linguistic and cognitive determinants of lexical acquisition in a second language

51
LINGUISTIC AND COGNITIVE DETERMINANTS OF LEXICAL ACQUISITION IN A SECOND LANGUAGE* I. Helene Ijaz The York Region Roman Catholic Separate School Board Richmond Hill, Ontario, Canada The meaning ascribed to select English spatial prepositions by adult EngliTh speakers was compared to that of advanced adult ESL learners. Two measures were used: a semantic-relatedness test and a cloze-type/sentence-completion tcst. The ESL learners differed substantially from the native speakers in the semantic bounda- ries they ascribed to the words. The differences derived from a different weighting of the words’ semantic dimensions. The weighting of the words’ semantic dimen- sions by the ESL learners was influenced by native-language transfer. The words’ usage was related to the degree of similarity in the linguistic classification of corres- ponding meanings in the L1 and the L2 and the consequent conceptual restructuring required in the acquisition of the L2 words. The meaning ascribed to central mem- bers of the semantic category investigated more closely approximated that of native speakers than the meaning assigned to noncentral ones. The findings are shown to be related to the similar linguistic classification of central concepts of semantic categories across languages and the language-specific classification of noncentral ones. The findings are interpreted as evidence that the principle of prototypicality underlies the structure of all languages as well as the cognitive processes involved in language acquisition. The central issue examined in this paper concerns the degree to which the meaning ascribed to words by advanced adult second-language (L2) learners is identical to that of adult native speakers. Observations on the lexical usage of advanced L2 learners (Harley and Swain 1978) suggests that such language learners overall, closely approximate native speakers in the use of L2 words but sometimes show slight differences. Second- language learners appear to experience particular difficulty with the use of terms that are closely related (Abberton 1968; Duskova 1970; Leven- ston and Blum 1977; Blum and Levenston 1978a, 1978b; Mougeon, Canale and Carroll 1979; Harley 1982)’ but neither the exact nature of the problem nor its sources has been identified. The present study com- pares actual advanced second-language learners’ interpretation of related English spatial prepositions to adult native English speakers’ treatment *This article is based on the author’s doctoral dissertation. 40 1

Upload: i-helene-ijaz

Post on 03-Oct-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

LINGUISTIC AND COGNITIVE DETERMINANTS OF LEXICAL ACQUISITION IN A

SECOND LANGUAGE*

I. Helene Ijaz The York Region Roman Catholic Separate School Board

Richmond Hill, Ontario, Canada

The meaning ascribed to select English spatial prepositions by adult EngliTh speakers was compared to that of advanced adult ESL learners. Two measures were used: a semantic-relatedness test and a cloze-type/sentence-completion tcst. The ESL learners differed substantially from the native speakers in the semantic bounda- ries they ascribed to the words. The differences derived from a different weighting of the words’ semantic dimensions. The weighting of the words’ semantic dimen- sions by the ESL learners was influenced by native-language transfer. The words’ usage was related to the degree of similarity in the linguistic classification of corres- ponding meanings in the L1 and the L2 and the consequent conceptual restructuring required in the acquisition of the L2 words. The meaning ascribed to central mem- bers of the semantic category investigated more closely approximated that of native speakers than the meaning assigned to noncentral ones. The findings are shown to be related to the similar linguistic classification of central concepts of semantic categories across languages and the language-specific classification of noncentral ones. The findings are interpreted as evidence that the principle of prototypicality underlies the structure of all languages as well as the cognitive processes involved in language acquisition.

The central issue examined in this paper concerns the degree to which the meaning ascribed to words by advanced adult second-language (L2) learners is identical to that of adult native speakers. Observations on the lexical usage of advanced L2 learners (Harley and Swain 1978) suggests that such language learners overall, closely approximate native speakers in the use of L2 words but sometimes show slight differences. Second- language learners appear to experience particular difficulty with the use of terms that are closely related (Abberton 1968; Duskova 1970; Leven- ston and Blum 1977; Blum and Levenston 1978a, 1978b; Mougeon, Canale and Carroll 1979; Harley 1982)’ but neither the exact nature of the problem nor its sources has been identified. The present study com- pares actual advanced second-language learners’ interpretation of related English spatial prepositions to adult native English speakers’ treatment

*This article is based on the author’s doctoral dissertation.

40 1

402 Language Learning Vol. 36, No. 4

of the same terms. It also explores factors that influence the acquisition of word meanings by second-language learners.

LEXICAL USAGE PATTERNS OF SECOND-LANGUAGE LEARNE-RS

Levenston and Blum (1977), in a study of the lexical usage of advanced adult L2 learners of Hebrew, found that, compared to native (Ll) speak- ers, the L2 learners used more words of general than of specific meaning, tended to generalize inappropriately, and failed to respond to constraints of register and collocation. Harley (1982), who compared the verb usage of French immersion students with that of native French speakers, found that many erroneous usages by the immersion students arose from L1 transfer at the incongruent boundaries of terms with otherwise overlap- ping meanings in the L1 and the L2. Harley and Swain (1978), in a study with a similar sample, noted that in the case of polysemous words, some meanings had been acquired while others had not. A similar finding was made by Kellerman (1978). The investigator asked Dutch university stu- dents and school children to judge the acceptability of English expres- sions involving break where Dutch can use the equivalent broken. He found that despite the respondents’ range of age and experience with English, a consistent pattern emerged with regard to the acceptability of the English expressions. The sentences He broke his leg and The cup broke were consistently rated as most acceptable in English. But senten- ces with more figurative, idiomatic, meanings of break such as She broke the world record and His fall was broken by a tree were rated as unac- ceptable in English by many L2 learners, although their literal transla- tions in Dutch are possible. According to Kellerman, the rank order of acceptable expressions in English seemed to be determined by the degree to which they represented the core meaning of break.

In a further study, Kellerman (1979) found that idiomatic or figurative meanings of L2 words are perceived to be highly marked by language learners. The investigator asked advanced Dutch learners of English to translate the Dutch sentence Zijn val werd door een boom gebroken which can be literally translated into English (His fall was broken by a tree). Most respondents avoided the direct translation and in many instances an inappropriate lexical item was used to render the meaning of gebroken.

Ijaz 403

The above findings suggest that L2 learners differ from adult native speakers in the meaning they ascribe to words in the target language, and that some meanings of L2 words are acquired more easily than others.

