supplement the utterance, which itself becomes part of the
ground in relation to which the inference is a figure. As Sperber and Wilson
say, "The speaker must make some assumptions about the hearer's cognitive
abilities and contextual resources, which will necessarily be reflected in the
way she communicates, and in particular in what she chooses to make explicit

Authenticated
or what she chooses to leave implicit" (1995: 281). Thus a question such as
Are you working this afternoon? might be understood as a request for a lift,
as conveying that the speaker did not expect the addressee to be working, or
indeed as virtually anything else that the utterance Are you working this
afternoon? taken together with the non-linguistic context or ground might
seem to support (Grundy 2000: 72). Thus the utterance itself is part of the
ground. Yet teachers are frequently anxious to avoid implicature, preferring
instead to deal only in entailment.
In a pedagogic situation, the contextual resources or ground are supplied
by an addressee who is also a language learner. The teacher who is able to
accurately estimate the contextual resources available will therefore be able
to test the learner's understanding of what is said if the learner demonstrates
this understanding by drawing the appropriate inferences, as demonstrated
by an ability to sustain the exchange.
The reality, however, is that speakers often use procedural encodings
alongside conceptual encodings to indicate to the addressee how a profiled
utterance is to be related to a previous part of the discourse, which is there-
fore to be treated as ground. Or where the available range of contexts might
be very great, a procedural encoding can be used to constrain the likely con-
texts that might be considered, which once again assists the addressee in
determining the ground in relation to which an utterance is to be understood.
Procedural encoding may also function to indicate how an utterance is to be
treated as a ground for the following utterance or turn - the Japanese sentence
final particles ne, yo, and yone are notable examples of procedurals which
function in this way. In recognizing the important function of procedural
encodings as grounding devices, we now see that they enable the right inter-
pretation of an utterance by pointing the addressee to the ground required
for understanding it. Thus, as we saw when we considered the talk exchange
in the previous section, exchanges without regulatory metalanguage such as
Text A are much more difficult to interpret than exchanges with a rich
accompaniment of metalinguistic cues. This is difficult for materials devel-
opers to accept because, in their natural focus on what is to be learnt, they
tend to see only figure and to overlook that without ground the figure is not
readily interpretable.
5.3. Contextualizing and Contextualized Pragmatic Phenomena: Deictics
In the final subsection of this part of the paper I will argue that as simulta-
neous encodings of figure and ground, indexicals should be prominent in a

Authenticated
134 Peter Grundy
language learning syllabus. This is already widely recognized in both com-

municative language teaching and in the wider range of approaches styled
as "humanistic" where learners are encouraged to anchor talk around their
own personal, spatial, and temporal base. It is notable that learners take very
easily to this sort of work since the underlying pragmatic dimension grounds
referential language in a way with which they are already familiar.
In relation to spatial grounding, learners can be provided with language
activities which encourage them to explore not only universal grounding
such as that implicit in the use of left and right but also culturally related
grounding such as that implicit in expressions like near-side and off-side; to
distinguish deictic and intrinsic uses of expressions like local (as in Murder
in New York - local man arrested) and last (as in last train); to explore
cases where the ground may shift from speaker to hearer (as in behind the
desk); and to explore the subtle differences between target language and
source language encodings of ground in movement descriptions as encoded
in predicates such as come, go, take, and bring. Furthermore, learners can
be encouraged to make non-Iocational or "blind" use of spatial deictics in
psychologically motivated encodings of emotional distance which explore
the difference between saying, for example, this is a problem and that s a
problem.
Because spatial grounding is visually monitored (hence the use of "blind"
to describe metaphorical extensions at the end of the previous paragraph), it
is much easier for the teacher to set up learning activities which draw
ground as well as figure into what the learner takes conscious account of
than in the case of temporal grounding. In temporal grounding, there is no
obvious counterpart to visual monitoring, so that expressions like tomorrow,
last week. and next month are barely perceived to be grounded in relation to
the time of utterance. One area in which this has significant implications for
second language learning, in English at least, is in the use of the perfect
aspect, a form with which language learners notoriously have difficulty,
often using it in contexts where the past would be appropriate. The resulting
anomalous utterances, such as I have come yesterday, seem to be a conse-
quence of the learners not understanding that the perfect aspect is used in
English where the event time and the reference time do not co-occur
(Reichenbach 1947). So when the learner perceives only the event time as a
figure and not both the event and the reference time as distinct figures in
relation to coding time, anomalous utterances may occur. It seems a likely
hypothesis that this learning problem would not occur if, instead of providing
discrete instances of perfect encodings which relate only event and coding
time (e.g., I've been to USA), teachers provided for comprehension and

Authenticated
The Figure / Ground Gestalt and Language Teaching Methodology l35
encouraged the production of sentences which also situated events in the

reference time of the utterance which contained them (e.g., I've just
returnedfrom USA and now I want to visit China).
The issue discussed in the previous paragraph reminds us how frequently
the ground is left to inference and raises the interesting question of the
extent to which lexicalizing what would usually be inferred is appropriate in
instructed second language acquisition contexts. This is a complex issue
taking in a wide range of phenomena. At one end of the spectrum there is
the obligatory lexicalization of subject pronouns in a language like English,
which will produce unexpected pragmatic effects for learners whose mother
tongues favour the null subject, or pro-drop, option. At the other, there are
cases when the lexical realization of a ground might equally be recovered as
an explicature (Sperber and Wilson 1995: 182). In such cases, the two utter-
ances will typically provide contrasting default inferences, as in simple
minimal pairs like I've been and I've been before. The teacher therefore has
to balance the need to provide properly grounded utterances with the real-
ization that specifying the ground lexically will sometimes flout the neo-
Gricean heuristic "Do not postulate more entities than necessary" (Levinson
1995: 102) and thus prompt the recovery of an implicature.
Finally, in our consideration of deixis, the use of temporal and spatial
expressions to link one part of a discourse to another indicates explicitly to
learners how two units of information are related. Thus, this paragraph
begins with the discourse deictic Finally and the fIrst sentence of the "last"
paragraph contained the deictic use of previous ("The issue discussed in the
previous paragraph"). For a learner struggling to understand the proposi-
tional meaning of a text, such discourse deictics may play a vital role in
ensuring that each new struggle to comprehend is assisted by the writer
making clear the relation of the current part of the text, seen as fIgure, to
each previous part, seen as ground.
6. The Figure I Ground Gestalt: Further Methodological Considerations
This section considers a number of general methodological issues which

considerations of fIgure and ground seem to raise and that go beyond the
discrete-item-syllabus issue discussed in Section 3 and the role of pragmat-
ics, and in particular indexicality, in second language instruction discussed
in the previous section.

Authenticated
136 Peter Grundy
6.1. The Easy / Difficult Issue
It is widely held that second or foreign language learners should start with
what is "easy" and then proceed to what is "difficult". Defining "easy" and
"difficult" is not unproblematic, however.
Most teachers of English would appeal to morphological criteria in rating
the present simple "easier" than the present perfect or the present continuous
(or even perhaps than the past). Similarly, it seems perverse to argue that the
comparative and superlative forms of adjectives are "easier" than the base
form. However, this position has been complicated by work on accessibility
hierarchies and linguistic universals which suggests that the existence of a
marked form implies the existence of an unmarked form within the same
hierarchy. The effects of teaching marked forms such as less-accessible rel-
atives before unmarked has been the subject of some experimentation
(Doughty, Eckman et al.). The results of these experiments do indeed sug-
gest that learners infer the existence of less-marked forms and that teaching
marked forms first may therefore be an economical way of helping learners
to acquire a language.
The figure / ground gestalt, at least as Wall ace interprets it, adds a new
dimension to the easy / difficult issue by suggesting that some linguistic cat-
egories are more salient than others. Should we therefore teach salient cate-
gories early and leave less salient ones until later? Thus we might want to
teach (or increase the frequency of the occurrence of) count rather than
mass nouns, definite rather than non-definite descriptions, first and second
rather than third persons, perfective rather than imperfective aspects, etc.
(Wallace 1982: 212). Or if it could be shown that there is in fact a salience
"hierarchy", then we might argue that less salient members of the hierarchy
should be taught early on the grounds that they imply the existence of more
salient members in the same hierarchy.
Although to base instructional practice narrowly on a salience hierarchy
hypothesis of this sort is to take a rather instrumental view of research find-
ings, we need to acknowledge that "take-away" understanding of this kind
is welcomed by busy teachers who necessarily rely on applied linguists to
provide the rationale for classroom practice.
6.2. The Materials Issue
Mukarovsky's thesis is that scientific writing automatizes or explains figures

that stand out from the background in order to make meaning clear. Poetic

Authenticated
language, on the other hand, foregrounds the means of expression at the

expense of propositional meaning. Jakobson builds on this insight when he
argues that the poetic function focuses on the code itself. Thus figure /
ground properties are most obviously revealed in poetic writing. If this
salience-orientation is a fundamental feature of our ability to understand
language, then it might be argued that teachers should privilege materials in
which it most obviously occurs (literature, advertisements, rhetorical texts,
song, etc.). And if one takes the view that second language learning should
focus not on the meaning conveyed (which is unlikely to be new to the learner
anyway), but on the language in which it is conveyed, then "poetic" texts
might seem to be a preferred material. This would certainly be revolutionary
for latter day grammar / translationists for whom textual "comprehension"
is typically conducted on quasi scientific writing in which propositional
meaning is privileged over the form in which it is conveyed.
So we see that considerations of figure and ground might cause us to
think about the function (in the Jakobsonian sense) most appropriate to
instructed second language acquisition. Should the referential function
which tries to minimize the figure / ground contrast continue to be the natural
function in language instruction or should the poetic function in which the
properties of the code itself are foregrounded have a wider role?
6.3. Which to Focus - Figure or Ground?
In our earlier discussion of the figure / ground gestalt, we noted that it is the
figure and not the ground which is remembered. If we extrapolate this to
language and in particular to Talmy's distinction of main sentence assertion
as figure and embedded sentence presupposition as ground, we have an
interesting question to resolve: which are we to practise, the salient item or
the less salient? (In asking this question, I make the perhaps insufficiently
questioned assumption that practice is a vital part of language learning.)
To take a concrete case, in audiolingual-type drills, the teacher has to
decide whether to vary the figure or the ground. Imagine a teacher prompt-
ing controlled practice around the structure I + <verb> + <adverb> +
<preposition phrase> and working away from the textbook model I work
hard at school, where I work hard is seen as figure (trajector) and at school
as ground (landmark).
The first question is whether this structure should be used to practise the
figure I work hard or the ground at school. And even if we knew how to
resolve this issue, there is a further question. Imagine the teacher decides to

Authenticated
138 Peter Grundy
cue a drill with flashcards containing words or pictures which indicate the
part of the sentence to be substituted. Should it be the figure which is
replaced (1 work hard at school->l write fast at school->l sit quietly at
school-> You work hard at school, etc.) or the ground (1 work hard at
school->l work hard at home->l work hard in the gym, etc.)? We may have
varying opinions about this simple example, but it raises a fundamental
question - what is the relationship between varied repetition, figure and
ground, and learning?
7. Conclusion
In this paper, I have argued the learnability position in instructed second

language acquisition, but have eschewed the familiar approaches. These tend
either to discuss the biological structures governing stages of the language
learning process (typically projected in the generative paradigm as the puta-
tive effect of prior knowledge of the Ll or the extent of the availability to
adult learners of universal grammar) or to consider largely system external
considerations such as motivation and institutional context. Rather, I have
tried to show how the internal properties of language, including especially
iconicity and indexicality, are a viable starting point for the teacher and
might well cause methodologists to rethink some of their assumptions.
Specifically, I have argued that successful language acquisition depends
on feeding cognitive processes and that the figure / ground gestalt when
applied to the study of language shows that decontextualised, discrete-item,
or segmental approaches to language teaching cannot work. I have argued
against the kind of naIvete which supposes that a description of a subset of
the formal properties of language constitutes a basis on which the language
teacher can proceed. Rather, an understanding of the acquisition process,
and in particular of the role that language as well as the learner plays in it, is
the key to deciding on the kinds of language encounter which are profitable.
The point is that language structures crucially depend, just as visual objects
do, on a background which shows their salience. We think that we see the
sun in the sky, but in fact the diffused light that we call "the sky" is part of
the same spectrum as the apparent object that we call "the sun". You cannot
have one without the other. Language teachers and materials writers have
been unwilling to understand the significance of this gestalt for language
learning classrooms.
The discrete-item-syllabus issue discussed in Section 4, the major identi-
fied areas of pragmatic investigation discussed in Section 5, and the three

Authenticated
pedagogic issues discussed in Section 6 in all probability constitute only a

small subset of the relevant aspects of instructed second language acquisition
for which the figure / ground gestalt has implications. But at this early stage
in our understanding of cognitive linguistics, much still remains for us to
discover.
Acknowledgement
The author wishes to acknowledge the encouragement and patience of the

editors of this volume, for which he is very grateful. The paper has been sub-
stantially revised in the light of the extremely insightful and justly critical
comments of two anonymous reviewers who pinpointed a very wide range
of faults and to whom all the improvements in both content and approach
(but not the still-outstanding faults) are due.

Authenticated
140 Peter Grundy
References
Blakemore, Diane
1987 Semantic Constraints on Relevance. Oxford: Blackwell
Bley-Vroman, Robert, Stephan Felix, and Georgette Ioup
1988 The accessibility of universal grammar in adult second language
learning. Second Language Research 4,1-32
Chafe, Wallace L
1994 Discourse, Consciousness and Time: The Flow and Displacement of
Conscious Experience in Speaking and Writing. Chicago: University
of Chicago Press.
Crandall, JoAnn
1999 Cooperative language learning and affective factors. In Jane Arnold
(ed.) Affect in Language Learning Cambridge: Cambridge University
Press.
Doughty, Catherine
1990 Second language instruction does make a difference: Evidence from
an empirical study of SL relativization. Studies in Second Language
Acquisition 13.4, 431-69.
Eckman, F.R., L. Bell, & D. Nelson
1988 On the generalization of relative clause instruction in the acquisition
of English as a second language. Applied Linguistics 9.1.
Grundy, Peter
2000 Doing Pragmatics (2nd ed.). London: Arnold.
Halliday, Michael A. K.
1976 System and Function in Language. Oxford: Oxford University Press.
Hanks, William H.
1992 The indexical ground of deictic reference. In Duranti Alessandro &
Charles Goodwin (eds.) Rethinking Context. Cambridge: Cambridge
University Press, 43-76.
Howatt, A.P.R.
1984 A History of English Language Teaching. Oxford: Oxford University
Press.
Hymes, Dell H.
1971 On communicative competence. In J. B. Pride & Janet Holmes (eds.)
Sociolinguistics. Harmondsworth: Penguin 1972, 269-93.
Jakobson, Roman
1960 Linguistics and poetics: Closing statement. In Thomas A. Sebeok
(ed.) Style in Language. Cambridge, Mass.: MIT Press, 350-77.
Krashen, Stephen D.
1985 The Input Hypothesis. London: Longman.
1987 Foundations of Cognitive Grammar: Volume 1, Theoretical Pre-
requisites. Stanford, Calif.: Stanford University Press.

Authenticated
1991a Foundations of Cognitive Grammar: Volume 2, Descriptive Appli-

cation. Stanford, Calif.: Stanford University Press.
1991b Concept, Image, and Symbol: The Cognitive Basis of Grammar.
Levinson, Stephen C.
1983 Pragmatics. Cambridge: Cambridge University Press.
1995 Three levels of meaning. In Frank Palmer (ed.) Grammar and
Meaning. Cambridge: Cambridge University Press.
Mukarovsky, Jan
1964 Standard language and poetic language (English version trans. Paul
L. Garvin). In Donald C. Freeman (ed.) Linguistics and Literary
Style. New York: Holt, Rinehart, Winston 1970,40--56.
Reichenbach, Hans
1947 Elements ofsymbolic logic. New York: Macmillan.
Richards, Jack C. & Theodore Rodgers
2001 (2nd ed.) Approaches and Methods in Language Teaching. Cambridge:
Rubin, E.
1915 Figure and ground. In Readings in Perception. D. C. Beardslee and
Michael Wertheimer (eds.) London: Van Nostrand, 1958, 194-203
Sperber, Dan and Deirdre Wilson
1986/1995 Relevance: Communication and Cognition. Blackwell: Oxford.
Talmy, Leonard
1978 Figure and ground in complex sentences. In 1. H. Greenberg (ed.)
Universals ofHuman Language 625-49.
Verschueren, Jef
2000 Notes on the role of metapragmatic awareness in language use.
Pragmatics 10/4439-56.
Wallace, Stephen
1982 Figure and ground. In Paul J. Hopper (ed.) Tense-aspect: Between
Semantics and Pragmatics. Amsterdam: John Benjamins 201-23.
Wilson, Deirdre & Dan Sperber
1993 Linguistic form and relevance. Lingua 90 1-15.
Wong Fillmore, Lilian
1989 Teachability and second language acquisition. In M. L. Rice and R. L.
Schiefelbusch (eds.) The Teachability ofLanguage. Baltimore: Paul
H. Brookes.

Authenticated
Authenticated
"Cultural Scripts": a New Medium for
Ethnopragmatic Instruction
Cliff Goddard
1. Linguistic Pragmatics and Cultural Analysis I
This paper is concerned with the potential applications of one stream of cog-
nitive linguistics - the cross-linguistic semantics of Anna Wierzbicka and
colleagues - to the teaching of ethnopragmatics (cultural pragmatics). To
begin with, therefore, some remarks on conventional approaches to linguistic
pragmatics are in order.
Pragmatic differences between languages have been addressed within
several frameworks, including the ethnography of communication, con-
trastive pragmatics, and linguistic anthropology. In general, work within
these different frameworks assumes that in any speech community there
will be certain shared understandings (variously termed "rules of speaking",
"norms of interaction", "preferred discourse strategies", etc.) about how it is
appropriate to speak in particular, culturally construed situations. The first
task in ethnopragmatic description is, therefore, to identify and describe
culturally preferred speech patterns. In most work there is a further step as
well, for it is widely recognised that language-specific speech patterns are
frequently, if not invariably, linked with the cultural values and attitudes of
the people concerned. The second task is, therefore, to spell out these links.
Both tasks are hindered, however, by an often unrecognised problem,
namely, weaknesses in the descriptive metalanguage which is being
employed (Wierzbicka 1991).
To describe culturally preferred styles of speaking, it is standard practice
to employ an inventory of terms such as "directness", "formality", "involve-
ment", and so on. Though such terms can be valuable and useful up to a point,
they are somewhat vague and tend to be used with rather different meanings
by different authors. For example, when Japanese speech patterns are con-
trasted with English ones, the Japanese are described as "indirect" and the
English as "direct"; but when English is compared with Hebrew, it is the
English speech patterns which are "indirect" and the Hebrew "direct" (cf. for
example, Mizutani and Mizutani 1987; KatrieI1986). There is no salvation

Authenticated
144 Cliff Goddard
in the idea that there is a "scale of directness" on which English is situated

between Japanese and Hebrew, because the relevant differences between the
three languages are not merely quantitative, but qualitative. They include
differences in how and when to be indirect, and, more importantly perhaps,
differences in why to be indirect.
To describe the cultural values and attitudes which explain preferred ways
of speaking, there are two standard approaches so far as the metalanguage
of description is concerned. One is the use of English terms such as "face",
"politeness", and "respect". Here there is an evident danger of terminological
ethnocentrism, because the meanings of such words do not correspond pre-
cisely to any words in the language being described and therefore presum-
ably do not designate precisely any emic cultural category (cf. Janney and
Arndt 1993; Matsumoto 1988). The ethnocentrism is inherent in choosing a
descriptive metalanguage which is language-specific and culture-specific -
whose fundamental tenns do not even have equivalents in the language being
described. Such an approach necessarily imposes an "outsider perspective".
The other approach is to use indigenous terms (such as enryo, wa, and
omoiyari, in the case of Japanese). In this case, we can be sure that they
designate cultural values which are "real" to the people concerned, but the
problem of translation remains - in reverse: How can the meanings of these
terms be explicated without distortion into English (or whatever other lan-
guage is the language of description)?
In my view, the solution to overcoming these problems is the semantic
theory developed by Anna Wierzbicka and colleagues over many years of
cross-linguistic semantic research (cf. Wierzbicka 1972, 1996a; Goddard
and Wierzbicka, eds. 1994,2002; Goddard 1998). The key idea is that at the
core of every language there is a small set of simple, basic meanings
(semantic primes) which cannot be further explicated without circularity.
Cross-linguistic research indicates the existence of at least 60 semantic
primes, including (using English exponents): substantive words like I, YOU,
SOMEONE/PERSON, SOMETHING/THING, and PEOPLE; verbal words such as
KNOW, WANT, FEEL, DO, HAPPEN, and SAY; determiner-like elements such
as THIS, OTHER, THE SAME, ONE, MANY, SOME, and ALL; spatial concepts
like WHERE/PLACE, HERE, ABOVE, BELOW, FAR, and NEAR; temporal con-
cepts like WHEN/TIME, NOW, AFTER, BEFORE; logical words like IF,
BECAUSE, NOT, MAYBE, and CAN; evaluational and descriptive concepts
like GOOD, BAD, BIG, and SMALL; intensifying and augmenting words like
VERY and MORE; and relational concepts such as LIKE, KIND OF, AND PART
OF. A table of proposed semantic primes is given in the Appendix.2 The the-
ory also allows that a semantic prime can be expressed by variant words

Authenticated
(allolexes) in different grammatical contexts; for example, in English else is

an allolex of OTHER and don't is an allolex of NOT. The vocabulary of
semantic primes has a basic grammar which governs how the primes can be
combined, so that, all in all, the system amounts to a sort of common mini-
language in terms of which one is supposed to be able to explicate all the
complex word-meanings found in any language: hence the designation
"natural semantic metalanguage" (NSM).
There is a large body of NSM research in lexical semantics, much of it
focusing on heavily culture-laden words in various languages from around
the world (cf. Ameka 1987; Ourst 2001; Goddard 1992, 1994, 1996; Harkins
1996; Peeters 2000; Travis 1998; Wierzbicka 1991, 1992, 1997, 1999; Ye
2001,2002). Equally, however, the metalanguage of semantic primes can be
used to state hypotheses about cultural norms of various kinds. Wierzbicka
proposes to call such statements "cultural scripts". 3 For applications of the
approach to Anglo-American and Anglo-Australian cultural norms, as well as
to those of Germany, Ghana, Malaysia, Japan, France, Poland, and Taiwan,
see Ameka (1994), Goddard (1997, 2000), Goddard and Wierzbicka (1996),
Hasada (1996), Peeters (1999, 2000), and Wierzbicka (1991, 1994a, 1994b,
1996b, 1998). I will illustrate what cultural scripts look like with two quick
examples.
Example 1: It can be argued that a wide range of "Anglo" ways of
speaking can be explained in terms of rather high-level cultural scripts
related to the ideal of "personal autonomy" (cf. Wierzbicka 1991). A very
general script of this kind is:
people think like this:
when I do something it is good if I do it
because I want to do it
This script is intended to represent a dominant or mainstream attitude in pre-

dominantly English-speaking countries such as the United States, Australia,
and Great Britain.
Example 2: At a much more specific level of detail, the following script
is intended to represent a cultural value or norm which is characteristically,
though not exclusively, Japanese (Wierzbicka 1994a):
people think like this:
if something bad happens to someone because of me, 1 have to say
to this person something like this: 'I feel something bad because of this'
It is linked, of course, with the often noted tendency of the Japanese to

"apologise" very frequently and in a broad range of situations.

Authenticated
146 Cliff Goddard
Why are these scripts better than saying, for example, that Anglo culture
values "personal autonomy" and that in Japanese culture one has to "apolo-
gise" a lot? There are several interrelated reasons which I will adumbrate
here but not fully justify, for reasons of space. First, cultural scripts are
more easily tested and refined than are generalisations framed in technical
terms, because they are more precise and predictive, and because they are
more directly accessible to the intuitions of ordinary native speakers. Second,
unlike complex terms such as "autonomy", "apology", and "sensitivity",
which lack exact equivalents in most non-English languages, cultural scripts
composed in simple terms can be readily translated between languages,
which makes them more practical for research across linguistic barriers.
Third, because cultural scripts are framed in the same metalanguage as can
be used for lexical semantics, they provide enhanced opportunities to demon-
strate links between discourse practices and sociocultural values, as embodied
in cultural key words, value terms, proverbs, etc.
2. Pedagogical Applications of Cultural Scripts
The cultural scripts approach is relatively new, and, as far as I know, it has
not been systematically applied or tested in language teaching (though it has
been used, with success, in some academic courses on intercultural commu-
nication). Nevertheless, there are reasons to think that cultural scripts could
provide significant pedagogical advantages in relation to the problem of
"bridging the culture gap" in language teaching.
In communicative and functional approaches to language teaching, and
in the emerging paradigm of Intercultural Language Teaching (Lo Bianco,
Liddicoat and Crozet, eds. 1999; Liddicoat and Crozet, eds. 2000), one of the
key tasks is to bring the learner to an understanding of those L2 concepts
and values which form the interpretive frame within which L2 individuals
negotiate meaning. In pursuit of this goal, a variety of ingenious pedagogical
techniques have been devised, which may involve analysis of texts, dia-
logues, and routines, use of role-plays, interviews, student projects, and other
activities (cf. for example, Valdes, ed. 1986, Kramsch 1993). It has often
been observed that cultural analysis in the classroom inevitably assumes a
contrastive dimension, as students become reflectively aware of how their
own concepts and values compare and contrast with those of the L2 culture.
According to a recent review of the field of interlanguage pragmatics
(Kaspar and Schmidt 1996), there has so far been little research into how
non-native speakers develop pragmatic competence, and even less into the

Authenticated
effectiveness of different methods of teaching it. Most research has concen-

trated on simply describing how the pragmatic behaviour oflanguage-Ieam-
ers differs from that of native speakers (with a particular view to the issue of
"pragmatic transfer"). Nonetheless, the small body of research into the peda-
gogical effectiveness of teaching pragmatic competence indicates that
explicit instruction yields results. Kaspar and Schmidt (1996: 160) say: "the
few existing studies are encouraging ... [F]ocusing on aspects of pragmatic
awareness through consciousness-raising activities and communicative
practice seems highly facilitative". For example, House (1996) found that a
teamer group given explicit instruction improved their pragmatic fluency
with linguistic routines much faster than a comparable group. According to
House (1996): "metapragmatic information is essential in counteracting neg-
ative pragmatic transfer" (p.24); and she notes that in follow-up interviews
students commented favourably on the benefits of a "cognitive-analytical
approach" .
Schmidt (1993) reviews data and theories about the leaming ofL2 prag-
matics from the perspective of experimental psychology. He argues that a
crucial condition for successfulleaming is selective "noticing" of significant
L2 features, which are often not obvious to the unassisted leamer. Further,
once the leamer has been sensitised to the relevant linguistic forms and con-
textual features, "attempting to analyse their significance in terms of deeper
generalisations" is also highly facilitative (p.35). Schmidt also notes that
insofar as Lt acquisition is a relevant point of comparison for L2 acquisition,
the literature on language socialisation makes it very clear that parents and
peers actively instruct children on pragmatic matters (e.g., by providing
"rules" and negative feedback, and by modelling routines).
I would like to suggest that cultural scripts can provide a valuable new
medium for pragmatic instruction: a medium which is both more intelligible,
less ethnocentric, and more precise than global labels such as "directness"
or "politeness", or the use of Lt translations of L2 lexical items. The fact
that only simple everyday words are required to phrase cultural scripts also
means that language students could be asked to work out their own versions
of scripts, both for the L2 culture and for their home culture, contrastively,
as an exercise in dialogic teaming (cf. Kramsch 1993).
A second advantage of the cultural scripts approach is that it could
encourage a more holistic approach to discourse practices, heightening the
leamer's appreciation of the fact that discourse practices are typically based
in a system of cultural values. In some current approaches, discourse practices
are presented virtually as being in a "world of their own", unconnected with
the larger cultural context. This danger is particularly acute, it seems to me,

Authenticated
148 Cliff Goddard
in that stream of research in contrastive pragmatics which concentrates on

speech-acts. All too often research of this kind is based on the assumption
(invalid, in my view; cf. Goddard 2002a) that there is an inventory ofuniver-
sal speech-acts such as "apology", "compliment", and "request", which just
happen to be "realised" differently in different cultures. Studies in this mode
tell us that, for example, "requests" are made more "directly" in Culture A
than in Culture B, or that "compliments" in Culture A are generally
"accepted" whereas in Culture B they are "rejected". Rarely is there any
attempt to relate these facts to the broader cultural framework, leaving them
to appear as isolated peculiarities.
Cultural scripts, in contrast, lend themselves to an approach in which
discourse practices are explored in tandem with important emic concepts, as
embodied in cultural key words, value terms, proverbs, and so on (cf.
Vamarasi 1995). Aside from being a good thing in itself, this may help reduce
students' frustration with what otherwise can appear as an unmotivated set
ofL2 behavioural "hangups". "Why", students often ask, "are 'they' so eva-
sive, so indirect, so crass, so insincere, etc?" It is not very helpful merely to
tell students that the L2 culture "values" it that way. What they need to
understand is why the L2 culture values it that way, i.e., how the preferred
speech styles fit in with the overall L2 model of human nature and "folk
philosophy". To the extent that a cultural scripts approach can promote this
kind of understanding, it can help develop positive attitudes to the L2 culture,
which would promote integrative motivation.
In the remainder ofthis study, I propose to illustrate the approach, begin-
ning with examples from English, and then with examples from Malay
(Bahasa Melayu).
3. Some Cultural Key Words and Cultural Scripts of English
Though Wierzbicka (1997) was the first to approach cultural key words using
a rigorous method of semantic analysis, their existence and importance has
been recognised by numerous scholars of culture studies. Essentially, cultural
key words are conceptual "focal points" for entire cultural domains, such
that studying them leads into a dense complex of cultural values, attitudes,
and expectations. With reference to Russian examples, Wierzbicka writes as
follows:
A key word such as dusa (roughly 'soul') or sud'ba (roughly 'fate') in

Russian is like one loose end which we have managed to find in a tangled

Authenticated
"Cultural Scripts": a New Mediurnfor Ethnopragmatic Instruction 149
ball of wool: by pulling it, we may be able to unravel a whole tangled "ball"
of attitudes, values, and expectations, embodied not only in words, but also
in common collocations, in set phrases, in grammatical constructions, in
proverbs, and so on ... (Wierzbicka 1997: 17)
It hardly needs to be argued that in modern Anglo-American culture the

wordfreedom (and the related word free) has key word status. What is often
not recognised, however, is that the modern Western concept of freedom is
culture-specific (Wierzbicka 1997: 125-155). It is not the same concept, for
example, as that designated by the Latin word libertas, which was, in a
sense, the contrary of the Roman institution of slavery; nor is it exactly the
same as, for example, the Russian concept of svoboda. The modern Anglo-
American concept offreedom has what political philosophers sometimes
term a "negative" orientation in the sense that the primary focus is on being
able to do what one wants to do without interference from other people.
According to Isaiah Berlin4 (1969: 122-127), the classical English political
philosophers, such as Jeremy Bentham, John Locke, Adam Smith, and John
Stuart Mill, understood the notion of[reedom in precisely in this way.
Berlin quotes Thomas Hobbes: "A free man is he that ... is not hindered to
do what he hath the will to do".
The meaning of this concept can be explicated as follows (cf.
Wierzbicka 1997: 129). This is an example of how a complex and subtle
concept can be "unpacked" using the small but expressive vocabulary of
semantic primes.
(1) this person (X) hasfreedom (isfree) =
when this person (X) wants to do something

this person doesn't have to think:
"I can't do it because someone else doesn't want me to do it"
it is bad if someone has to think like this
this person can think:
"I can do it if it is not bad for other people"
it is good if someone can think like this
Various grammatical facts are consistent with this explication. For example,
one can speak offreedom OF (e.g.,freedom of action, of opinion, of trade)
orfreedom TO (e.g.,freedom to emigrate), referring to something which is
desirable. One can also speak offreedom FROM (e.g.,fi-eedom from persecu-
tion, from harassment, from coercion, from external control), referring to a

Authenticated
150 Cliff Goddard
situation when other people do something which prevents us from doing

what we want to do. 5 Notice that explication (1) does not presentj1-eedom as
an unqualified right. There is explicit exception made in relation to goals
which would be 'bad for other people', which implicitly also rules out inter-
fering with other people's freedom. In general, however, the semantics of
j1-eedom is based around the idca of "non-interference".
The cultural value of freedom can be aligned with some very general
cultural scripts, which can be stated as in (2) and (3) below. These proscribe,
as it were, particular ways of speaking which could be seen to negate or to
disregard another person's individual freedom of action and of opinion.
(2) people think like this:

no-one can say to someone else:
"I want you to do something
you have to do it because of this"
this is good

no-one can say to someone else:
"I think like this about something
you have to think the same because of this"
this is good
Arguably, a wide range of Anglo-American ways of speaking can be linked

with these and similar cultural assumptions (Wierzbicka 1991). For example,
compared to speakers of many languages of the world, speakers of main-
stream English are inhibited from using the bare imperative, because it
could be seen as expressing the kind of message close to that proscribed by
script (2), i.e., 'I want you to do something; you have to do it because of
this'. Similarly, compared with speakers of many other languages, speakers
of mainstream English tend not to express their opinions in the "absolute"
fashion characteristic of, say, Russian or Hebrew, which could be seen as
conveying the kind of message close to that proscribed by script (3), i.e. 'I
think like this about something; you have to think the same'.
How then are such messages conveyed in the Anglo-American "lingua-
culture"? Essentially, speakers are guided by more specific scripts into using
more elaborated locutions and phraseology, which embed the potentially
sensitive messages into non-threatening contexts. In relation to directives, it
is well-known that English has a brace of elaborate interrogative directive
formulas (also called whimperatives), such as:

Authenticated
Can you do this?