CONCEPTUAL AND LINGUISTIC CONSTRAINTS ON THE MEANING ASCRIBED TO L2 WORDS

Words do not carry meaning by themselves, but only in relation to concepts. Words provide linguistic labels for concepts, and meanings constitute the associate links between concepts and words. A word’s meaning refers to the total communicative value expressed by the word (Leech 1974). The instances to which a word’s meaning can be applied in linguistic usage define the word’s semantic boundaries. Within the total linguistic/semantic system of the language a word’s semantic boundaries are defined by the word’s semantic relations to other words. These, in turn, are defined in terms of different degrees of similarity and contrast.

PROTOTYPICAL STRUCTURE OF CONCEPTUAL CATEGORIES

Conceptual categories are structured around central members or best examples (Rosch 1973; Rosch, Mervis, Johnson and Boyes-Braem 1976). The central member of a category best exemplifies the category as a whole in that it has the largest number of attributes associated with the category. Other members may belong to the category to varying degrees, depending on the number of attributes they share with the central member. Category members resemble each other to varying degrees and the degree of resemblance of a given item to other members of the category correlates negatively with its membership in other categories (Rosch and Mervis 1975). The more attributes an item shares with other category members, the less likely it is to have elements in common with other categories. Conversely, the fewer characteristics it shares with other category members, the more likely it is to share attributes with other categories. Category members may differ with regard to the number and nature of their attributes as well as in the relative salience and importance of attributes (Mervis 1980).

A number of studies have demonstrated that adults and children of all ages agree on the best examples of categories, but that children at different age levels differ in the boundaries they attribute to categories

404 Language Learning Vol. 36, No. 4

(Saltz, Soller and Siege1 1972; Rosch 1973; Neimark 1974; Rosch 1975; Anglin 1977; Hayes 1977). Disagreements between and within age groups essentially concern the poorer exemplars of categories but, with increasing age, children approximate the judgments of category membership of adults.

Rosch et al. (1976) have shown that in natural object categories best examples are the first categorizations made in the perception of the environment, the categories first learned and named by children, and the most inclusive for which a concrete image of the entire category can be formed. They concluded that best examples represent categories at a level of abstraction at which the most basic category cuts are made.

WORD MEANINGS AS SETS OF DIFFERENTIALLY WEIGHTED SEMANTIC DIMENSIONS

Coleman and Kay (1981) have shown that the principle of prototypicality underlies the meaning structure of polysemous words. They have demonstrated that the meaning of the English verb lie consists of a cognitive prototype to which various instances of its use may correspond to varying degrees. They proposed that the verb lie represents a semantic category characterized by three elements: the falsity of the statement, the speaker’s belief that the statement is false, and the speaker’s intention to deceive. They noted that in different instances of the word’s use these three elements may have different importance. Coleman and Kay have shown experimentally that category membership of instances to which the verb lie applies can be determined by the judgment of language users.

The findings by Coleman and Kay suggest that the meaning of a given sense or “meaning feature”’ of a word is defined by a particular combination of semantic dimensions that are variable in weight. The weight of a given semantic dimension is indicative of the degree of membership of the meaning in the semantic category defined by that dimension. Word meanings become thus defined as multidimensional networks of individually weighted semantic dimensions or components.2

The meaning features of words and the meanings of related words may differ from each other in the nature and weight of their semantic dimensions. Perfect synonymy between the meanings of two words implies overlap in all the meaning features the words can denote. Related words may overlap in their meaning features to varying degrees. Their

Zjaz 405

degree of overlap in meaning features determines their mutual substitutability in different linguistic contexts.

NATIVE LANGUAGE CONSTRAINTS AS POSSIBLE DETERMINANTS OF MEANING ASCRIBED TO L2 WORDS

Second-language acquisition, unlike first-language acquisition, involves the mapping of two lexical and conceptual systems onto each other. Across two languages many words may roughly correspond in meaning, but few word pairs completely overlap in all their lexical functions. Corresponding words in a language learner’s native language and the target language may differ in their semantic boundaries. The two languages may differ in the number and nature of distinctions made within a common, shared concept or in the linguistic distinctions made between semantic categories.3 In many instances, a lexical item in the L2 cannot be directly mapped onto a concept existing in the L1, and the L2 learner has to restructure existing L1 concepts or develop a new concept that corresponds to a lexical item in the L2.

Restructuring L 1 concepts and defining new semantic boundaries for them involves permeating and breaking down native language conceptual structures. The degree to which L2 learners are able to do this is unknown. Transfer of lexical meanings from the native language to the target language is a familiar phenomenon with second-language learners at the beginners’ level. Concepts underlying words in the L1 are transferred to the L2 and mapped onto new linguistic labels, regardless of differences in the semantic boundaries of corresponding words. Whether the influence of the mother tongue decreases in proportion to the learner’s growing proficiency in the target language or whether its influence is simply less noticeable as it affects progressively more complex areas of the target language is not known.

The findings from a number of studies with bilinguals (Schachter 1974; Hakuta 1975; Harley and Swain 1978; Kellerman 1978, 1979, 1983; Harley 1982) suggest that native language conceptual structures may often exert their influence on target language structures in a somewhat indirect fashion: they may not necessarily result in overt transfer of L1 structures, but rather in the avoidance or underrepresentation of L2 structures. D’Anglejan and Tucker (1973) and Bates and McWhinney (198 1) have shown that native-language constraints may influence conceptual patterns and semantic, pragmatic, and perceptual strategies in the L2 without becoming easily apparent in linguistic usage.

406 Language Learning Vol. 36, No. 4

HYPOTHESES

On the basis of the above empirical and theoretical considerations, it was hypothesized that

1 . advanced ESL learners would differ from adult native speakers in the semantic boundaries they ascribe to English words;

2. the meaning ascribed to typical or central members of semantic categories4 by ESL learners would correspond more closely to that of adult native speakers than the meaning ascribed to less typical members; and

3 . the semantic boundaries ascribed to English words by advanced ESL learners were influenced by native language conceptual constraints.

THE MEANINGS OF THE WORDS INVESTIGATED

The meanings of the following English spatial prepositions were inves- tigated: on, upon, onto, on top of, over and above. Prepositions, as parts of speech, may perform two functions: semantic and/or syntactic. In their semantic function, prepositions represent a relation between two entities. For example, in the sentence “The book is on the shelf,” the book and the sherf constitute the two entities and on expresses their relation to one another. The nature of the spatial preposition appropriate in a sentence is determined by contextual constraints, particularly the verb preceding it. In some instances, one of two or more prepositions may be acceptable in a given context and the term used may differ with different speakers.

The words investigated in this study fall into two groups: the semantic category of on-which also includes the words on top OL upon and onto-and the category of over, which includes above. In a pilot study we attempted to define the typical as well as the noncentral meaning features of the six words. Based on prototype theory (Rosch 1973; Rosch et al. 1976) we assumed that the central and noncentral meaning features of the words could be inferred from the frequency with which, in a task involving the free generation of sentences with the words, meanings with certain componential combinations occurred.

The method of semantic analysis used to define the meanings was based on the identification of the relevant semantic dimensions while treating each dimension as variable in weight. Two types of semantic

Zjaz 407

dimensions were distinguished: (1) bipolar dimensions that were shared by all the six words investigated and that were, to some degree, defining in terms of membership in the semantic domain encompassed by the words as a whole; and (2) semantic elements that were not shared by all the words, but only formed part of the individual meaning features of some of the words. In the analysis of each of the meanings described below, dimensions that were shared by all words are included. They are described by polar distinctions. The prefix “-” in the notation of such dimensions implies the absence of that dimension in a given componen- tial combination. By contrast, the prefix “7” implies not only the pres- ence of the relevant dimension, but also its variability in weight. Seman- tic elements that were not shared by all words included bipolar dimensions as well as factors of semantic consequence such as extremi- ties, parts of location, or the shape of entities. They are only included when they are present. In our notation system, they will be separated from shared dimensions and from each other by a slash (/).

We asked seventeen adult native speakers of English to generate sen- tences with on, upon, onto, on top OL over and above and analyzed the sentences produced in terms of the meanings involved. It was found that

1 . each of the six words has more than one meaning feature; 2. the words’ meaning features derive from different combinations of semantic

components; 3. each word has one componential combination that is more typical than others;

and 4. some meaning features are shared by more than one word.

For all words, sentences with concrete, spatial meanings as well as sentences with nonspatial meanings were generated. All the words’ spa- tial meanings involved the semantic dimensions f CONTACT, f VER- TICALITY, AND f MOVEMENT. That is, these dimensions are shared by all the words. All the uses of on involved contact between two entities, and in all but one sentence contact occurred with a surface. TCON- TACT combined with either T MOVEMENT or -MOVEMENT: static position, on one hand, and *VERTICALITY on the other. A compo- nent of T MOVEMENT was usually accompanied by one of 7 DIREC- TION. Other components could also be present, such as a SURFACE with which contact occurred, ELEVATION of that surface (The books are on the shelf), ATTACHMENT (The keys are hanging on the hooks), a FIGURATIVE MEANING, etc. The most typical meaning of on was

408 Language Learning Vol. 36, No. 4

defined by the componential combination T CONTACT T VERTICAL- ITY -MOVEMENT. That of upon and onto was found to be one of T CONTACT T VERTICALITY 7 MOVEMENT/ T DIRECTION and that of on fop of one of TCONTACT TVERTICALITY -MOVEMENT/ T ELEVATION/ 7 EXTREMITY. The meaning of on top of appears to be marked in terms of a high degree of ELEVATION of the entity with which contact occurs: it involves an EXTREMITY of ELEVATION.

All meaning features of over share a component of 7 VERTICALITY which can combine with either TCONTACT or -CONTACT and T MOVEMENT or -MOVEMENT. The most typical meaning of over is characterized by the componential combination -CONTACT T VERTI-

ARY TRAVERSAL can occur in either a horizontal or vertical direction, with the former being the more typical of the meaning of over. The most typical meaning of above was found to be one of -CONTACT T VERTI- CALITY -MOVEMENT. Table 1 shows the most typical componential combination of each of the six words along with a representative sen- tence. In order to avoid lengthy and cumbersome notations of the words’ meanings, the abbreviations used in Table 1 will henceforth be employed.

On, on top ofand upon share the componential combination TCON- TACT TVERTICALITY -MOVEMENT and on, on top OL upon and onto, T CONTACT T VERTICALITY T MOVEMENT/ T DIRECTION. Over and above overlap in their componential combi- nations -CONTACT T VERTICALITY -MOVEMENT, -CONTACT T VERTICALITY T MOVEMENT, and -CONTACT 7 VERTICAL- ITY T MOVEMENT/ T VERTICAL BOUNDARY TRAVERSAL. The meaning features shared by on on one hand, and on top OL upon and onto on the other, and those shared by over and above are nontypical of the meaning of on and over, respectively.

Obviously, our pilot study did not exhaust all the meaning features the six words can denote, but it uncovered their most typical meaning feature and some of their less typical ones.

CALITY 7 MOVEMENT/ T BOUNDARY TRAVERSAL. BOUND-

METHOD

The meaning of each word was defined by a multidimensional seman- tic space based on the word’s semantic relations with each of the other

Tabl

e I

Com

pone

ntia

l Com

bina

tions

of

the

Typi

cal M

eani

ng F

eatu

res o

f O

N,

ON

TO

P O

F, U

PO

N, O

NTO

, OV

ER

and

AB

OV

E

Wor

d C

ompo

nent

ial C

ombi

natio

n Ty

pica

l C

onte

xt

ON

7 C

ON

TAC

T V

ERTI

CA

LITY

-MO

VEM

ENT

I sa

t on

the

law

n.

(7 C 7 V-

M)

UPO

N

7 CO

NTA

CT

VER

TIC

ALI

TY 7 M

OV

EMEN

T 7 D

IRE

CT

ION

( 7

c 7

v 7 M

/ 7 D

) H

e pl

aced

his

han

d up

on h

er s

houl

der.

ON

TO

7 C

ON

TAC

T V

ERTI

CA

LITY

M

OV

EMEN

T/ T

Sh

e ju

mpe

d on

to th

e ho

rse’

s ba

ck.

7 DIR

EC

TIO

N ( 7 c 7 v T

M/ T

D)

ON

TO

P 7 C

ON

TAC

T 7

VER

TIC

ALI

TY-M

OV

EMEN

T/

7 D

IRE

CT

ION

/ 7 EL

EVA

TIO

N/ 7

EX

TREM

ITY

( 7 C

7 V

-M/

7 EX

TR

)

Whe

n yo

u st

and

on t

op o

f th

e hi

ll,

you

can

see

in a

ll di

rect

ions

. O

F

OV

ER

-CO

NTA

CT 7 V

ERTI

CA

LITY

7 M

OV

EMEN

T/

7 HO

RIZ

ON

TA

L B

OU

ND

AR

Y T

RA

VER

SAL

(-c T

V T

MI

7 H

BT

)

The

dog

jum

ped

over

the

fen

ce.

AB

OV

E -C

ON

TAC

T 7

VER

TIC

ALI

TY 7 M

OV

EMEN

T (-

C 7

/V-M

) Th

ere

is a

smal

l she

lf ab

ove

my

desk

.

410 Language Learning Vol. 36, No. 4

words. The group average semantic space assigned to each word by adult native speakers of English served as the baseline by which the semantic boundaries ascribed to the words by the second-language learners were measured.