Could you do this?
Will you do this?
Would you do this?
Would you mind doing this?
and the like. These constructions clearly convey the message 'I want you to
do this', but at the same time they embed the potentially confronting message
into a question-form, with the effect of inviting the addressee to say whether
or not he or she will comply, thereby acknowledging, in form at least, the
addressee's autonomy. Part of acquiring mainstream Anglo-American
English as a child (or learning it as an adult) is internalising a personal
script like (4).
(4) when I want to say to someone

"I want you to do something"
it is good to say something like this at the same time
"I don't know if you will do it"
Incidentally, this explanation is quite different from the conventional story

that English speakers use interrogative directives for reasons of "polite-
ness". From a factual point of view, politeness can hardly explain the use of
interrogative directives with young children, for example, or the fact that
they can be used to deliver abrupt, non-polite messages (e.g., Can you just
shut up and listen?), and even be combined with swearing (e.g., Will you
bloody well hurry up?). From a theoretical point of view, it is not very illu-
minating simply to say, for example, that native speakers of English are
more "polite" than speakers of German (cf. House and Kasper 1981).
Similarly, in my view, learners of mainstream Anglo-American English
internalise a personal script like (5) below, which guides them towards leav-
ing "room for disagreement" when they express opinions (for example, by
frequent use of lexical hedges such as sort of and rather, and introductory
epistemic formulas such as I think, I reckon, and I guess; (cf. Moon 1998;
Scheibman 2001).
(5) when I want to say to someone

"I think like this about something"
it is good to say something like this at the same time
"I don't know if you think the same"

Authenticated
152 Cliff Goddard
As argued by Wierzbicka (1991) the cultural script in (5), together with a

cluster of related Anglo-American attitudes, also helps to account for the
fact that, seen from the point of view of many other languages, English is
characterised by an "all-pervasive presence of tag questions, highly diversi-
fied in fonn and function"(p.4O). To be sure, not all tag questions have to do
with matters of opinion or evaluation (some, for example, accompany
imperatives, others genuinely seek factual confinnation). Nevertheless,
Wierzbicka's general point seems clear:
The very fact that tag questions have come to play such a major role in
English seems to reflect the same cultural attitudes which have led to the
expansion of interrogative forms elsewhere, and to the restrictions on the
use of the imperative, the same emphasis 011 possible differences of opinion.
o.lpoint olview. (Wierzbicka 1991: 40; emphasis added)
Contrasting English with her native language, Polish, in this respect,

Wierzbicka continues as follows. Similar points could be made about many
other languages .
... Polish cultural tradition does not foster constant attention to other people's
'voices', other people's points of view, and tolerates forceful expression of
personal views and feelings without any consideration of other people's
views and feelings. In fact, the basic Polish tag, prawda? 'true?', presents
the speaker's point of view not as a point of view, but as an objective 'truth';
and it doesn't seek agreement but an acknowledgement of this 'truth'.
(Wierzbicka 1991: 40)
These examples (the profuse use of interrogative directive fonnulas and tag
questions in English, the preference for "hedging" in opinion-giving, and so
on) illustrate the point that cultural values can underlie even the minutiae of
everyday fonns of talk, with the implication that, as Liddicoat and Crozet
put it:
[E]ven very simple language offers genuine opportunities for cultural under-
standing in the language classroom. In fact, it is often this very simple for-
mulaic language which is most culturally loaded. (Liddicoat and Crozet
2000: lO)
Of course it would be wrong to assume that all varieties of English share the
same cultural scripts and associated repertoires of fonnulas, routines and
characteristic phrasings. For example, it can be argued that American English

Authenticated
is more concerned with projecting enthusiasm and a "positive attitude" than

many other varieties of English, and that Australian English is more tolerant
of sarcasm and expressions of negativity than American English
(Wierzbicka 1999:240-272; 2002).
4. Some Cultural Key Words and Cultural Scripts of Malay
Malay (Bahasa Melayu) is the national language of Malaysia. By way of

orientation, one can say that traditional Malay culture is a meld of an indige-
nous Austronesian tradition (the adat, embodied in hundreds of traditional
sayings and maxims), a substantial Indic legacy, and Islamic culture and
religion (the Malays have been Muslims for over four centuries) (Tham
Seong Chee 1990).6
Two of the more salient and "untranslatable" of Malay concepts are
embodied in the words sabar and hormat (and its verbal form menghormati).
I propose to explicate these words and to link them with cultural scripts. It
is not possible here to fully demonstrate their key word status, to justifY the
precise phrasing of the explications, or to explore the similarities and differ-
ences between the Malay words and their nearest English glosses (see
Goddard 2000).
Sabar is usually glossed as 'patient' or 'forbearing'. The range of use not
only includes contexts such as sabar menunggu 'waiting patiently', but also
resisting negative emotions such as anger and sorrow (JanganZah marah!
Sabar! 'Don't be mad! Calm down! '; Sabar, jangan menangis kuat sangat
'Be calm, don't cry so loud'). This is an Islamic virtue: Sabar separuh
daripada iman 'Being patient is half of faith'. Muslims need to be sabar in
order to maintain the strict guidelines of Islamic life (including, for example,
praying five times a day, and observing the annual month of fasting), and in
order to maintain their faith in the face of criticism and hardships. Further-
more, in the Islamic view personal misfortunes and sufferings should be
seen as tests from God. This perspective helps explain why in difficult per-
sonal situations - situations which in Western culture would be likely to
attract advice such as 'do something about it' or 'express how you feel' - it is
quite characteristic Malay advice to bersabar 'endure it, forbear'. According
to explication (6), being sabar is having the mental discipline or self-control
to stay calm and not to form impulsive intentions in the face of suffering or
affliction.

Authenticated
154 Cliff Goddard
(6) X is sabar [at this time] =
at this time, X felt something bad

because of this, X could have thought like this:
I don't want this
I want to do something now
X did not think like this, because X didn't want to think like this
it is good if a person can be like this
Moving now to cultural scripts, I would like to begin with the claim that
Malay culture encourages concern for careful and premeditated action.
Expressions such asfikir dulu 'think first' andfikir panjang 'think long' are
heard daily. Impulsive or wilful actions (ikut suka hati 'doing what you feel
like') are viewed as dangerous and immature. As the saying goes: lkut rasa
binasa, ikut hati mati 'Follow (your) feelings suffer, follow (your) heart
die'.

I don't want something bad to happen because I do something
because of this, when I want to do something
it is good if! think about it for some time before I do it
Script (8) takes this a step further, spelling out the cultural priority placed on
verbal caution and premeditation, particularly in relation to hurting the feel-
ings of others: "[Malay values include] showing consideration and concern,
anticipating the other ... and, above all, being sensitive to the other person"
(Wilson 1967: 132); cf. the expressionsjaga hati orang 'watch over other
people's feelings', memilihara perasaan 'look after feelings'. The motives
here are not purely altruistic, in virtue of the strong cultural theme that people
are likely to retaliate, possibly in subtle and calculated ways (cf. the concept
of dendam 'revenge, payback'), against any offence to their maruah 'dignity,
standing' or nama 'reputation'. As the saying goes: Rosak badan kerana
mulut 'The body suffers because of the mouth'.

it is not good if when I say something to someone,
this person feels something bad
because of this, when I want to say something to someone,
it is good if! think about it for some time before I say it

Authenticated
"Cultural Scripts": a New Medium/or Ethnopragmatic Instruction l55
Needless to say, the complex of attitudes spelled out, in part, in scripts (7)
and (8) are consistent with the status of sabar 'patient, forbearing' as a cul-
tural key word.
The second key word I want to consider, hormat (verbal form: menghor-
mati), is usually glossed as 'show respect' or as 'proper politeness'. One
sociologist (Kessler 1992: 148) has described it as "deference that is owed to
a social position", stressing its role as a focus of cultural elaboration. As testi-
mony to the orthodoxy of these attitudes, another sociologist Geert Hofstede
(1997: 28) found Malaysia to be the country which ranked highest of 53
countries and regions on the so-called "power distance index", which indi-
cates the extent to which the less-powerful individuals in societal roles and
relationships expect and are willing to obey authority. (For comparison, the
USA comes in at rank 40 on this scale, and Australia at rank 36.)
Hormat involves being on one's "best behaviour", including speaking in
a nice tone of voice, not saying too much, not opposing or contradicting.
Linguistic etiquette is also very much an issue: avoiding the pronouns aku
'1', kaulawak 'you', use of various honorific words, and as far as possible
using a refined (halus) speech style. According to explication (9), the idea
behind menghormati is to show the "respected" person that you recognise
his or her higher standing and that you want to avoid his or her disapproval.
To this end, you will behave in a deliberately selective fashion - in terms of
what you do, what you say, and how you say it. 7
(9) X menghormati person-Y =
X thinks things like this about Y:

Y is someone above me
I don't want Y to think anything bad about me
X wants Y to know this
because of this, when X is with Y
X does some things, X doesn't do some other things
X says some things, X doesn't say some other things
X says some words, X doesn't say some other words
In relation to the component 'Y is someone above me', it should be noted

that using the ''vertical dimension" to speak of differences in social standing
is very common and natural sounding in Malay. For example, people can be
said to occupy a tinggi 'high' or rendah 'low' position, rulers and leaders are
said to be di atas 'above' their subjects or followers, who are in turn di bawah
'below'them, and so on.

Authenticated
156 Cliff Goddard
Finally in this abbreviated exposition, the cultural script in (10) addresses

the significance of relative age, which is evident in many ways in Malay
life. Within the family, for instance, one is supposed to show respect not
only to one's parents and grandparents, but to one's older brothers (abang)
and older sisters (kakak). Outside the family, people much older than oneself
are usually addressed as pakcik 'uncle' or makcik 'auntie'; those who are
older (but not much older) are abang and kakak (or kak). Youngsters of both
genders can be addressed as adik (or dik) 'younger sib, young one'. Younger
people are supposed to menghormati older people generally.

when I am with someone older than me, it is good to think:
this person is someone above me
I want this person to think good things about me
I don't want this person to think anything bad about me
The phrasing of this script goes beyond vague and English-based generali-
sations about "respecting" older people, 8 while at the same time linking
closely with the explication of menghormati. Taken together, the script and
the lexical concept account for the self-evident status (from a Malay per-
spective) of the injunction to menghormati orang tua 'respect old people'.
5. Ethnopragmatics in Language Instruction
I hope that this brief review of some cultural key words and related cultural
scripts in English and Malay has illustrated some of the points made in the
introductory sections of this paper; in particular, that the cultural scripts
approach lends itself to an integrative treatment of discourse style which
draws connections between broad cultural themes, key lexical items, proverbs
and common sayings, linguistic routines, and many other things besides
(such as child-raising practices, to mention only one matter not touched
upon in this paper). As such, the scripts approach has obvious potential for an
integrated pedagogy of pragmatics and culture. I hope also to have illustrated
the advantages of the cultural scripts approach over conventional descriptors
such as "indirectness", "politeness", "sensitivity", "respect", and so on.
Cultural scripts are clearer, more articulated, and potentially much more
accurate than vague and language-specific labels of this kind.
Because the same non-ethnocentric metalanguage can be used both for
cultural scripts and for explicating the meanings of cultural key words - the

Authenticated
"Cultural Scripts": a New Medium for Ethnopragmatic instruction 157
vocabulary of values, social ideals, emotions, and so on - it becomes simpler

to demonstrate the interdependence between them. There are strong peda-
gogical advantages to an approach which focuses on and problematicises the
"words people live by". By focusing on cultural key words and unpackaging
their conceptual content, leamers can see how cultural discourse practices -
their own, as well as those of the target language - are grounded in cultural
values and assumptions. Though limitations of space prevent detailed dis-
cussion here, I would like to suggest that this kind of approach combines and
reconciles the emerging Intercultural Language Teaching (ILT) approach
with the Lexical Approach (Lewis 1997).
In my view, the cultural scripts approach provides language teachers with
a valuable new medium for ethnopragmatic instruction. Whether this opti-
mism is justified will only become clear after a sufficient number oflanguage
educators have applied the approach in practical teaching situations.

Authenticated
158 ClifJGoddard
Appendix: Proposed Semantic Primes
(after Goddard and Wierzbicka 2002)
Substantives: Time:
I, YOU, SOMEONE, SOMETHING, WHEN/TIME, NOW, BEFORE, AFTER,
PEOPLE, BODY A LONG TIME, A SHORT TIME, FOR
SOMETIME
Mental predicates: Space:

THINK, KNOW, WANT, FEEL, SEE, WHERE/PLACE, HERE, ABOVE,
HEAR BELOW, INSIDE, ON (ONE) SIDE,
NEAR, FAR, TOUCHING
Speech: Logical concepts:

SAY, WORDS, TRUE BECAUSE, NOT, MAYBE, CAN, IF
Actions, Events, Movement: Evaluators and Descriptors:

DO, HAPPEN, MOVE GOOD, BAD, BIG, SMALL
Existence, Possession, and Life: Intensifier and Augmentor:

THERE IS, HAVE, LIVE, DIE VERY,MORE
Determiners and Quantifiers: Taxonomy, Partonomy, Similarity:

THIS, THE SAME, OTHER, ONE, TWO, KIND OF, PART OF, LIKE
ALL, SOME, MANY/MUCH

Authenticated
"Cultural Scripts"; a New Medium for Ethnopragmatic Instruction 159
Notes
1. The parts of this paper which are devoted specifically to Malay are based on
Goddard (2000,2001). I would like to thank Norlinda Hasan, Vicki Knox, Lee
Mee Wun, Mohamad Mokhtar Hassan, and Anna Wierzbicka for helpful com-
ments and information with these studies. For comments specifically on this
paper I would like to thank two anonymous reviewers.
2. For a number of detailed studies of the proposed metalanguage in a sample of six
genetically and geographically disparate languages see Goddard and Wierzbicka,
eds. 2002. In a previous set of smaller studies (Goddard and Wierzbicka, eds.
1994), sixteen languages were surveyed, only one of which was Indo-European.
A full description of the Malay version of the natural semantic metalanguage
can be found in Goddard (2002b).
3. It is, of course, important to recognise that cultural norms are not static or ahis-
torical: they change, interact, and vary in space and time. In order to be able to
understand this variation, however, we must first be able to describe "local"
norms with rigour and precision, without getting bogged down in complex,
language-specific, and culture-specific vocabulary. It is also important to
acknowledge that cultural scripts are not necessarily "binding" on individuals.
The claim is merely that they furnish an interpretive backdrop against which
individuals construct and negotiate meaning.
4. Unfortunately, in Berlin's discussion the English words freedom and liberty are
used interchangeably, which is confusing because the two words do not mean
the same (cf. Wierzbicka 1997: 132-134).
5. Freedom from X can also be used in situations when some condition (such as
hunger or poverty) prevents us from doing what we want to do. Expressions like
freedom from hunger and freedom from poverty have a political undertone
because they imply that hunger, poverty, and so on are conditions imposed by
other people.
6. Present-day Malaysia is a multi-ethnic country. In this paper I confine myself to
the "culture of communication" of the Malay people, who form the majority of
the population.
7. Explication (9) applies to menghormati when it is used about people. Like English
'respect' in this regard, the word menghormati can also be used in relation to non-
human entities such as one's country, rules, the environment, etc. A different
(related) explication would be required to cover such cases.
8. The relevant concept is not really relative age (being older) as such, but, roughly
speaking, who was born earlier: one should menghormati those who have 'lived
for some time before me'.

Authenticated
160 Cliff Goddard
References
Ameka, Felix
1987 A comparative analysis of linguistic routines in two languages: Ewe
and English. Journal ofPragmatics 11(3): 299-326.
1994 Areal conversational routines and cross-cultural communication in a
multilingual society. In: Heiner Piirschel (ed.), Intercultural Communi-
cation. 441-469. Bern: Peter Lang.
Berlin, Isaiah
1969 Four Essays on Liberty. Oxford: Clarendon Press.
Durst, Uwe
2001 Why Germans don't feel "anger". In: Jean Harkins and Anna
Wierzbicka (eds.), Emotions in Cross-linguistic Perspective. 115-148.
Berlin: Moutin de Gruyter.
Goddard, Cliff
1992 Traditional Yankunytjatjara ways of speaking-A semantic perspective.
Australian Journal ofLinguistics 12: 93-l22.
1994 The meaning of lah: Understanding "emphasis" in Malay (Bahasa
Melayu). Oceanic Linguistics 33(1): 145-165.
1996 The "social emotions" of Malay (Bahasa Melayu). Ethos 24(3):
426-464.
1997 Cultural values and "cultural scripts" of Malay (Bahasa Me/ayu).
Journal ofPragmatics 27(2): 183-201.
1998 Semantic Analysis: A Practical Introduction. Oxford: Oxford Uni-
versity Press.
2000 Communicative style and cultural values - Cultural scripts of Mal ay
(Bahasa Melayu). Anthropological Linguistics 42 (1): 81-106.
2001 Sabar, ikhlas, setia - patient, sincere, loyal? A contrastive semantic
study of some "virtues" in Malay and English. Journal ofPragmatics
33: 653-681.
2002a Directive speech-acts in Malay: An ethnopragmatic perspective. In:
Les Cahiers de Praxematique 28 (Special Issue 'Language, discours,
culture', edited by Christine Beal), 113-143.
2002b Semantic primes and universal grammar in Malay (Bahasa Melayu).
In: Cliff Goddard andAnna Wierzbicka (eds.), Meaning and Universal
Grammar - Theory and Empirical Findings, Volume I, 87-172.
Amsterdam: John Benjamins.
Goddard, Cliff and Anna Wierzbicka
1996 Discourse and Culture. In: TeunA. van Dijk (ed.), Discourse: A multi-
disciplinary introduction. 231-257. London: Sage Publications.

Authenticated
Goddard, Cliff, and Ann a Wierzbicka (eds.)

1994 Semantic and Lexical Universals - Theory and Empirical Findings.
2002 Meaning and Universal Grammar - Theory and Empirical Findings.
Vol I and Vol TT. Amsterdam: John Benjamins.
Harkins, Jean
1996 Linguistic and cultural differences in concepts of shame. In: D. Parker,
R. Dalziell and I. Wright (eds.), Shame and the Modern Self. 84-96.
Melbourne: Australian Scholarly Publishing.
Hasada, Rie
1996 Some aspects of Japanese cultural ethos embedded in nonverbal
communicative behaviour. In: F. Poyatos (ed.), Nonverbal Communi-
cation in Translation. 83-103. Amsterdam: John Benjamins.
Hofstede, Geert
1997 Cultures and Organisations: Software of the mind. New York:
McGraw Hill.
House, Juliane
1996 Developing pragmatic fluency in English as a foreign language.
SSLA 18: 225-252.
House, Juliane and Gabriele Kasper
1981 Politeness markers in English and German. In: F. Coulmas (ed.),
Conversational Routine. The Hague: Mouton.
Janney, Richard and Horst Amdt
1993 Universality and relativity in cross-cultural politeness research: A
historical perspective. Multilingua 12(1): 13-50.
Kasper, Gabriele and Richard Schmidt
1996 Developmental issues in Interlanguage Pragmatics. Studies in Second
Language Acquisition 18: 149-169.
Katriel, Tamar
1986 Talking Straight: Dugri speech in Israeli Sabra culture. Cambridge:
Kessler, Clive S.
1992 Archaism and Modernity: Contemporary Malay political culture. In:
Joel S. Kahn and Francis Loh Kok Wah, (eds.) Fragmented Vision.
133-157. Sydney: Allen & Unwin.
Kramsch, Claire
1993 Context and Culture in Language Teaching. Oxford: Oxford Uni-
versity Press.
Lewis, Michael
1997 Implementing the Lexical Approach. Hove: Language Teaching
Pub li cati ons.

Authenticated
162 Cliff Goddard
Liddicoat, Anthony, J. and Chantal Crozet (eds.)

2000 Teaching Languages, Teaching Cultures. Melbourne: Language
Australia.
Lo Bianco, Joseph, J. Anthony Liddicoat, and Chantal Crozet (eds.)
1999 Striving for the Third Place: Intercultural Competence through
Language Education. Melbourne: Applied Linguistics Association of
Australia and Language Australia. Matsumoto, Yo
1988 Reexamination of the universality of face: Politeness phenomnena in
Japanese. Journal ofPragmatics 12: 403-426.
Mizutani, o. and N. Mizutani
1987 How to be Polite in Japanese. Tokyo: Japan Times.
Moon, Rosamund
1998 Fixed Expressions and Idioms in English. Oxford: Clarendon Press.
Peeters, Bert
1999 'Salut! 9a va? Vous avez passe un bon weekend?' Journal of French
Language Studies 9: 239-257.
2000 "S'Engager" vs. "To Show Restraint": Linguistics and Cultural
Relativity in Discourse Management. In: Susanne Niemeier and Rene
Dirven (eds.), Evidence for Linguistic Relativity. 193-222.
Schmidt, Richard
1993 Consciousness, Learning, and Interlanguage Pragmatics. In Gabriele
Kasper and Shoshana Blum-Kulka (eds.), Interlanguage Pragmatics.
21-42. New York: Oxford University Press.
Scheibman, Joanne
2001 Local patterns of subjectivity in person and verb type in American
English conversation. In: J. Bybee and P. Hopper (eds.), Frequency
and the Emergence ofLinguistic Structure. 61-89. Amsterdam: Jolm
Benjamins.
Tham Seong Chee
1990 A Study of the Evolution ofthe Malay Language. Singapore: Singapore
University Press.
Travis, Catherine
1998 Omoiyari as a core Japanese value: Japanese-style empathy? In: A.
Athanasiadou and E. Tabakowska (eds.), Speaking of Emotions:
Conceptualisation and Expression. 55-82. Berlin: Mouton de Gruyter.
Valdes, Joyce Merrill (ed.)
1986 Culture Bound. Bridging the cultural gap in language teaching.
Vamarasi, Marit
1995 Bridging the Southeast Asian language-culture gap through teaching
key words. Journal of Southeast Asian Language Teaching. IV (Dec):
1-13.

Authenticated
"Cultural Scripts ": a New Medium for Ethnopragmatic Instruction 163
Wierzbicka, Anna
1972 Semantic Primitives. Translated by Anna Wierzbicka and John
Besemeres. FrankfurtlM.: Athenaum Verlag.
1991 Cross-Cultural Pragmatics. Berlin: Mouton de Gruyter.
1992 Semantics, Culture and Cognition. Oxford: Oxford University Press.
1994a "Cultural scripts": A semantic approach to cultural analysis and cross-
cultural communication. In: L. Bouton and Yamunu Kachru (eds.),
Pragmatics and Language Learning. 1-24. Urbana-Champaign:
University of Illinois.
1994b "Cultural Scripts": A new approach to the study of cross-cultural
communication. In: M. Piitz (ed.), Language Contact and Language
Conflict. 67-87. Amsterdam: John Benjamins.
1996a Semantics, Primes and Universals. Oxford: Oxford University Press.
1996b Contrastive sociolinguistics and the theory of "cultural scripts":
Chinese vs. English. In: M. Hellinger and U. Ammon (eds.),
Contrastive Sociolinguistics. 313-344. The Hague: Mouton.
1997 Understanding Cultures Through Their Keywords. Oxford: Oxford
University Press.
1998 German "cultural scripts": Public signs as a key to social attitudes
and cultural values. Discourse & Society 9(2): 241-282.
1999 Emotions across Languages and Cultures: Diversity and Universals.
Oxford: Oxford University Press.
2002 Australian cultural scripts - 'Bloody' revisited. Journal ofPragmatics
34(9): 1167-1209.
Wilson, Peter J.
1967 A Malay Village in Malaysia. New Haven: HrafPress.
Ye,Zhengdao
2001 An inquiry into "sadness" in Chinese. In: Jean Harkins and Anna
Wierzbicka (eds.), Emotions in Cross-linguistic Perspective. 361-406.
2002 Different modes of describing emotions in Chinese: Bodily changes,
sensations, and bodily images. Pragmatics and Cognition 10(1/2):
321-356.

Authenticated
Authenticated
Grammatical Instruction in the Natural Approach:
a Cognitive Grammar View
Michel Achard
1. Introduction
The place of grammatical instruction in communicative methodologies has

been at the core of the pedagogical literature since the late seventies. The
debate centers on the perceived discrepancy between communicative com-
petence and accuracy in language use. Most practitioners agree that
communicative activities should constitute the most important focus of the
foreign language lesson, and that students' enthusiasm to express themselves
should not be stifled by undue expectations placed upon accurate grammatical
production. On the other hand, however, data from the Canadian immersion
programs (Swain and Lapkin 1989 among many others) have generated fear
that exclusive attention on communication encourages fossilized production,
and some fonn of grammatical instruction is usually deemed desirable in
most communicative approaches. Consequently, the integration of grammar
in communicative models currently constitutes one of the hardest pedagogical
challenges foreign language teachers face, especially at the beginning levels
of instruction.
The nature of grammatical instruction obviously depends critically on
the teacher's view ofthe nature of rules and overall organization of the target
system, as well as her beliefs about the specificity of grammatical knowledge
to language acquisition. Consequently, because they present different views
of what a grammar is and (sometimes) how it is learned, different linguistic
models make different predictions as to what kind of grammar should be
taught. In this paper, I argue that the linguistic model of Cognitive Grammar
(henceforth CG Langacker 1987,1991) offers a way out of the pedagogical
dilemma presented above. I illustrate this position by showing how the reex-
amination of grammatical instruction in the light of CG principles affords a
kind of grammatical instruction fully congruent with the communicative
principles defended by the Natural Approach (henceforth NA) model
(Krashen and Terrell 1983). I argue that the adoption of the CG principles
enhances the model's overall coherence by i) allowing the natural integration

Authenticated
166 Michel Achard
of grammatical instruction into the pedagogical sequence of the NA lesson,

and ii) providing the necessary methodological guidelines to form instructors
to teach communicative grammar.
This paper is organized in the following fashion. Section two presents
some background notions about the relationship between Cognitive Linguis-
tics, acquisition theories, and language pedagogy. Section three summarizes
the basic tenets of the Natural Approach. Section four briefly introduces the
relevant aspects of CG that make it ideally suited for communicative gram-
matical instruction. Section five presents two practical examples of CG's
pedagogical contribution to the NA model. Section six summarizes the
results.
2. Cognitive Linguistics, acquisition, and pedagogy
Because its main concern is the description oflanguage in use (that is, in the
full richness of its psychological and social context) the cognitive linguistics
movement has from its beginnings been interested in the relation between
linguistic description and language learning. The very foundations of the
movement make it well suited for that task (Taylor 1993). For example, the
role of meaning in determining the form of grammatical constructions pro-
vides an intuitively appealing way of teaching those constructions. As we
shall see later in this paper, the knowledge of an expression's meaning and
communicative function greatly facilitates the design of activities to present
it in class. Also, the constructs used in the various cognitive linguistics
models (for example schemas, metaphors, idealized cognitive models,
blends) are socially and culturally rich, and thus lend themselves well to
cross-cultural comparisons that prove useful in the foreign language class-
room. However, in the absence of a fully developed cognitive linguistics L2
acquisition model, the movement's contribution to language pedagogy will
not reach its full potential if cognitive linguists do not incorporate their
descriptive insights into well-established models of acquisition and peda-
gogy.
Second language pedagogy is a truly multidisciplinary endeavor, because
any activity proposed in the classroom (a matter of pedagogy) makes an
implicit assumption about the way in which the target system is organized
(a matter of linguistics), as well as how the target construction is learned (a
matter of acquisition). However, despite the collaboration one would hope
to find given the dependence of each one of those fields on the others, just
the opposite obtains. Each discipline remains surprisingly separate from the

Authenticated
Grammatical instnlction in the Natural Approach 167
others. Linguistic theories are largely unconcerned with matters of acquisi-

tion and teaching. Models of first language acquisition are silent on second
language issues (a notable exception is Bates and MacWhinney 1981). The
inspiration for second language instruction comes from specific L2 acquisi-
tion models as Krashen and Terrell (1983) or VanPatten (1996), among
others. These models are obviously well informed by work in linguistics
and first language acquisition, but their relation to specific models in those
areas is quite indirect.
This situation has created a great deal of frustration over the years.
Linguists often complain that the methods they see being used in language
classrooms do not conform to what they know of language, and they regret
that their expertise is overlooked in the design of methods and activities. On
the other hand, language teachers complain that the linguists' expertise is
simply of little help with practical classroom related matters, because they
do not find in current linguistic theories the elements which would help
them in their daily activities by providing strong theoretical guidelines on
which to base a teaching methodology.
In this climate, it is important to keep separate the issues of language
description, acquisition, and pedagogy, and be very specific about the area
in which claims are being made. As a way of teasing these issues apart, let
us consider VanPatten's (1996) model of L2 acquisition, and particularly of
the processes involved in L2 acquisition. The model is presented in Figure 1.
Il III
Input - - Intake - - Developing system - - Output
I: Input processing Il: Accommodation, restructuring Ill: Access
Figure 1. Processes Involved in L2 Acquisition (from VanPatten 1996: 164)
VanPatten posits three kinds of cognitive processes that account for the pro-
cessing of L2 structures. The first set of processes transforms the input
(what the learner hears) into intake (what she understands). The second set
of principles organizes the intake with respect to the learner's developing
system in L2 (her emergent grammar), and the third kind of processes
extract the appropriate elements of that grammar for specific conversational
purposes. We can use the model in Figure 1 to show precisely how CG can
contribute valuably to L2 acquisition and pedagogy. Let us first note that if

Authenticated
168 Michel Achard
VanPatten is most explicit about acquisition, he also implicitly addresses

pedagogical issues. The pedagogical activities he recommends for the class-
room include providing students with processing guidelines for accurate
comprehension. Such practices are obviously designed to facilitate the
processes at level I, that is, the transformation from input into intake.
CG can most profitably contribute to L2 research at level II, and to some
extent at level Ill. Let us begin with level H. CG provides a linguistic
description of the target system, as well as a way of capturing the patterns
of regularity in that system (the rules of the language). The specific CG
conception of a grammar is presented in Section 4 of this paper. This is, of
course, valuable in L2 pedagogy because instructors constantly need to com-
pare the students' developing system to the native grammar. Furthermore,
CG provides a hypothesis about the way in which linguistic structures are
processed and learned, so the speaker masters the full panoply of choices at
her disposal. I will consider the learning/processing aspect of the theory
more specifically in Section 4.3, but we shall see that the general cognitive
abilities of generalization, categorization, viewing focus, and selection play
a critical role. In terms of the model in Figure 1, that aspect ofCG directly
relates to the passage from intake to the structure of the developing system.
With respect to level Ill, CG directly addresses the speaker's choice to rep-
resent her conceptualization with a specific subset of the expressions she
has at her disposal. The centrality of speaker choice (construal) and the
problem it poses for grammar instruction will be presented in Section 5.2.'
As it now stands, Cognitive Grammar therefore provides both a particular
conception of language organization and learning, as well as a potential
framework for SLA investigation (Achard 1997). The CG SLA research
program involves the precise study of the inner workings of the generalization
mechanisms (such as schema formation, for example) that can be isolated in
the acquisition of the grammar of a second language. The papers by Cadiemo,
Waara, and Lowie and Verspoor in this volume provide successful examples
of CG-inspired SLA research. Ideally, a pedagogical CG would directly
implement the results obtained by SLA investigations. Although a lot of
work still remains to be done, the three aforementioned papers represent a
hopeful sign that it will soon be possible. In the meantime, however,
Cognitive Grammar can still valuably contribute to second language teach-
ing by integrating its descriptive insights into compatible well-established
models of L2 pedagogy. In this paper, I argue that the nature of the target
system as it is described in CG, as well as the learning mechanisms posited
by the model, favor a particular form of grammatical instruction that is fully
congruent with the principles and practices of the Natural Approach. The

Authenticated
Grammatical Instruction in the Natural Approach 169
scope of the paper is therefore exclusively contained within the area ofped-
agogy.
3. The Natural Approach
In addition to its basic compatibility with Cognitive Grammar (which will be

established in Section 4.4), there are several reasons for selecting the NA as a
beginning language method. Firstly, it is one of the original communicative
methods of second language instruction. The model gained prominence in
the early eighties and is still going strong today. Its basic idea is that the
natural conditions of language acquisition outside the classroom should be
reproduced as much as possible in class. Its three main principles are as
follows. One, comprehension precedes production, and students should not
be pushed to speak before they are ready. Two, production is allowed to
emerge in stages, starting with nonverbal responses, and the mistakes that
do not impede communication are not corrected. Three, the syllabus of each
lesson is communicative, not grammatical in orientation. Understandably,
the model received strong endorsement from the proponents of the profi-
ciency movement in L2 teaching in the US who wanted the focus of class-
room practice to shift from noncommunicative grammar-driven activities to
communicative task-driven activities. Secondly, the communicative syllabus
and overall approach provide a great deal of pedagogical flexibility, where
each instructor can find her own teaching style. Finally, the method is fully
worked out, that is, its working constructs and techniques (different kinds of
activities, input techniques, error correction techniques, for example) are
well understood and described. This makes it easier to ensure the model's
pedagogical consistency by forming instructors to use a well-tested method-
ology.
3.1. The Natural Approach View of Acquisition
The model's pedagogical method is designed to follow a specific view of

acquisition. That view centers around the five hypotheses I will be summa-
rizing in this section. The order in which they are presented here differs from
Krashen and Terrell (1983) for expository purposes.
The Input Hypothesis "states that we acquire (not learn) language by
understanding input that is a little beyond our current level of acquired
competence" (Krashen and Terrelll983: 32). Speaking ability is viewed as

Authenticated
170 Michel Achard
naturally emerging from exposure to written (reading) and spoken (listening)

input. It is therefore not taught formally, rather, "it emerges after the acquirer
has built up competence through comprehending input" (Krashen and Terrell
1983: 32). This hypothesis stresses the use of context and extra linguistic
information to acquire structures the students do not yet possess. Importantly
for our purposes here, Krashen and Terrell note: "the Input Hypothesis thus
claims that we use meaning to help us acquire language" (1983: 32).
Perhaps the most important corollary to this hypothesis is that input does
not need to be finely tuned. The teacher is not restricted to using structures
that the student is familiar with already. To the contrary, the input presented
must be somewhat new, although comprehensible. The authors' famous i+ 1
formula states that: "an acquirer can 'move' from a stage i (where i is the
acquirer's level of competence) to a stage i+1 (where i+l is the stage imme-
diately following i along some natural order) by understanding language
containing i+ I" (Krashen and Terre1l1983: 32). Consequently, as long as the
students comprehend the input, there is no restriction on the kind of structures
that can be used to present it. Obviously, the nature of what the students can
understand places restrictions on the input, and instructors learn to alter their
natural speech to produce "caretaker speech" that caters to the general level
oftlleir audience.
The Affective Filter Hypothesis states that attitudinal variables relating
to success in second language acquisition may directly relate to language
acquisition. Based on the work of Dulay and Burt (1977), Krashen and
Terrell suggest that successfulleamers have a low affective filter, meaning
that they are more open to the input. Conversely, a high affective filter pre-
vents input from being processed properly. Krashen and Terrell's represen-
tation of the effect of the affective filter is given below.
Filter
Language
Acquired
Input Acquisition
Competence
Device
Figure 2. Operation of the "Affective Filter" (From Krashen and Terre1l1983: 39)
This diagram is interesting because it contains one of the very few references
to a Generative model with the mention of the Language Acquisition Device
(Chomsky 1965). The Chomskyan orientation of the NA acquisition model

Authenticated
is seldom explicitly stated, but it is undeniably present in the remaining

hypotheses. This will be considered in the next section.
The Acquisition-Learning Hypothesis and the Monitor Hypothesis go
together, and they largely determine the kind of grammatical instruction
proposed in the NA. They are therefore crucial to the issues considered in
this paper. The Acquisition-Learning Hypothesis states that adults have two
ways of developing competence in a second language. They can acquire the
language unconsciously, by merely using it to communicate like children
do. The result of that acquisition is also an unconscious process by which
adults only have a feeling of correctness or error at production, but they
have no conscious knowledge of how they achieve production.
The second way to develop L2 competence is through learning, a con-
scious process that utilizes rules and generalizations to produce and correct
speech production. Through learning, the learner gains explicit access to the
rules of the language, and she can use that knowledge to reflect upon her
own production. However, the Monitor Hypothesis determines the scope of
the rules that are consciously available to the learner, and can therefore fos-
ter learning. It states that "learning has only an extremely limited function
in adult second language performance: it can only be used as a Monitor or
an editor" (Krashen and Terre111983: 30, emphasis in the original). It can
therefore act only on those areas of language that are conscious.
In order for the Monitor to be effective, the learner has to have enough
time to be thinking about correctness, or be focussed on form, and, obvi-
ously, she has to know the rule. Furthermore, the Monitor is restricted to
specific areas of grammar, namely, to simple rules. Simple rules are defined
as "rules that do not require elaborate or complex movements or permuta-
tions", or are "difficult due to their semantic properties" (Krashen and
Terrell 1983: 31). The example Krashen and Terrell give ofa Monitor-sensi-
tive rule is the -s third person agreement in English, which is better mas-
tered when the learners have enough time to think about the correctness of
their discourse.
Finally, the Natural Order Hypothesis reinforces the fact that acquisition
is largely opaque to teaching and follows its natural course given a sufficient
amount of comprehensible input. It states that grammatical structures are
acquired in predictable and set order, regardless of their order of presentation
in the curriculum. Evidence used in favor of this position comes mostly
from Krashen (1981), who provides the order presented in Figure 3 for
English:

Authenticated
172 Michel Achard
ING (progressive)
PLURAL
COPULA (to be)
1
AUXILIARY (progressive)
ARTICLE (a, the)
1
IRREGULAR PAST
r
REGULAR PAST
III SINGULAR (-s)
POSSESSIVE (-s)
Figure 3. Average Order of Acquisition of Grammatical Morphemes for English as

a Second Language (Children and Adults). From Krashen and Terrell
(1983: 29).
Taken together, these five hypotheses constitute a model of second language

acquisition. It can be summarized in a nutshell in the following way.
Language acquisition occurs naturally provided the learner is exposed to a
sufficient amount of comprehensible input in a low anxiety environment.
The NA teaching method implements this position by proposing a teaching
model based on a high level of input in the target language, communicative
activities, and a relaxed atmosphere.
3.2. Grammar in the Natural Approach
The teaching of grammar in the NA is problematic. First, it is necessarily

limited in scope by the learning-acquisition dichotomy. By definition, gram-
matical activities can only be reserved to the rules that can be accessed by
the Monitor. However, these rules are difficult to isolate precisely. Further-
more, it is unclear how the Monitor interacts with the Language Acquisition
Device. This leaves the instructor without any foundation on which to base

Authenticated
grammatical instruction. Consequently, grammar is often simply not taught,

and the NA has carved its niche in the market of methodologies as a gram-
marless method, where grammatical instruction is viewed as a possible
home activity for learners. The sole emphasis of class time is communica-
tive competence.
As indicated earlier, however, this position has become increasingly sus-
pect because of the fear of fossilized production. Terrell himself in his later
work (TerrelI1991) acknowledges some place for grammatical instruction,
provided that it is congruent with the NA principles. This is where the prob-
lem lies. The Generative flavor of the posited acquisition device renders the
reintroduction of grammar difficult, because the LAD is largely opaque to
teaching. In the reminder of this paper, I show that this problem vanishes if
we replace the Generative conception of language and acquisition by the
CG perspective. Importantly, this move does not alter the pedagogical
integrity of the NA model; it merely changes its assumptions. The primary
focus on input remains unchanged, and a great part of the acquisition still
occurs unconsciously. However, the learning mechanisms posited by CG
are not opaque to teaching, and they can be enhanced by specific pedagogi-
cal practices. I therefore argue that by adopting different views about the
nature of the target language and the way in which it is learned (the
processes at level II in Figure 1), we can provide the kind of grammatical
instruction that is fully congruent with the model's basic pedagogical orien-
tation. Section five will show how these new views can be put into effect in
the everyday routine of a foreign language lesson. Before that, however, we
need to familiarize ourselves with the aspects of CG that are the most criti-
cal to the purposes of this paper.
4. Cognitive Grammar
In this section, I briefly present the aspects of the CG framework that make it
directly compatible with the view of L2 pedagogy defended by the Natural
Approach, and thus enables it to provide a satisfactory frame for communi-
cative grammatical instruction.
4.1. Construal
The main objective of CG is to describe the semiological function of lan-

guage, that is, the symbolic association between meanings and forms. All

Authenticated
174 Michel Achard
linguistic expressions are symbolic units that contain a phonological and a

semantic pole. All units are thus meaningful, and the description of their
meaning constitutes an essential aspect of CG investigation. In the CG con-
ception of language, meaning is equated with conceptualization, to be expli-
cated in terms of cognitive processing. It is thus anthropomorphic and sub-
jective, and includes, besides the objective properties of the described
object, the way in which the conceptualizer chooses to present it. Each
expression is characterized by the specific profile its presence imposes on a
conceptual base, that is, how precisely it structures that base. For example,
the meaning of Tuesday is characterized as the appropriate subsection of our
social organization of time. The profile of an expression represents the way
in which a given conceptualizer construes the relevant base. Consequently,
because they present a different profile on the same base, two competing
expressions express alternative construals of that base.
The meaning of grammatical constructions is also characterized as the
kind of profile they impose on a base. Let us, for example, consider two
French causative constructions that differ from each other in their word
order. VV and VOV are respectively introduced in (1) and (2).
(1) Marie a laisse partir Jean

Mary has let leave John
'Mary let John leave'
(2) Marie a laisse Jean partir

Mary has let John leave
'Mary let John leave'
In previous work (Achard 1993, 1996) I argued that the two constructions
have their own specific semantics, characterized by the specific profile they
each impose on a dynamic scene. The choice between the constructions
depends on whether the causee (John) is presented as the energy source that
initiates the process coded by the infinitive in the complement. If John is
presented as the initiator of the leaving, VOV (illustrated in 1) is chosen. If
John is not construed as the initiator of the infinitival process, VV (illus-
trated in 2) represents the favored choice.
The notion of construal is extremely important for the purposes of this
paper. In Section 5, I will consider in detail how this notion critically impacts
the teaching of grammar. For now, I will simply note that since the meaning
of a construction is a matter of conventionalized choice, its distribution in dis-
course is determined by the speaker's decision rather than by the system itself.