Two tests were used: (1) a semantic-relatedness test and ( 2 ) a sentence- completion/cloze-type test. On the semantic-relatedness test, the seman- tic relationship between word pairs involving the six terms was to be expressed in terms of spatial distance along a straight line 12 cm in length. On the left end, the line was bounded by the word in relation to which the other word-printed somewhat above and to the centre of the line - was to be judged:

on

Subjects were asked to indicate by an X along the line where, in their estimation, the word above the line should be placed in terms of its semantic similarity in relation to the word on the left end of the line.

The sentence-completion test (SCT) consisted of 26 sentences with 28 required insertions. The sentences involved different meaning features of on and over. Some contexts corresponded to the terms’ typical or central meanings and others to noncentral ones. The test items fell into seven categories of meaning which differed from each other in either the pres- ence of one or more of the components CONTACT, VERTICALITY and MOVEMENT and/or in that they referred to a concrete or figura- tive level of abstraction. There were four sets of componential combina- tions with concrete, contiguous meanings ( 7 C 7 V-M; T C 7 V + M; T C-V-M; T C-V T M), two sets with concrete, noncontiguous mean- ings (-C T V-M; -C 7 V T M), and two sentences with figurative mean- ings. Contexts with meanings of the same combination of shared compo- nents (CONTACT, VERTICALITY, MOVEMENT) differed from each other in the weight of those components and/or in the presence of addi- tional components.

The first eight sentences were related to a picture. The remaining eighteen sentences represented independent and unrelated contextual units. All sentences consisted of simple sentence structures with subject and predicate, with the required insertions to be made in the predicate. Table 2 lists the SCT items arranged in order according to the seven

Iiaz 411

major categories of meaning. It also lists the complete componential combination of the relational meaning involved in each sentence as well as the item number of each sentence on the SCT, SCT1, 2, 3, 5, 6, 7, 12, 15, 17, 24 and 26 involved static, contiguous meanings; SCT8, 9, 14, 19, 20,21,22 and 25 motional meanings with resultant contact. SCT4 and 18 static, noncontiguous meanings; SCT10, 11, 13, 16 and 23 motional, noncontiguous, meanings; and SCT27 and 28 figurative meanings, one a meaning of on (SCT27) and the other of over and above (SCT28).

Of the contexts with contiguous meanings, some had a component of VERTICALITY (SCTI, 2, 3, 7, 8, 9, 12, 14, 17, 19, 22, 24, 25 and 26) and others did not (SCT5, 6, 15, 20 and 21). Some referred to contact with a flat surface (SCT1, 2, 3, 5, 7, 8, 9, 17, 19 and 25), others did not (SCT6, 15, 20 and 21), and in a third set of contexts the nature of the entity with which contact occurred was ambiguous (SCT12, 14, 22, 24 and 26). Two contexts were clearly marked in terms of an EXTREMITY of ELEVATION of the entity with which contact occurred (SCT1 and 24), whereas others were ambiguous with regard to the degree of ELE- VATION of the entity involved (SCT2, 7, 8, 9, 14, 22 and 25). Some meanings involved additional semantic components such as ASTRIDE- NESS (SCT12), ATTACHMENT (SCT6, 15, 19 AND 21), COLLISION (SCT20), or a FIGURATIVE MEANING (SCT17). The motional, non- contiguous meanings differed from each other in the weight of the com- ponent of MOVEMENT and/or in the direction of the movement. The sentences with figurative meanings were included in order to explore the validity of Kellerman’s (1978, 1979) finding that of the polysemous meanings of words, figurative meanings are more marked for second- language learners than concrete meanings.

SCT1, 2, 3, 7, 12, 24 and 26 all involved meanings of static contact from a vertical direction ( T C T V-M), presumably typical or central meaning features of on, and SCT5, 6, 8, 9, 14, 15, 19, 20, 21, 22, 25 and 27 nontypical ones. SCT10, 11, 13 and 23 involved the components -C 7 V 7 M/ 7 BT and presumably typical or central meaning features of over and SCT4, 18, and 28 nontypical ones. SCT16 was ambiguous as to whether or not a component of BOUNDARY TRAVERSAL was rele- vant and, as a result, with regard to its degree of resemblance to the typical meaning of over.

As was shown above, the degree of typicality of a meaning in relation to a semantic category hinges (a) on the nature of the relevant semantic dimensions and (b) on their weight. Because each semantic dimension is a

Tabl

e 2

Sem

antic

Ana

lysi

s of t

he M

eani

ngs of t

he S

CT

Con

text

s

Shar

ed

Com

pone

ntia

l C

ombi

natio

n C

onte

xt

TC

TV

-M

The

re is

a b

aske

t --

- th

e fl

oor.

Two

wat

ches

are

---

the

tabl

e.

The

boo

ks a

re -

-- th

e sh

elf.

She

sat --

- a

pillo

w -

-- th

e ch

air.

The

chi

ldre

n w

ere

sitti

ng -

-- th

e se

e-sa

w.

The

cat

is -

-- t

he r

efri

gera

tor.

Whe

n yo

u st

and

---

the

hill,

you

can

see

in

all

dire

ctio

ns.

Jane

’s a

part

men

t is

---

the

seco

nd f

loor

.

TC

-VT

M

TC

-VT

M

Mar

y cl

imbs

---

the

foo

tsto

ol to

reac

h th

e sh

elf.

Som

e w

orke

rs c

limbe

d --

- th

e ro

of.

I pu

t th

e di

sh -

-- th

e ta

ble.

Sh

e ju

mpe

d --

- th

e ho

rse’

s ba

ck.

He

plac

ed h

is ha

nd -

-- h

er s

houl

der.

Jim

nai

led

the

plaq

ue -

-- th

e bo

ard.

The

pic

ture

is -

-- th

e w

all.

The

key

s ar

e ha

ngin

g --

- the

hoo

ks.

Dog

s m

ust

be k

ept --

- th

e le

ash.

She

sew

ed t

he b

utto

n --

- th

e sl

eeve

. H

e st

oope

d to

avo

id h

ittin

g hi

s he

ad -

--

the

lam

p.

Com

plet

e C

ompo

nent

ial

SCT

Com

bina

tion

Item

7 C 7 V

-M/

7 S

3 7

c ?v

-M/

T S/

7~

2 T

c~v-

M/ 7 S/

TE

7 T

CTV

-MI 7%

7~

26

7 c 7

V-M

/ 7 E

/ 7 A

STR

12

7 c 7

V-M

I 7

S/ 7

E/ 7 EX

TR

7 c 7

V-M

I 7

E TE

XTR

1 24

7 c7

V-M

I 7

FIG

17

7 c7

v 7

MI 7

D/ 7 S/

7 E

8

7c7

v ?M

/ 7

D/ 7 SI

TE

7 c 7

~7

~1

TD

/ 7 S/

TE

7

c 7 V

TM/ 7

D/ T

E

7c

7v

~M

/TD

7

c 7

v TM

/ TA

7 C-

V-M

/ 7 S

+ C-V

-M/ 7 A

I

T C

-V-M

/ 7 A

Tc-

v?M

/~A

7 c 7

v 7 M

/ ~C

OL

9 25

22

14

19 5 6 15

21

20

-C 7

V-M

-CT

VT

M

7 FI

G

Jane

’s a

part

men

t is

--- th

e se

cond

flo

or,

dire

ctly

---

the

vari

ety

stor

e.

The

pic

ture

is -

-- th

e so

fa.

The

hel

icop

ter

is f

lyin

g --- t

he

apar

tmen

t bu

ildin

g.

The

dog

jum

ped

--- t

he f

ence

. T

he k

ite r

ose

high

er a

nd h

ighe

r, --

- th

e ho

uses

...

and

---

the

tree

s.

The

dan

cer

rais

ed h

er a

rm h

igh

--- h

er

head

.

Are

you

---

the

ball

team

thi

s su

mm

er?

The

pro

fess

or w

as l

ectu

ring

far

--- o

ur

-C 7

V-M

-C 7

V-M

/ 7 P

OST

-CT

VT

M

-c 7

v 7 M

/ 7 H

BT

-C +

V + M

/ + V

BT

-c7

v7

M/ 7

VB

T -c

7 v 7

MI

7 V

BT

7 C-V

-M/

7 FI

G

-C 7

v 7

MI

7 FI

G

18 4 16

13

10

I1

23

21

28

head

s.

Lege

nd:

A

AST

R

C CO

L D

E EX

TR

ATT

AC

HM

ENT

AST

RID

ENES

S C

ON

TAC

T C

OLL

ISIO

N

DIR

ECTI

ON

EL

EVA

TIO

N

EXTR

EMIT

Y

FIG

M

H

BT

POST

S V

V

BT

FIG

UR

ATI

VE

MEA

NIN

G

MO

VEM

ENT

HO

RIZ

ON

TAL

BO

UN

DA

RY

TR

AV

ERSA

L PO

STER

IOR

ITY

SU

RFA

CE

VER

TIC

ALI

TY

VER

TIC

AL

BO

UN

DA

RY

TR

AV

ERSA

L

414 Language Learning Vol. 36, No. 4

continuum that is weighted differently in different contexts and under the influence of other relevant dimensions, the distinction between cen- tral and noncentral meaning features of a word may not always be clear- cut. For example, a meaning may have a combination of components typical of the meaning of on or over, but the weight of the different semantic components may be nontypical of the target word and thus account for the relative nontypicality of the meaning involved in relation to the target.

The SCT contexts included sentences that clearly required the use of on or over and others in which the use of either on or over as well as that of one or more related near-synonyms was acceptable. In all sentences but two the grammatical function of the words was that of prepositions. In the two remaining sentences (SCT10, 11 and 23) the target words functioned as adverbs.

SUBJECTS

The sample consisted of six groups of subjects:

1 . A group of adult native speakers of English whose parents’ native language was English (EA;n = 87).

2. A group of adults whose native language, according to their own statement, was English, but whose parent(s) had a native language other than English (EB; n = 33).

3 . A group of adult German immigrants to Canada (GI; n = 45). 4. A group of German high school teachers of English, resident in Germany (GN;

5 . A group of adult immigrants to Canada whose native language was Urdu (UI;

6 . A group of subjects with nineteen different native languages who were either

n = 16).

n = 50).

immigrants to Canada or foreign students in Canada (MI; n = 38).

All ESL learners participating in the study had advanced to native-like oral proficiency in English. All participants except those in GN resided in English-speaking Canada. EB was treated as a group distinct from EA because in their childhood, subjects in that group may have been exposed to a language other than English and/or a nonnative-like form of English that may have influenced the nature of the meaning they ascribed to English words. EB was included among the L2 learner groups. Subjects in MI were included in the study to obtain a more comprehensive data

Zjaz 415

base from which to determine whether language learners, regardless of native language, more closely approximate native speakers in the mean- ing they ascribe to typical or central members of semantic categories than to noncentral ones.

The essential prerequisites for participants in the study were a mini- mum of twelve years’ formal education and, in the case of the L2 learn- ers, acceptable responses on at least 80 percent, 22 of the 28 test items of the SCT. Based on the finding of Oller and Inal (1971) that the preposi- tion usage of second-language learners is a valid measure of their sum- mative proficiency in English, it was assumed that by using the selection criterion of a score of 80 percent on the SCT for the L2 learners, only subjects with a high level of English proficiency would be included.

DATA ANALYSIS

The similarity ratings of the word pairs on the semantic-relatedness test by the native English speakers and the ESL learner groups were subjected to an analysis of variance and subsequently to multidimen- sional scaling. The multidimensional scaling procedure used was Young and Lewyskyj’s ALSCAL 4 (1981) which takes as input the group mean ratings of the similarity relations of each word and yields as output a Euclidean solution in n dimensions. The analysis yielded a representation of the semantic structure attributed to each term by the different groups in three-dimensional geometric space. This space was defined by the coordinates of the language groups along each of the three most salient semantic dimensions of each word.

In analyzing the data obtained from the SCT, all responses given on each test item were established. The frequency of each response item by members in each group was computed. Acceptable responses that were given by more than three native English speakers were treated as separate response options. Acceptable responses produced by three or fewer sub- jects in EA were combined as other acceptable responses. Acceptable responses given by more than three members of EA will henceforth be referred to as major acceptable responses (MARS) as opposed to other acceptable responses. In the classification of acceptable response terms, the cut-off point of three occurrences was somewhat arbitrary, but it appeared meaningful in view of the data at hands (a) in order to avoid an inordinately large number of indiscriminate other acceptable responses,

416 Language Learning Vol. 36, No. 4

and (b) to allow a meaningful comparison of the language learner groups with the native English speakers.

The number of MARS for each test item was different. Response terms that could not be easily classified as acceptable or unacceptable were presented to eleven judges who were native English speakers: (either linguists or teachers of English) and who had not participated in the study. The judges were asked to rate sentences, which had been com- pleted with the terms in question being underlined, as either acceptable or unacceptable in English. The data thus obtained for each test item were subjected to a variety of analyses using the Chi square test of statistical significance which compared the responses given by native English speakers with those of the L2 learner groups as well as with those of the combined ESL learners.