Authenticated
4.2. A structured Inventory of Conventionalized Linguistic Units
In the CG view, the internal structure of a language is very homogenous.

Langacker (2000: 8-9) writes (emphasis in the original):
In Cognitive Grammar, a language is described as a structured inventory

of conventional linguistic units. The units (cognitive routines) comprising
a speaker's linguistic knowledge are limited to semantic, phonological, and
symbolic structures which are either directly manifested as part of actual
expressions, or else emerge from such structures by the processes of abstrac-
tion (schematization) and categorization (this restriction is called the con-
tent requirement). In describing these units as an inventory, I am indicating
the non-generative and non-constructive nature of a linguistic system.
Linguistic knowledge is not conceived or modeled as an algorithmic device
enumerating a well-defined set of fonnal objects, but simply as an extensive
collection of semantic, phonological, and symbolic resources which can be
brought to bear on language processing.
This conception of the grammar of a language represents an important con-

tribution to foreign language teaching because it enables the teacher to treat
structural learning (the learning of grammatical constructions) on par with
lexical learning. There are no special devices that interpret grammatical
forms and remain invisible to teaching. Nor are there specific rules that yield
linguistic expressions as their output. Rather, all linguistic expressions,
regardless of their level of complexity, represent a form/meaning pair, with a
semantic and phonological pole. Each one is individually learned, and there-
fore amenable to the same presentational pedagogical techniques. Let us
come back to our previous example of French causative constructions.
Because VV and VOV are meaningful, their form can be accessed through
their meaning in a way identical to lexical learning. Activities can thus be
designed where students learn to associate VV with situations of low levels
of agentivity of the causee, and VOV with situations where the causee is
more agentive. This pedagogical approach is exactly identical to the one
used to present two competing lexical items.
Although all linguistic expressions are similar in their symbolic nature,
they obviously differ in their level of complexity. Grammatical rules take the
form of constructional schemas. These schemas can be viewed as templates
that generalize over existing expressions and sanction their felicity.
Importantly, they are also symbolic in nature, i.e., they have a semantic and
a phonological pole. The content of these poles is, however, more abstract
than that of the actually occurring expressions which instantiate them.

Authenticated
176 Michel Achard
Importantly, these rules or templates cohabitate in the grammar with actually

occurring expressions they abstract over. The cohabitation of the schema
(the rule) and the occurring expressions validated by the rule is illustrated in
Figure 4 by the rule of plural formation (-s) (from Langacker 1988: 131).
SCHEMA
~0
0CIJ QUAGMIRE
quagmire
§J0
~G
~
REE
I'
tree
BjL
,
-8
11
FIXED EXPRESSIONS NOVEL EXPRESSION
Figure 4, Plural Formation
Figure 4 represents the cohabitation in the grammar of the schema (the

rule), occurring expressions, and novel expressions, In the CG model, the
generalization over a set of data (knowledge of the rule) is not incompatible
with the specific learning of individual pieces of data. A schema and its
instantiation represent two different facets of linguistic knowledge. Their
coexistence in the grammar represents different ways for the speaker to
access a complex but regular expression with unit status.
4,3, Processing and Learning
As was mentioned earlier, no pedagogical decision can be made in the

absence of a learning theory, Consequently, it is not enough to describe the
CG grammar to validate methodological choices, but we need to consider
more specifically how learning is done in the model.
The linguistic structures and relations mentioned above ultimately reside
in cognitive processes, identified as neurological activity, CG, like other
models of language and linguistic acquisition such as the Competition

Authenticated
Grammatical Instnlction in the Natural Approach 177
Model (MacWhinney 1987) for example assumes a connectionist style of

computation, based on PDP (Rumelhart and McCleeland 1986, Gasser
1990), where learning resides in the adjustment of the connection weights
between input and output units.
The most basic notion of CG, entrenchment (what gives unit status to a
particular expression) can be identified by an adjustment of the connection
weights, brought about by the occurrence of a specific pattern of activation
which renders more likely the occurrence of the same or a similar pattern.
In dynamic terms, entrenchment represents the emergence of an attractor.
Patterns of regularity (schemas) are extracted from the data according to con-
nectionist principles. Each expression or structure is equated with a point in
space state (or a trajectory through it) (Elman 1990). If two expressions are
similar, their patterns of activation will be neighbors in space state, (or two
lines with similar trajectories). The patterns representing similar structures are
therefore grouped together to form a specific region of space state. The occur-
rence of a given pattern renders more likely the activation of neighboring
patterns. The repeated use of similar patterns therefore facilitates the occur-
rence of any pattern in the general area. This amounts to extracting a schema.
Another important construct of CG, categorization, is identifiable as
capture by an attractor. Presenting the system with a certain input tends to
activate different patterns. When input X results in full activation of pattern
Y (which may have won over other patterns), we can say that Y is used to
categorize X. Any structure is available to serve as categorizing structure,
and thus, via categorization, an increasing number of structures are organ-
ized in complex relational networks (allophones of a phoneme, related
metaphors, variants of a grammatical construction). Importantly, each target
for categorization tends to activate different established units, anyone of
them being capable of categorizing it (the activation set of the target).
Members of the activation set compete for being the categorizing structure
of the target. The winning one, or active structure, categorizes the target.
These principles of processing and learning must be applied to the com-
plexity of linguistic production. We can now see the processing dynamics
by which a language event can be described by a set of linguistic expres-
sions. Any language event has many facets that can be categorized by many
established linguistic units. These units are generally organized in compli-
cated relational networks. By virtue of overlapping content, each facet of
the language event serves to activate a set of potential categorizing struc-
tures (the activation set), which compete to categorize it. The winners of
each competition are determined by the dynamic processes of activation,
entrenchment, and inhibition considered previously.

Authenticated
178 Michel Achard
4.4. CG and the Natural Approach
It should be obvious that the view oflearning defended by CG is incompati-

ble with the one advocated by the NA. To take just one example, the distinc-
tion between learning and acquisition has no possible correlate in a system
where grammar is composed of a structured inventory of symbolic units.
Therefore, this paper's claim that the CG view of language can valuably be
incorporated in the NA teaching method can only be maintained if we can
show that i) the NA classroom practices are developed with a certain degree
of independence from the model's position on acquisition, and ii) the CG
view is fully compatible with the NA's pedagogical position. I will consider
these two points in turn.
The separation between the theory and practice of the NA is rendered
necessary by the eclectic response the model has received over the years. Its
pedagogical style "touched a chord for many practitioners" (Omaggio Hadley
1993: 54), but its stance on acquisition has been heavily criticized. First, the
similarity between L 1 and L2 learning has come under heavy attack.
Whereas the child's acquisition of a linguistic system is accompanied by the
acquisition of the conceptual categories of her community, the adult learning
a second language already possesses a fully-developed set of conceptual
categories, along with a linguistic system to code them. Consequently, the
NA is not equipped to properly evaluate the influence of transfer phenomena,
that is, the interference of the native systems on the developing target sys-
tems. The Acquisition / Learning Hypothesis has also received its share of
criticism. Munsell and Carr (1981) question the distinction between the two,
as well as the notion of conscious versus unconscious rules. McLaughlin
(1987) argues that it is too vague, and that the claim that learning does not
become acquisition is thus untestable. Finally, the Natural Order Hypothesis
is based on too few morphemes from English alone to be considered univer-
saL Furthermore, its predictive power is greatly reduced by methodological
problems. McLaughlin (1987: 56) writes: "If the Natural Order Hypothesis
is to be accepted, it must be in a weak form, which postulates that some
things are learned before others, but not always."
Perhaps unexpectedly, the model's popUlarity among teachers was virtu-
ally unaffected by the aforementioned criticisms. Language practitioners
were attracted by its communicative flow, and interpreted the five hypothe-
ses with common sense rather than theoretical zeal. Any good teacher will,
for example, naturally try to make her students feel as comfortable as possible
in class. The hypotheses that resisted common sense interpretation (such as
the Natural Order Hypothesis) were simply disregarded. Even the acquisition

Authenticated
Grammaticallnstruction in the Natural Approach 179
learning distinction was welcome as a license to focus on communicative

activities rather than a serious statement about how people learn a second
language. If we exclude the issue of grammatical instruction, we can thus
reasonably say that the Natural Approach enjoyed its pedagogical success
independently from its theoretical position.
In a way that mirrors this separation between theory and practice, the NA
relation to its Generative roots is also quite indirect. For example, in the for-
mulation of the model found in Krashen and Terrell (1983), the Generative
hypothesis is merely used as an assumption about language organization and
learning that validates the model's stance on acquisition. There is no critical
reliance on specific constructs of the theory such as the parameter setting
mechanism, for example, just a general mention of the Language Acquisition
Device (see Figure 2). In these conditions, given i) the independence of the
NA's pedagogical approach from its theoretical foundations, ii) the heavy
criticism aimed at the model's position on acquisition, and iii) the loose rela-
tion existing between that position and the Generative conception oflanguage
upon which it is based, it is indeed possible to replace the Generative hypo-
thesis by the CG conception of language. This move will obviously change
the assumptions upon which the model's acquisition theory is built, but also
(and more importantly for our purposes here) strengthen its pedagogical
force by making it possible to teach grammar in a communicative manner.
Because in CG the learning mechanisms are not opaque to teaching, class-
room practice can be constantly enriched by new fmdings in acquisition. 2
The second stage of the integration of the CG view oflanguage and the
NA teaching model involves showing the close connection between the two.
In fact, Cognitive Grammar is not only congruent with the Input Hypothesis
(the most important one for our pedagogical agenda), but it explains why
input should be so critical to language learning. Since linguistic production
is largely a matter of construal, the user exercises her choice to use whatever
expression best fits her conceptualization for expressive purposes. Conse-
quently, she is at the center of a "usage event". Langacker (2000: 9-10,
emphasis in the original) writes:
It is not the linguistic system per se that constructs and understand novel
expressions, but rather the language user, who marshals for this purpose the
full panoply of available resources. In addition to linguistic units, these
resources include such factors as memory, planning, problem solving ability,
general knowledge, short and long-term goals, as well as full apprehension
of the physical, social, cultural, and linguistic context. An actual instance of
language use, resulting in all these factors, constitutes what I will call a
usage event.

Authenticated
180 Michel Achard
The notion of a usage event explains why input is so critical to L2 learning.

Because language is largely a matter of conventionalized choice, input in
the language represents the only window on the conditions of that choice.
To place it in the learner's perspective, witnessing the live conditions under
which real language events occur represents her only chance to reproduce
similar events on her own. Learning a language is thus not only a matter of
learning a system (which can be done through exposure to the rules of that
system), but understanding the conditions that preside over the choice of
different constructions. This can only be done by witnessing such choices
being made. By placing the psychological and social context of an utterance
at the core of linguistic production, both the CG and NA models render the
language learning act indissociable from the full richness of witnessed lan-
guage use. Despite the differences between the two models considered pre-
viously, that common position alone justifies putting them together.
5. CG Inspired Grammatical Instruction
We are now ready to consider the practical applications of the adoption of a

CG view of language in the NA teaching method. I will more specifically
concentrate on two areas, namely i) the place of grammatical activities in
the natural cycle of a NA lesson, and ii) the particular pedagogical problems
posed by the centrality of the notion of construal.
5.1. Grammatical Activities in the NA
The adoption of the CG conception of language influences both the general

form of the grammatical instruction dispensed in the method, as well as the
sequencing of the activities in the classroom.
In general terms, CG-inspired grammar instruction will respect three
important communicative principles. First, it will not be metalinguistic. The
purpose of grammatical instruction is to allow the learner to focus on the
conditions that motivate specific structural choices. The most efficient way of
achieving that goal is therefore not to explain those choices from the outside
in a metalinguistic manner, but to illustrate them as much as possible. Since
the rules are generalizations over specific instances, the best explanation for
them consists in facilitating the recognition and access to the individual
instances. Secondly, grammatical instruction will be inductive to allow for the
natural process of schema extraction. Since grammatical rules are emergent,
grammar is necessarily an inductive phenomenon. Inductive instruction is

Authenticated
Grammatical Instrnction in the Natural Approach 181
likely to be more effective than deductive instruction because constructional

schemas are extracted from a mass of occurring expressions. Note that it is
also clearly more communicative than a deductive presentation because
constructional schemas are extracted out of real language use. The meaning-
ful input that "drives" acquisition can thus successfully be applied to gram-
matical constructions as well. Inductive grammatical instruction is hardly
new, and it is recommended in numerous works on grammar instruction, by
researchers working independently of Cognitive Grammar (Herron and
Tomasello 1992, VanPatten 1992), for example. However, CG constitutes a
theoretical validation for those positions, and provides the necessary tools
to investigate the constructions' functional import. Once a construction's
meaning is available, the instructor's role is to find the best way to help the
student inductively grasp it and associate it with a particular form.
Thirdly and expectedly, the teaching of grammar will be done in the target
language. The greatest temptation to use the source language occurs when
the instructor is faced with the difficult description of abstract concepts,
most typically metalinguistic grammatical notions. In this model, this temp-
tation is greatly reduced because grammatical instruction involves present-
ing the input in such a way as to stimulate schema formation. This is not to
say that explicit instruction is never possible. As we will see later in this
section, the level of explicit instruction is largely a matter of individual pref-
erence, and it can be done succinctly and efficiently in the target language.
Beyond these general guidelines, the CG model allows grammatical
instruction to take its proper place in the course of the NA lesson. Unlike the
mechanisms of the LAD, the operations of schema extraction, entrenchment,
and categorization that account for learning in CG can be facilitated by such
pedagogical techniques as order of presentation, isolation, repetition, etc.
Furthermore, grammatical knowledge does not constitute the absolute core
of language learning, but merely represents one dimension of linguistic
knowledge. Consequently, the teaching of grammar will not hold a privi-
leged position in the CG-inspired communicative classroom. Rather, it will
take its natural place in the sequence of pedagogical activities that consti-
tutes the NA classroom.
A typical NA lesson starts with an input phase that triggers the acquisition
process. The students are exposed to as much comprehensible input as pos-
sible, loosely centered around the targeted semantic field. The input phase is
designed for the students to recognize the relevant vocabulary and under-
stand most of what relates to the targeted semantic domain. Following that
input phase, a series of communicative activities gradually guide the students
from comprehension to full production by placing them in progressively less

Authenticated
182 Michel Achard
and less structured situations in the target area. I propose that grammatical
activities simply follow the activities phase within each targeted semantic
domain. This allows grammatical instruction to benefit from all the acquisi-
tion that took place in the beginning of the lesson. Since the input phase, the
students have been exposed to specific instances of the targeted grammatical
constructions, but they have not focused on their commonalties to extract
patterns of generalization. Grammatical focus is thus most appropriate when
the students are comfortable with the semantic field of the lesson and can
concentrate on extracting patterns of regularity from that field.
The transition between the activities phase and the grammar phase should
be so smooth that even though the instructor needs to work at it very hard,
the students should not even realize they are doing grammar. In fact, gram-
matical activities are communicative activities with one added particularity.
Because the meaning of grammatical constructions is more abstract than
that of lexical items, grammatical activities should be more repetitive and
focused than other activities. The learners need to be exposed to more
instances of the structure in order to learn it, and they also need more
focused activities to successfully associate the construction's form to its
meaning. This represents the most difficult aspect of grammatical instruc-
tion. The activities need to be naturally repetitive and focused. The idea is to
find situations where the repetition and focus are both natural and functional.
Let us take an example to illustrate the principles of communicative
grammatical instruction proposed in this paper. One of the first grammatical
points considered in the NA French textbook Deux Mondes (Terrell et al.
2001 Chapitre 1) concerns the generalization that some verbs, most notably
the verbs of preference such as aimer 'like', detester 'hate', or prejerer
'prefer' are directly followed by an infinitive (J'aimejouer 'I like to play').
The context for this presentation (the theme of the lesson) is Mafamille et
moi 'My family and myself', where learners are exposed to new vocabulary
and structures concerning the hobbies and pastimes families enjoy together.
Consistent with the NA practice, no grammar is presented in the course of
the lesson proper, but a special grammatical section in English (in different
colored pages) is cross-referenced with the relevant activities. This format
sets up grammatical instruction as a side activity, its remote physical loca-
tion clearly indicative of its peripheral role within the method.
By contrast, in the kind of methodology proposed in this paper, gram-
matical instruction needs to be fully integrated in the lesson. In fact, it is
important to know precisely what the students have already acquired from
other parts of the lesson to know both what the grammar presentation can
assume as background knowledge, and what it needs to concentrate on. The

Authenticated
Grammatical Instntction in the Natural Approach 183
activities phase, in particular, provides very good preparation for the stu-
dents' later focus on form. For example, in Chapitre 1 activity 5, (Terrell et
al. 2001: 50), one student reads a statement expressing her like or dislike for
specific activities, and another student agrees or disagrees with that state-
ment by saying oui or non. The model for the activity goes as follows.
Student 1: Pendant les vacances,j'aime voyager 'during the vacation I like
to travel'. Student 2: Moi aussi,j'aime beaucoup voyager 'I like to travel a
lot too' (Moi, non! Je n 'aime pas voyager 'I don't like to travel'). The goal
of this activity is for students to acquire the meaning of the activities rather
than the form of their expression. However, they are obviously exposed to a
lot of "grammar". The narrow range of the activity allows for repetition,
and the infinitival forms are clearly available for students to infer their gen-
eralization. The only missing element is some form of emphasis placed on
that generalization. 3 By placing the students in situations where they prac-
tice individual instances repetitively, such activities provide the necessary
background for the grammatical activities where the focus is on what those
instances have in common.
After going through a series of similar activities, and prior to the gram-
matical presentation of the V+INF rule, the students are able to understand
almost everything that relates to likes and dislikes, generate some likes and
dislikes, and recognize the forms of aimer 'like'. The goals of the grammat-
ical activity are thus to i) bring to their attention the fact that some verbs are
directly followed by an infinitive, ii) show that French infinitives are of 3
kinds (-er, -ir, -re), and iii) show that most verbs have an -er infinitive. 4
In order to achieve these goals, the grammatical activities need to be nat-
ural, focused, and repetitive enough to favor the emergence and entrenchment
of the three infinitival schemas, but nonetheless involve a valid communica-
tive task. 1 propose the following as a possible example, simply because it
has already been used with success in beginning classes. The details of the
activity do not matter much. Language instructors possess the necessary
resourcefulness and creativity to design their own activities. What is more
important is the spirit in which the activity is created, as well as the mindset to
incorporate similar ones in the flow of the NA class. The instructor "plants"
a backpack in the classroom filed with objects chosen to evoke specific
activities. For example, a book evokes reading, a tennis ball evokes playing
tennis. At the relevant point of the lesson (following the communicative
activities where the students have practiced different likes and dislikes), the
instructor notices the backpac~ and asks whom it belongs to. When nobody
claims it, she opens it to find out if it contains a clue that would identify its
owner. Obviously no such cue exists, but the objects inside indicate what

Authenticated
184 Michel Achard
the owner does (and can thus be assumed to like). The instructor pulls out
the objects one by one, and the students indicate what the unknown person
likes to do. For example, when being shown a running shoe, they usually
volunteer elle aime courir 'she likes to run', or elle aime faire du sport 'she
likes to play sports'. Student production is then written on the board in three
columns that clearly indicate the relevant morphological groupings. The
length of the -er column shows how prevalent that pattern is in French. At
this point, additional explicit instruction exclusively depends on the instruc-
tor's taste. If deemed necessary, it is quite easy to briefly comment on the
patterns on the board in French to further focus the student' attention on the
formal aspects of their own production. This grammatical activity is obvi-
ously quite similar to the other activities proposed in class daily. However,
its repetitive nature, narrow scope (it's all in the bag!), and order of presen-
tation allow the students to focus on form, which contributes to the entrench-
ment ofthe V+INF schema and its -er, -ir, and -re instantiations.
Beyond this particular example, the practice of grammatical instruction
as it is defended in this paper involves three steps. First, the selection of a
particular point that deserved to be shown in a grammatical light. Secondly,
the design of an activity that will enhance the form of the targeted construc-
tion. Finally, the planning of the sequence of activities, so that grammatical
instruction most efficiently draws on the practice provided by both the input
and activities phase.
To briefly summarize this section, we can note that communicative
grammatical instruction impacts the NA lesson in a minimal way, because it
makes use of similar kinds of activities instructors routinely use to teach
lexical items. Grammatical activities are simply more narrowly focused and
naturally repetitive than the others, and their strategic integration in the
course of the lesson allows the students to focus on the relevant patterns
most efficiently. From this perspective, grammatical instruction does not
involve the presentation of a different domain of knowledge. The students
simply consider the same semantic domains they are exposed to in the
course of their communicative syllabus through a narrower lens. This is, of
course, consistent with the CG position that the rules of grammar cohabitate
with their specific instances in our linguistic knowledge.
5.2. Teaching Construal
The second aspect of CG-inspired grammatical instruction is perhaps more

radical and difficult to implement because it calls to question the velY way in

Authenticated
Grammatical instruction in the Natural Approach 185
which learners approach the rules of a foreign language. Recall from Section
four that linguistic production is primarily a matter of speaker choice or
construal. The centrality of construal makes strong implications for teach-
ing. It is not the system per se that is responsible for the form of specific
constructions just by being in a certain way, but the shape of novel expres-
sions is detennined greatly by speaker choice. In that sense, the study of the
mechanisms of that choice belongs to the access processes (level HI) pre-
sented in Figure 1.
The centrality of the speaker to the form of the utterance seems trivial.
However, it is important for pedagogical purposes. Focusing on the speaker
rather than on the system represents the difference between teaching set pat-
terns of lexical associations and teaching construal, that is, the convention-
alized way of matching certain expressions to certain situations, as well as
the flexibility of using the available alternatives to express specific semantic
nuances. This difference is significant for the instructor. Grammatical rules
traditionally given in a language class are considered a property of the sys-
tem, and not a result of the speaker's choice. As an example, learners of
Romance languages are taught that the subjunctive mood follows a specific
list of verbs. Beside its specific morphological shape, learning the subjunc-
tive mood therefore amounts to learning the verbs it occurs with. The fact
that a given verb takes the subjunctive is viewed as a lexical property of that
verb. Selectional restrictions are thus marked in the lexicon, and the stu-
dents learn them at the same time as the verb's meaning. Such rules are
deeply entrenched in our pedagogical history. They are also easy to state (if
not to follow) because they yield a stable system of co-occurring expres-
sions that can easily be apprehended through memory. Difficulties only
arise when learners become aware of the numerous exceptions that creep
into the supposedly regular system. Teaching construal, on the other hand,
is much harder because of its inherent flexibility. It is methodologically dif-
ficult to provide clear guidelines that teach people how to exercise choice.
In the case of a foreign language, in addition to providing the students with
the whole range of conventionalized options, instructors need to be able to
show them the precise conditions under which natives would use a particu-
lar construction.
In order to illustrate the differences between a model of lexical associa-
tion, where the constructions are considered part of the system, and the con-
strual-based approach recommended by CG, Twill now consider the case of
the distribution of definite and partitive articles in French. In addition to the
definite le, la, les 'the', and indefinite un, une, des 'a', French has a series of
partitive articles du, de la, which are often translated in English by 'some',

Authenticated
186 Michel Achard
and indicate an idea of quantity. In the overwhelming majority of beginning

French textbooks, these articles are presented in the chapter dealing with
food, and the way in which they are presented emphasizes their co-occur-
rence with specific verbs. Deux Mondes is no exception, and in Chapter 7
the students are given the following advice: (Terrell et al. 2001: 245,
emphasis in the original):
To choose the appropriate article, look at the kind of verb used in the sen-
tence. With verbs describing likes or dislikes, such as aimer, adorer,
detester, preferer, use the definite article because you are talking about
things in a general sense, Nathalie aime beaucoup les carottes et les petits
pois, mais elle deteste les epinards 'Nathalie likes carrots and peas, but she
detests spinach', Je n'aime pas le cafe fort 'I don't like strong coffee'.
On the other hand, if the verb deals with having, obtaining, or consuming,
use du, de la, de l', or des, because you are talking about some amount of a
thing. Such verbs include avoir, acheter, manger, boire, prendre, and
many others. Les Fran~ais boivent du cafe apres le diner 'the French drink
coffee after dinner'. Nous mangeons de la pizza tous les vendredi soir 'We
eat pizza every Friday night'.
This quote is addressed to the students, and can therefore be interpreted as a

helpful suggestion to take advantage of a general tendency. However, the
instructor receives similar advice as to the best kind of presentational tech-
nique to use to present article distribution (Terrell et al. 2001: 245): "Point
out that it is the verb of which the given noun is the object that determines
the article choice, e.g., j'aime le cafe, but j'aime boire du cafe tous les
matins".
As it is presented in Chapter 7 of Deux Mondes, article selection is
clearly a property of specific verbs, as illustrated by the strength of the verb
"determines" in the advice to the instructor. The problem with this approach
is not only that it does not cover the whole range of French article selection.
The students who use the book are, after all, just beginning to learn the lan-
guage. The association of the definite and partitive articles with the liking
and consumption verbs respectively would be perfectly acceptable if the
exceptions to the distribution rules were only arcane expressions not to be
encountered until much later. This, however, is not the case, and students
rapidly become confused when they realize that the French system of article
selection is actually much more flexible and subtle than its presentation had
led them to believe. In fact, the very example used in the instructor's edition
J'aime boire du ca,fe tous les matins could very well be J'aime boire le cafe

Authenticated
tous les matins with a slight difference in meaning to which we will come
back later in this section. This is particularly important in a method such as
the NA that places such strong emphasis on input. The instructor is expected
to use the widest possible range of comprehensible expressions in her
speech in order to enrich the students' understanding of a given domain.
When talking about drinks and socializing, the students are thus likely to
hear bothj'aime boire du cafe andj'aime boire le caFt Admittedly, they
will not experience a total breakdown in comprehension, but they may be
confused, or simply miss the cultural dimension of the meaning distinction
between the two forms.
The reality of usage is such that in numerous cases, the expected co-
occurrence of the consuming verbs with the partitive on the one hand, and
the liking verbs with the definite article on the other hand, is violated. The
following are a few random examples:
(3) Two roommates discuss their shopping habits:

On achete toujours de la biere et du vino Lui il boit la biere, et moi je
bois le vino
'We always buy beer and wine. He drinks the beer, and I drink the
wine'
(4) On achete toujours la biere au Casino

'We always buy the beer at Casino'
(5) A la cantine, hier, pour la premiere fois j'ai pris le poulet, c'etait pas
mauvais
'At the cafeteria yesterday for the first time I took the chicken, it was
not bad'
(6) - 11 n 'y a plus rien au Jrigo!

- Oui, je sais, j 'ai bu la biere et le vin, deso!e mais j 'avais trop soif!
- 'There is nothing left in the fridge'
- Yes I know, I drank the beer and the wine, sorry, but I was too
thirsty!'
(7) a. Tu prends le cafe?

b. Tu prends du cafe?
c. Tu prends un cafe?
'Do you take the/some/a coffee?'

Authenticated
188 Michel Achard
Note that in (3) and (6), the consumption verb boire is unexpectedly fol-
lowed by the definite article. In the same way, prendre is also followed by
the definite article in (5). Furthermore, some verbs such as acheter in (3)
and (4), and prendre in (7) can felicitously be followed by different articles.
Finally, partitive articles can also follow the liking verbs, as shown in (8):
(8) Moi, ce quej'aime bien, c'est de la conjiture avec du pain!

'Me, what I really like is some jam with some bread'
This small sample of examples should suffice to show that pure lexical
association is incapable of rendering the complexity of article distribution in
French, even from the beginning. To explain the exceptions, teachers have
to backpedal from the convenient generalization expressed earlier, and try
to explain each case on an individual basis. This wastes useful class time,
often involves explanations that need to resort to the source language, and
eventually does little to alleviate confusion.
Rather than imputing the distribution of mticles to the main verb, it seems
more appropriate to explain it in terms of the way in which the entity evoked
by the verbal complement is construed. The CG account of article distribution
involves the recognition of the meaning of both finite and partitive articles,
characterized by the way in which they relate the object nominal to the
speech situation. It is far beyond the scope of this paper to propose an exact
characterization of either of the articles, but some broad generalizations will
suffice to give the flavor of the CG analysis and how its adoption would
alter current teaching practices in the Natural Approach.
The French definite article marks generic nouns (as in J'aime le cafe), as
well as nouns which are judged identifiable by the hearer. With the partitive,
the identifiability of the nominal is not at issue; the focus is on the fact that
it represents a certain quantity of the substance referred to by the nominal
(Je bois du cafe). Competition between the two articles arises when a given
situation can be construed in two alternative ways, i.e., when a nominal can
be considered with respect to its identifiability, or with respect to its mere
quantity. In certain contexts, even with the verbs that usually favor the con-
strual of the nominal as a quantity, some discourse conditions render an
alternative construal possible, where the identifiability of the nominal is
emphasized. Generally, the referent of the entity has already been men-
tioned, or it is obviously available to the hearer as part of her background
knowledge. For example in (3), because beer and wine are mentioned first,
they can then be treated later as identifiable entities, i.e., the ones that were
just recently mentioned. Similarly in (6), the contents of the refrigerator is

Authenticated
part of the roommates' background knowledge. They can both easily identify
each item in it. Although the person who drank the beer and the wine neces-
sarily drank a specific amount of both, the emphasis in the utterance can
therefore be placed on the items in the refrigerator as identifiable entities.
The quantity consumed is irrelevant; what matters is that both entities have
now disappeared. In a different discourse context, where the focus is on the
quantity of liquid ingested, the partitive would constitute a more appropriate
choice. The case is similar in (5). Although customers in restaurants are
served a certain quantity of chicken, the chicken is not construed as a quantity
of a specific animal but as an item on a restaurant menu, and thus identifiable
within that schema.
It is therefore clear that the data in (3)-(8), as well as other numerous
examples not mentioned here, do not constitute exceptions to a rule of co-
occurrence. Rather, they illustrate cases where despite the lexical semantics
of the main verb, some aspect of the situation favors the construal of the
nominal as an identifiable entity. The pedagogical challenge consists in
teaching those situations, so that, when students are placed in the appropriate
situation, they can make the same choices the natives make, and enjoy the
same flexibility of expression. From a pedagogical standpoint, there are two
aspects of construal that need to be treated separately, especially at the begin-
ning level of instruction. The first one concerns the linguistic choices that
are socially conventionalized and therefore stable. The second one concerns
the more flexible, on-line choices made by specific speakers as they react to
the particularities of a given situation. Since my focus here is on the first
year of instruction, I will mostly be concerned with the first aspect.
The basic idea behind the teaching of construal is that students should be
placed in situations where native speakers are the most likely to exercise a
specific choice. The best place to start therefore is where that choice is
detennined by the presence of a visible cultural schema. Cultural schemas
constitute strong pedagogical assets because they are both meaningful
(therefore culturally rich) and easily teachable (see in particular Goddard
this volume). As an example, let us use the utterance in (7a). Prendre le cafe
is an unexpected collocation because the consumption verb prendre is
expected to be followed by a partitive article. However, the expression
refers to much more than just drinking coffee. It evokes a specific part of a
large food and drink schema in which people drink certain beverages at dif-
ferent times of the day. This ritual does not only involve coffee. The other
beverages that also mark the social rhythm of the French day also appear in
the same Def. Art -I- Noun construction. Combinations such as prendre
l' aperitif' take a drink before lunch!dinner' , prendre le digestif 'take an

Authenticated
190 Michel Achard
after dinner liquor', prendre le the 'take a 5 o'clock snack' almost have the
status of fixed expressions. They are limited in number, and not particularly
productive. For example, *prendre le jus d'orange 'take the orange juice',
or *prendre le vin 'take the wine' are infelicitous in that sense, although
these forms would indeed be acceptable if the speaker were reading off a
grocery list.
Communicative activities can be created to link these apparently excep-
tional uses of the definite article to the culturally marked episodes in which
they appear. Consequently students will associate the whole expression
prendre fe cc~fe to a specific social act, and not analyze the verb and article
combination as a confusing sequence. Successful activities will focus on the
context of the drinking episode, namely, the rhythm of a person's day and
the social circumstances of the drink, and not on what is actually consumed.
F or example, a guessing game can be designed with two students working
in pairs. One of them reads off a specific set of circumstances that surround
the social activity someone is involved in, without mentioning any bever-
age. The other student guesses what that activity is. A possible sequence
could be as follows. Student 1: Mt: Dumas est au bar. 11 parle avec ses amis.
Il est 11 heures et demie du matin 'M Dumas is in a bar, he is speaking with
his friends, it is 11: 30 a.m.' Student 2: If prend ['aperitif 'He is taking a
drink before lunch'. 5
The same kind of analysis can be given for situations where the unex-
pected use of the definite article follows a precise schema, such as a restau-
rant menu or a shopping list, where prendre le + noun is also found, but
marks a different situation. Activities can be designed where menus are
passed to students who need to order from a set of dishes, hence justifying
the use of the definite article. Note that in such tight context (and yet mean-
ingful if the activity is done properly), an answer such as C'est sur la liste
'It's on the list' to a student's query as to why la and not de la is used
constitutes a legitimate grammatical explanation. I will not present a whole
lesson here, but the idea is to associate a particular linguistic usage to its
communicative function to show that conveying that function represents the
meaning of the construction. Attention to meaning necessarily entails atten-
tion to form, because students need to have access to the proper form in
order to convey the intended meaning. Teaching the whole range of a con-
struction is accomplished by targeting each meaning separately, and inte-
grating the activities designed to enhance its awareness to the general flow
of the NA lesson.
As indicated earlier, this kind of teaching does not address the other
aspect of construal, namely, the individual flexibility speakers enjoy in their

Authenticated
linguistic expression. Complete flexibility is not a reasonable goal for the

first year of instruction. It should, however, constitute a pedagogical focus
for the intermediate and advanced levels. Students will, hopefully, be suc-
cessful in time, when they gain more and more exposure to the natives'
reaction to specific situations. The purpose of the beginning level is to get
them to realize that speaking a foreign language involves a succession of
choices, and begin to investigate the most visible social parameters of those
choices.
6. Conclusion
This paper addressed the issue of grammatical instruction in communicative

methods of L2 teaching, and particularly in the Natural Approach. It was
argued that the adoption of the Cognitive Grammar perspective on language
organization and learning allows the instructor to introduce the teaching of
grammar in the NA lesson without compromising the principles and prac-
tices of the model.
The grammatical activities have been shown to take their natural place in
the rhythm of the NA lesson. They are quite similar to other activities, but
narrower in focus and more repetitive in scope. Furthermore, they follow
the other communicative activities in the relevant semantic domain, in order
to capitalize on the student's exposure to the individual instances.
CG-inspired grammatical instruction fills up a pedagogical gap for two
reasons. First, it allows the instructor to teach grammar in a communicative
way. This is desirable in the light of the risk of fossilization that grammar-
less methods of language teaching have inspired. Secondly, it provides a
theoretical frame for the design of pedagogical activities. This is particu-
larly important in the NA because the techniques used in the classroom are
so well worked out. The kind of instruction proposed in this paper affords
pedagogical consistency to the model by treating lexical and grammatical
instruction in the same way.
The CG emphasis on the centrality of construal to linguistic production
was also shown to radically affect grammatical instruction. Rather than
teaching patterns of lexical co-occurrence, the instructor will try to place the
students in the very situations where natives make specific and possibly
unexpected choices. This involves the use of social schemas to show that
the choice of a specific expression is a reflection of social usage. In the gen-
eral climate of L2 instruction that favors cultural awareness, the teaching of
construal represents the ultimate cultural contextualization.

Authenticated
192 Michel Achard
Notes
1. In the CG view of acquisition, the different processes are not as clearly delin-
eated as they are in Figure 1. For example, we will see that construal is at the
same time crucial for the way in which a system is organized (level IT), as well
as for the selection of a given structure in a given situation (level TTT).
2. Note importantly that the adoption of the CG model can easily account for the
facts that the NA hypotheses are attempting to capture. For example, even
though the distinction between acquisition and learning is not compatible with a
CG view of language, it is quite possible that different levels of entrenchment
and abstraction of their meaning allow different constructions to be learned
differently. The precise way in which each construction is learned must be
investigated in a rigorous acquisition research program. Similarly, CG would
not posit an existence of a Monitor, but different constructions may well benefit
from direct explicit instruction at different levels, depending among other things
on their own entrenchment, generality, and clarity of meaning. Finally, the order
of acquisition of specific morphemes constitutes an empirical issue that needs to
be confirmed by multiple studies in different languages.
3. Some NA practitioners would argue that this exposure is sufficient, that no further
exposure on form is desirable, because the students will extract the necessary
generalizations in due time. This paper contends, however, that specific gram-
matical activities can help the students by bringing into focus the generalized
patterns that constitute grammatical rules, while preserving the spirit of the NA
approach.
4. The systematic presentation of the -er verb paradigm should be done in a separate
activity.
5. The success of this activity depends on its integration in the general course of the
lesson. The considerations presented in Section 5.1 obviously also apply here.

Authenticated
References
Achard, Michel
1993. "French causative constructions, word order followingfaire,laisser,
andforcer." In Proceedings of the 19th Berkeley Linguistics Society.
Berkeley, University of California: 1-12.
1996. "Two causation/perception constructions in French". Cognitive
Linguistics 7: 315-357.
1997. "A Cognitive Grammar view of second language acquisition." Joumal
ofIntensive English Studies 11: 157-177.
Bates, Elizabeth, and B. MacWhinney
(1981). "Second-language acquisition from a Functionalist perspective:
Pragmatic, semantic, and perceptual Strategies". In H. Wnitz (ed.),
Native Language and Foreign Language Acquisition. Annals of the
New York Academy of Sciences, vo!. 379. New York, New York
Academy of Sciences: 190-214.
Cadierno, Teresa
This volume. "Expressing motion events in a second language: A cognitive
typological approach".
Chomsky, Noam
1965. Aspects ofthe Theory of Syntax. Cambridge, Mass.: MIT Press.
Dulay H, and M. Burt
1977. "Remarks on creativity in language acquisition". In M. Burt, H. Dulay
and M. Finocchiaro (eds.), Viewpoints on English as a Second
Language. New York: Regents: 95-126.
Elman, Jeffrey
1990. "Finding structure in time". Cognitive Science 14: 179-211.
Gasser, Michael
1990. "Connectionism and universals of second language acquisition".
Studies in Second Language Acquisition 12: 179-199.
Goddard, Cliff
This volume. "Cultural scripts: A new medium for metapragmatic instruction?"
Herron, Carol and Michael Tomasello
1992. "Acquiring grammatical structures by guided induction." The French
Review 65: 708-718.
Krashen, Steven
1981. Second Language Acquisition and Second Language Leaming.
Oxford: Pergamon Press.
Krashen, Steven, and D. Terrell Tracy
1983. The Natural Approach: Language Acquisition in the Classroom. San
Francisco: Alemany Press.
1987. Foundations of Cognitive Grammar. Vo!. 1: Theoretical Prerequisites.

Authenticated
194 AfkhelAchard
1988. "A usage-based model". In B. Rudzka-Ostyn (ed.), Topics in

Cognitive Linguistics. Amsterdam and Philadelphia: John Benjamins:
127-161.
1991. Foundations of Cognitive Grammar. Vol. 2: Descriptive Application.
2000. "A Dynamic usage-based model". In M. Barlow and S. Kemmer
(eds.), Usage Based Afodels ofLanguage. Stanford, CSLI: 1-63.
Lowie, Wander and Marjolijn Verspoor
This volume. "Making sense of prepositions: The role of frequency and simi-
larity in the acquisition ofL2 prepositions".
McLaughlin, Barry
1987. Theories of Second-Language Learning. London: Edward Amold.
MacWhinney, Brian.
1987. "The competition model". In B. MacWhinney (ed.), Afechanisms of
Language Acquisition. Hillsdale, NJ: Erlbaum: 249-308.
Munsell, Paul, and Thomas Carr
1981. "Monitoring the Monitor: A review of Second Language Acquisition
and Second Language Learning". Language Learning 31: 493-502.
Omaggio Hadley, Alice
1993. Teaching Languages in Context. Boston, Mass: Heinle and Heinle.
Rumeihart, David, and James McCleeland (eds.)
1986. Parallel Distributed Processing: Explorations in the Aficrostructure
of Cognition, vol. 1, Foundations. Cambridge, Mass. and London:
MITPressn3radford.
Swain, Merrill, and Sharon Lapkin
1989. "Canadian immersion and adult second language teaching: What's
the connection?" The Afodern Language Journal 73: 150--159.
Taylor, John
1993. "Some pedagogical implications of cognitive linguistics". In R.
Geiger and B. Rudzka-Ostyn (eds.), Conceptualizations and Afental
Processing ofLanguage. Berlin, Mouton de Gruyter: 201-223.
Terrell, Tracy
1991. "The role of grammar instruction in a communicative approach." The
Afodern Language Journal 75: 52-63.
Terre1l, Tracy et al.
2001. Deux Afondes. A Communicative Approach 4th edition. McGraw Hill
Publishers.
VanPatten, Bill
1992. "Structured input and structured output: Explicit grammar instruction
for the communicative classroom". Presentation given at ACTFL 92.
Chicago, Illinois.
1996. Input Processing and Grammar Instruction in Second Language
Acquisition. Norwood, NJ: Ablex.
Waara, Renee.
This volume. "Construal, Convention, and Constructions in L2 Speech".

Authenticated
Teaching Temporal Connectors and their
Prototypical Non-temporal Extensions
Angeliki Athanasiadou
1. Introduction
Teaching complex events has shown that there is no one-to-one relation

between the interclausal relations and the connectives that express them. It
is not at all unusual to come across complex sentences in grammar books,
dictionaries, and corpora containing a temporal link instead of a causal or a
concessive one: While she is a likeable girl she can be extremely difficult to
work with (Longman). In these cases the interclausal relations are left
underspecified, i.e., the semantics of the connective that is used to indicate
the link does not fully match the semantics of the relation that is intended
(Spooren 1997: 150). Underspecified relations can be recognized by ex-
changing an underspecified connective by an explicit counterpart. Here the
temporal while can be replaced by the concessive even if. One might expect
that these instances create interpretive problems, and the plausible question
is how are they going to be taught when it is not known what relation is
being hinted at if it is not indicated explicitly?
Moreover, why is it the case that certain temporal connectors, apart from
their temporality, extend to specific abstract relations and certain other tem-
poral connectors to different abstract relations? What is it, in other words,
that drives while to concession, as long as to conditionality, and since to
causality?
The aim of this paper is to establish the links between the temporal
nuances of specific temporal connectors of English, namely when, as long
as, as soon as, as, since, while and their various non-temporal, abstract uses.
It will be concluded that the type of abstract extensions the above temporal
connectors are driven to is not random. On the contrary, it is systematically
based on their temporal idiosyncracies. A systematic organisation of the
above connectors will be attempted in an effort to provide an effective tool
for their teaching.

Authenticated
196 Angeliki Athanasiadou
2. Some Theoretical Assumptions
One way of analysing underspecified interc1ausal relations is in the frame-

work of pragmatic implicatures. According to Grice it is possible "for what
starts life ... as a conversational implicature to become conventional" (1975:
58). Traugott and K6nig, based on the above assumption, state that "distinct
new polysemies of a form are new conventional meanings" (1991: 193). They
argue that the implicature analysis of underspecified relations can account
for meaning changes in connectives. On the basis of the fact that meanings
shift from what is said to what is meant, they describe these meaning
changes as a process of conventionalising conversational implicatures. Since
semantic change involves specification achieved through inferencing, the
kind of inferencing that is dominant in the development of connectives is
strengthening of informativeness as a conversational implicature becomes
conventionalised.
What the implicature analysis does not describe is why, for instance, rea-
son or causal relations are possible interpretations of the temporal connec-
tives as and since and why contrast relations are possible interpretations of
the temporal connectives when and while. So the relationship between con-
nectives and expressed relations is restricted. In practice, this means that,
whereas the connective while can be used to express contrastive relations, it
is difficult to express a conditional relation using it.
The only way to account for the restrictions in the relationship between
connectives and interclausal relations is to examine the temporal idiosyn-
cracies of the above mentioned connectives. Since time is a domain of
physical nature there is a tendency to go beyond this physical setting and to
extend into all sorts of abstract, rational links. Thus, temporal links are very
easily extended into links of contrast, causality, or conditionality. But in
order to find out by means of which connectives abstract extensions are
construed, one has to thoroughly examine their temporal specifications.
This will prove essential for the idiosyncratic tendencies of each one of the
above temporal connectives. The process underlying the relationship
between the temporal domain which is the source and the extended abstract
domains which are the targets is metaphorical. Time can be viewed as a
domain of conceptualisation which is important for structuring experience.
So the temporal idiosyncracies of the physical domain systematically lead us
to the construal of the extended domains of contrast, causality or reasoning,
and conditionality. During this process, the source meaning may disappear
completely or it may be replaced by what appears to be totally unrelated
meanings or functions. Through metaphor, however, one attempts reconcili-

Authenticated
Teaching Temporal Connectors 197
ation with seemingly unrelated meanings. Reconciliation should not be

regarded negatively. What is actually involved is specification achieved
through inferencing, and one way of inferencing may be through the process
of metaphor. It actually has to do with understanding and experiencing one
thing in terms of another and the directionality of transfer is from a basic,
usually concrete, meaning to one more abstract (for instance, cf. Lakoff and
Johnson (1980), Claudi and Heine (1986), Sweetser (1990».
In the present paper I will focus (i) on the type of temporal structure of
the complex event, (ii) on its extension which is shifted from the external
temporal into the internal, abstract, evaluative, cognitive situation, (iii) on
the type of link expressed by the connective which is dependent on the tem-
poral organisation of the event structure, and (iv) on the assumption of the
prototypicality of the abstract domains which are superimposed on the tem-
poral domain.
3. When
Although when may be visualized as determining a point in time in examples

(1)-(3), it may also be necessary to specify a time relationship between
events, i.e., between the event of the when-clause and the event of the main
clause. If we check the time settings of every event in every example we
realize that the time settings are given in the order of the two events. When
may then be visualized as denoting anteriority (1), simultaneity (2), and
posteriority (3). From this viewpoint, when is a very rich temporal connective
in the sense that it can denote all three major divisions of time relationship.
(i) previous to given time reference (first I get back home and then I tell all
the news):
(1) When I get back home I'll tell you all the news. (Quirk and Greenbaum
1973:231)
(ii) simultaneous with given time reference (when you come next week we
shall be talking about solar energy):
(2) Join us next week when we shall be talking about solar energy.
(CCELD)

Authenticated
198 Angeliki Athallasiadou
(iii) subsequent to given time reference but only if it denotes a landmark in

time without the implication of duration (first he was born and then the war
broke out):
(3) He was already born when the war broke out. (Radden and Dirven, to
appear: 84)
So in purely temporal terms when generally expresses either the succession

(4) or the simultaneity (5) of two states of affairs:
(4) He went away when 1 visited her. (Quirk and Greenbaum 1973: 339)
(5) When he returnedfram work, his wife was in the kitchen. (Quirk and
Greenbaum 1973: 340)
The two states of affairs may exhibit every possible combination of time
point or time duration.
The wide occurrence of when in all time settings, in all possibilities of
representation of events on the time axis, i.e., point or duration, and in all
combinations of figure-ground reversal (the when-clause may be either pre-
posed or postposed to the main clause) make it a very useful but also a very
vague connective. Due to this wide variety of occurrence, apart from pure
time, it can extend to a wide variety of abstract domains and express contrast,
condition, reason, adversativeness.
3.1. Contrast
When introduces a subclause to say that something happened which stopped

one from doing what had been planned or intended:
(6) 1 was going out when there was a knock at the do07" (CCELD)
Although the time concept is dominant, the notion of contrast between the
plan of going out, which is a durational event, and its being stopped by the
knock on the door, a time point event, is moreover evoked by the figure-
ground reversal. When introduces a subclause that contains the important
new event which happens suddenly and unexpectedly; and this is the figure,
while the main clause constitutes the ground. There are two reasons why in
(6) the contrast is highly dramatic and superimposed on the time concept.
First of all there is a clash between what is normally the ground (the when-

Authenticated
clause) and what happens here in the when-clause. The current analysis in a
cognitive-linguistic perspective is that complex sentences display a figure-
ground configuration, whereby the main clause is the figure and the subclause
is the ground. In purely temporal uses of when-clauses both the ground and
the figure can be preposed. But in unmarked, in terms of time, cases the
ground is given first so that the figure can be projected against it as in (6).
This explains why in cases like these the when-clause cannot be preposed
(* When there was a knock at the door I was going out). The second reason
of contrast is that a stretch of time, namely the activity of going out, is inter-
rupted by an abrupt time point event. When is then equivalent to 'and sud-
denly' and it can be paraphrased as 'I was going out, and suddenly there
was a knock at the door'. So (6) mainly has a contrastive meaning which is
due to the temporal combination of the two events, a time duration+a time
point.
The contrast is not so dramatic and it is not superimposed on the time
concept in a scene where the two events can be taken as simultaneous
points in time:
(7) We were having a chat when she entered.
Here the temporal and the contrastive meanings go together. (7) has two
readings: a purely temporal one which can be seen if we apply the test of
the reversal When she entered we were having a chat and it can be para-
phrased as 'At the moment she entered we were having a chat' and a con-
trastive one as in (6) 'We were having a chat and suddenly she entered'.
One might, then, wonder whether we have a situation similar to (7) if we
change (6) to I was going out when she knocked at the door and, moreover,
if we can have the reversal of the events When she knocked at the door, I
was going out. Even if we change the type of the verb (from there was a
knock at the door to she knocked at the door) the meaning is still the same
because we have the succession of the two events: the event of going out is
prior to the event of knocking. However, in (7) the two events occur simul-
taneously - which explains the possibility of their reversal. Furthermore,
the difference between (7) and (6) is that a knock at the door is more
momentary than entering a house. That is why the contrast is so dramatic in
(6) - another explanation of the fact that the contrastive meaning of (7) is
milder than in (6).

Authenticated
3.2. Condition and Reason-giving
The notion of contrast gets the upperhand, the time factor still being present,
but now it is further shifted into condition or reason-giving.
(8) How can I ever get a job in America when I can't remember any
English? (Radden and Dirven, to appear: 85)
The contrast expressed by when in (8) is between getting a job in America

and not remembering any English. But now the contrast can be further over-
laid with a conditional (8a) or a reason-giving meaning (8b):
(8a) How can I ever get ajob in America if I can i remember any English?
(8b) How can I ever get a job in America since I can i remember any
English?
The when-clause in (8) is postposed because, apart from its conditional or

reason-giving implications, it keeps its contrastive sense by indicating why
a particular opinion has been given or why a particular comment has been
made, providing thus the explanation of not being able to find a job which
follows naturally. The contrast is more dramatic with when than with since
or if. So the fronting of when is not really possible (*When I can i remember
any English how can I ever get ajob in America?) because it does not have
an explicit conditional or reason-giving sense, but it only implies a condi-
tion or a reason. But ~f and since can be preposed because they present a
hypothesis on the condition of which an eventual result follows. In (8a) a
hypothetical state of affairs is asserted 'If! can't remember any English, I
cannot get a job in America' while in (8b) the reason is made explicit '1
can't remember any English and therefore I cannot get ajob in America'.
The temporal setting described in (8) is a time duration+a time duration
(not get ajob+not know English). Both events run in parallel and the contrast
of their content is best expressed by when. Once the shift from time to pure
contrast has been conventionalised, it is easily further shifted into condition
and reason-giving and it can be paraphrased by 'considering that' or 'given
that' . Both domains of condition and reason do not concern the events them-
selves but the way speakers interpret them and act upon their interpretation.
They involve hypothetical or explanatory circumstances of events and can
thus be relativised by the speakers. So it seems to be the case that when we
have two time duration events, when is extremely vague and, apart from
time, it expresses contrast, condition, reason-giving. The durational aspect

Authenticated
of the two events allows a great many inferences when we interpret exam-
ples like these, which is evidence for the fact that a sentence is more than
the sum ofthe interpretations of the individual elements.
3.3. Adversativeness
When is almost non-temporal when it introduces a fact or a comment which

makes the other part of the sentence rather surprising or unlikely. In such uses
it acquires the strong adversative meaning of while. whereas or although.
(9) You describe this policy as rigid and unflexible when in fact it has
been extremely flexible. (Radden and Dirven, to appear: 85)
In (9) we have the combination of a time point+a time duration (you

describe+it has been flexible) , but the temporal sense is vaguely present in
the way the two events are combined. The contrastive sense of when is further
loaded with adversativeness. The reason when is so adversative in (9) lies at
the way the two events are organised. The time duration event cannot be
preposed since the explanation contained in the subclause follows naturally
only after the statement of the time point event. So these adversative settings
are postposed since they provide justifications or explanations of the events
described in the main clause. This meaning is much stronger than that of
reason-giving in (8) which also involves explanations or interpretations.
From the above discussion it can be seen that when is one of the most
useful and, possibly, most frequent temporal connectives, but also a very
vague one. It denotes the succession or the simultaneity of two events that
can be either points or stretches on the time axis. However, mere temporal
ordering of events is hardly relevant, hence the variety of possible time
combinations with when that express degrees of contrast:
time duration+time point dramatic contrast (6)

time point+time point mild contrast (7)
time duration+time duration hypotheticality or reason-giving (8)
time point+time duration adversativeness (9)
Language use has shown that relations between events are not spelled out
by explicitly using the connectives that are considered to represent them.
The explicit use of the temporal connective when is considered as a break
on the contrast continuum which varies from mild to adversative. So if two

Authenticated
events are related merely temporally, this frustrates the processing of the
information, it reduces the enrichment of the interpretation, and, moreover, it
deprives the speakers or the hearers to infer in a variety of ways and to decide
whether they prefer to leave the relation between events underspecified or
not.
So learners of English face a double difficulty: on the one hand they have
to master the concepts and the conceptual relations and on the other they
have to master the linguistic forms that can express this conceptual informa-
tion. For this reason they must be taught to infer the relationship intended as
well as to leave relations between events underspecified. This means that
they must be exposed to the wide variety of the uses of when which will
give them an indication of its vague character and of the non-iconic, though
prototypical, use of the continuum of contrast which seems to be superim-
posed on its temporal characteristics.
In sum, two simultaneous time duration events linked with when allow
many inferences and are intended as such by speakers. It is also the case
that prototypically non-sequential or both sequential and non-sequential
when can be enriched by a causal relation:
(10) I couldn't work when the television was on. (Traugott and Konig
1991: 197)
An additional reason for causality is the fact that we have a sequence of two
states. According to Traugott & Konig (1991: 197) states are more likely to
give rise to causal interpretations than sequences of events. The temporal
succession of two events, namely a time point + a time duration event leads
to an adversative interpretation; adversative settings are postposed since
they provide justifications or explanations of the events described in the
main clause.
4. As long as, as soon as
The two conjunctions as long as and as soon as have extended their temporal
meanings into a conditional one. In its temporal meaning as soon as presup-
poses a time point:
(11) As soon as she got out ofbed the telephone stopped ringing. (CCELD)

Authenticated
When a second event happens as soon as a first event happens, the second
event happens immediately after the first event. In its temporal meaning as
long as presupposes time duration:
(12) You couldn't turn the heat off as long as the system was operating.
(CCELD)
If something happens or is the case as long as something else happens or is

the case, it happens or is the case all the time that the second thing is happen-
ing or is the case.
The immediacy of as soon as and the durational character of as long as
can also be seen in their extended abstract uses:
(13) As soon as we get the tickets we'll send them to you. (CCELD)
(14) We were all right as long as we kept our heads down. (CCELD)
In (13) as soon as can be paraphrased by 'the time when' and 'if' and in (14)
as long as can be paraphrased by 'during the time that' and 'on the condition
that'. The time point setting and the 'immediately after' interpretation of as
soon as are more obvious when the as soon as-clause is preposed. In fact it
could be replaced by when. Both conjunctions, however, express a prospec-
tive temporality which naturally leads to the abstract use of the domain of
condition.
So the choice of the adequate linguistic form to express the relation of
condition can be seen as a skill learners have to master. Of course, the
capacity to choose a specific connective develops over the years and the
above specific connectives may be acquired at a later stage than more general
and vague ones like when. Research on first language acquisition, reported
by Bloom (1991), Bloom and Capatides (1987), among others, indicates
that specific connectives such as although are acquired at a later stage than
general ones like but or and. So older children are expected to use specific
forms, whereas younger children are expected to use general forms. The lat-
ter group uses underspecified relations more often than older children. One
might then expect second language learners, due to insufficient language
proficiency, to be explicit, i.e., not to leave relations implicit. But this is only
speculative and must be tested on the basis of a wide variety of parameters
such as age, situation of use, background knowledge oflearners. Tt is true
that the specific use of the connectives as soon as and as long as occurs
only when learners master the relations between events to such a degree that
they are able to refine their language according to their needs and the needs

Authenticated
of the hearers. They should, however, become suspicious of why both con-
nectives express prospective temporality which leads to conditionality. As
long as basically links two simultaneously durational events while as soon
as links two consecutive ones.
The examples met in Collins Cobuild and in Longman dictionaries have
shown that the relationship of the events described by both connectives is
future oriented. The prospective character of as soon as as well as the imme-
diate consequence or succession of another event can be paraphrased as
'from the moment when something happens something else will happen
immediately'. This naturally leads to the future hypothetical interpretation
of ~f As long as is also characterised by prospective temporality but the
durational nature ofthe events described leads to more general hypothetical,
namely, to the conditional interpretation: 'one thing can happen only on the
condition that another thing happens'.
5. While, as, since
Things are different with while, as, and since: they either express purely
temporal settings or they are used in a non-temporal, more abstract way. But
the two meanings seldom overlap. Purely temporally, while denotes two
simultaneous duration events:
(15) He stayed with Mom and me while Dad sat with Dr. Leon in the living-
room. (CCELD)
or a time point event within a duration event:
(16) While he was turning the key in the lock, someone opened a door on
the other side of the corridor. (CCELD)
As denotes two duration events:
(17) As time passed, things seemed to get worse. (Longman)
a time point event within a time duration event:
(18) I saw Peter as I was getting off the bus. (Longman)
or two successive time point events:

Authenticated
(19) As men retire they are replaced on the board. (CCELD)
Since only denotes posteriority. It means 'from the time that' and marks the
lower temporal boundary of the event in the main clause. It may also signal
an overlap with some point in an earlier event.
(20) They haven ~ met since the wedding last year. (Longman)
Both while and as denote two duration events that occur simultaneously.
However, one cannot replace the other: (15) *He stayed with Mom and me
as Dad sat with Dr. Leon in the living-room, or (17) *While time passed,
things seemed to get worse. This is due to the fact that as is conceptually
more delimited or bounded (e.g., (19» than while. Moreover as can also
indicate two consecutive time point events (19), whereas while always pre-
supposes at least one duration event which, as a result, makes it less
bounded than as. Since is far more specific than both as and while because
it implies a duration event after a specifically given time point.
5.1. Concession, Adversativeness
The above-mentioned temporal idiosyncracies play a very important role in

the abstract extensions of the three conjunctions. While extends into conces-
sion (21) or adversativeness (22), but not into reason, exactly because of its
simultaneously durational component:
(21) While the application may not be directly in your area of expertise,
your assessment would be appreciated.
(22) So while I have sympathy for these fellows who reacted against the
formality of their predecessors, I think they went too far. (CCELD)
So with while when two stretches of events occur simultaneously they are
contrasted. The same happens with as where the two simultaneously dura-
tional events extend to adversativeness:
(23) Flattered as I was by his attention, I somehow knew that he wasn ~ the
man for me. (CCELD)
So mere simultaneity of events is hardly relevant unless the two events are
counter to expectation and extend to concession or adversativeness. We can

Authenticated
see this with other connectives as well, e.g., and, an explicit additive con-
nective which has a concessive reading:
(24) He can play the Beethoven sonatas and he s only seven years old.
(Traugott & Konig 1991: 200)
Two simultaneous stretches of events can be contrasted and express con-

cession and/or adversativeness. While and as are connectives originally
expressing simultaneity or temporal overlap, characteristics which have
been retained. The source of concession or adversativeness is due to their
original meanings of concomitance, correlation, or co-occurrence. So the
temporal connectives while and as can be replaced by even if or although
because the situation described in the first event is assumed not normally to
co-occur with the situation in the second event, i.e., there is no agreement
between the two events. But it would be blunt and very factive to use even if
or although to an assessor (21), to reactionists (22), or to boyfriends (23).
So the main extension of while is from temporality to contrast; it further-
more leads to the inference of surprise which, of course, derives from con-
cession. In prototypical concessive settings the main clause is surprising in
the light of the dependent one. This is basically a situation of simultaneous
events.
5.2. Reason, Cause
Both as and since share the component of consecutiveness which drives

them into the extension of reason or cause:
(25) A cup of tea? 1 hardly think so as I'm going out in about two minutes.
(Longman)
(26) Since it was Saturday, he stayed in bed an extra hour. (CCELD)
The main clause indicates the consequence of the as- and the since-clause.
Stating reasons with as and since can occur before or after the main clause
either because speakers may explain why something is the case (27) or they
may introduce a reason for something especially when this reason involves
something that is already known to the person one is talking to (28).
(27) a. As we're both tired, let sjust grab a takeaway. (Longman)

b. Since you are unable to answer perhaps we should ask someone
else. (Longman)

Authenticated
(28) a. As you are here, why don't we discuss our plans? (Radden and
Dirven, to appear: 88)
b. The days were short since it was now December. (Radden and
Dirven, to appear: 89)
In (27) and (28) reasons seem to be pragmatic operations that give explana-
tion after the information of the main clause. They contain invitations for
action (27a, b, 28a) by addressing the hearer directly and by appealing to
old information already known to both participants of the situation (27a, b,
28a, b).
It was stated before that these abstract extensions have no temporal ele-
ments left. However, the conjunction as does not just express time (17, 18, 19)
or just reason (25, 27a, 28a) or just adversativeness (23), but a combination
of the above domains:
(29) She wept bitterly as she told her story. (CCELD)
The two domains that overlap are time and reasoning: the simultaneous
duration of the two events, i.e., weeping and telling her story, leads to the
meaning extension of time into reason. So apart from the component of time
duration of as, the domain of reason is also present since the explanation of
why she was weeping bitterly is provided. The fact that in (29) both domains
of time and reason overlap, but especially the presence of the domain of
time, makes the replacement of as by since impossible. As was said above,
since is temporally unspecified.
Whereas simultaneous events are interpreted as having a link of reason
(29), consecutive events are interpreted as having a causal link, apart from
the temporal one:
(30) As he o.ffered to help heT; she began to cry.
Causality is not only due to the fact that the as-clause is preposed, which
iconically mirrors the cause-effect relationship, but also because of the
physical circumstances that surround an event: the fact that he offered to
help her, caused her crying. Thus, the general effect of consecutiveness is
causal interpretation. Events, then, that follow one another consecutively
can be interpreted as having a causal link. Here consecutiveness is addition-
ally stressed by the verb begin which explicitly marks a new event. Since
could not replace as here either, as we would get a totally different interpre-
tation.

Authenticated
6. Concluding Remarks
Utterances like the ones treated in this paper are rather frequent in natural
discourse, and despite the fact that they constitute instances ofunderspecifi-
cation, they do not seem to create interpretive problems. When native
speakers were asked, they were perfectly able to identify the intended hidden
meaning or the meaning that was superimposed on the temporal one. The
question, of course, remains as to how they can be taught. On the basis of
the process of metaphor, involving shifts from the external temporal to the
internal abstract domains, learners can be presented with the following sum-
marising table of the temporal connectives:
when as long as as/since while

as soon as
simultaneous prospective simultaneous / simultaneous

consecutive posterior
CONTRAST CONDITION REASON CONCESSION

CAUSE ADVERSATIVENESS
On the basis of the above table, it can be seen that two events that temporally
follow one another or occur simultaneously and are linked by the above
connectives tend to express degrees of contrast as well as degrees of causality.
This is natural since mere succession or co-occurrence of two events is rarely
highly relevant information unless the succession or the co-occurrence is
either contrary to expectation or a relation of cause and effect is implied.
These specifications are gradually incorporated into the lexicon. The situation
must be as follows: in the beginning, first or second language learners are
willing to produce and link complex events but they master few and general
temporal connectives, hence the large number of underspecified relations.
Later, when they master language sufficiently, they make a conscious
choice between explicit and implicit use of connectives. Being aware of the
needs of their addressees, learners may be advised to be explicit or implicit.
This shows that learners must become aware of or must look for contrastive
and causal connections because these two types of connections play a central
role in our mental knowledge. They must be taught to rely heavily on as-
sumptions concerning contrast and causality when processing information,
and, moreover, to be able to distinguish degrees of contrast and causality. In

Authenticated
particular, the contrast between two events may be mild or dramatic or may
extend to concession or adversativeness whereas causality may be objec-
tively construed and thus reflect the cause and effect link. It may also be
subjective and then reflect the reason and consequence setting. Condition-
ality is linked to causality and concession: a conditional relation can be seen
as a hypothetical variant of a causal relation, a concessive relation as an
inoperant cause. The affinity between degrees of causality and degrees of
contrast reflects their deep-lying relationship which is apparent in cognition
and discourse (Couper-Kuhlen & Kortmann eds. 2000: 2).
Learners, then, must be taught to rely on the different, more specific,
non-temporal, meanings of the above connectives which basically involve
assumptions of causality and contrast. This is facilitated when there exists a
continuum situation, i.e., when there are no breaks but the succession or
simultaneity of the described events on the temporal axis. This indicates that
the explicit use of the temporal connectives hinders or even stops the further
processing of information. So the implicit contrastive or causal readings of
the temporal connectives must be favoured and must be emphasised when
taught, because many possibilities are open for specific inferences. Psycho-
linguistic research on causality and contrast might prove useful in the way
events linked by temporal connectives are processed, in the way they are
retained and can be recalled best. But these are aims beyond the scope of the
present paper, which only points out the potential difficulties of connectors
for learners and provides some speculative suggestions for their teaching.
Practising teachers must anticipate the difficulties learners will experience
and try to find principles which will predict learning problems (Taylor 1993:
213). One such principle might be the fact that the conceptualisation and the
construal of the non-temporality of the temporal connectives is central and
prototypically lies in the learners' consciousness.