RESULTS

DIFFERENCES IN THE SEMANTIC BOUNDARIES ASCRIBED TO THE WORDS BETWEEN NATIVE SPEAKERS AND L2 LEARNERS

All the L2 learner groups were found to differ from native English speakers in the semantic relations they ascribed to the words on the semantic-relatedness test. They also differed from each other in the nature and degree of such differences. EB differed significantly from EA in their similarity judgment of 13 percent of the word pairs, GI of 40 percent, GN of 46.7 percent, and UI and MI of 73.3 percent each. The combined ESL learners differed significantly from EA in their similarity rating of 80 percent of the word pairs. Table 3 shows the group mean ratings for each word pair as well as the values for standard deviation and T-probability obtained in a comparison of each L2 learner group with EA.

The differences between EA and the L2 learner groups in the semantic boundaries ascribed to the words derived from a different weighting of the words’ semantic dimensions. Figures 1 and 2 show the configurations obtained by multidimensional scaling for the terms on and over and each of the L2 learner groups in relation to EA.6 Each figure consists of two plots of the three-dimensional subject weight space recovered by ALSCAL. One plot shows the values for dimensions I and I1 and the

rjaz 417

Tcble 3 Analysis of Variance of Similarity Ratings of Word Pairs by EA and the L2 Learner

Groups on the Semantic Relatedness Test

Word Language Standard Pair Group N Means Deviation T-Prob.

on- over

onto- above

on top of- on

above- over

onto- on top of

on top of- upon

EA EB GI GN UI MI

EA EB GI GN u1 MI

EA EB GI GN UI MI

EA EB GI GN UI MI

EA EB GI GN UI MI

EA EB GI GN UI MI

87 32 45 16 50 38

87 33 45 16 49 38

87 33 45 16 50 38

87 33 45 16 50 38

87 33 45 16 50 38

86 33 45 16 50 38

9.97 9.14 8.55 8.26 5.33 7.95

10.24 9.76 9.43 9.79 7.50 8.78

2.67 2.28 1.77 2.86 2.98 2.46

3.32 3.03 2.24 3.01 2.78 2.87

4.46 5.30 4.61 6.43 6.16 5.89

3.06 3.73 4.08 5.07 4.41 4.65

1.80 2.66 3.19 2.46 3.56 3.14

1.58 2.07 2.62 1.95 3.04 2.39

2.04 2.22 2.04 2.59 3.05 2.38

3.11 3.08 2.02 2.83 2.41 2.48

3.43 3.41 3.08 3.46 3.55 3.59

2.87 2.99 3.68 3.54 3.52 3.35

,111 .008* .016* .ow* .001*

,231 ,062 ,391 .ooo* .001*

,377 .018* ,783 ,528 ,633

.643

.017* ,694 ,254 .380

,231 .814 .036* .006* .033*

,313 .090 .024* .021* .013*

(Continued)

418 Language Learning Vol. 36, No. 4

Table 3 Analysis of Variance of Similarity Ratings of Word Pairs by EA and the L2 Learner

Groups on the Semantic Relatedness Test, continued

Word Language Standard Pair Group N Means Deviation T-Prob.

over- onto

upon- above

on top of- over

on- above

upon onto

on top of- above

EA EB GI GN UI MI

E A EB GI GN U I MI

E A EB GI GN U I MI

EA EB GI GN UI MI

E A EB GI GN UI MI

EA EB GI GN UI MI

86 32 45 16 49 38

86 33 45 I5 50 38

86 32 45 16 50 38

86 33 45 16 50 38

87 31 45 16 50 38

86 31 45 16 50 38

9.53 9.09 8.71 8.85 6.79 8.39

9.45 8.65 8.04 7.20 5.96 7.34

8.46 7.83 8.12 7.16 5.02 6.34

9.57 8.72 7.90 6.46 5.02 6.77

4.08 5.58 4.42 4.04 5.12 5.56

7.91 6.23 6.58 5.11 4.14 5.81

2.22 2.66 2.70 1.77 3.48 2.56

2.37 2.79 3.40 2.68 3.36 3.04

2.87 3.64 3.27 3.62 3.47 3.51

2.44 3.09 3.39 2.78 3.44 3.34

3.44 3.86 3.53 3.16 3.37 3.54

3.10 3.97 3.61 2.27 3.20 3.28

,407 ,086 ,190 .ooo* .021*

,146 .015* .007 * .ooo* .ooo*

.351 ,574 ,148 .000* .001*

,174 ,003 * .ooo* .ooo* .ooo*

.041* ,593 .965 .094 .030*

.016*

.030*

.002*

.ooo* ,001 *

Konlinued)

Zjaz 419

Table 3 Analysis of Variance of Similarity Ratings of Word Pairs by EA and the L2 Learner

Groups on the Semantic Relatedness Test, continued

Word Language Standard Pair Group N Means Deviation T-Prob.

upon- EA 87 2.88 2.88 on EB 32 1.78 1.93 ,058

GI 45 2.66 2.55 ,673 GN 16 2.29 2.68 ,436 UI 50 3.58 3.30 ,163 MI 38 3.44 2.83 ,308

over- EA 87 9.02 2.30 upon EB 31 8.66 3.34 .581

GI 45 8.38 2.99 .210 GN 16 7.63 2.07 ,024' UI 50 6.12 3.48 .mo* MI 38 7.51 3.15 .010*

onto- EA 86 2.79 2.63 on EB 32 3.35 3.50 ,415

GI 45 3.56 3.55 ,204 GN 16 2.01 1.20 ,065 UI 50 4.69 3.74 .002* MI 38 3.83 3.08 ,074

other those for dimensions I1 and 111.7 Also shown are the values for Kruskal's stress and squared correlation in distances.

The different semantic boundaries ascribed to the words by the ESL learners as opposed to the native English speakers was reflected on the SCT in the erroneous responses given as well as in a different distribution of alternative acceptable terms. Linguistic usages that are unacceptable in a given context represent violations of the semantic boundaries of the words used. They reflect the erroneous assumption on the part of the users that a meaning feature of the word fits and complements the meaning of the respective context. That is, a meaning is erroneously attributed to the word that does not form part of its meaning but that of another word.

In many contexts with two or more MARS, the L2 learners differed from the native speakers in the distribution of the MARS.* Unlike erroneous usages, differences in the semantic boundaries of words that result in differences in the distribution of alternatively acceptable terms are very subtle. They do not have a disruptive effect on communication

420 Language Learning Vol. 36, No. 4

+ GN

VERTICALITY (11) +

GN. - -

CONTACT (I)

.uI

GI. M 1.