Authenticated
2lO Angeliki Athanasiadou
References
Bloom, L.
1991 Language development from two to three. Cambridge: Cambridge
University Press.
Bloom, L. and Capatides, J.
1987 Sources of meaning in the acquisition of complex syntax: The sample
of causality. In: Journal of Experimental Child Psychology, 43,
112-128.
Claudi, Ulrike and Bemd Heine
1986 On the metaphorical base of grammar,.In: Studies in Language lO:
297-335.
Collins Cobuild English Language Dictionary (CCELD).
1988 London and Glasgow: Collins.
Couper-Kuhlen, Elizabeth, Bemd Kortmann (eds.)
2000 Introduction. In: Elizabeth Couper-Kuhlen and Bemd Kortmann
(eds.) Cause-Condition-Concession-Contrast, 1-8. Berlin, New York:
Mouton de Gruyter.
Grice, Paul
1975 Logic and conversation. In: Syntax and Semantics Ill: Speech Acts,
Peter Cole and Jerry Morgan (eds.), 41-58. New York: Academic Press.
1980 Metaphors we Live by. Chicago: Chicago University Press.
Longman Dictionary of Contemporary English.
1995 Third Edition.
Quirk, Randolph and Sidney Greenbaum
1973 A University Grammar ofEnglish. Longman.
Radden, Giinter and Rene Dirven
to appear Cognitive English Grammar. John Benjamins Publishing Company.
Spooren, Wilbert
1997 The Processing of Underspecified Coherence Relations. In: Discourse
Processes, 24,149-168.
Sweets er, Eve
1990 From Etymology to Pragmatics. Cambridge: Cambridge University
Press.
Taylor, John
1993 Some pedagogical implications of cognitive linguistics. In: Richard
Geiger and Brygida Rudzka-Ostyn (eds.) Conceptuaiizations and
Mental Processing in Language. CLR 3. Berlin, New York: Mouton
de Gruyter.
Traugott, Elizabeth Closs and Ekkehard K6nig
1991 Semantics-Pragmatics of Grammaticalization Revisited. In: Elizabeth
Closs Traugott and Bemd Heine (eds.) Approaches to Grammati-
calization, Volume 1, 189-218. John Benjamins Publishing Company.

Authenticated
Expanding Learners' Vocabulary Through
Metaphor Awareness:
What Expansion, What Learners, What Vocabulary?
Frank Boers
1. Introduction
One of the central tenets ofCogmtive Semantics (e.g., Gibbs 1994, lohnson
1987, Lakoff 1987) is that the meaning of many conventional figurative
expressions, including figurative idioms, is motivated rather than arbitrary.
This runs counter to the "traditional" conception of figurative idioms as
"dead" metaphors whose meaning has become completely arbitrary. While
the "traditional" view maintains that idiomatic language is largely unsys-
tematic, Cognitive Semantics endeavours to look for systematicity in figu-
rative language. When applied to FLA (foreign language acquisition), the
fonner view implies that most idioms can only be learned through random,
blind memorisation. In comparison, the Cognitive Semantic proposal looks
more appealing, as it carries the potential of alternative learning strategies
(e.g., Boers and Demecheleer 1998; Lazar 1996; Low 1988).
These alternative strategies rest on an enhanced metaphor awareness on
the part of the language learner. Such an enhanced metaphor awareness
involves (i) recognition of metaphor as a common ingredient of everyday
language; (ii) recognition of the metaphoric themes (conceptual metaphors
or source domains) behind many figurative expressions; (iii) recognition of
the non-arbitrary nature of many figurative expressions; (iv) recognition of
possible cross-cultural differences in metaphoric themes; and (v) recognition
of cross-linguistic variety in the linguistic instantiations of those metaphoric
themes.
In an FLA context, an enhanced metaphor awareness can be applied to
the acquisition of conventional figurative expressions in roughly three
ways: the imagery behind idioms can be made explicit by refelTing to the
literal (or original) meaning; learners can be encouraged to invest cognitive
effort in trying to figure out the meaning of idioms independently; and
idioms can be grouped under common metaphoric themes. The results of

Authenticated
212 Frank Boers
several controlled experiments have shown that these strategies can indeed
be very beneficial to learners' retention of unfamiliar figurative expressions.
The three proposed learning strategies and the related experimental findings
are outlined in the following section.
2. Metaphor Awareness and Vocabulary Retention
Firstly, language learners can be made aware of the metaphoric nature of

many expressions by referring to their literal (or original) senses. "Dead
metaphors" can thus be resuscitated by associating the idiom with a more
vivid or concrete scene. For example, showing someone the ropes can be
exemplified by the scene of an experienced sailor instructing a novice. The
advantage of this technique is that it lends concreteness to the figurative
expressions, and concreteness is known to facilitate vocabulary learning
(e.g., Solanen 1997). This effect, in turn, can be explained by reference to the
dual coding hypothesis (e.g., Clark and Paivio 1991), which maintains that
concrete items are actually stored twice, verbally as well as non-verbally
(i.e., as an image).
The results of controlled experiments have confirmed this beneficial
effect. In one experiment (reported in detail in Boers 2000a), 73 French-
speaking university students were given a list of English expressions to
describe upward and downward trends in economics. While the control
group was presented with the expressions as depicting either gradual or fast
change (which we might call a "traditional" presentation), the experimental
group was made aware of the literal or original meaning of the expressions.
For example, the verbs soar, skyrocket, and crash were associated with the
image of rockets or airplanes; the verbs plunge and dive were associated
with diving; and the verbs mount, creep up, go downhill, slide, and peak
were associated with mountaineering. In a subsequent language production
task (a description of graphs), the experimental group displayed a signifi-
cantly better retention of the studied lexis than the control group (p < .001).
A follow-up on this experiment will be described below. In another experi-
ment (reported in detail in Boers 2001), 54 Dutch-speaking college students
were asked to consult a dictionary with a view to explaining the meaning of
10 figurative idioms (pass the baton; champ at the bit; a poisoned chalice;
a chink in someone 50 armour; haul someone over the coals; go off at half
cock; a steady hand on the tiller; gird your loins; run someone ragged; and a
dummy run). The control group were given the supplementary task to invent
contexts in which each of the idioms could be used, while the experimental

Authenticated
group were given the task to hypothesise about the etymological origins of
each of the idioms. In subsequent tests measuring participants' retention of
form as well as meaning, the experimental group scored consistently and
significantly better than the control group (p.<OOl).
A second way in which an enhanced metaphor awareness can be benefi-
cial to vocabulary retention exploits learners' problem-solving skills. If it is
true that the meaning of many idioms is motivated, then learners may be
encouraged to figure them out independently before turning to the teacher
or to a dictionary for help (e.g., Lennon 1998). Inferring the meaning of
unfamiliar figurative expressions requires cognitive effort and involves
deep cognitive processing, which raises the probability of memory storage
(e.g., Ellis 1994). Guessing the meaning of "imageable" idioms such as
keeping something under one s hat and crying over spilt milk appears not to
be beyond the capacities of many language learners, even at lower-interme-
diate levels of proficiency. In an experiment (reported in detail in Boers and
Demeche1eer 2001) 78 French-speaking university students were presented
with a range of unfamiliar English imageable idioms and asked to "guess"
their meaning. Despite the absence of any contextual cues, about 35 % of the
participants' responses overall were correct. Contextual cues (which are
usually available in "normal" learning conditions) obviously add to the
probability of successful inferences (e.g., Cooper 1999).
A third major application of metaphor awareness to vocabulary expansion
rests on the existence of general metaphoric themes. What I call metaphoric
themes largely corresponds to the notion of conceptual metaphors in Cogni-
tive Semantics (e.g .. Lakoff 1987). They are conventionally presented in
capital letters, while their instantiations (i.e., the actual linguistic expres-
sions) are listed in italics. For the present purposes I prefer using the tenn
metaphoric theme, because the use of the term conceptual metaphor might
carry the risk of luring us into the ongoing theoretical debate around the
exact conceptual status and cognitive functioning of metaphor. That debate is
probably beyond the scope of a chapter which desperately tries to maintain
an applied linguistics focus. For detailed information about the different
camps in the theoretical debate I refer the reader to Engstrom (1999),
Fauconnier and Turner (1998), Glucksberg, Manfredi and McGlone (1997),
Katz (1998), Murphy (1996), Vervaeke and Green (1997), and Vervaeke and
Kennedy (1996). In an SLA context, metaphoric themes may offer a frame-
work for the integration of knowledge. Since numerous expressions can
often be traced back to a common source domain, they can be presented to
learners as belonging to the same category. Grouping idioms under a shared
metaphoric theme is a way of lending structure and organisation to a world

Authenticated
214 Frank Boers
of vocabulary which may at first sight appear to be completely unsystem-

atic. When new items are well integrated in a larger network, they become
more easily retrievable from memory (e.g., Baddeley 1990). Associating
sets of idioms through metaphor can help such integration. This knowledge
that organised vocabulary is easier to learn than random lists has started to
inspire practice books for learning idioms also (e.g., Wright 1999).
The beneficial effects of a lexical organisation along metaphoric themes
have also been confirmed by the results of controlled experiments (reported
in detail in Boers 2000a). One of the experiments, carried out with the par-
ticipation of 118 Dutch-speaking secondary school pupils, focused on
expressions to describe anger. While the control group was presented with
the vocabulary organised along "traditional", "functional" lines (e.g., how
to describe sudden anger; how to describe angry people), the experimental
group was presented with the same lexis organised along metaphoric
themes as identified by K6vecses (1986). These categories were ANGER IS
A HOT FLUID IN A CONTAINER (/ was boiling with anger; she was all
steamed up; sheJlipped her lid; etc.); ANGER IS FIRE (addingfuel to the
fire; she exploded; he kept smouldering for days; etc.); and ANGRY
PEOPLE ARE DANGEROUS ANIMALS (he has aferocious temper; don't
bite my head off; don't snap at me; etc.). In a subsequent cloze test, the
pupils in the experimental group were more likely than those in the control
group to reproduce the targeted expressions (p < .05).
Another experiment, which was carried out with the participation of 74
French-speaking university students, focused on multi-word verbs. The
control group was given a list of such verbs as presented and explained in A
Practical English Grammar (Thomson and Martinet 1980: 295-339), i.e., a
"traditional" alphabetical listing of the verbs. The experimental group
received the same explanations, but here the multi-word verbs were grouped
under the headings of orientational metaphors, as identified by Boers (1996),
Lindner (1981) and Lindstromberg (1997). These categories included MORE
IS UP (cut down prices; turn down the radio; etc.); ACTIVE IS UP (set up a
business; close down afactory; etc.); GOOD IS UP (feeling down; cheer up;
etc.); and VISIBLE IS OUT and UP (come up with an idea; look up infor-
mation;.find out; etc.). Again, a subsequent cloze test revealed better vocab-
ulary retention in the experimental group than in the control group (p < .01).
Summing up, it has been suggested that an enhanced metaphor awareness
can serve as a vehicle for vocabulary acquisition, the advantages of which
have been pointed out by the results of several controlled experiments (also
see Boers 2000b, and K6vecses and Szabo 1996, for additional empirical
support).

Authenticated
The aim of the present article is to frame the encouraging results of those
learning experiments by (i) establishing the conditions under which raising
learners' metaphor awareness is likely to be most fruitful, (ii) defining the
role of the teacher in this process, and (iii) identifying areas that invite fur-
ther exploration. While strongly advocating the use of metaphor awareness
to expand learners' vocabulary, this article will nevertheless try to fine-tune
expectations along three dimensions: (i) the kind of expansion that we may
realistically hope for, (ii) the kind oflearners who are most likely to benefit,
and (iii) the kind of vocabulary that may most suitably be seasoned with a
touch of metaphor awareness.
3. What Kind of Expansion?
3.1. Short-term or Long-term?
A first practical concern for language teachers may be how much time and
effort should be invested in raising students' metaphor awareness in order to
get long-term returns. Could a single eye-opener suffice, or do learners con-
stantly need to be reminded of the metaphoric nature of the expressions they
encounter? For some linguists it has been sufficient to read the first couple
of pages of Metaphors we live by (Lakoff and 10hnson 1980) to perceive
metaphors everywhere for the rest of their lives. However, we can hardly
take this prolonged effect of a one-off eye-opener in all language learners
for granted.
Firstly, to my knowledge there is as yet no hard empirical evidence of
beneficial effects on vocabulary retention in the long term. In all the afore-
mentioned learning experiments, the participants were tested rather ShOlily
(within two weeks) after the novellexis had been presented to them. Secondly,
there is no guarantee that a learner's enhanced metaphor awareness when
processing one set of figurative expressions will lead that learner to transfer
this insight to her processing of other figurative expressions in the future.
The question of a potential long-term effect on vocabulary retention as well as
the question of potential transfer to new figurative expressions invite further
research.
In a very modest attempt at finding preliminary answers to these research
questions, a follow-up test was administered to the university students who
had participated in the aforementioned experiment on up-down expressions
(soar; plunge; slide; etc.), after a year had passed.

Authenticated
216 Frank Boers
In the period between the first experiment and the follow-up, these stu-
dents of business and economics had continued their common English
courses. The programme was divided over two tenns, intemlpted by a two-
month summer break. Each term contained about 30 hours of English
tuition, given by different teachers to small groups of students. We managed
to track down 30 students who had received the experimental treatment
(i.e., a one-off eye-opener concerning the metaphoric nature of certain up-
down expressions) and 19 students who had received the more traditional
treatment. Another 9 students who had been in that control group now had
to be considered as a separate, third category, for the following reason.
During the term immediately following the first experiment, these individuals
had received tuition from a teacher (the author) with an almost unhealthy
interest in metaphor. This teacher had not been able to resist the temptation of
regularly drawing his students' attention to the metaphoric nature of various
expressions. Whenever figurative expressions in the course materials (such
as a heavy debt burden; getting back onto the right track; pruning expenses;
beingfed up; etc.) could be motivated, their meaning was clarified by refer-
ring to a concrete scene or source domain.
The follow-up task was similar to the task in the first experiment: the
participants were asked to write a short essay describing graphs. The essays
were then collected and screened for the occurrence of the up-down expres-
sions that had been studied the previous year (skyrocket; dive; climb; etc.).
The results did not show a significant difference between the range of
up-down expressions produced by the experimental group and the control
group anymore. The average numbers of different targeted expressions were
3.67 and 3.16, respectively. This means that either the experimental group
had regressed or the control group had caught up in the one-year period.
Interestingly, however, the average number of different targeted expressions
(soar; slide; etc.) used by the third group (the students who had received 30
hours of tuition by a metaphor addict) was 4.89, which did turn out to be
significantly higher (p .007). In other words, the lexical resources tapped for
this patiicular task appeared larger in students whose metaphor awareness
had been raised on several occasions.
Considering the small scale of the experiment and the many variables
involved, we can draw only very tentative conclusions from these findings,
of course. It appears that a one-off eye-opener about the metaphoric nature
of certain expressions is not sufficient to yield a long-term advantage in
retention. On the other hand, as might be expected, it appears that recurring
awareness-raising activities do have a long-tenn beneficial effect on memory
storage. One hypothesis is that the students whose metaphor awareness was

Authenticated
raised in connection with various expressions as they came up in class may

have applied this insight to their processing of other figurative expressions,
including the previously studied up-down lexis. Much more research is
obviously needed to collect more conclusive evidence of such potential
transfer of metaphor awareness (also see Kovecses and Szabo 1996).
3.2. Receptive or Generative?
A second practical concern for language teachers may be the degree of moni-
toring that is required when students apply metaphor awareness to expand
their language resources. Iflearners have become aware of the prevalence
of a certain metaphoric theme in the target language, can they then use this
awareness to "generate" figurative expressions independently or should they
carefully confine their usage to expressions they have actually encountered in
the target language? So far, the proposal has been to use metaphor awareness
only in the latter sense, i.e., as a channeling device to comprehend, store, and
reproduce figurative language input.
The problem with having learners use an identified metaphoric theme to
generate figurative expressions is that the result may not be standard, con-
ventionallanguage. After all, it may be feasible to motivate idiomatic expres-
sions by relating them to a common metaphoric theme or source domain,
but it remains impossible to predict exactly what the idioms belonging to
that metaphoric theme will look like in a given language. In other words, it
is impossible to predict exactly how a particular language will instantiate a
given conceptual metaphor. If learners employed their recognition of that
metaphor to generate figurative expressions in the target language inde-
pendently, they would have to fully appreciate the chances of producing
"marked" language. For example, it does not follow from the well-docu-
mented existence of the metaphor ANGER IS HEAT in English that state-
ments like After all those insults, smoke came out of his ears and After this
morning s argument, she 50 been a barbecue all day are standard English
phrases, even though they may be quite comprehensible.
An additional argument in favour of confining a learner's output to the
figurative expressions s/he has actually encountered is the risk ofLI inter-
ference. When two languages are closely related, they will probably share
many metaphoric themes. On the one hand, recognising such shared
metaphoric themes may be helpful to the learner, because transfer from Ll
to the target language can then speed up the learning process. On the other
hand, transfer inevitably involves the risk of erroneous direct translations

Authenticated
218 Frank Boers
(e.g., Cornell 1999; Swan 1997), and this risk appears especially high in lan-
guage learners who actually perceive the two languages as close (Kellerman
1987). Even "siblings" of idioms vary across languages. The Dutch and
French equivalents of the English idiom biting someone s head oj]; for
example, are iemand zijn neus afbijten and se faire bouJJer le nez, which both
depict noses rather than heads being bitten off. Knowledge of the existing
metaphoric themes of the target language does not entail mastery of its stan-
dard linguistic instantiations.
On the other hand, figurative language is (fortunately) not made up exclu-
sively of conventional expressions, and the above arguments in favour of
carefully monitoring students' language output are perhaps rooted in language
learning programmes that aim at native-like linguistic accuracy. It might be
worthwhile to investigate what would happen if learners were actually
encouraged to "take risks" and to exploit their knowledge of a prevalent
metaphoric theme in the target culture to "coin" figurative expressions
through creative thinking. Such an experiment would inevitably result in a
fair number of unconventional expressions, some of which would undoubt-
edly be considered by native speakers as wrong. Others, however, would
possibly be considered merely as original or perhaps even as poetic. A
research question here is to what extent native speakers comprehend and
tolerate such unconventional instantiations of a familiar metaphoric theme.
In a pilot experiment, English teachers at the University of Birmingham
were asked to rate expressions as (i) wrong English; (ii) deviant, but accept-
able English; or (iii) correct English. The expressions were presented in the
context of two fictitious e-mail messages:
Dear Bill,
I'm afraid I won't be able to join you on that hiking trip next weekend. I got
into trouble with Paula and I think I'd better spend the weekend with her to
make up, you see. The thing is, I forgot her birthday last Monday, which was
stupid because you know what a ferocious temper she has. To make matters
even worse, I went out for a couple of beers after work and I got home rather
late that night. So she wasn't exactly in a good mood when I fmally turned
up - without a birthday present. In fact,
A. she really let her anger out of its cage.

B. she really unchained her anger.
C. she really unleashed her anger.

Authenticated
Dear Bill,
I'm afraid I won't be able to join you on that hiking trip next weekend. I got
into trouble with Paula and I think I'd better spend the weekend with her to
make up, you see. The thing is, I forgot her birthday last Monday, which was
stupid because you know what an inflammatory temper she has. To make
matters even worse, I went out for a couple of beers after work and I got
home rather late that night. So she wasn't exactly in a good mood when I
finally turned up - without a birthday present. In fact,
A. she's been a flame-thrower ever since.

B. she's been breathing fire ever since.
C. she's been breathing flames ever since.
Two of the options were existing English idioms (unleashing one;" anger
and breathing fire), two were close to these existing idioms (unchaining
one's anger and breathing flames) and two were "completely novel" cre-
ations (letting one 's anger out olits cage and being aflame-thrower). Please
notice that in each e-mail message the three options instantiated a
metaphoric theme that had been introduced earlier on through the phrases
ferocious temper and inflammatory temper, respectively.
The informants' responses to the non-standard phrases varied consider-
ably, as shown in table 1.
Table 1. Rating of non-standard expressions (9 respondents)
Wrong Acceptable Correct
Breathing flames 2 7
Being a flame-thrower 7 2
Unchaining one's anger 5 3

Letting anger out of a cage 3 3 3
Most respondents rated the close equivalents of standard idioms either as

wrong (7 responses) or as deviant, but still acceptable English (10 responses).
Respondents may have assessed these phrases by implicitly contrasting
them with the standard idioms. The respondents' appreciation of the two
"completely novel" expressions appeared more ambivalent, however. Being
a flame-thrower was rated either as wrong (7 responses) or as correct English

Authenticated
220 Frank Boers
(2 responses). Letting anger out ~r its cage was judged acceptable or correct
by no fewer than six respondents. The difference in appreciation may be
due to the fact that the scene of letting an animal out of its cage is a more
salient source domain experience for most people than a scene involving a
flame-thrower. Put simply, the flame-thrower phrase may have come across
as too far-fetched. Nevertheless, these preliminary data seem to suggest that
completely novel instantiations of an established metaphoric theme may
more easily be considered by native speakers as correct (but possibly
"poetic") language than slightly deviant versions of standard idioms.
Considering the fair degree of tolerance on the part of native speakers
attested in the above pilot experiment, it seems that it may be feasible to take
metaphor awareness beyond the realm of receptive vocabulary learning, after
all. Language learners might be aided by their understanding of a prevalent
metaphoric theme in the target culture to actively communicate abstract
notions through creative figurative language. The success of such an
endeavour would clearly depend on the learners' awareness of (i) the proba-
bility of cross-linguistic variation in conventionalised instantiations of the
given metaphoric theme and (ii) the possibility of cross-cultural variation in
the popularity of the metaphoric theme itself. After all, even when a meta-
phoric theme is shared by two linguistic cultures, it may be more popular in
one culture than the other (e.g., Boers and Demecheleer 1997; Boers 1999;
Emanatian 1995).
The decision whether metaphor awareness should serve only as a recep-
tive channel for vocabulary retention or whether it could also be used to
inspire students' language generation will ultimately depend on the objectives
of the language learning programme. If the programme aims principally at
native-like accuracy, then careful monitoring oflearners' figurative language
output will obviously be required. On the other hand, if the programme is
oriented towards using the language as a medium for intercultural commu-
nication (e.g., Byram 1997), then learners could possibly exploit their
knowledge of attested metaphoric themes to create figurative language
themselves. Moreover, overextension of a learned metaphoric theme may
still be rectified as the learner progresses and becomes more sceptical of L 1
borrowing (see below).

Authenticated
4. What Kind of Learners?
4.1. Elementary, Intennediate, or Advanced?
A third practical concern for language teachers may be what level of profi-
ciency their students need in order to successfully apply metaphor aware-
ness to vocabulary learning. For teachers who still consider metaphor to be
an ornamental device associated with poetry and literature, the use of meta-
phor in the classroom will probably appear appropriate only when teaching
advanced students. However, when metaphor is recognised as a basic cogni-
tive process which is at the root of a plethora of linguistic phenomena (poly-
semy, idioms, proverbs, etc.), then it may very well be relevant to all levels
of proficiency. Thinking and communicating abstract thoughts requires
metaphor. Although certain teaching methods, such as Total Physical
Response (Asher 1977), do tend to postpone abstraction at elementary levels,
it is difficult to imagine a complete ban on all abstract thought in a learning
situation involving (near-)adults. Whenever genuine communication takes
place, abstract ideas may be expressed, and thus figurative language may be
needed. Moreover, metaphor provides learners with a tool to extend the
meaning of simple, concrete words to denote more complex, abstract con-
cepts for which they have not yet acquired the precise tenns.
The aforementioned experiments about metaphor awareness and vocab-
ulary retention were all carried out with the participation of intennediate
learners of English. On the one hand, these Dutch- and French-speaking
students may have been aided in their processing of the presented meta-
phors by the close cultural relationship with English. On the other hand,
they may also have been impeded in their processing of the presented figu-
rative language by gaps in their lexical knowledge. Further experimental
research is clearly needed to measure the effects of an enhanced metaphor
awareness in (i) elementary or advanced language learners and (ii) learners
of a language that is more distant from their own.
It is obvious that elementary language learners would face most difficulty
in applying a strategy of metaphor awareness to newly encountered expres-
sions, quite simply because they would often lack the lexical knowledge
needed to interpret such expressions in the first place. For instance, in order
to independently make sense of she was fuming, the learner would need to
understand the word fuming. In order to recognise the metaphoric nature of
many expressions, one needs knowledge of the literal (or prototypical)
senses of polysemous items. For example, an elementary learner who has
never encountered any spatial sense of the preposition beyond may not be

Authenticated
222 Frank Boers
aware of its being used figuratively in it s beyond my understanding. In that

case, the learner will not benefit from the imaging techniques which have
been shown to enhance memory storage. In order for elementary learners to
reap any rewards of metaphor awareness, a considerable amount of guid-
ance by the teacher would therefore be indispensable.
It is evident that, at the other extreme, advanced learners will least often
be blocked by gaps in their lexical knowledge. Consequently, they would
probably be in the best position to independently recognise the metaphoric
nature of newly encountered expressions. Their lexical knowledge would
often allow them to associate the figurative expression with its source
domain, or to trace the figurative sense of a polysemous word to its literal
origin. Such associations lend concreteness to newly encountered expres-
sions, which has been shown to facilitate retention.
As far as production of figurative language is concerned, however, ad-
vanced learners might not be the most gratifying (or grateful) target group,
because advanced students appear to be most sceptical about the acceptabil-
ity of Ll equivalents in the target language (Kellerman 1978). In other
words, advanced learners may be most hesitant to "take risks" and they may
be most likely to shy away from trying out figurative expressions whose
"correctness" in the target language they are not sure of.
If one wanted to exploit established metaphoric themes as a source of
inspiration for learners to actively generate figurative language (which, as
explained above, is a controversial issue), then intermediate learners could
therefore make up the most responsive target group. Again, of course, any
decision to encourage "risk-taking" behaviour would depend on the extent
to which a particular educational programme puts the emphasis on intercul-
tural communication as opposed to native-like linguistic accuracy.
4.2. Cognitive Styles?
A fourth practical concern for language teachers may be whether the technique
of metaphor awareness is powerful enough to accommodate all individual
learners, even if they share the same level of proficiency. While the afore-
mentioned learning experiments about vocabulary retention through metaphor
awareness reveal beneficial effects in the experimental groups as a whole, they
also show considerable differences between the scores of individual learners.
F or example, in the experiment set up to measure learners' expansion of
vocabulary to describe upward and downward trends (soar, plunge, etc.), the
scores of individual participants ranged from 3 to 14 reproduced expressions.

Authenticated
Such variation calls for caution to never treat a group of learners as a homo-
geneous population. Individual differences are inevitable along multiple
affective and cognitive parameters. One parameter is that of cognitive styles.
A cognitive style can be roughly defined as an individual's consistent
strategy to organise and process information. A number of cognitive style
dimensions or continua have been identified in psychology (e.g., Schmeck
1988; Witkin and Goodenough 1981) and two of these appear to be of special
interest to our discussion: the analytic / holistic continuum and the verbaliser /
imager continuum (e.g., Riding and Cheema 1991).
The analytic / holistic dimension (e.g., Bever 1975) refers to an individ-
ual's preference for processing information either as separate parts or as
large integrated chunks. Analytic individuals tend to break down a problem
into distinct parts, whereas holistic individuals tend to maintain a more
global picture. The verbaliser / imager dimension refers to a subject's pref-
erence for thinking either in words or in pictures (e.g., Paivio and Harshman
1983). Both of these dimensions may also have a profound impact on the
way individuals process metaphors and figurative language.
In a dual experiment (described in detail in Boers and Littlemore 2000),
71 university students were asked to explain the existence of a couple of
well-established metaphoric themes. The respondents differed in their pre-
ferred strategies to explain the metaphors and these strategies correlated
with their cognitive styles as "diagnosed" by means of the Riding (1991)
computer-assisted test of cognitive styles. Very briefly, this test measures
analytic processing by presenting the participant with a simple shape and a
complex shape, placed side by side. The participant's task is to decide as
quickly as possible whether the simple shape is contained within the complex
shape. To measure holistic processing, the participant is asked to decide
whether two shapes are identical. Verbal and imagery processing are meas-
ured by presenting pairs of words which the participant has to assess
whether these are the same type or the same colour.
Firstly, participants who were found to have an analytic cognitive style
appeared best at tracing a given metaphor back to a distinct, concrete source
domain. They clearly distinguished literal from figurative usage. Holistic
respondents, on the other hand, appeared to find it harder to identify a dis-
tinct source domain at the origin of a given metaphor. In their explanations
they were more likely to "blend" their conception of different domains
without making a clear distinction between literal and figurative.
Secondly, imagers were more likely than verbalisers to explain the given
metaphors by referring to concrete scenes. For example, they explained the
metaphoric theme ECONOMIC COMPETITION IS RACING (lagging behind;

Authenticated
224 Frank Boers
the race for market share; etc.) by calling up the scene of a busy businessman
who is in a hurry to get to prospective clients or to new markets before his
competitors do. Likewise, imagers typically explained the ECONOMIES ARE
MACHINES metaphor (exchange-rate mechanisms; monetary tools; etc.) by
calling up the scene of a Fordist assembly line with machinery.
Although the aims of this experiment were not directly related to FLA,
the observed variation in individuals' processing of metaphors does suggest
that language learners' are also likely to have different approaches to unfa-
miliar figurative expressions in the target language. Due to their cognitive
style characteristics, some learners may be more susceptible to the strategy
of metaphor awareness than others. The findings lead us to hypothesise that
an enhanced metaphor awareness will probably be most beneficial to learners
with an analytic and imager cognitive style. Analytics are probably most
capable of recognising the metaphoric nature of an expression or the figura-
tive usage of a polysemous word by comparing it with a distinct source
domain or literal usage. The identification of distinct source domains (or
metaphoric themes) behind sets of expressions provides a framework of
vocabulary organisation, which is known to facilitate memory storage.
Imagers are probably most capable of associating a novel figurative expres-
sion with a mental picture or concrete scene, and imaging and concreteness
have also been shown to facilitate retention.
Again, it is clear that a lot of experimental research would be required to
test these hypotheses. One major complication in such research is the likeli-
hood of gradual shifts in individuals' learning styles. After all, a learning
process is supposed to have an effect on the learner's cognitive processing.
Furthermore, even if future research turns up more conclusive evidence of
cognitive style differences in learners' susceptibility to the strategy of en-
hanced metaphor awareness, then many questions will remain as to possible
implications for educational practice. It goes without saying that a language
course should ideally cater to all individual learning styles present in a given
target group. In practice, this means that different approaches to learning
can complement one another.
5. What Kind of Vocabulary?
5.1. Degrees of Transparency
A fifth practical concern for language teachers may be that not all idioms
lend themselves equally well to explicit imaging techniques. While the

Authenticated
semantics of some idioms may be pretty transparent to most language users

(e.g., barking up the wrong tree) many other idioms may be quite opaque
(e.g., paying through the nose). In an SLA context, it will be easier to guess
the meaning of transparent idioms than that of opaque ones. It will also be
easier to motivate the semantics of transparent idioms than that of opaque
ones. The more opaque the idiom, the more the learner will have to rely on
contextual cues to figure it out.
The degree of semantic transparency of an idiom is determined by the
interplay of various factors. One factor which contributes to semantic trans-
parency is the presence of a common metaphoric theme or conceptual
metaphor (e.g., Gibbs 1993). For example, letting off steam derives a fair
degree of transparency from the fact that (along with she was fuming; she
erupted; etc.) it reflects the conceptual metaphor ANGER IS HEAT. In the
aforementioned learning experiments on vocabulary retention through
metaphor awareness, the targeted expressions were generally clear instances
of presented metaphoric themes. Therefore, further research would be
required to measure the effect of metaphor awareness on learners' retention
of other categories of idioms, such as expressions which are only indirectly
associated with a general metaphoric theme or expressions which do not
seem to belong to a general metaphoric theme at all.
Idioms that are directly associated with a metaphoric theme tend to be
more transparent than more "peripheral" ones. For example, she wasfuming
is a more "central" instance of ANGER IS HEAT than he hit the ceiling
(LakoffI987: 384-385). In an SLA context, we may assume that motivating
expressions which are only indirectly associated with a general metaphoric
theme and which may thus be conceived as belonging to the periphery of
the category will tend to require more cognitive effort on the part of the
learner. This need not imply that such expressions are unsuitable candidates
for explicit metaphor processing, however. After all, problem-solving tasks
which require cognitive effort or in-depth cognitive processing are known
to enhance memory storage also.
Many idioms are less easily categorised under a general metaphoric
theme. Selling someone down the river; having egg on one sface; and kick-
ing the bucket may be cases in point. Fortunately, relating an idiom to a
more general metaphoric theme is but one way of clarifying its semantics.
When confronted with idioms that are not straightforward instantiations of
any general metaphoric theme, the language teacher or learner may still
resort to other techniques to enhance imaging and concreteness (e.g., Irujo
1986). A simple reference to the literal usage from which the figurative
usage is derived may sometimes suffice (e.g., Boers 2000b). For example,

Authenticated
226 Frank Boers
the figurative usage of hurdle in the government is facing major hurdles can
be clarified by referring to the literal sense of the word (e.g., a steeplechase
context).
A complementary method to motivate an idiom is to retrace its likely
etymological origin (e.g., Boers 2001). For example, it is feasible for teach-
ers in a classroom context to quickly clarify the use of under in under one 50
own steam by referring to the era of steam propulsion on ships and trains.
Still, we have to acknowledge the existence of idioms whose origins have
become too obscure to be of any use to the average learner. For instance,
tracing the origin of under way hardly seems worth the effort in general
EFL classrooms.
Further empirical research is clearly required if we want to help teachers
(and learners) decide whether a given idiom is a suitable candidate for
imagery processing. A research team at the University of Antwerp and the
Erasmus College of Brussels has recently embarked on just such a project.
As part of a computer-assisted vocabulary acquisition programme, large
numbers of idioms are processed by students along alternative cognitive
channels, including a cognitive channel that exploits mental imagery. By
assessing the students' ease at using a particular channel for each idiom, and
by assessing the students' rate of retention of each idiom after using a par-
ticular cognitive channel, the study hopes to generate a practical repertoire
of figurative idioms for which imagery processing can be recommended on
the basis of empirical evidence. At a more theoretical level, the collected
data may allow us to identify more precisely what features of idioms con-
tribute to "imageability" in the first place.
5.2. Degrees of Vagueness
A final concern (closely related to the previous one) may be the problem of
categorising figurative expressions under precise metaphoric themes. After
all, it has been argued that one of the advantages of an enhanced metaphor
awareness is the mental framework it provides for organising figurative lan-
guage input. Categorising idioms under their common metaphoric themes
lends structure to the influx of figurative language which may at first sight
seem completely arbitrary and unsystematic.
Unfortunately, different conceptual metaphors inevitably intersect, so
that different ones may be at play simultaneously. For example, an expres-
sion such as slimmingjlabby workforces may be presented by teachers as an
instantiation of HEALTH, FITNESS, and SPORTS metaphors. Moreover, rather

Authenticated
specific metaphoric themes can be subsumed under more generic ones. For
example, an expression like We are catching up with our competitors can be
presented as an instantiation of a RACING metaphor as well as a more general
SPORTS metaphor. The identification of conceptual metaphors is an issue of
ongoing debate in cognitive semantics (e.g., Grady 1997, 1999) and an in-
depth discussion is clearly beyond the scope of this applied linguistics article.
The relevance for SLA, though, is that recurring difficulties in matching
figurative expressions to precise and unique metaphoric themes could result
in confusion and scepticism on the part of the learner. The advantage of
establishing some kind of order in the complex world of figurative language
could be lost. To avoid this, it seems advisable (in initial stages) to draw
learners' attention principally to "clear" cases, i.e., expressions whose source
domain can be pinpointed pretty unambiguously.
Unfortunately, for numerous conventional figurative expressions the link
with the original source domain has become very vague. For example, few
language users are likely to recognise economic recovery as being derived
from the domain of health, unless the context points to this source domain
(e.g., After a long period of illness the economy is recovering at last).
Checklists for the recognition and categorisation of figurative expressions
in real discourse are being developed (e.g., Steen 1999a, 1999b), but to my
knowledge these have not yet been geared to FLA needs.
Apart from serving as a framework of organising figurative language
input, the strategy of metaphor awareness serves to enhance imaging and con-
creteness. As has been mentioned above, metaphoric themes (or conceptual
metaphors) vary in their degrees of abstraction: abstract, generic metaphors
subsume more concrete, specific ones. While the former contain "bare" image
schemas (PATH, CONTAINER, etc.) (e.g., Lakoff 1990), the latter contain
more elaborate or "richer" source domain information. In order to facilitate
imaging in an SLA context, it seems advisable to give preference to the
more specific metaphoric themes over the more generic ones. For example,
an expression such as Go on with your work, which reflects the PATH
schema contained in the generic ABSTRACT ACTIVITY IS MOTION
metaphor, is not likely to evoke rich imagery. On the other hand, expres-
sions such as The enterprise has been blown off course, The enterprise has
been derailed, and The enterprise hasjaken off, which reflect more specific
motion metaphors involving ships, trains, and airplanes, appear much better
candidates for explicit metaphor processing.