EA

.EB

i r l

k l

.EB

- - U I 'GN

+

G I

EA.

G I

.MI

VERTICALITY (11)

Kruskal's Stress ,027 Squared Correlation ,999

'The dimensions are indicated between brackets.

Figure 1. Multidimensional Scaling of the Meaning Attributed to ON by EA and the L2 Learner Groups

VERTICALITY (11) +

CONTACT (111) +

G I

Kruskal's Stress ,026 Squared Correlation ,999

Figure 2. Multidimensional Scaling of the Meaning Attributed to OVER by EA and the L2 Learner Groups

Ijaz 421

and become apparent only in the usage patterns of larger groups of speakers rather than in the linguistic usage of individual speakers.

Our findings from the semantic-relatedness test and the SCT clearly support Hypothesis 1 that advanced ESL learners would differ from native speakers in the semantic boundaries they ascribe to English words,

LINGUISTIC USAGE PATTERNS BY NATIVE ENGLISH SPEAKERS ON THE SENTENCE COMPLETION TEST

With some exceptions, native English speakers agreed more strongly in their response pattern in contexts with typical or central meanings of on and over than in contexts with noncentral meanings. A11 the contexts with a meaning of static contact from a vertical direction ( T C 7 V-M), which had been presumed to involve typical or central meanings of on, elicited a response with on from 85 percent or more of the subjects in EA, with the exception of the following:

SCT

24 26

I The cat &--the refrigerator. When you stand---the hill, you can see in all directions. She sat---a pillow on the chair.

SCT26 already contains an on in its second prepositional phrase and, for stylistic reasons, the use of another on may have been avoided. SCTl ( 7 C 7 V-M/ T S/ T E/ 7 EXTR) ( 7 C T V-M/ 7 E/ TEXTR) are highly marked in terms of an EXTREMITY of ELEVA- TION of the entity with which contact occurs. Linguistically, this devia- tion from the typical meaning of on became reflected in the relatively low frequency of responses with on and the high frequency of responses with on fop of which was produced by more than one third of the subjects in EA. It appears that, despite the fact that SCTl and SCT24 had meanings of static contact from a vertical direction, the meanings involved- by virtue of the strongly weighted component of EXTREMITY- were more noncentral to the meaning of on than central. The meanings of SCTl and 24 will henceforth be treated as nontypical of the meaning of on.

In contexts with meanings of on that, according to their combination of shared components, were clearly noncentral to on, the frequency of responses with on varied substantially. In four of the twelve sentences, 85 percent or more of the native speakers used on:

and SCT24

422 Language Learning Vol. 36, No. 4

SCT

15 20 25

5 A mirror is---the wall. Dogs must be kept---a leash. He stooped to avoid hitting his head---the lamp. I pul the dish-the table.

Of these, SCT5, 15 and 20 involved nonvertical meanings of on and SCT20 and 25 motional meanings. The frequency of responses with on by EA was the lowest in certain contexts with a meaning of movement with resultant contact from a vertical direction (SCT8, 9, 19 and 22: T C T V T M ) . In these contexts, the majority of native speakers pre- ferred the use of other words, particularly upon and onto, whose typical meaning, unlike that of on, is motional.

The frequency of responses with on by EA was partly, though not exclusively, related to the degree to which the relational meaning involved in a context corresponded to the typical meaning of on. Differe- nces in the frequency of responses with on were particularly related to the presence or absence of the component of MOVEMENT. Overall, in contexts with static, contiguous meanings (7 C-M), the frequency of responses with on by EA was high and that with other prepositions was low. By contrast, in contexts with motional, contiguous, meanings ( T C T M) the frequency of responses with on was much lower, but that with other prepositions was higher. A considerably greater variety of terms was used in contexts with motional meanings of on than in con- texts with static meanings. In contexts with nonvertical meanings, the frequency of responses with on was, overall, high whereas in those with vertical meanings it varied substantially, depending on whether a motional or static meaning was involved. Evidently, in contexts with relational, contiguous meanings, differences along the semantic dimen- sion 1- MOVEMENT are made linguistically more salient in English than differences along the dimension f VERTICALITY.

In contexts with noncontiguous meanings, the frequency of responses with over by EA was closely related to the degree to which the relational meaning involved corresponded to the typical meaning of over. It proved to be particularly related to the degree of dynamic movement involved. Close to unanimous agreement on the use of over was found in the sentence The dog jumped- the fence which involves strong dynamic action. By contrast, the fewest responses with over were produced in the

Ijaz 423

sentence There is a picture-the sofa which involves a meaning of static position. The frequency of responses with over in contexts with noncon- tiguous meanings was inversely proportional to that of above. In the case of noncontiguous, vertical meanings as in that of contiguous meanings, linguistic distinctions centered predominantly around differences in weight along the semantic dimension k MOVEMENT.

THE LINGUISTIC USAGE OF THE L2 LEARNERS ON THE SENTENCE COMPLETION TEST

Like the native English speakers, the L2 learners associated the use of on and over, respectively, most readily with contexts with typical or central meanings of the target words. Table 4 shows the distribution of responses by the L2 learner groups in relation to that by EA in the SCT contexts, expressed in percentages,q as well as the relevant statistical information also included in Table 4 are the respective values for the combined L2 learners.

The combined second-language learners did not show a significant difference in responses from EA in any of the contexts with a meaning of static contact from a vertical direction ( T C T V-M) with three exceptions: SCTl and 24 which, because of their strong component of EXTREMITY of ELEVATION, were earlier shown to have meanings nontypical of on, and SCT7 (The books are-the table) which involved the componential combination T C 7 V-M/ 7 S/ T E. But in all contexts involving a motional, contiguous meaning with resultant contact from a vertical direction ( 7 C T V T M) and a clearly noncentral meaning feature of on, the combined ESL learners differed significantly in their response pattern from EA. These contexts included the four (SCT8, 9, 19 and 22) in which EA produced the lowest frequency of responses with on. Of the three contexts with a nonvertical, static meaning of contiguity ( 7 C-V-M), which represents another noncentral meaning feature of on, the combined ESL learners differed significantly in their responses from EA in two instances:

SCT 6 I5

The keys are hanging---the hooks. Dogs must be kept---a leash.

Tabl

e 4

Dis

trib

utio

n of R

espo

nses

in th

e SC

T C

onte

xts

by E

A a

nd th

e L

2 Le

arne

rs

on

top

SCT

Con

text

L

G

N on

up

on

of

onto

ov

er

abov

e fr

om

OA

R

UR

X

df

p

The

cat

is -

--

the

refr

iger

ator

.

Two

wat

ches

ar

e --

- th

e ta

ble.