Authenticated
228 Frank Boers
6. Conclusions
In this chapter I have tried to establish the conditions under which an en-
hanced metaphor awareness is likely to be beneficial to language learners'
vocabulary expansion. On the way, we have briefly visited six unexplored
areas of investigation to gauge the following research questions.
(i) How much metaphor instruction is needed to yield long-term gains in

learners' vocabulary expansion?
(ii) Is the technique of metaphor awareness confined to its use as a recep-
tive channelling device for the organisation, storage, and reproduction
of figurative language input, or can it be stretched to active generation
of figurative language?
(iii) Do language learners of different levels of proficiency respond to
(and benefit from) the technique of metaphor awareness in comparable
ways?
(iv) Do individual learners with different cognitive styles respond to (and
benefit from) the technique of metaphor awareness in comparable
ways?
(v) What degree of semantic transparency do figurative expressions need
in order to be suitable candidates for explicit metaphor processing?
(vi) What kind of metaphoric themes should learners be made aware of and
how precise should learners' categorisation of figurative expressions
be?
The preliminary answers to these questions suggested in this chapter have

of necessity been very tentative. The reader is kindly invited to treat them as
hypotheses waiting to be put to the test in ongoing and future research.

Authenticated
References
Asher, James
1977 Learning Another language through Actions: The Complete Teacher s
Guidebook. Los Gatos, CA: Sky Oaks Productions.
Baddeley, A.
1990 Human Memory: Theory and Practice. Needham Heights, MA:
Allyn and Bacon.
Bever, T. G.
1975 Cerebral asymmetries in humans are due to the differentiation of two
incompatible processes: holistic and analytic. Developmental Psycho-
linguistics: Communication Disorders 26(3): 251-262.
Boers, Frank
1996 Spatial Prepositions and Metaphor: a Cognitive Semantic Journey
along the Up-Down and the Front-Back Dimensions. (Language in
Performance 12.) Ttibingen: Gunter Narr Verlag.
1999 When a bodily source domain becomes prominent: the joy of counting
metaphors in the socio-economic domain. In: Raymond W. Gibbs
and Gerard J. Steen (eds.), Metaphor in Cognitive Linguistics, 47-56.
(Current Issues in Linguistic Theory 175.) AmsterdamlPhiladelphia:
John Benjamins.
2000a Metaphor awareness and vocabulary retention. Applied Linguistics
21(4): 553-571.
2000b Enhancing metaphoric awareness in specialised Reading. English
For Specific Purposes 19(2): 137-147.
2001 Remembering figurative idioms by hypothesising about their origins.
Prospect 16(3): 35-43.
Boers, Frank and Murielle Demecheleer
1997 A few metaphorical models in (western) economic discourse. In:
Wolf Andreas Liebert, Gisela Redeker and Linda Waugh (eds.),
Discourse and Perspective in Cognitive Linguistics, 115-129.
(Current Issues in Linguistic Theory 151.) Amsterdam! Philadelphia:
John Benjamins.
Boers, Frank and Murielle Demecheleer
1998 A cognitive semantic approach to teaching prepositions. English
Language Teaching Journal 52(3): 197-203.
2001 Measuring the impact of cross-cultural differences on learners' com-
prehension of imageable idioms. English Language Teaching Journal
55(3): 255-262.
Boers Frank and Jeannette Littlemore
2000 Cognitive style variables in participants' explanations of conceptual
metaphors. Metaphor and Symbol 15(3): 177-187.
Byram, Michael
1997 Teaching and Assessing Intercultural Communicative Competence.
Clevedon: Multilingual Matters.

Authenticated
230 Frank Boers
Clark, J.M. and A Paivio

1991 Dual coding theory and education. Educational Psychology Review
3(3): 149-21
Cooper, T.
1999 Processing of idioms by L2 learners of English. TESOL Quarterly
33(2): 233-262.
Cornell, Alan
1999 Idioms: an approach to identifying major pitfalls for learners. IRAL
37(1): 1-21.
Ellis, Nick C.
1994 Vocabulary acquisition: the implicit ins and outs of explicit cognitive
mediation. In: N. C. Ellis (ed.), Implicit and Explicit Learning of
Languages, 211-282, London/San Diego: Academic Press.
Emanatian, Michele
1995 Metaphor and the expression of emotion: the value of cross-cultural
perspectives. Metaphor and Symbol 10(3): 163-182.
Engstrom, A
1999 The contemporary theory of metaphor revisited. Metaphor and Symbol
14: 53-61.
Fauconnier, G. and M. Turner
1998 Conceptual integration networks. Cognitive Science 22: 133-187.
Gibbs, Raymond W.
1993 Why idioms are not dead metaphors. In: C. Cacciari and P. Tabossi
(eds.), Idioms: processing, structure, and interpretation, 57-76,
Hillsdale/London: Lawrence Erlbaum Associates.
1994 The Poetics of Mind. Figurative Thought, Language, and
Understanding. Cambridge: Cambridge University Press.
Glucksberg, S., D. Manfredi and M. McGlone
1997 Metaphor comprehension: How metaphors create new categories. In:
Ward, T.B, S.M. Smith and J. Vaid (Eds.), Creative Thought: An
investigation of conceptual structures and processes, 327-350.
Washington, DC: American Psychological Association.
Grady, Joseph E.
1997 Theories are buildings revisited Cognitive Linguistics 8(4): 267-290.
1999 A typology of motivation for conceptual metaphor: correlation vs.
resemblance. In: Raymond W. Gibbs and Gerard 1. Steen (eds.),
79-100.
Katz, AN.
1998 Figurative language and figurative thought: A review. In: AN. Katz,
C. Cacciari, R.w. Gibbs and M. Turner (eds.), Metaphor: Implica-
tions and applications, 3-43. New York/Oxford: Oxford University
Press.
Irujo, Suzanne
1986 A piece of cake: learning and teaching idioms. English Language
Teaching Journal 40(3): 236-242.

Authenticated
Johnson, Mark
1987 The Body in the Mind: The Bodily Basis ofMeaning, Imagination and
Reason. ChicagolLondon: University of Chicago Press.
Kellerman, Eric
1978 Giving learners a break: native language intuitions as a source ofpre-
dictions about transferability. Working Papers in Bilingualism 15:
59-92.
1987 Aspects of transferability in second language acquisition.
Unpublished PhD. Dissertation, University ofNijmegen.
K6vecses, Zoltan
1986 Metaphors of Anger; Pride and Love: a Lexical Approach to the
Study of Concepts. AmsterdamlPhiladelphia: John Benjamins.
K6vecses, Zoltan and Peter Szabo
1996 Idioms: a view from Cognitive Semantics. Applied Linguistics 17(3):
326-355.
Lakoff, George
1987 Women, Fire and Dangerous Things: What Categories Reveal about
the Mind. ChicagolLondon: University of Chicago Press.
1990 The Invariance Hypothesis: is abstract reason based on image-
schemas? Cognitive Linguistics 1(1): 39-74.
1980 Metaphors we Live by. Chicago/London: University of Chicago
Press.
Lazar, Gillian
1996 Using figurative language to expand students' vocabulary. English
Language Teaching Journal 50(1): 43-51.
Lennon, Paul
1998 Approaches to the teaching of idiomatic language. IraI36(1): 12-30.
Lindner, Susan
1981 A lexico-semantic analysis of verb-particle constructions with Up
and Out. Unpublished PhD dissertation, University of California San
Diego.
Lindstromberg, Seth
1997 English Prepositions Explained. Amsterdam/Philadelphia: John
Benjarnins.
Low, Graham
1988 On teaching metaphor. Applied Linguistics 9(2): 125-147.
Murphy, G.
1996. On metaphoric representation. Cognition 60: 173-204.
Paivio, A. and R. Harshman
1983 Factor analysis of a questionnaire on imagery and verbal habits and
skills. Canadian Journal ofPsychology 33 (1): 17-28.
Riding, R. J.
1991 Cognitive Styles AnalYSis. Birmingham: Learning and Training
Technology.

Authenticated
232 Frank Boers
Riding, R. J. and I. Cheema

1991 Cognitive styles: an overview and integration. Educational Psychology
11: 193-215.
Schmeck, R. (ed.)
1988 Learning Strategies and Learning Styles. New York: Plenum Press.
Sokmen, Anita 1.
1997 Current trends in teaching second language vocabulary. In: N. Schmitt
and M. McCarthy (eds.), Vocabulary: Description, Acquisition and
Pedagogy, 237-257, Cambridge: Cambridge University Press.
Steen, Gerard J.
1999a From linguistic to conceptual metaphor in five steps. In: Raymond
W. Gibbs and Gerard J. Steen (eds.), 57-77.
Steen, Gerard 1.
1999b Metaphor and discourse: towards a linguistic checklist for metaphor
analysis. In: Lynne Cameron and Graham Low (eds.), Researching
and Applying Metaphor, 81-104, Cambridge: Cambridge University
Press.
Swan, Michael
1997 The influence of the mother tongue on second language vocabulary
acquisition and use. In: N. Schmitt and M. McCarthy (eds.), pp.
156-180.
Thomson, A.J. andA.V Martinet
1980 A Practical English Grammar (Third edition). Oxford: Oxford
University Press.
Vervaeke, J. and C.D. Green
1997 Women, fire and dangerous theories: A critique of Lakoff's theory of
categorization. Metaphor and Symbol 12: 59-80.
Vervaeke, J. and 1. Kennedy
1996 Metaphors in language and thought: Falsification and multiple mean-
ings. Metaphor and Symbolic Activity 11: 273-284.
Witkin, H. A. and D. R. Goodenough
1981 Cognitive styles: Essence and origins. New York: International Uni-
versities Press.
Wright, John
1999 Idioms Organiser: Organised by Metaphor, Topic and Key word.
Hove: Language Teaching Publications.

Authenticated
A Cognitive Linguistic View of Polysemy in English
and its Implications for Teaching
Szilvia Csabi
1. Introduction
The present paper deals with the method of vocabulary teaching, using the
principles developed in cognitive linguistics. The focus is on polysemy and its
implications for teaching English as a second language. Cognitive linguistic
research on polysemy (i.e., "the association of two or more related senses
with a single linguistic form" [Taylor 1989: 99]) has suggested that the
meaning structure of polysemous words is motivated and can be accounted
for in a systematic way. This insight can be employed in applied linguistics
and can be put to use in teaching as well as learning foreign languages.
The claim proposed here is that teachers' and learners' awareness of the
cognitive mechanisms making up the network of senses is useful in the lan-
guage learning process. It is not claimed here that traditional vocabulary
learning processes, such as memorization, should be replaced. Rather, the
point of the study is to support the idea that, besides memorization, awareness
and acquisition of the cognitive structure of word meanings aids teaching
and learning.
Words with several senses often cause problems for teachers and learners
of English since they are often seen as unrelated and unsystematic. This belief
probably discourages teachers and students to deal with polysemous words
in the classroom. The strategy commonly employed by teachers is to not
deal with the various senses of a certain word all at once, but to explain the
specific senses when they turn up. However, a more fruitful strategy, result-
ing in the easier mastery of the target language, would be to provide the
explanations and motivations for the related senses of given words together.
In order to support the appropriateness of the present approach to vocab-
ulary teaching and learning, two experiments are reported on examining
whether the explicit knowledge of conceptual metaphors, metonymies, and
conventional knowledge present as motivational factors in the target lan-
guage facilitate learning word meanings in the target language. Classroom

Authenticated
234 Szilvia Csabi
research on teaching polysemous words such as hold and keep and idioms in
which they appear to secondary school students appears to support the appli-
cation of cognitive linguistic principles in language teaching and learning.
In sum, this paper claims that learners who know how certain conceptual
mechanisms (such as conceptual metaphors and metonymies) structure the
meanings of certain polysemous words and idioms will acquire their mean-
ings more easily, compared to other learners who are not familiar with these
mechanisms. On these grounds, it is suggested that the motivations of word
meanings should be clarified to students in the language classroom to
achieve better results.
2. Vocabulary Teaching and Cognitive Linguistics
Researchers in applied linguistics have developed serious interest in the

study of word meanings and their role in language teaching. Lewis (1993)
recognizes the importance of conceptual metaphors and metaphorical pat-
terning in language teaching primarily because metaphors constitute "power-
fully generative pattern systems" (113). Thus, it is beneficial for students to
be aware of, acquire, and use metaphors. This is also suggested by Low
(1988), who uses the concept of metaphorical competence, and who claims
that students should develop this kind of competence in order to understand
and produce metaphors. MacLennan (1994) also emphasizes the importance
of metaphor in the language classroom since it is a central component of
language that should not be neglected. MacLennan (1994) stresses that the
knowledge of metaphors can aid students in vocabulary acquisition as well
as in learning grammar. In connection with teaching words and their mean-
ings, Lennon (1996) examines high-frequency "easy" verbs like put, go, take
and concludes that, although learners may have "a broad outline of word
meaning", they still have hazy lexical knowledge of polysemous items and
phrasal verbs (35). Lennon (1996) proposes that there is a need for detailed
work in the classroom on high-frequency, "easy," and "simple" verbs - which
are most often polysemous items that participate in a number of idiomatic
expressions - in order to clarify their meaning-range and limitation.
Within the cognitive linguistic framework, Deignan, Gabrys, and Solska
(1997) describe a study concerning the problem of how metaphorical expres-
sions differ in the native and target languages. The authors suggest that raising
students' awareness regarding similar and different metaphors in Ll and L2
is worthwhile to aid students' comprehension and production of metaphors.
The study by Kovecses and Szab6 (1996) focuses on using cognitive lin-

Authenticated
Polysemy in English and its Implications for Teaching 235
guistics in the teaching and learning idioms, more specifically, phrasal verbs
with up and down. Their findings support the idea that it is advantageous to
use the cognitive paradigm in the teaching and learning of idioms in L2.
The use of the cognitive paradigm in teaching and learning builds upon
the principle of motivation, which is a central phenomenon in cognitive lin-
guistics. Lakoff (1987) defines motivation as follows: "The relationship
between A and B is motivated just in case there is an independently existing
link, L, such that A-L-B 'fit together'. L makes sense of the relationship
between A and B" (448, italics in the original). The reason for its centrality is
that "[ijt is easier to learn something that is motivated than something that
is arbitrwy. It is also easier to remember and use motivated knowledge than
arbitrary knowledge" (Lakoff 1987: 346, italics in the original). Possible
motivations for word meanings can be provided by conceptual metaphor
(CONCEPT A IS CONCEPT B), conceptual metonymy (CONCEPT A FOR CON-
CEPT B), conventional knowledge, image schemas (e.g., UP-DOWN schema),
and construals (Lakoffand Johnson 1980; Lakoff 1987; Langacker 1987).
Motivation plays an important role in a number of phenomena in cognition
and language. Lakoff observes that polysemous words and most idioms arc
motivated, since their meanings make use of and are consistent with certain
already existing patterns (1987: 438). Naturally, not all word meanings and
idioms are motivated, but in the majority of cases we can find a link between
form and meaning. As Lakoff (1987: 316) argues,
Polysemy occurs when a single word has more than one meaning - and when
those meanings are systematically related. Systematic relationship is crucial
here. The two meanings of bank - place where you put your money and the
edge of a river - are not systematically related. Such cases are called hom-
onyms. Cases of poly se my are cases like warm, which refers to both to tem-
perature and to clothing that makes you feel warm. Another example would
be newspaper, which can name either what you read at the breakfast table or
the company that produces it.
The cognitive view of polysemous words can be used in the examination of

words such as hold and keep. These words have highly motivated senses
that are also important in the idioms containing them. In the following,
experiments using the sense analysis of these highly polysemous words and
idioms containing them are presented. Naturally, the limitations ofthe study
do not allow for a very detailed and complete linguistic analysis, so the
examination concentrates only on the major ways in which the senses of
and the idioms with hold and keep are motivated since these are in the focus
of the experiment.

Authenticated
236 Szilvia Csabi
I f we adopt Lakoff's view on motivation, we can assume that the language

learners' knowledge (i.e., awareness and acquisition) of the motivations of
meanings probably helps learners master the meanings more easily. Thus,
they will be in a more advantageous situation concerning their ability to
speak the language more fluently and in a more native-like manner. In this
way, it can be predicted that if language learners acquire the motivations of
the senses of polysemous words and of the idioms in which they appear, they
will learn, remember, and use them more easily. These learners will achieve
better results compared to results of those who lack explicit knowledge of
motivated meanings.
This also means that a teacher working in accordance with this approach
to vocabulary teaching should make the links between words and their mean-
ings explicit in the classroom whenever it is possible, and apply cognitive
linguistic principles and the results of cognitive linguistic research on word
meanings. In other words, the teacher should "facilitate" the students' meta-
phorical competence in class (cf. Low 1988) in order to make students be
aware of and acquire the knowledge that metaphor provides them with. This
may be a challenging and intellectually demanding task both for the teacher
and the learner - but, as MacLennan (1994: 105-106) also claims, "[l]earn-
ers should not be protected from the difficulties inherent in metaphor and
other nonliterallanguage" since "[s]everal advantages are to be gained from
teaching learners about metaphor and several disadvantages are evident
when it is not taught".
In order to test the applicability of the cognitive approach to vocabulary
teaching, two experiments were conducted. The purpose of the experiments
was to investigate whether the explicit knowledge of conceptual metaphor,
metonymy, and conventional knowledge present in the target language
facilitate learning word meanings. More precisely, the question is whether
the explicit knowledge of the motivated senses of two polysemous words
hold and keep helps learners to learn, remember, and use these words better
than other learners who only memorize the words and their senses. It is
hypothesized that explicating the motivations for the senses of polysemous
words and the idioms in which they occur promotes better learner perform-
ance; that is, better results can be achieved.
The main reasons for choosing hold and keep are as follows: Hold and
keep occur very frequently in English, they are both within the first 200
most frequently used lexical items in English according to the Brown
Corpus (Francis and Kucera 1982). Both can express basic actions, they
have metaphorical senses that have already been conventionalized, and they
can be found in a number of idioms. Their usage possibly presents a problem

Authenticated
for the learners of English since they can easily be confused. They appear to
be used in a variety of senses, which are probably difficult to acquire
because they are seemingly unrelated. In addition, learners are usually only
taught their most frequent meanings (as in She was holding an umbrella or
You can keep the change). However, gaps in knowledge of further senses
may often turn out to be impediments to fluent interaction with and clear
understanding of the speech of native speakers.
3. Method
The study designed to investigate the effects of learners' (lack ot) knowl-
edge of sense motivations consisted of two experiments. The experiments
employed the same materials and similar, though not identical, procedures.
First, a pilot study was conducted in the fall of 1997 in a Budapest secondary
school (the same school where the later experiments were carried out). The
subjects of the pilot study were 1ph-graders (age 17) who were on an ad-
vanced level of English, and who had four English lessons a week. The aim
of the pilot study was to see whether the experiment is conductible, and if
so, then which test items can be used in the redesigned experiments. After
the pilot study, the materials and the procedures were changed and their
final version was used in the experiments reported below.
3.1. Subjects
The classes selected for the study performed in the winter of 1998 included
fifty-two students enrolled in a Budapest eight-class secondary school. Half
of them were in their 8tll grade (ages 13-14) and half of them were in their 9th
grade (ages 14-15). The students belonged to four learner groups (in what
follows: Group A, B, C, and D) with thirteen students in each at the time of
the actual research. All the students involved in the study were native
speakers of Hungarian and learned English as a second language. Groups A
and B - the groups in the 8th grade - were of roughly equal proficiency
(intermediate level); they had been studying English for two and a half
years, and they had 4-5 English lessons a week. Groups C and D - the
groups in the 9th grade - were also of roughly the same level of proficiency
in English (intermediate level). The members of these two groups had been
studying English for three and a half years and had 4-5 lessons a week.
Group A was the experimental group and Group B was the control group in

Authenticated
238 Szilvia Csabi
the first experiment. Group C was the experimental group and Group 0 was
the control group in the second experiment.
3.2. Procedures
3.2.1. Instruments
The instrument used in the research was a test containing twenty-two items
related to hold and keep (see Test in the Appendix). Tt focused on (a) hold
and keep as polysemous items; (b) phrasal verbs including hold and keep;
(c) idioms containing hold and keep. Naturally, students in every group
were familiar with the most frequently used meanings of hold and keep, as
illustrated in He was holding a kn~fe in one hand and a fork in the other, She
held her daughter s hand as they crossed the road, I held the baby in my
arms or in Here s a jive-pound note - you can keep the change and I keep
all her letters. That is, the central meanings of hold and keep were already
known to students - nevertheless, they were also included in the teaching and
testing process since further, figurative meanings and idiomatic meanings
largely depend upon these meanings. The information provided by the teach-
ers indicated that all the other items in the test were unknown to the students.
The senses of hold and keep and the idioms in which they occur were all
pretaught to each group. The language used during the explanations was
Hungarian for practical reasons. Since the researcher had no previous teach-
ing experience with the groups involved in the study, it would have created
an uncontrollable factor to use English words in the explanations that were
not known to the students.
The general teaching procedure and the administration of the test were
divided into three parts in each group: First, some senses of hold and keep
were presented to the groups. Students were instructed to memorize the
words and their meanings. Following this, they were required to complete
the first ten sentences of the test. Either hold or keep had to be placed in the
gaps. Secondly, phrasal verbs containing hold and keep were presented to
the classes, whose meanings students had to memorize. Afterwards, the sec-
ond part of the test had to be completed, in which the phrasal verbs had to
be inserted. Thirdly, multiword idioms containing hold and keep were pre-
sented to students, who memorized their meanings. After this activity they
had to fill in the third part of the test, which contained more than one sen-
tence describing a situation appropriate for the use of a given idiom. The
items to be inserted in the gaps were not written on the test sheet, so students

Authenticated
had to retrieve them from memory. Shortly after the teaching and testing
procedure described above, a post-test was administered to each group, that
is, students were required to complete the same test as they had to fill in on
the day of the treatment.
The only difference between the experimental and the control groups
was the presentation of the material to be taught. Groups A and C, the experi-
mental groups, were taught in accordance with the cognitive linguistic
approach to meaning described in detail below. This means that instead of
teaching them the English expressions with their Hungarian equivalents,
only the most important motivating factors were explained to them, and they
memorized the items on the basis of their knowledge of sense motivations.
In contrast to this, Groups Band D, the control groups, were only taught the
English words and their Hungarian equivalents, and they relied only on this
information when memorizing the expressions and their meanings.
The theoretical background for the systematic explanation of the sense
motivations of hold and keep was provided by Lexical Network Theory
(Norvig and Lakoff 1987). This theory sees "a lexical item as a network of
minimally differing senses, with links of a small number of types" (Norvig
and Lakoff 1987: 195). Hold and keep have several senses that are related to
each other, and no abstract, general meaning can be discovered which covers
all possible senses. A plausible explanation for the systematicity in the dif-
ferent senses of both hold and keep may be given in terms of a network
"such that each sense is a minimal variant of some other sense" (Norvig and
Lakoff 1987: 197, italics in the original). This means that if sense A is a min-
imal variant of sense B, there is only a single significant difference between
them from which their other differences can be predicted. According to
Norvig and Lakoff(1987: 197-198), the following links exist which connect
the various senses of polysemous words: image-schema transformation
links, metaphoric links, metonymic links, frame-addition links, semantic
role differentiation links, and profile shifts.l
Lexical network analysis provides a fine-grained structure of word mean-
ing and lends itself to the graphical representations of meaning structures.
Experiments concerning prepositional meanings reported by Sandra and
Rice (1995) seem to prove the legitimacy of lexical network analysis since
statistically significant results suggest that "language users make rather fine-
grained distinctions, much in the way that prepositional network modellers
do" (124). The empirical evidence obtained from the experiments suggests
that language users are able to perceive certain usage types as homonymous
and others as related to each other, similar to the way proposed by linguists
(Sandra and Rice 1995). This supports the validity of the analysis.

Authenticated
240 Szilvia Csabi
3.2.2. Some Senses of hold and Idioms with hold 2
The primary concrete, or central, meaning of hold (hold-I) is illustrated in

She held the purse in her right hand. Here the agent is a human whose hand
is the focus of attention, and the patient is usually a concrete, relatively
small, light, and easily manipulated object that is not difficult to grasp. Here
no movement is involved. Minimally related senses are given in The lovers
held each other tight, where the patient is another person instead of an
object; in He held the rope in his teeth, where teeth are in the focus of the
action instead of the hand; and in The dog held the newspaper between its
teeth, where the agent is a nonhuman animate being instead of a human.
Hold implies the interaction of forces between the agent and the patient
(the Agonist and the Antagonist). The patient would tend to move (due to
gravitational forces the patient would fall to the ground), but the agent
exerts some force towards having the patient remain in the given position
(in the hands, etc.). This is a pattern which emerges in Leonard Talmy's
work as force dynamics. Force dynamics deals with how entities interact
with respect to force, and it encompasses notions like the exertion of force,
resistance to exertion, overcoming of resistance, blockage of force, removal
of blockage, and others (Talmy 1988: 49). Thus, in the cases mentioned
above, force-dynamic patterns are perceptible. An example where this force
dynamic pattern is foregrounded is Will this branch hold me?
A further component common to all senses of hold is a durative compo-
nent. It implies continuous action since the agent's main force tendency is to
prolong the present state in which the patient is, that is, to keep it in a cel1ain
position. This means that there is a stronger Antagonist exerting force to
keep the Agonist from moving. This durative component is profiled in We 'Il
hold this bookfor you, Hold still while I take your picture, and Would the
weather hold?
Hold-2 is connected to hold-l by a metaphoric link. This minimal vari-
ant is motivated by the metaphor POSSESSING (SOMETHING) IS HOLDING
(IT IN THE HAND). The rationale for the existence of this sense of hold is the
following: When a person holds something, it usually belongs to him or her,
it is his or her property. In this way, we can have hold in She has never held
ministerial office, He did not hold a firearm cert~ficate, The Fisher family
holds 40% of the stock, where the thing held is conceptualized as the agent's
possession.
A further variant (hold-3), which is linked to the possession sense, is the
sense of controlling, which is motivated by the metonymy THE HAND
STANDS FOR CONTROL and the related metaphor CONTROL IS HOLDING

Authenticated
SOMETHING IN THE HAND. These cognitive mechanisms make it possible

to use hold in The terrorists held them hostage, Demonstrators have been
holding the square since Monday, The town was held against frequent
enemy attacks.
This analysis of some senses of hold shows that the senses boil down to
a systematically ordered network in which they are interrelated primarily by
metaphor and metonymy. Idioms with hold also exhibit highly motivated
meanings. Several cognitive linguistic studies (Gibbs 1990, 1994; Gibbs and
O'Brien 1990; Kovecses and Szab6 1996) show that idioms are not neces-
sarily lexicalized units, and that conceptual metaphors motivate idiomatic
meanings. Gibbs and O'Brien (1990) examine whether people have tacit
knowledge of the metaphorical basis of idioms and conclude that people
have consistent mental images for various idioms. Thus, there is supportive
evidence for the claim that idiomatic meanings are extended primarily via
metaphor from the meanings of the particular words in them.
On these grounds, we can describe the motivations of the meanings of
several idioms containing hold. Idiomatic meanings largely depend on the
meanings of the constituent parts, such as the verb and the particle in the
case of phrasal verbs. The phrasal verbs hold back, hold down, and hold up
used in the experiment manifest various cognitive mechanisms that motivate
their meanings. They strongly evoke the force dynamic situation described
above since they focus on the exertion of force in order to not let something
(change or movement) happen. This is the basis for their metaphorical
meanings as well.
Thus, hold back is motivated by the force dynamic pattern of hold and
by the direction implied by back, as in Judy held her back. Hold back can
be used metaphorically in connection with emotions since emotions are
seen as metaphorical forces acting within the self (Kovecses 2000): Most of
us were doing our best to hold back tears.
Hold down foregrounds the aspect of control in addition to force tenden-
cies, as in I was trying all the time to hold down the lid of the box with one
hand. Its metaphorical uses are motivated by the metaphors CONTROL IS
HOLDING SOMETHING IN THE HAND and CONTROL IS UP / LACK OF CON-
TROL IS DOWN. This is the possible motivation for hold down in There
would still not be enough forces to hold down the previously subject people.
In hold down a job, we again have several sources of motivation. In addi-
tion to the aspect of control, the aspect of continuity is foregrounded, and at
the same time we also have cultural motivation since hold down is often
used by cowboys referring to the use of force to keep cows still when
branding them.

Authenticated
242 Szilvia Csabi
Besides the force dynamic pattern applied in the physical sense of hold up
(Ralph held up his hand), the metaphorical sense of hold up is motivated by
the metaphor OBSTRUCTION IS UP. Thus, we can use hold up in The whole
thing was held up about half an hour and The criminals held up the train.
F or the purposes of the experiment, the following idioms were chosen:
hold one 50 tongue, hold one 50 temper, and hold one 50 head up. Hold one 50
tongue implies force dynamic patterns (the tongue would move, but it is not
allowed to move) as well as control through the metaphor CONTROL IS
HOLDING SOMETHING IN THE HAND. We also have a conventional image
of a person holding his or her own tongue. Evidently, tongue metonymically
stands for speaking. Hold one 50 temper is similar in some respects to hold
one 50 tongue since it also indicates force dynamic patterns and control, due
to the fact that emotions are often understood as forces that have to be con-
trolled (Kovecses 2000). Finally, hold one s head up receives motivation
from the emotion metaphor THE CONCEITED PERSON IS UP/HIGH (Kove-
cses 1990). This is again an easily imaginable idiom, with a conventional
image in which a person's head is held high up in the air.
The meanings of the above idioms appear to be motivated primarily by
metaphorical extension. The senses of hold render it possible that a number
of idiomatic expressions develop whose meanings are also motivated, in
part by the meaning structure of hold.
3.2.3. Some Meanings o/keep and Idioms with keep
The primary concrete, or central, meaning of keep is illustrated in You can

keep the change, You can keep your things in here, These old clothes are not
worth keeping (keep-I). Keep implies temporary possession in the central
sense. The durative component is strongly emphasized, since keep usually
refers to a lasting state. The force dynamic pattern which is at work in the
case of hold is present in keep as well. As Talmy (1988: 62) notes, keep is
"the key force-dynamic word". There are conflicting force tendencies, one
towards rest or remaining in the given position, and another towards motion
or change. In the central sense, keep does not involve the use of hands as
hold does. Rather, the things kept are temporarily in the possession of the
agent - not necessarily in his or her hands. Thus, the most important differ-
ence between hold-I and keep-I is that keep implies a longer and stronger
duration of the force tendency, it focuses on possession, and it does not nec-
essarily indicate the use of hands.

Authenticated
Further senses of keep are linked primarily by profile shift, that is, something
which is backgrounded in one sense is fore grounded in another, minimally
related sense (Norvig and Lakoff 1987: 190). In the minimally different
sense keep-2 explicated by She kept a diary for over 20 years, continuity,
the durative component, is foregrounded. In keep-3, illustrated in She kept
her promise / her word / the secret both the force dynamic pattern and the
durative component are profiled (e.g., the person does not let anyone else
know the secret). Keep-4, a further variant of keep-I, profiles not only the
aspect of continuity but also the aspect of temporary possession and main-
taining a state as well as in It costs more each year to keep a house, He
scarcely earns enough to keep himself and his family.
Idioms with keep such as keep in and keep out (of); keep one 50 fingers
crossed, keep somebody at arm 50 length, keep something under one 50 hat
profile the force dynamic pattern. The meanings of the phrasal verbs keep in
and keep out relate to the meaning of keep in which the durative and the
force dynamic components are profiled. The particles in and out evoke the
image of a container, a bounded area, with things that can be either in or out
of it. On the one hand, keep in can metaphorically refer to emotions as in
He could scarcely keep in his indignation since human bodies are often
understood as containers for emotion, and therefore the CONTAINER meta-
phor is often utilized in talking and thinking about emotions (K6vecses
2000). On the other hand, keep in can also refer to a culturally motivated
situation in school illustrated in She was kept in for an hour for talking in
class. In this case, the building of the school, more specifically the class-
room itself, is the metaphorical container with the pupil inside it. In line
with the above, keep out focuses on the outside of the container, that is,
what is out of the container. Usually there is a boundary that the relevant
entity should not cross, as in The sign said "Ministry ofDeji:mse - Danger -
Keep out!" Keep that dog out of my study! and That child seems incapable
of keeping out of mischief
Keep one 50 fingers crossed, keep somebody at arm 50 length, and keep
something under one 50 hat also focus on force tendencies and continuity
implied by keep. The aim of the agent is to maintain the state or position of
the fingers, somebody, or something, respectively. Keep one 50 fingers
crossed implies the image of having the fingers in a special position for a
long time, which stands metonymically for being anxious or worried about
somebody. The idiom keep somebody at arm 50 length has metaphorical
motivation as well. Lakoff (1987: 447) argues that a conventional image
(the arm, which is chest high, is oriented forward with respect to the body;
the hand is open and the palm is facing away from the subject; the angle of

Authenticated
244 Szilvia Csabi
the hand relative to the forearm is roughly 90 to 135 degrees; the ann muscles
are tense; and the person being kept at arm's length is facing the subject), and
two metaphors (INTIMACY IS PHYSICAL CLOSENESS and SOCIAL/PSYCHO-
LOGICAL HARM IS PHYSICAL HARM) motivate the meaning of this idiom
("keep someone from becoming intimate, so as to protect oneself from social
or psychological harm"). Keep something under one s hat also implies the
existence of a conventional image with a person and his or her hat, under
which there is something hidden. This image is complemented by the
metaphor KNOWING IS SEEING. The rationale is that if something cannot be
seen, it cannot be known either - this is why the secret is hidden under the
person's hat. This is the logic that two other idioms with keep (keep the lid
on and keep something under wraps) also utilize.
3.3. Experiment 1
The first experiment involved Groups A and B. The presentation of the

material differed in the following way in the two groups:
Within the framework of a 45-minute lesson, students in the experimental
group (Group A) were presented with the senses of hold and keep given
above, and motivations were illustrated and explained along the lines of the
above analysis. The lesson consisted of three parts. In the first part, the
meanings of hold and keep were demonstrated to the students in the follow-
ing way: keywords (like hand, control) indicating the motivation for the
meanings were put on the blackboard with representative example sentences
chosen from dictionaries. On the basis of the analysis, the motivations for
the selected senses were explained to students. That is, senses of hold and
keep were linked to each other with the help of explicating the connections
between them - though without explicit reference to the linguistic concepts
of metaphor, metonymy, and profile shift. Thus, the meaning of hold in hold
an office was explained as follows: When a person holds something, it usu-
ally belongs to him or her, it is in his or her possession. Students were then
required to memorize the items and their meanings on the basis of these key-
words. This was followed by the completion task in which students had to fill
in ten missing words, either hold or keep. The explanation, memorization,
and the gap-filling exercise took twenty minutes altogether.
In the second part of the lesson, phrasal verbs such as hold back/up/
down/down a job were taught with the help of example sentences. The moti-
vation for the meanings was illustrated by what the students had already
learned about the meanings of hold and keep, and schematic drawings (e.g.,

Authenticated
Polysemy in English and its Implications/or Teaching 245
a circle with an X inside and outside indicated the phrasal verbs keep in/out).
After the explanation part, students had to memorize the expressions and
their meanings, and they had to complete the six sentences given with the
phrasal verbs they had just learned. The procedure took ten minutes.
In the last part of the lesson, idiomatic expressions such as hold one S
tongue or keep something under one s hat were presented to the class. The
expressions and their schematic, but representative, drawings were sketched
on the blackboard. These illustrated the mental images inferred from the
literal meanings of the idioms (thus, the meaning of keep something under
one s hat was indicated by a hat and something under it). The meanings
were explicated and memorized, and afterwards the students were instructed
to complete the last task of the test.
During the explanation parts, body language was also used to strengthen
the students' conception of meanings. The illustrative part of the explana-
tions (the use of drawings in addition to body language) appeared to be very
useful for the students' better understanding of meanings. In several cases,
students themselves found out the exact meanings of the expressions in
question.
In the control group (Group B), the lesson also consisted of three pat1s.
The difference in the procedure was that the motivation for the meanings
was not given. Instead, the words and expressions were written on the black-
board with their Hungarian equivalents. In the first part of the lesson, the
selected meanings were put on the blackboard in Hungarian. They were clar-
ified in a few words, but nothing was said about the way the meanings are
connected. Students had to memorize the words and their senses, and after-
wards they had to complete the first ten sentences of the test. This took fifteen
minutes altogether. 3 Afterwards the phrasal verbs were written on the black-
board with their Hungarian equivalents. Again, the meanings were clarified,
but motivations were not given. After the explanation and the memorization
procedure, students were instructed to complete the second gap-filling exer-
cise. Having spent ten minutes on this task, idioms and their Hungarian
equivalents were presented and written on the blackboard, which the stu-
dents had to memorize. Finally, they were asked to fill in the last part of the
test using the idioms learnt just before. This last task lasted fifteen minutes.
One day later the test was re-administered to the two groups - naturally,
it was not known to them that they would have to work with the learned
material again. They were instructed to fill in the tests in ten minutes with-
out prior mention or recall of the given expressions.

Authenticated
246 Szilvia Csabi
3.4. Experiment 2
In the second experiment Groups C and D were involved. Experiment 1 and

Experiment 2 differed only in minor ways. The material taught and tested
was presented in two lessons, thus the first part dealing with the senses of
hold and keep was presented separately from the second and the third parts
dealing with idioms. The reasons behind this practice were twofold. The
main reason was to reduce the possibility of feeding too many things for
students to learn. Furthermore, it was hypothesized that separating the
material to be learned produces better results in the long run. In order to see
the effect on the results, the test was re-administered not one but two days
after the treatment.
4. Results
The tests and the post-tests of all groups were analyzed, and the number of
correct answers was counted. Expressions in the second and third parts of
the test which were not entirely correct (e.g., keep this under hat or keep fin-
ger) but in which keywords (hold and keep) were given correctly were
counted as correct answers, although with indication that these answers
were not entirely correct. The reason for not excluding these answers was to
see whether students were able to differentiate between the senses of hold
and keep. The percentage values for the total number of correct answers on
the tests written by Groups A, B, C, and D are given in Table 1 and 2 in the
Appendix. The percentage values of the distribution of entirely correct and
not entirely correct (keyword-type) answers are supplied in Table 3.
The data from the two pairs of groups were then statistically compared
separately for each pair of tests. The tests of Group A were compared to the
tests of Group B (the first test of Group A to the first test of Group B, and the
post-test of Group A to the post-test of Group B). The same procedure was
applied in the case of Groups C and D. Not all the assumptions of the t-test
were met in the research; therefore the Wilcoxon Rank Sums test, a non-
parametric test, was chosen to make comparisons between Groups A and B,
as well as C and D. The null hypothesis was the starting point, according to
which there was no significant difference in the ranks assigned to the per-
formance of the groups, Groups A and B on the one hand, and Groups C and
D on the other hand, in any pair of tests.
The nondirectional Rank Sums test indicates that the differences between
the tests of Groups A and B as well as those between the post-tests of Groups

Authenticated
A and B are significant. The differences between the tests and between the
post-tests of Groups C and D are also statistically significant. In the test
written by Groups A and B the a level is at a < .01, and in the post-test of
these two groups the a level is at a < .01 again. In the test completed by
Groups C and D the a level is at a < .02, and in the post-test of these groups
the a level is at a < .01. The probability, then, is less than 2 % in general
that the differences between the group pairs occurred by chance alone.
Therefore, the null hypotheses of no difference can be rejected. Thus, the
statistically significant difference between the experimental group and the
control group in both experiments clearly shows that students in the experi-
mental groups outperformed those in the control groups.
In addition to being significant, the differences are meaningful because
they are consistently large in the case of both experiments. Ifwe take into
consideration the percentage values, the divergence between the two tests
taken by Group A (one on the day of the experiment and the other a day
later) shows that their performance decreased by 9.44%, whereas the per-
formance of the control group, Group B, decreased by almost the double of
this value, 18 %. The same tendency is observable in the case of Groups C
and D, where the difference of the performance of Group C was 8% after
two days, and where the difference of the control group performance
decreased by almost the double of this value after two days. The trend that
the post-test values are somewhat higher in the case of Groups C and D
(which they took two days after the first test) than in the case of Groups A
and B (which they took one day after the first test) can be attributed to the
fact that the separation of the material to be learned was beneficial to their
performance.
Table 3, which contains the percentage values of entirely correct and
keyword-type answers for each exercise of the test, shows that both experi-
mental groups did better in all three exercises than the matching control
groups as far as the total number of correct answers is concerned. In the sec-
ond exercise of the first test, both experimental groups gave good answers
in 100 % of the cases, and in the first exercise of the first test Group A wrote
good answers in 99.2% of the cases - which is very close to 100%. This
never happened in the case of the control groups. The highest value that
Group C achieved was 97.3% in exercise 2 of the first test, which is admit-
tedly a high value. The lowest values reached on any of the exercises are
attributable to the control groups: Group B achieved 52.5 % in exercise 3 on
the post-test (which is the lowest value of all), and Group D got only 68.4 %
of the correct answers in the first exercise on the post-test. Generally speak-
ing, the third exercise, where the multiword idioms had to be filled in,

Authenticated
248 Szilvia Csabi
appeared to be the most difficult exercise since the number of correct results
was the lowest there. Possibly the strings of words in the idioms were too
long to remember precisely since students only had a very short time to see,
understand, and memorize them. Still, the tendency was that the experimen-
tal group did better on this part of the test also - although it has to be noted
that group D, the control group, had a higher number of entirely correct
answers in exercise 3 than Group C, the experimental group; however, the
total number of good answers was still higher in the case of Group C.
5. Discussion
The statistically significant differences in Experiments I and 2 seem to sup-

port the idea that the explicit knowledge of motivated meanings in the target
language helps learners to learn, remember, and use polysemous words such
as hold and keep better than other learners who only memorize the words and
their Hungarian equivalents. The hypothesis, according to which motivation
for the meanings of polysemous words and the idioms in which they occur
promotes better learner performance, and better results can be achieved, is
proved.
This finding is relevant in several respects. It raises questions concerning
the deeply entrenched traditions of memorizing words. The validity of mem-
orization is not subject to query, but its effectiveness can be promoted by
applying new principles such as using keywords referring to conceptual
metaphors, or schematic drawings to indicate meanings. It is not questioned
that good results can be achieved with the help of memorization only, but it
is argued that these results can be further improved by means of employing
cognitive semantics. The study also shows that it makes sense to teach
vocabulary in a systematic way since students can acquire and use expres-
sions more easily.
A further advantage is that this method can arouse students' interest in
and motivation for language learning since they themselves can be involved
in guessing meanings and actively participate in creative mental work. The
reactions of students in the experimental groups were positive and encour-
aging throughout the experiments.
Thus, the method of teaching polysemous items and idioms proposed
here appears to be a good way of complementing traditional ways of teaching
vocabulary. Naturally, this method can be improved and developed in many
directions to make the teacher's and the students' work even more interest-
ing and motivating. In any case, the emphasis lies on what MacLennan

Authenticated
(1994: 106) also claims to be of great importance, namely, that teachers

need to aid students "to develop special learning techniques for new
words", and that "[l]earning that utilizes metaphor ... could be one of these
techniques" .
The study also has implications for idiom teaching. Since teaching
idioms is generally thought to present difficulties in the language class-
room, idioms often play a rather neglected role in the curriculum, although
they occur frequently in the speech of native speakers of English, and their
acquisition is advantageous for learners. Probably language learners would
benefit from the use of the methodology advocated here since it possibly
provides them with a more meaningful approach to link the forms and
meanings of expressions than simple memorization. By means of employ-
ing the cognitive approach to vocabulary learning, students' desire to know
why things are the way they are in the target language (i.e., why X means Y)
can probably be satisfied to a greater extent than by using other methods.
Students in the experimental groups were explicitly taught the motivation
of word and idiom meanings, and they were instlUcted to learn items with
the help of keywords and visual images. Thus, mental pictures provided con-
nections to meanings, and these were stored in memory. The keywords and
visual images were associative mediators (Stevick 1976), which provided
meaningful links between expressions and their meanings, thus promoting
the storage and recovery of items that belonged together (i.e., a form and
the corresponding meaning). The use of mediators produced better retention
than simple repetition, and the creative mental work required by this kind of
teaching beneficially influenced the memorization and retention process.
As the results of the first tests show, the perfornmnce of the experimental
groups was better than the control groups, that is, the learned items were
better recalled from short-term memory. The differences between the per-
fonnance of the groups became greater in the post-tests. Thus, traces in the
long-term memory were probably stronger in the case of the experimental
groups, and fewer items were forgotten within one or two days, as results of
the post-tests show.
The fact that Groups C and D performed better after two days may be
attributed to the influence of distributed learning process. The separation of
the material to be taught and the smaller number of items to be learnt at the
same time affected the recall of items in a positive way. It takes longer to
learn a larger number of items because of the effect of croWding. That is, if
the number of the items to be learned at once is smaller, memorization and
retention is more successful. This may be the reason why Groups C and D
had better results on the post-tests.

Authenticated
250 Szilvia Csabi
A further question that the study bears on is the effect of the strategy of
transfer from LI to L2. Hungarian has two words (jog and tart) that have
similar meanings. However, the meanings of the Hungarian words are not at
all identical with those of hold and keep. Rather, their meanings only partly
cover those of the English words. The network of meanings structuring fog
and tart appears to be different, and there are different metaphorical links
also. Thus, there are no one-to-one correspondences between the meanings of
word pairs. Also, the motivations for the English expressions do not always
and necessarily match the motivations for their Hungarian equivalents.
Nevertheless, the fact that the meanings of the selected idioms may make
use of similar conceptual motivation in English and Hungarian does not
ensure that their linguistic forms will be exactly the same. Thus, the idioms
do not necessarily correspond word for word in English and Hungarian.
Consequently, students could not simply translate the Hungarian expres-
sions into English.
Thus, students' awareness of conceptual links is beneficial in compre-
hending and using the target language adequately. This is in line with
Carter's suggestion (1993: 148) that "learning a language involves under-
standing something of that language [which] has to be explicitly taught".
Carter also claims that language awareness aids the development of inferen-
tial, interpretative, and analytic skills. These claims also support the view
taken by Widdowson (1990) according to which it is probable that "a con-
scious awareness of how language works and the subjection of their experi-
ence to analysis would suit [learners'] cognitive style, increase motivation by
giving added point to their activities, and so enhance learning". Widdowson
also adds that the skill of comparing L 1 with L2 would engage students in
rational inquiry advocated in the curriculum. Thus, a possible answer to
these needs of language teaching may be provided by applying the cognitive
approach to language, and within that, vocabulary teaching.
6. Conclusion
This paper aimed at showing that cognitive linguistics can offer language
teachers and learners a helpful approach to deal with polysemous words and
idioms. This approach can explicate motivations of senses and systemati-
cally link them. Some meanings of the polysemous words hold and keep and
some idioms used in the experiments were shown to be motivated within
the framework oflexical network analysis. In order to see whether language
learning is more fruitful and permanent if it operates with conceptual mech-

Authenticated
anisms than learning founded upon pure memorization, two experiments

were carried out. The statistically significant and meaningful results of the
experiments support the hypothesis and suggest that there is a difference
between the perfonnance of the experimental and the control groups.
Naturally, the limitations of the study make it impossible to draw defini-
tive and general conclusions concerning language teaching. The study could
be developed and replicated in several ways. Nevertheless, the study has
certain implications for teaching. One of these is the beneficial effect of
learning sense motivations, since this can be more stable than memorizing
words and their LI equivalents. Another is the significance of awareness
raising and acquisition of metaphors and metaphorical networks. Although
this process is intellectually demanding, it seems to be useful in the class-
room since it probably produces better results in the long run. In addition,
students are in a more advantageous position to understand (at least parts
of) the operation of language. The cognitive approach can probably be
employed from the very start of language teaching; that is, it can be effec-
tively used with beginners as well as intennediate and advanced learners.
The approach can presumably be useful if employed throughout the whole
teaching process. However, we should not exclude this approach with stu-
dents who started learning English some years earlier.
Therefore, research concerning the application of cognitive linguistics in
teaching and learning foreign languages deserves attention. Employing cog-
nitive linguistic principles in the classroom produces statistically significant
results which researchers and teachers should not ignore. In the long run,
the contribution of cognitive linguistics to language teaching is likely to be
beneficial since it may improve and enrich teaching and learning methods.
Acknowledgements
I am especially grateful to Zoltan K6vecses, my consultant, who provided

me with advice, criticism, and encouragement. I wish to thank Zoltan
06rnyei, Giinter Radden, and Marianne T6th for their comments and sug-
gestions. I would also like to thank the students and the teachers of Radn6ti
Mikl6s Secondary School, Budapest, who took part in the experiments.

Authenticated
252 Szilvia Csabi
Appendix
Test
I. Fill in tire gaps witlr tire inflected form of hold or keep.

1. Kate is _ _ a book in her right hand.
2. Ronald Reagan _ _ the office of President for 8 years.
3. Jim hasn't got enough money to _ _ his family.
4. The terrorists _ _ them prisoner.
5. Jack _ _ Joe's secret for 5 years.
6. You can _ _ the book I lent you; I don't want it back.
7. Police are _ _ two men because of the jewel robbery.
8. Jane _ _ a diary for 15 years.
9. It is very expensive to _ _ a large house.
10. Mr. Smith does not _ _ the right degree for this job.
11. Fill in tire gaps witlr tire learned phrasal verbs.

1. Jim cannot stay at the same workplace for a long time. He cannot _ _ __
a job for more than a month.
2. The teachers _ _ John _ _ at school on Monday to finish his homework.
3. The heavy traffic _ _ us _ _ and we couldn't get to the airport in time.
4. Can't you read? The notice says "Danger- _ _ _ _ !"
5. The joke was very funny and James could not _ _ _ _ laughter.
6. Our king has not got enough soldiers to _ _ _ _ these people.
Ill. Fill in tire gaps witlr tire learned idioms.

1. John is usually friendly and polite with his colleagues, but he has not yet made
friends with any of them. He _ _ them _ _ .
2. Jane: I want to tell you something, Dad. I'm getting married next month.
Dad: What?! You're marrying Jim? I can't believe it! I don't want you to do it!
Jane: Don't be so angry, Dad. Just _ _ _ __
3. So, you'll participate in a competition on Saturday. I hope you will win. I'll
_____ for you.
4. Bill: I don't know what to do now. Jane will tell my wife and my family every-
thing about our love affair. I am so embarrassed and ashamed. I will never be
able to again.
5. Bob: Yesterday my daughter told me that she broke my favourite and very
expensive vase. I was very angry with her and I wanted to scold her but I didn't.
1 _ _ _ __
6. Kate: I have something to tell you, Sue. But don't tell it to anyone.
Sue: All right. I will _ _ this _ _ .
Kate: Fine. I am getting married to Peter next month.

Authenticated
Tables
Table 1. Percentage values for the total number of correct answers (Groups A & B)
Group A GroupB
Test Post-test Test Post-test
96.15 % 86.71 % 88.1 % 70.6%
Table 2. Percentage values for the total number of correct answers (Groups C & D)
GroupC GroupD
Test Post-test Test Post-test
95.71 % 87.7% 88% 71.2%
Table 3. Percentage values for the number of entirely correct answers (ECA) and
keyword-type answers (KTA)
Exercise 1 (%) Exercise 2 (%) Exercise 3 (%)

Group & Test
ECA KTA ECA KTA ECA KTA
Group A: Test 99.2 0 91 9 70.5 15.3
Correct answers 99.2 100 85.8
Group A: Post-test 90.7 0 71.7 19.2 56.4 19.2
Correct answers 90.7 90.9 75.6
Group B: Test 91.53 0 80.7 7.6 60.2 21.7
Group B: Post-test 78.4 0 64.1 10.2 42.3 10.2
Group C: Test 96 0 89.7 10.3 67.9 23
Correct answers 96 100 90.9
Group C: Post-test 93 0 71 18 50 26.9
Correct answers 93 89 76.9
Group D: Test 84.6 0 85.8 11.5 69.2 14.1
Group D: Post-test 68.4 0 53.8 21.7 60.2 11.5

Authenticated
254 Szilvia Csabi
Notes
1. Description of the links used in the analysis (cf. Norvig and Lakoff 1987):
- Metaphoric links (i.e., links established by metaphoric mappings that exist
independently of the given lexical item), e.g., The baby took the toy jrom its
mother and He took a glance at the book: PERCEIVING IS RECEIVING metaphor.
- Metonymic links (i.e., links established by metonymic mappings that exist
independently of the given lexical item), e.g., The love between them is strong:
LOVE FOR THE RELATIONSHIP IT PRODUCES metonymy.
- Profile shift (i.e., instances where what is backgrounded in one sense is fore-
grounded in a minimally related sense), e.g., 1 took a punch at him and 1 took a
punch from him: result profiled in the second example.
2. It is outside the scope of this study to consider each meaning of hold and keep-
rather, those meanings that are made use of in the experiments are focused on.
3. The shorter time spent on this task is due to the fact that the explanation proce-
dure was shorter and less detailed.

Authenticated
References
Carter, Ronald
1993 Language awareness and language learning. In: M. Hoey (ed.), Data,
description, discourse, 139-150. London: HarperCollins Publishers.
Deignan, Alice, Danuta Gabrys, & Agnieszka Solska
1997 Teaching English metaphors using cross-linguistic awareness-raising
activities. ELT Journal 51: 352-360.
Francis, W. N. & H. Kucera
1982 Frequency analysis ofEnglish usage: Lexicon and grammar. Boston:
Houghton Miffiin Company.
Gibbs, Raymond W.
1990 Psycholinguistics ofidiomaticity. Cognitive Linguistics 1-4: 417-51.
1994 The poetics ofmind. Cambridge: Cambridge University Press.
Gibbs, Raymond W. & J. O'Brien
1990 Idioms and mental imagery: The metaphorical motivation for idiomatic
meaning. Cognition 36: 35-68.
K6vecses, Zoltan
1990 Emotion concepts. New York: Springer Verlag.
2000 Metaphor and emotion. Language, culture, and body in human feeling.
K6vecses, Zoltan & Peter Szab6
1996 Idioms: A view from cognitive semantics. Applied Linguistics 17:
326-355.
Lakoff, George
1987 Women, fire, and dangerous things. Chicago: Chicago University
Press.
Lakoff, George & Mark lohnson
1980 Metaphors we live by. Chicago: Chicago University Press.
1987 Foundations of cognitive grammar, Volume 1. Stanford: Stanford
University Press.
Lennon, Paul
1996 Getting 'easy' verbs wrong at the advanced level. IRAL XXXIVI1:
23-36.
Lewis, Michael
1993 The Lexical Approach. London: Language Teaching Publications.
Low,G.
1988 On teaching metaphor. Applied Linguistics 9: 25-49.
MacLennan, Carol H. G.
1994 Metaphors and prototypes in the learning and teaching of grammar
and vocabulary. IRAL XXXIIIl: 97-110.
Norvig, P. & George Lakoff
1987 Taking: A study in Lexical Network Theory. Berkeley Linguistics
Society 14: 195-206.

Authenticated
256 Szilvia Csabi
Sandra, D. & Sally Rice

1995 Network analyses of prepositional meaning: Mirroring whose mind-
the linguist's or the language user's? Cognitive Linguistics 6: 89-130.
Stevick, E. W.
1976 Memory, meaning & method. Massachusetts: Newbury House Pub-
lishers.
Talmy, Leonard
1988 Force Dynamics in language and cognition. Cognitive Science 12:
49-100.
Taylor, John
1989 Linguistic categorization: Prototypes in linguistic theory. Oxford:
Clarendon Press.
Widdowson, H. G.
1990 Aspects of language teaching. Oxford: Oxford University Press.
Dictionaries
1989 Collins Cobuild Dictionary ofPhrasal Verbs.

1995 CoWns Cobuild English Dictionary.
1995 CoWns Cobuild. English Guides 7: Metaphor.
1987 Longman Dictionary o.lContemporary English. 2nd ed.
1989 Oxford Advanced Learner 50 Dictionary.
1989 The New Lexicon Webster 50 Dictionary ofthe English Language.
1989 The Oxford English Dictionary. Volumes VII & VlIl.
1987 The Random House Dictionary of the English Language. 2nd ed.

Authenticated
Applying Cognitive Linguistics to Pedagogical
Grammar: The Case of Over
Andrea Tyler and JYvyan Evans
1. Introduction
Language learning is one of the most complicated feats that human beings
accomplish. Any number of very real reasons exist as to why L2 learning
presents tremendous challenges. However, instructed L2 learning has been
further complicated by the fact that important elements of systematicity that
exist in language have not been appropriately captured by the pedagogical
grammars which underlie modem foreign language teaching textbooks and
materials. For instance, lexical classes, such as English prepositions, are
represented in the grammars (and the textbooks based on them) in piece-
meal fashion. When students (and their teachers) encounter varying uses of
these forms, the systematic relations between the multiple uses remain
unexplained. For example, traditional analyses have not offered any expla-
nations for why the four different meanings found in the sentences in (la-d)
are all associated with the form over:
(1) a The picture is over the mantle
b The teller at the central bank switched the account over to a local
branch
c The film is over
d Arlington is over the river from Georgetown
In sentence (la) over is interpreted as roughly 'located higher than'; in sen-
tence (lb) over is interpreted as roughly 'transferred'; in sentence (lc) over
is interpreted as roughly 'completed or finished'; and in sentence (ld) over
is interpreted as roughly 'on the other side of'. Such varying meanings are
typically presented, if they are addressed at all, as an unorganized list of
unrelated meanings that are accidentally coded by the same phonological
form. This results in a fragmented picture of the lexical class, leaving the
learner with the impression that the various uses are arbitrary. Indeed, learn-
ers of English as a second language and many teachers of ESL have noted
that acquiring the semantics of English prepositions is very difficult (e.g.,
Celce-Murcia & Larson-Freeman, 1998). In spite of the recognized difficulty,

Authenticated
258 Andrea Tyler and JY"Yan Evans
a survey of ten currently used English Language Teaching (ELT) textbook

series found that none even discuss this issue.
In the last 20 years, a new paradigm in linguistics, Cognitive Linguistics
(CL), has revealed that much that has been deemed idiosyncratic and arbitrary
under the traditional view of language is, in fact, systematic. CL provides a
unified, accessible account of how many grammatical constructions and
lexical items work and how varying uses of these forms are systematically
related to one another. Because CL adopts a usage-based approach to lan-
guage, it is mindful of the contexts in which lexical items and grammatical
constructions occur. Context-based analyses have revealed that speaker
choice of a grammatical construction, such as passive rather than active, is
meaning based. This insight is coherent with the basic CL tenet that syntax
and morphology are meaningful and governed by many of the same cogni-
tive principles as lexis. For the teacher, this approach has the potential to
provide rich insights into the organization of and motivation for the core and
"exceptional" uses associated with aspects of lexis and grammar. Ultimate-
ly, these insights offer language learners a more coherent and explanatory
description of the language. In this paper, we illustrate the usefulness to lan-
guage teaching of taking a CL approach through a brief examination of the
semantics of the English preposition over.
Traditional accounts have represented the semantics of English prepo-
sitions as highly arbitrary (e.g., Bloomfield, 1933; Chomsky, 1995; Frank,
1972). However, a number of cognitive linguists, such as Brugman (1988),
Dirven (1993), Kreitzer (1997), Lakoff (1987), and Lindner (1981) have
argued that a good deal more systematicity exists in the semantics of English
prepositions than has traditionally been assumed. Following up on that earlier
CL work and incorporating recent refinements in cognitive metaphor theory
(e.g., experiential correlation, Grady, 1999), Tyler & Evans (2001, 2003)
have argued that many of the multiple uses associated with a preposition,
such as over, are related in relatively straightforward, systematic ways.
Tyler & Evans (2001, 2003) demonstrate that by following a few basic
assumptions about the nature of language and applying a highly constrained
set of cognitive principles, a more systematic picture of the semantics of
English prepositions emerges.
The purpose of the present paper is to demonstrate the insights into the
semantics of English prepositions that arise from this model, and to illus-
trate how these insights might be applied in the language classroom.
Because of space limitations, this paper presents only the outlines of the
model as illustrated through an analysis of a limited number of the mean-
ings regularly associated with the preposition over.l

Authenticated
2. The Outline of the Model
The basics of the model include three fundamental assumptions about

human language, a schematic representation of the central sense associated
with the preposition, and a limited set of cognitive principles.
2.1. Basic Assumptions
We start by considering three fundamental assumptions upon which the

model rests.
2.1.1. The Principled Polysemy Network
Our first basic assumption is that the multiple meanings associated with each
preposition form a principled polysemy network organized around a central
sense, rather than a list of unrelated meanings. Two lines of argumentation
support this assumption.
First, work in experimental psychology (e.g., Rosch, 1975) has estab-
lished that humans organize their mental representations of categories around
a central exemplar that can be represented at various levels of abstraction or
generality (lohnson-Laird, 1983). Cognitive linguists (e.g., Dewell, 1994;
Langacker, 1987, 1991, 1992; Lakoff, 1987; Lakoff and Johnson, 1999;
Taylor, 1995; and Vandeloise, 1991) have extended this understanding of the
general organization of human cognition to the mental lexicon. Their work
offers strong evidence that the mental lexicon is not organized like a dic-
tionary in which each meaning associated with the same phonological form
represents an unrelated word. Rather, lexical items are better understood as
forming natural categories that participate in organized semantic networks,
or polysemy networks, organized around a central sense. Work in psycho-
linguistics (e.g., Sandra and Rice, 1995) offers empirical support for this
position.
The second line of argumentation in support of polysemy networks rep-
resents the view that the basic purpose for humans using language with each
other is communicative in nature. As a result, in naturally occurring com-
municative events, lexical items occur in sentential context, not in isolated,
citation form. Assuming that a lexical item is initially used to indicate one
established meaning, we posit that a speaker attempting to communicate
with a listener would use that lexical item to mean something new or different

Authenticated
260 Andrea IYler and Jryryan Evans
from the established meaning only if they believed the listener had a reason-
able chance of understanding the new meaning. This understanding presum-
ably would come from inferences arising from the situated or contextualized
use of the lexical item as it occurs in the ongoing discourse. This suggests
that the additional meanings that have come to be associated with over orig-
inally arose from situated uses and the inferences that were derivable from
context. With repetition across a number of similar contexts, the inferences
come to be independently associated with the lexical form as additional
senses; following Traugott (1989) we term this process of extending mean-
ing pragmatic strengthening.
To summarize, our first assumption is that the multiple meanings associ-
ated with a preposition are not accidental, but rather that they are related to
each other in systematic ways represented by an organized semantic network.
2.1.2. The Non-propositional Nature ojConcepts
We next turn to our second assumption, which concerns the non-propositional

nature of concepts. Cognitive linguists argue that 1) human conceptual
structure is crucially shaped by our human perceptions of and interactions
with the real world, i.e., the external physical-social world, and 2) language
is a reflection of human cognitive structure. Concepts deriving from human
interaction with the spatio-physical world, such as the spatial relations
coded by prepositions, are better represented as being more gestaltlike and
schematic in nature, often crucially involving sensory-motor imagery, rather
than as linguistic propositions or semantic feature bundles (e.g., Johnson,
1987; Kosslyn, 1980; Langacker, 1987).2 Mandler (1988, 1992, 1996) argues
that beginning at a very young age, through a process of reanalysis of per-
ceptual information, humans create mental representations of their recurring
sensory-motor experiences with the spatio-physical world. These conceptual-
izations involve spatial scenes, or highly abstract, schematic generalizations
established in memory in response to observing or experiencing physical
entities in a number of similar events or similar spatial relationships.
Since it is highly unlikely that our perceptions of entities and events in
the real world are interpreted in terms of bundles of linguistic propositions
or semantic features, it is also unlikely that our sensory-based conceptual-
izations are represented in memory in terms of linguistic propositions or
semantic feature bundles (Mandler, 1988; 1992; 1996). For instance, when
one encounters the lexical item bird, the mental representation prompted for
is probably not [+feathers, +wings, +sits on a nest, etc.], but something based

Authenticated
The Case afOver 261
on sensory-motor imagery, which is more holistic in nature. As Langacker

(1987) notes, it is virtually uncontestable that our understanding or concep-
tualization of objects involves imagery concerning their shape. Similarly,
we believe that the conceptualized spatial relations coded by prepositions,
and the situations or spatial scenes in which they are involved, are not likely
to be represented by linguistic propositions and semantic features.
Because the spatial relations coded for by prepositions have their origins
in our experiences with the spatio-physical world, particularly visual expe-
riences, we represent them here through diagrams. However, we want to
emphasize that by utilizing such diagrams we are not making any serious
claim as to how these concepts are actually represented in the human con-
ceptual system; they are simply attempts at characterizing the information
associated with prepositions and the utterances in which they occur, in non-
propositional terms.
2.1.3. Language Radically Underdetermines the Interpretation of

Utterances
We now turn to our third assumption. Interpretation of an utterance is always

richer than the content supplied by lexical items and the syntactic configu-
rations in which they appear (Green, 1989; Grice, 1975; Langacker, 1987;
Sperber & Wilson, 1986). Interpretation of utterances inevitably involves
inferencing and background knowledge. Moreover, in line with our assump-
tion of the non-propositional nature of meaning, we assume that linguistic
utterances that refer to actions or events in the spatio-physical world prompt
for gestaltlike conceptualizations of situations or scenes rather than a series
of discrete dictionary-type definitions strung together (Langacker, 1987).
3. Illustrating the Model
Now let us turn to a consideration of how these assumptions influence our

representation of the central sense of a preposition.
3.1. The Central Sense for Over
We propose that prepositions code for conceptual spatial relations between

two entities, one in focus and one in background. Following Langacker
(1987), we will call the focus element the Trajector (TR) and the back-

Authenticated
262 Andrea Tyler and ~)ryan Evans
ground element the Landmark (LM). We take this basic spatial relation
associated with each preposition as the central sense from which the various
additional meanings have ultimately been derived.
Conceptual content can be abstracted away from recurring spatial scenes,
giving rise to a highly abstract and schematized representation, which we
tenn a proto-scene. A proto-scene can be equated with the primary meaning
associated with a particular preposition, and thus includes infonnation relat-
ing to the TR and LM, as well as the spatial relation mediating the two. As
proto-scenes are idealized, they do not contain detailed information about
the nature of either the TR or the LM, nor detailed metric infonnation con-
cerning notions such as the exact shape of the LM or the degree of contact
between the TR and LM.3
3.2 The Proto-Scene for Over
•
Figure 1. The proto-scene for over
Figure 1 represents the proto-scene denoted by the English preposition over.

In figure 1 the dark sphere represents the Trajector (TR); the Landmark
(LM) is represented by a bold line. The dotted line indicates that the TR is
construed as being within potential reach of or being conceptually proximal
to the LM.4 The notion of the TR being within potential reach of the LM
represents a functional element which appears to be an important aspect of
the infonnation denoted by each preposition. The functional element arises
as a consequence of the spatio-configurational properties associated with a
particular preposition (cf. Tyler & Evans, 2001; 2003: chapter 7; Vandeloise,
1994). Another way to articulate the functional relation denoted by over is
to say that the TR and LM are within each other's sphere of influence, a
consequence of the TR being in a region conceived as proximal to the LM
(Dewell, 1994). In many scenes in which the configuration between the TR
and the LM is denoted by over, the influence of the LM on the TR is in the

Authenticated
form of the LM acting as a (potential) obstacle to the forward movement of

the TR.5
3.3. The Cognitive Principles
We noted earlier that in naturally occurring discourse, lexical items always

occur in context. In addition, we noted that work in pragmatics (e.g., Green,
1989; Grice, 1975) has established that interpretation of any utterance inevita-
bly involves inferencing. Therefore, in addition to representing prepositions
as non-propositional conceptualizations of spatial scenes, our model also
posits a limited set of cognitive principles that constrain and guide the infer-
ences which arise during the normal interpretation of utterances. These in-
clude a number of inferencing strategies, and ways of viewing a scene
(Langacker, 1987, 1992).
3.3.1. In/erencing Strategies: Real World Force Dynamics
In our full model presented in Tyler and Evans (2001; cf. Tyler and Evans
2003), we introduce several inferencing strategies. For present purposes we
will discuss only one.
As a default, speakers assume that all elements in a conceptual spatial
scene are subject to real-world force dynamics (Tal my, 1988, 2000).
Vandeloise (1991) discusses this in terms of a naIve theory of physics that
applies to how humans conceptualize spatial relations and how they use lan-
guage to express those conceptualizations. In other words, as listeners are
interpreting utterances, they assume the objects being discussed are subject
to force dynamics such as gravity.
3.3.2. Ways o/Viewing a Scene
In addition to the various inferencing strategies which guide situated inter-

pretation, spatial scenes are conceptualized from a particular vantage point.
The conceptualizer represents the default vantage point and is usually "off-
stage". However, any spatial scene can potentially be viewed from a variety
of vantage points. Langacker (1987, 1991a, 1992) argues that: "Grammar
(like the lexicon) ... structures a scene in a particular way for purposes of
linguistic expression, emphasizing certain facets of it ... , viewing it from a

Authenticated
264 Andrea Tyler and fYvyan Evans
certain perspective ... " (1987: 39). That is, the physical vantage point on a
spatial scene will determine how we conceptualize that scene, and no two
vantage points offer the same view. Hence, as the vantage point changes,
the exact interpretation ofthe scene changes.
Consider the following example. In a scene in which a large cloth is
positioned in relation to a table so that the cloth covers the top of the table,
the scene can be construed by focusing on contact between the cloth and the
table. In this case, the scene is likely to be coded in English by the sentence:
The tablecloth is on the table. Alternatively, the relationship between the
cloth and the table can be viewed as the cloth hiding or obscuring the table
from the observer's view. In this case, the scene might be coded as: The
cloth is over the table. A less typical, but perfectly acceptable view would
be to place the table in focus, in which case the coding would be something
like: The table is under the tablecloth. Hence, the same basic scene affords
several distinct ways of being viewed and interpreted.
Although this example involves changes in lexical items in order to
signal a change in vantage point, shifts in vantage point are not necessarily
coded by changes in lexical items. We will see several examples of this in
the following section.
4. Extending Meaning Beyond the Proto-scene
Now we turn to a consideration of how the proto-scene associated with over,

in conjunction with our basic assumptions and our cognitive principles, can
account for how several, distinct meanings came to be associated with the
preposition over. To illustrate, consider the interpretation of the very straight-
forward sentence: The cat jumped over the wall.
4.1. Overview
If we limit ourselves to only the information provided by the lexical items,

we know that in general there is an entity, a cat, involved in a particular
kind of motion, jumping, and at some point in this motion the cat was
located higher than a particular landmark, a wall. More specifically, we
know: 1) The lexical item jumped specifies a motion of pushing off from a
solid surface; we can interpret jumped as including information about the
starting point of the cat's motion; 2) The lexical item jumped also adds the
information that the cat used enough force to propel itself off the ground,

Authenticated
The Case afOver 265
thus creating momentum; and 3) The preposition over prompts for a partic-
ular conceptualized spatial relation between the TR (the cat) and the LM
(the wall, which is understood as a barrier which the cat must overcome if
its forward motion is to continue),6 as depicted in the proto-scene in figure l.
When interpreted within the context of the sentence, it tells us that at some
point in its jump, the cat was located higher than the wall.
The diagram in figure 2 captures the conceptualization that might arise if
the interpretation of the utterance were based solely on the information pro-
vided by the lexical items.
•
A
Figure 2. Schematization of a literal interpretation, i.e., one involving no inferences,

of sentences of the type: The cat jumped over the wall
In this conceptualized spatial scene, the beginning of the motion coded by

jump is represented as point A, the information coded by over is represented
as point B. Notice the LM (the wall) is represented by a vertical line.
Finally, the dashed arc represents the trajectory the cat followed to move
from the starting point of its jump to the position denoted by over the wall.
. I ..
......./ .........................................................
f\~
A C
Figure 3. Schematization of normal interpretation, i.e., one involving inferences, of
sentences of the type: The cat jumped over the wall

Authenticated
266 Andrea Tyler and Ji5Jvyan Evans
Clearly, this is not the normal understanding of the full motion the cat would
be expected to engage in when we interpret the sentence. Rather, native
speakers normally interpret the sentence to mean something like the cat fol-
lowed a trajectory that approximates the diagram in figure 3, in which the
cat comes back down to earth, ending its jump at approximately point C.
What we want to emphasize is that nothing in the linguistic information
provided in the sentence The cat jumped over the wall specifically codes
point C. In other words, point C is inferred. The question is how this infer-
ence arises.
We argue that the interpretation of this utterance, including the inference
of point C, comes from integrating our knowledge of: 1) the real world (for
instance, our knowledge of the action of jumping which involves an ani-
mate entity creating enough momentum to push itself off the ground and
propel itself to a position higher than the wall, and our knowledge of cats -
that they cannot stay suspended indefinitely in space the way, say, a hum-
mingbird can; 2) the key spatial configuration between the TR and LM
coded by over which tells us that at some point in its movement, the cat was
positioned higher than the wall and that the wall represents a potential
obstacle to the eat's forward motion; and 3) the force dynamics of gravity
and momentum which tell us that the cat, having reached point B, must
come back to earth, point C. Thus we argue that full interpretation of the
sentence The cat jumped over the wall crucially involves the inference that
the cat ends its jump at point C.
Furthermore, we propose that repeated observations of entities engaging
in similar motion (i.e., motion that involves the entity pushing off from a
starting point, reaching a point in its movement in which it is located higher
than a LM, and then returning to ground at point C) and exposure to utter-
ances which prompt for conceptualizations of entities involved in such
motion (for instance, The girl stepped over the branch in the path, The rab-
bit hopped over the stone, The horse jumped over the fence) result in a
highly abstract schematization being established in memory. The diagram in
figure 3, which we call the A-B-C trajectory, constitutes an attempt to repre-
sent this schematization.
We hypothesize that repeated encounters with utterances involving a
particular preposition, here over, and a particular inference, that the motion
that the TR engages in involves point Q, can result in the inference itself
becoming a distinct meaning associated with the lexical item or can give
rise to secondary inferences which become distinct meanings associated
with the preposition (Traugott, 1989).

Authenticated
The Case afOver 267
4.2. Inferences associated with the A-B-C trajectory and establishment of

extended meanings
In essence, we are arguing that the inevitable inferences that occur as a part
of normal, everyday interpretations of prepositions, as they occur in senten-
tial contexts, provide a powerful mechanism for extending the meanings
associated with a preposition. To illustrate this point, we will consider three
distinct meanings associated with over, which appear to arise from natural
inferences that result from interpreting sentences which involve an A-B-C
trajectory. We saw these three distinct meanings - transfer, completion, and
on-the-other-side - illustrated at the beginning of the paper. They are
repeated again in the examples in 2-4. The diagrams in figures 4-6 are
meant to represent the schematized spatial scene that is prompted for by
each of these distinct senses associated with over. In each case, they ulti-
mately arise from the inference of C in the A-B-C trajectory, but have been
changed in particular ways in line with the two cognitive principles dis-
cussed previously. The diagrams in figures 4-6 do not look exactly like the
diagram of the A-B-C trajectory in figure 3. Each diagram in figures 4-6
reflects some change or addition to the original spatial scene which is
important in the ultimate establishment of the distinct meanings associated
with over.
A key point we will attempt to convey is that at first glance, and if we
only attend to propositional definitions, there appears to be little relation
among the three meanings: transfer, completion, and on-the-other-side.
However, if we focus on the spatial scene prompted for by the preposition
as it occurs within sentential context and on the inferences that inevitably
arise during everyday interpretation of utterances, systematic relations
among these distinct interpretations reveal themselves. With all three of
these distinct meanings, the original spatial configuration of the TR being
higher than the LM is no longer active. The crucial point we are making is
that the inference of C arose from interpreting a sentence that does involve
the proto-scene.
4.2.1. The Transfer Sense
Recall sentence (lb) reproduced below as (2):
(2) The teller at the central bank switched the account over to a local
branch.

Authenticated
268 Andrea Tyler and JYvyan Evans
In this sentence, over has a meaning approximately equivalent to 'transfer'.

We suggest that this meaning has arisen because often when an object moves
from point A to point C, and there is a potential recipient located at point C,
then transfer of the object occurs. The notion of transfer is particularly
salient in a sentence such as: Andre Agassi hit the ball over the net to Pete
Sampras. We believe that after a speaker encounters several sentences such
as this one, which involves the interpretation that the movement of an object
through the A-B-C trajectory results in a transfer of the entity from point A to
point C, that the distinct meaning of transfer is added to the semantic network.
The diagram in figure 4 characterizes the spatial scene prompted for by
sentences ofthis kind:
...............................
«(/ I
Figure 4. Transfer sense: The TR has been transferred from the left side of the imped-
iment to the right side, as represented by the dark sphere which is in focus.
The TR has been transferred from point A to point C, as represented by the

sphere, which is highlighted. Highlighting is one of the changes in vantage
point that has been identified in the CL literature (Langacker, 1992). As we
noted in our earlier discussion, whenever there is a shift in vantage point, a
shift in meaning is involved. Thus we argue that there are two sources for
the addition of the extended meaning of transfer to the semantic network of
over - the situated inference of the object ending up at point C and the shift
in vantage point such that point C is highlighted.
4.2.2. The Completion Sense
The second distinct sense is illustrated in (lc) reproduced here as (3):

(3) The film is over
In this sentence, over is interpreted as something along the lines of 'complete'

or 'finished'. We suggest that this distinct meaning arises from the inference

Authenticated
that when the TR lands at point C, the process the TR is involved in is com-
pleted. In our interpretation of the film, we understand that a process is
completed, i.e., the viewing or showing of the film is completed. 7
The diagram in figure 5 represents the conceptualized spatial scene
prompted for by this utterance:
I
Figure 5. Completion sense: The dark sphere on the left represents the location of
the TR at the beginning of the process. The large sphere on the right,
which is in focus, represents the end-point or completion.
The sphere on the left represents the location of the TR at the beginning of the
action or process. The large sphere on the right is highlighted and represents
the location of the TR at the completion of the action or process. A key to
this interpretation is that the end point of the trajectory is highlighted or
given special focus. As we noted with the transfer sense, highlighting is one
of the changes in vantage point that has been identified in the CL literature,
hence establishment of the completion sense, with highlighting on point C,
involves a change in vantage point from the original scene depicted by the
A-B-C trajectory. As with all senses related to the A-B-C trajectory, point C
is taken to arise from an inference involving our knowledge of force
dynamics. Thus, we again see that the principle that a scene can be viewed
from many vantages, in conjunction with background knowledge of force
dynamics, combine to give rise to a new interpretation.
It is important to point out that in this use of over the focus is on point C,
the point at which the action is completed. The spatial configuration associ-
ated with the central sense of over is no longer strictly associated with this
sense. The location, point C, has been re-interpreted as providing information
about the action or process, not the spatial relation between the TR and the
ting a process and as such is now acting as an adverb. Through repetition
and entrenchment in memory - the process we are identifying as pragmatic
strengthening - the repeated inference of "completion" has come to be inde-
pendently associated with over as a distinct sense.

Authenticated
270 Andrea IYler and ~ryan Evans
4.2.3. The On-the-other-side Sense
Now reconsider the sentence in (1 d) reproduced below as (4):
(4) Arlington is over the river from Georgetown.
In this sentence, over provides the interpretation of 'on-the-other-side'. This

distinct meaning appears to arise from the inference that when the TR moves
through an A-B-C trajectory and ends at point C, it is located on the other
side of the LM from where it originally started. The final, resultant location
of a TR that has moved through an A-B-C trajectory has been reconceptual-
ized as a stable locative state. Moreover, the interpretation seems to involve
a shift in vantage point. Recall that in the proto-scene the vantage point of
the conceptualizer is "off-stage" (Langacker, 1987). However, sentences
such as Arkansas is over the bridge and (4), are only felicitous if the
speaker/conceptualizer is understood to be at roughly point A in the A-B-C
trajectory. Consider the following exchange:
(5) A: Where's Arlington from here?

B: It's just over the river.
This exchange would only be felicitous if the Arlington is on the opposite

side of the river from where the interlocutors are located. Thus, the interpre-
tation involves the interlocutors being located "on-stage", at point A.
The diagram in figure 6 represents the conceptualized spatial scene
prompted by this sentence:
« e» I•
Figure 6. On-the-other-side-of sense: The eye icon on the left represents the van-
tage point, the verticalline the impediment and the dark sphere the TR.
The vertical line represents the LM. The shaded (and hence highlighted)
sphere represents the TR, which is construed as being at the completion
point of the action. The eye icon on the left represents the vantage point,
which locates the conceptualizer as being onstage and represents a shift
from the basic A-B-C schematization.

Authenticated
Tt is important to note that in our understanding of this sense, neither the

central spatial configuration of one entity being located higher than another
entity represented in the proto-scene associated with over, nor the action of
moving from point A to point C is strictly associated or involved. The aspect
of the scene that is receiving focus is the final spatial positioning of the TR.
The meaning has come to be completely associated with the inferred point
C. Again, we see that a shift in the way the scene is viewed gives rise to a
distinct interpretation.
In sum, by assuming a non-propositional representation of the preposi-
tion, our proto-scene, in conjunction with clearly established principles such
as inferencing and knowledge of gravity, we have a rather straightforward
explanation of how these three seemingly unrelated meanings can be sys-
tematically related to over.
5. The Network of Senses Associated With Over
The diagram in figure 7 represents the polysemy network we have established

for over. The network involves 14 senses, including the proto-scene. For
each sense, by using a highly limited set of cognitive principles, we have
been able to trace how the distinct meaning could arise from interpretation
of the proto-scene associated with over. Moreover, we have been able to
construct similar networks for 17 of the most commonly occurring English
prepositions (cf. Tyler and Evans 2003).
Filled circles indicate a distinct sense in the network. Open circles indicate
a spatial scene which gives rise to a cluster of senses.
The following are illustrative sentences for each of the 14 senses:
1. The picture is over the mantel.
2A. Arlington is over the river from Georgetown.
2 B. Your article is over the page limit.
3. Joan nailed the board over the hole in the wall.
4. Frank looked over the train's undercarriage.
4 A. The committee agonized over the decision.

Authenticated
272 Andrea Tyler and !i5Jvyan Evans
above-and-beyond
(excess I) completion
2.B 2.C
transfer focus-of-attention
covering 4.A
3
2.A ....1 - - - - - - -__ 4

on-the-other- examining
side-of
trajectory
cluster
6 reflexive
5.A
more 5.B 5.C 6.A
control preference repetition
5.A.l
over-and-above
(excess JI)
Figure 7. The semantic network for over.
5 A. My mother never drives over the speed limit.
5 A 1. The child was overtired and so had difficulty falling asleep.
5 B. My neighbor always has control over his pit bull.
5 C. I prefer coffee over tea.
6. The fence fell over.
6 A. Marty keeps making the same mistake over and over.

Authenticated
6. Applying the Model to the Classroom
We believe that the approach to prepositions we have outlined here has the
potential to provide a number of benefits for the second language learner.
First, the model represents the various, extended senses associated with over
as being clearly motivated by a relatively small number of principles. Although
we acknowledge that some uses of over (and other English prepositions) are
bound to be idiosyncratic, especially, for instance, when they combine with
verbs in verb-particle constructions, the amount of arbitrariness is significantly
reduced under the current approach. Thus the model provides a more sys-
tematic, explanatory account of the semantics of English prepositions than
traditional approaches, cutting down considerably on the amount of arbitrari-
ness in the representation and hence reducing the need for rote learning on the
part of the second language learner. Second, because the model draws heavily
on the notion of the experiential basis of meaning and represents the extended
senses as arising from observations of the external, spatio-physical world, it
reflects the learners' own experiences with the world. Understanding the mo-
tivation behind the extended senses as experientially motivated and coherent
with the learners' own observations of the world would seem to make these
senses easier to acquire. Third, the various senses are represented as gestalt-
like conceptualizations of situations or scenes which are systematically con-
nected, rather than a series of discrete dictionary-type definitions strung to-
gether in a list. The systematic connections for over are modeled in the graphic
representation of the network, as illustrated in figure 7. Such graphic repre-
sentations of polysemy networks provide visual rubrics that may be useful
presentational tools for the language teacher and useful aids for the second
language learner. Finally, the constrained, principled nature of the model
would seem to provide a solid foundation for the learners from which to infer
the meanings of unfamiliar uses of over when they are encountered in context.
In the remainder of this section we offer a few suggestions concerning
how the proto-scene and two of the extended senses might be taught. These
ideas and materials have been piloted in a small, quasi-experimental class-
room intervention (Winke & Kim, 2002). These lessons are aimed at inter-
mediate-level English language learners who presumably have already been
exposed to some version of the central sense of over. The teaching activities
themselves and their sequencing are motivated by the model we have out-
lined. They draw on the notions that observations of the external, spatio-
physical world provide cognitive framing for the internal, conceptual world
and that cognitive representations of observations of events in the word
involve a scene complete with participants engaged in the event.

Authenticated
274 Andrea Tyler and ~ryan Evans
1. Start by introducing a visual representation of the proto-scene and empha-

sizing the spatio-physical configuration between the TR and LM. (Much
of this should be familiar, but the notion of the TR and LM being within
each other's spheres of influence will be new.) Briefly show several pic-
tures, accompanied by appropriate language, which illustrate the central
use of over.
2. Move to scenes involving the A-B-C trajectory:
a. The point to emphasize is that in this scene, over codes one crucial
point in the overall movement, the point at which the TR is higher
than but interacting in some way with the LM. Often the interaction
involves the LM being an obstacle.
b. Use a flip book which shows a cat jumping over a wall. Stop at the
series of pictures in which the cat is at point A, then go to pictures
where the cat is at point C. Ask how the cat got from point A to point
C. Point out that the wall is an obstacle in the cat's forward motion.
Next, stop at the series of points where the cat is best described as
being over the wall. Note that there are many points in the entire
event in which the cat is not over the wall, but that over picks out the
key points which tell us that the cat jumped such that it was higher
than the wall and within the wall's sphere of influence.
c. Continue through the pictures. Emphasize the notion that the cat must
land on the other side of the wall.
d. Emphasize that because English speakers use over when they
describe the scene involving movement from point A to B to C, a
strong connection has developed between over and this entire A-B-C
scene.
- Alternatively, the points could be made with clips from movies or
cartoons showing everyday actions which involve an object moving
from one side of an obstacle to another. Using Power Point, freeze
the frames which illustrate points A, B, and C (as in the discussion of
the flip book).
3. Now move to the presentation of the Completion sense.
a. Using flip book andlor video clips, stop at point C. Ask whether the
'jump' is completed. Has the cat finished jumping?
b. Present a visual representation of the A-B-C trajectory with C high-
lighted. Emphasize the notion that everyday actions of moving from
one side of an obstacle to the other side require the moving object to
finish the action at point C and that since over is used in the descrip-
tion of this whole scene, it has developed the additional meaning of
'finished, completed'.

Authenticated
The Case afOver 275
c. Give several examples, again using pictures, flip books, etc. E.g., 'The
jump is over'.
d. Then explain that once over became associated with completion of
physical movement, it could be extended to mean completion in gen-
eral. Give several examples of non-physical uses such as 'Class is
over'.
4. Move to the Transfer sense.
a. Ask a student to come to the front of the room. Stand on one side of
the desk (point A); ask the student to stand on the opposite side of the
desk (point C). Throw some large, silly object (like a nerf ball) over
the desk to the student. Repeat the throw, but before you throw the
object ask, 'Who has the X?' Toss the object, when the student has
caught the object ask, 'Now who has the X? How did the X get from
me to student? I threw it OVER the desk. I tossed it OVER the desk.'
b. Emphasize that: 1) because when an object moves from point A to
point B to point C, the object is transferred from A to C; and, 2)
because English speakers use over to describe the whole A-B-C scene,
over has taken on the additional meaning of 'transfer'. Reinforce with
several examples of physical transfer over an obstacle while intro-
ducing common phrases such as hand over, pass over, toss over, etc.
c. Explain that once over was commonly used to describe transfer of
physical objects, it was extended to indicate transfer in general: sign
over, turn over, win over, take over, etc. E.g.,
- The Beatles immediately won the hearts of millions of teenagers.
- The Beatles eventually won over the hearts ~f many of their parents
as well. {Note how the use of over raises the implication of an obsta-
cle that had to be overcome.}
- After long debate, George Bush managed to win over a few govern-
ments to his position on Iraq.
7. Conclusion
We have analyzed the mUltiple senses associated with each English preposi-
tion as forming a polysemy network organized around a central sense, the
proto-scene, which is made up of a TR and a LM in a specified spatial con-
figuration and a functional element. Each proto-scene is understood to con-
stitute the primary meaning representation associated with a particular
preposition, from which additional meanings have been systematically
derived. Thus, each preposition and the multiple uses associated with it are

Authenticated
276 Andrea Tyler and ~ryan Evans
represented as an organized, connected network of related meanings, rather

than arbitrary lists of distinct meanings that happen to share the same phono-
logical form.
We further argue that distinct, mUltiple senses can be accounted for by a
highly limited set of pragmatic and cognitive principles. In this paper we
have focused on the importance of non-spatial, extended meanings of the
prepositions and the interpretations that arise from the preposition as it
occurs in sentential context. In addition, we have considered the inferencing
strategy of using real world force dynamics in interpreting prepositions in
context; and we have examined the cognitive principle that a conceptual
scene can be viewed from a number of vantage points and that each change
in viewing can give rise to a change in interpretation ofthe scene. Inferences,
which are an unavoidable aspect of sentential interpretation, in conjunction
with shifts in vantage point, are argued to ultimately give rise to additional
meanings in the semantic network associated with a preposition.
We believe that such an analysis has great potential in offering a more
teachable account of the multiple interpretations assigned to each preposi-
tion. The suggested lesson illustrates how this understanding of prepositional
meanings can be presented to L2leamers with a minimum of technical jargon
or grammatical explanation. The language teacher, armed with the insights
provided by this account, can provide more coherent, insightful explanations
of the various meanings associated with English prepositions, and thus move
beyond the instruction to simply commit the various meanings to memory.

Authenticated
Notes
1. In the full analysis of over, 14 senses are identified, (cf. Tyler and Evans, 2001).
Moreover, we emphasize that the model is based on a thorough analysis of 17 of
the most commonly occurring English prepositions (cf. Tyler & Evans, 2003).
2. Langacker (1987) argues "sensory imagery is a real phenomenon whose role in
conceptual structure is substantial. We can plausibly suppose that a visual image
(or a family of such images presupposing different orientations and levels of
specificity) figures in our knowledge of the shape of an object; and certainly one
aspect of our conception of a trumpet assumes the form of an auditory image
representing the sound it makes" (pp. 111). He emphasizes that a commitment to
the importance of sensory imagery in the shaping of conceptual structure does
not imply a position that sensory imagery is an exclusive or essential facet of all
meaning of linguistic expressions. Neither should sensory images be confused
with the naIve view that a sensory or even a conceptual visual image is analogous
to a photograph or a picture. As the experimental psychologist Kosslyn (1980)
argues, "Image representations are like those that underlie the actual experience,
but in the case of mental imagery these representations are retrieved or formed
from memory, not from immediate sensory stimulation" (p. 18).
3. In our full model, the notion of a functional element plays a crucial role. We
hypothesize that in addition to the spatial configuration between a TR and a LM,
the concept prompted for by a preposition also involves a functional element
(Tyler and Evans, 2003; Vandeloise, 1991). In the case of in, for example, the
functional element involves the notion of containment. Johnson (1987), for in-
stance, has argued that the functional element of containment includes location,
confinement, protection, and potential obscuring of the element(s) being con-
tained. For instance, if a young child is in a playroom, the caretaker knows where
the child is located, the actions of the child are limited to those which can take
place within the space of the playroom, the child is protected from certain out-
side threats (e.g., the hot stove in the kitchen), and, for the most part, the child is
physically obscured from entities outside the playroom. The container and its
interior region also form the physical environment, which surrounds the TR. In
the case of the child in the playroom, the interior region of the room largely
determines the temperature, lighting, ambient sounds, etc, in other words, the
general physical environment which surrounds the child.
Our analysis has revealed that the functional element is key to appropriately
characterizing the distinction between the prepositions over and above. However,
the points made in the present discussion do not crucially refer to the functional
element.
4. The TR being potentially within reach of the LM allows over to depict spatio-
physical configurations in which there is contact between the TR and the LM as
well as those in which there is no contact. This is a crucial distinction between
over and above (whose functional element is distal in nature). The difference in

Authenticated
278 Andrea Tyler and Ji'j''vyan Evans
the functional elements of over and above is illustrated in pairs of sentences

such as:
a. Mary skated over the icy pond.
b. MalY skated above the icy pond.
The normal interpretation of sentence (a) is that there is contact between
Mary (i.e., the skates she is wearing) and the icy pond. The interpretation of
sentence (b), in contrast, does not allow for contact between the two. Mary might
be skating on a stream that is located at some distance from the pond or she may
have unearthly powers of levitation, but her skates are not understood to come
into contact with the icy pond.
5. Representing prepositions as spatio-physical relations between two entities
whose relationship involves a functional element is one of the important ways
in which this model differs from Brugman (1980) and Lakoff. As mentioned in
note 3, our representation of over differs crucially from that of above. Although
our representation of the proto-scene for over strongly resembles that of above,
it is misleading to "translate" the "meaning" of the protoseene for over with the
'above' sense, as Lakoff (1987) does. Such prepositional translating leads to the
erroneous conclusion that using the verb jump plus the 'above' sense of over
allows NSs of English to interpret the sentence The cat jumped over the wall as
the cat landing on the top of the wall, or on a spot slightly higher than the wall.
This is at odds with the interpretation normally assigned by native speakers.
6. Recall this notion of potential obstacle is a result of the functional element
denoted by over.
7. Here film is actually a metonymy in which the name of the physical entity (the
film or the movie) is understood as the process of showing the film.

Authenticated
References
Bloomfield, Leondard
1933 Language. New York: Holt, Rinehart and Winston.
Brugman, Claudia
1981 The story of over. MA Thesis. Linguistics Dept., UC Berkeley.
Published [1988] as The story of over: Polysemy. semantics and the
strncture of the lexicon. New York: Garland Press.
Celce-Murcia, Marianne and Diane Larsen-Freeman
1998 The Grammar Book: An ESLlEFL Teacher s Course. Rowley, Mass.:
Newbury House.
Chomsky, Noam
1995 Categories and transformations. In The minimalist program, 219-394.
Cambridge, MA : MIT Press.
Dewell, Robert
1994 Over again: Image-schema transformations in semantic analysis.
Cognitive Linguistics, 5, 351-380.
Frank, Marcella
1972 Modern English: A Practical Reference Guide. Englewood Cliffs, NJ:
Prentice-Hall.
Grady, Joseph
1997 Foundations ofmeaning: Primary metaphors and primary scenes. Lin-
guistics Dept., University of California at Berkeley. Ph. D Dissertation.
1999 A typology of motivation for conceptual metaphor: Correlation vs.
resemblance. In G. Steen and R.Gibbs (eds.). Metaphor in Cognitive
Linguistics (pp. 27-47) Philadelphia, PA.: John Benjamins.
Green, Georgia
1989 Pragmatics and Natural Language Understanding. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Grice, H. P.
1975 Logic and conversation. In P. Cole & J. Morgan (eds.) Syntax and
Semantics. vol. 3, 41-58. New York: Academic Press.
Johnson, Mark
1987 The Body in the Mind: The Bodily Basis of Reason. Thought and
Imagination. Chicago: Chicago University Press.
Johnson-Laird, Philip
1983 Mental Models. Cambridge, MA: Harvard University Press.
Kosslyn, Stephen
1980 Image and Mind. Cambridge, MA: Harvard University Press.
Kreitzer, Anatol
1997 Multiple Levels of Schematization: A Study in the Conceptualization
of Space. Cognitive Linguistics, 8,4,291-325.
Lakoff, George
1987 Women. fire and dangerous things: What categories reveal about the
mind. Chicago, IL: Chicago University Press.

Authenticated
280 Andrea Tyler and JYvyan Evans
Langacker, Ronald
1987 Foundations of Cognitive Grammar, vol. 1. Stanford, CA.: Stanford
University Press.
1992 Prepositions as grammatical(izing) elements. Leuvenese Bijdragen,
81,287-309.
Lindner, Susan
1981 A lexico-semantic ana~vsis ofEnglish verb particle constructions with
out and!l/2.. Doctoral dissertation. Linguistics Dept., UC San Diego.
Distributed [1983] by the Indiana University Linguistics Club.
Lindstronberg, Seth
1998 English prepositions explained. Amsterdam: John Benjamins.
Mandler, Jean
1988 How to Build a Baby: On the Development of an Accessible Repre-
sentational System. Cognitive Development, 3, 113-136.
1992 How to Build a Baby: H. Conceptual Primitives. Psychological
Review, 99, 4, 587-604.
1996 Preverbal Representations and Language. In P. Bloom et al. (eds.)
Language and Space, 365-384. Cambridge, Mass.: The MIT Press.
Rosch, Eleanor
1975 Cognitive representations of semantic categories. Journal of Experi-
mental Psychology: General, 104, 192-233.
Sandra, Dominiek & Sally Rice
1995 Network analyses of prepositional meaning: Mirroring whose mind-
The linguist's or The language user's? Cognitive Linguistics, 6, 1,
89-130.
Sperber, Daniel and Dierdre Wilson
1986 Relevance: Communication and Cognition. Cambridge, MA: Harvard
University Press.
Talmy, Leonard
1978 Figure and ground in complex sentences. In J. Greenberg (ed.).
Universals in human language, vol. 4, 625-649. Stanford, CA.:
Stanford University Press.
1988 Force dynamics in language and cognition. Cognitive Science, 12,
49-100.
2000 Toward a Cognitive Semantics. Cambridge, MA: MIT Press.
Traugott, Elizabeth
1989 On the rise of episternic meanings in English: An example of subjec-
tification in semantic change. Language, 65,1,31-55.
Tyler, Andrea and Vyvyan Evans
2003 The Semantics of English Prepositions: Spatial, Embodied Meaning
and Cognition. Cambridge: Cambridge University Press.
Vandeloise, Claude.
1991 Spatial prepositions: A case study in French. Chicago: The University
of Chicago Press.
Winke, Paula and YiYoung Kim
2002 Poster presented at Second Language Research Forum. Toronto,
Canada. (October 2002).

Authenticated
Index
AcquisitionlLearning Hypothesis 171 , constructions 1, 5-7, 51-53, 55-57,

178 61-64,66-73,122,125,131-132,
activation 66, 77-78, 80-82, 89, 177 149,151,166,174-175,180-182,
Affective Filter Hypothesis 170 185,192,258
as 195-196,204-208 cultural key word 146, 148, 153,
as long as 195,202-204, 208 155-157
as soon as 195, 202-204, 208 cultural script 2, 143, 145-148, 150,
awareness 8, 96-97, 99-103, 105, 152-154,156-157,159
107,110-113,127,147,190-191,
211-217,220-222,224-228,233- discourse style 156
234,236,250-251 discrete-item-syllabus 7, 118, 122-123,
127, 135, 138
bilingual lexicon 79, 81, 89
English 6, 8,15-17,20-21,24,28,32,
categorization 15, 52, 55, 103-107, 34, 56, 64, 73, 77, 80-81, 83-87,
113,168,175,177,181 89-91, 100-101, 104-105, 107,
causal/reason-giving link 207,209 109,111, 115, 121, 134-136, 142,
central sense 8-9,83,242,259,261- 144-146,148-153,156,159,171-
261,269,273,275 172, 178, 182, 185, 195, 200, 202,
Cognitive Grammar 1-2, 51, 165, 168- 212-214,216-219,221,233,236,
169,173,175,179,181,191 239,249-251,257-258,262,264,
conceptual metaphor 108-111, 211, 271,273-278
213,217,225-227,233-236,241, entrenchment 76, 177, 181, 183-184,
248 192
conceptual metonymy 108, 111-113, ethnocentrism 144
234-235 experiential correlation 258
conceptual representation 78-80, 82, experiment 15, 22-23, 83-88, 90-92,
89-90 104,123,136,147,212-216,218,
conceptualization process 51, 53, 67, 220-225,233,235,239,241-242,
71 244,246-252,254,259,273,277
concessive/adversative link 195
connectors 195,209 figure/ground gestalt 7,119,121, 123-
construal 3, 6, 14, 51, 55-56, 168, 124, 127, 130, 135-139
173-174,179-180,184-185,188- force dynamics 240, 263, 266, 269,
192,196,209,235 276

Authenticated
282 Index
foreign language teaching methodol- learner constructions 51-53, 64, 66,

ogy 8, 95-96, 99,101-103 69-73
frequency 6, 17,25,52-53,59-61,64, learning patterns 7
77-78, 80, 82-86, 88-91, 136, lexical fossilization 80
234 light verbs 52, 57, 61,64, 73
functional element 262, 275, 277-278 linguistic relativity 8, 95, 102
link of contrast 196
Grammatical activities 172, 180, 182-
184,191-192
Grammatical instruction 7, 165-166, Malay (Bahasa Melayu) 8, 148, 153-
168,171,173,179-182,184,191 156, 159
Grammatical rules 175,180,185,192 metaphor 4,8,80,103, 108-110, 112,
ground 6-7,16-17,19-20,23,31-34, 123,196-197,208,213-217,220-
36,41,71,115,119,121,123,139, 228,234-236,240-244,249,254,
198-199 258
metapragmatic ground 127
hedges 127, 151 Monitor Hypothesis 171
Hungarian 8,237-239,245,250 Motion events 6-7,13,15-17,19-20,
23,25,41-43,73
idioms 5, 211-214, 217-221, 224-226, motivation 8-9, 98, 111, 138, 148,
234-236,238,240-250,252 233-237,239,241-245,248-251,
indexicality 131, 135, 138 258,273
inferencing 196-197, 261, 263, 271,
276 Natural Approach 165-166, 168-169,
Input Hypothesis 169-170, 179 172-173,178-179,188,191
interactive activation model 76, 81, 89 Natural Order Hypothesis 171, 178
intercultural awareness 100-102, 111 network 3-5,9,113,177,214,233,
interlanguage pragmatics 146 239,241,250-251,259-260,268,
intelTogative directives 151 271-273, 275-276
L 1 transfer 30, 64, 71
over 9,257-258,260-262,264-275,
landmark 2, 131, 137, 198,262,264 277-278
language acquisition see also second
language acquisition 1, 5-7, 13-15,
18-19,26,29-30,39,41,43,51, polysemy 8, 22, 233, 235, 259, 271,
55-56, 115, 119-122, 127, 130, 273,275
135,137-139,165,167,169-170, polysemy network 259,271
172,179,203,211 pragmatic strengthening 260, 269
language awareness 96-97, 99, 102, pragmatics 8,126,131,135,143,146-
105,250 148, 156, 263

Authenticated
Index 283
prepositions 6, 77, 83-86, 88-91, 113, tag questions 152

257-258,260-261,263,267,271, temporal link 195
273,276-278 test 24-26,32,37,39,57-59, 83-84,
principle of contrast 80 92, 199, 214-215, 223, 228, 237-
proto-scene 262,264-265,267,270- 239,245-248,252
271,273-275,278 trajector 124,131,137,261-262
prototypicality 103-107,113,197 translation equivalence 82-83, 91
radial category 8
retention 8,212-216,220-222,224- Usage event 55,179-180
226,249 usage-based 4-6,51, 54-56, 77, 122,
124,258
second language acquisition 6, 13, 18, usage-based approach 5, 51, 54-56,
26, 115, 119, 121-122, 130, 135, 77,258
137-139,170,172
semantic extension 4, 8
semantic prime 144 vocabulary 8, 22, 26, 100, 112-113,
Semantics of English prepositions 145,149,157,159,181-182,211-
257-258,273 215,220-222,224-226,228,233-
similarity 6, 77-78, 81-82, 109, 158, 234,236,248-250
178
since 195-196, 200, 204-208 ways of speaking 144-145,150
spatial scenes 261-263 when 195-203, 205,208
speaker construa151, 55-56 while 195-196, 201, 204-206, 208

Authenticated
Authenticated
Authenticated
Authenticated

Cognitive Linguistics Second Language Acquisition and Foreign La 2004

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cognitive Linguistics Second Language Acquisition and Foreign La 2004

Uploaded by

Copyright:

Available Formats

Cognitive Linguistics, Second Language Acquisition,

and Foreign Language Teaching

Brought to you by | The University of Texas at Dallas (UTDALLAS)

Brought to you by | The University of Texas at Dallas (UTDALLAS)

Library of Congress Cataloging-in-Publication Data