The

re is

a

bask

et -

-- th

e fl

oor.

The

re is

a

pict

ure

--- th

e so

fa.

EA

86

54.7

EB

32

50

.0

GI

45

42.2

G

N

16

81.3

U

I 50

84

.0

MI

37

62.2

To

tal

L2s

180

62.8

EA

86

94

.2

EB

30

86.7

G

I 45

95

.6

GN

16

93

.7

UI

50

90.0

M

I 36

10

0.0

Tota

l L2

s 17

7 93

.2

EA

86

10

0.0

EB

31

100.

0 G

I 45

97

.8

GN

16

10

0.0

U1

50

96.0

M

I 37

10

0.0

Tot

al

L2s

179

98.3

EA

87

EB

33

GI

45

7.0

34.9

43

.7

2.2

53.3

18

.7

14.0

29

.7

.6

32.8

21.8

73

.6

18.2

75

.7

22.2

73

.3

2.3

3.1

2.2

2.0

5.4

2.8

5.8

13.3

6.

3 6.

3 2.

0

4.5

2.2 .6

1.2

3.1

3.41

5.

12

4.38

13

.31

2.7

4.17

1.1

9.88

.86

. 00

. 00

8.0

7.98

.9

5

2.3

2.14

. 00

.ll

.OO

4.0

1.28

.O

O

1.1

1.46

4.6

6.1

.27

4.4

.OO

4 4 4 4 4 4 1 1 1 2 1 2 1 1 1 1 1 2 2

,492

.2

75

,357

.0

10*

,382

,043

*

,352

1.

000

1.00

0 .0

19*

,329

,342

1 .om

.741

1 .

Ooo

,2

89

.482

,482

,873

L

,9

98

GN

U

I M

I T

otal

L2

s

5 A

mir

ror

is -

--

EA

th

e w

all.

EB

G

I G

N

UI

MI

Tot

al

L2s

6 T

he k

eys

are

EA

ha

ngin

g --

- th

e E

B

hook

s.

GI

GN

U

I M

I T

otal

L2

s

7 T

he b

ooks

are

E

A

--- th

e sh

elf.

EB

G

I G

N

UI

MI

Tot

al

L2s

16

50

38

182 83

33

45

16

50

38

182 86

33

45

16

50

38

182 84

33

45

16

50

38

182

96.4

97

.O

93.3

87

.5

88.0

10

0.0

93.4

76.7

54

.5

82.2

43

.7

74.0

73

.7

69.8

98.8

84

.8

88.9

93

.8

70.0

94

.7

84.6

18.7

75

.0

6.3

.O1

18.0

50

.0

32.0

19

.27

15.8

73

.7

10.5

1.

94

18.7

67

.6

13.7

5.

16

2.4

1.2

3.0

1.25

2.

2 4.

4 1.

34

12.5

6.

14

12.0

8.

35

1.40

.5

6.0

4.74

22.1

1.

2 36

.4

9.1

7.99

11

.1

6.7

4.99

25

.O

31.2

22

.86

10.0

16

.0

13.1

7 15

.8

10.5

6.

28

17.6

12

.6

9.60

1.2

12.1

3.

0 9.

66

8.9

2.2

6.65

6.

3 5.

47

4.0

26.0

25

.90

5.3

4.92

5.5

9.9

12.1

8

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

,932

.O

Oo*

,379

,076

,535

,5

12

.046

* .O

lS*

.495

,093

.018

* ,0

83

.om*

,001

* .0

43*

.008

*

.008

* .0

36*

,065

.0

00*

.086

.002

* (C

ontin

ued)

Tabl

e 4

Dis

trib

utio

n of

Res

pons

es i

n th

e SC

T C

onte

xts

by E

A a

nd i

he L

2 Le

arne

rs, c

ontin

ued

on

top

SCT

C

onte

xt

LG

N

on

upon

of

on

to

8 M

ary

clim

bs -

--

the

foot

stoo

l to

reac

h th

e sh

elf.

9 So

me

wor

kers

cl

imbe

d ---

roof

.

10

The

kite

ros

e hi

gher

and

hi

gher

, --

- th

e ho

uses

. . .

EA

87

EB

33

G

I 45

G

N

16

u1

47

MI

38

Tota

l L2

s 17

9

EA

87

E

B

33

GI

44

GN

16

U

I 50

M

I 37

To

tal

L2s

180

EA

87

EB

33

GI

45

GN

16

U

I 49

M

I 36

To

tal

L2s

179

27.6

23

.0

5.7

31.0

36

.4

12.1

15

.1

30.3

46

.7

4.4

11.1

31

.1

12.5

6.

3 12

.5

50.0

46

.8

8.5

4.2

14.9

50

.0

7.9

7.9

18.4

42.5

7.

8 9.

5 25

.7

19.5

10

.3

4.6

39.1

18

.2

6.1

15.1

42

.4

29.5

4.

5 20

.4

27.3

18

.7

62.5

30

.0

2.0

6.0

8.0

32.4

5.

4 21

.6

27.8

3.

9 9.

4 26

.7

over

ab

ove

6.9

3 .O

4.5

2.0

8.1

3.3

abov

e

from

5.

7 3.

0 4.

4

-

2.6

2.2

10.3

3.

0 6.

8

2.0

2.7

3.9

OA

R

UR

X

df

p

2.3

4.6

3.0

5.82

2.

2 11

.98

18.7

10

.68

25.5

25

.10

13.1

12

.97

12.3

25

.38

6.9

2.3

9.1

3.0

6.48

2.

3 4.

5 11

.94

18.7

15

.53

26.0

24

.0

42.4

8 10

.8

18.9

18

.90

11.7

13

.3

25.2

4

37.9

55

.2

36.4

54

.5

42.2

48

.9

31.2

62

.5

40.8

42

.9

44.4

50

.0

40.2

49

.7

3.4

3.4

3.0

6.1

.42

4.4

4.4

.52

6.3

1.12

4.

1 12

.2

4.70

5.

6 1.

92

3.4

6.7

1.52

6 6 6 6 6 6 7 7 7 7 7 7 3 3 3 3 3 3

,443

,0

62

,099

.oo

o*

.043

*

.ow*

,485

,1

03

.030

* .oo

o*

.009

*

.ool

*

,936

91

6 ,7

73

,195

,5

90

,678

11

and

---

the

tree

s.

12

The

chi

ldre

n w

ere

sitti

ng --

- th

e se

e-sa

w.

13

The

dog

jum

ped

---

the

fenc

e.

14

He

plac

ed h

is

hand

---

her

BA

86

E

B

32

GI

44

GN

16

U

I 49

M

1 36

T

otal

L2

s 17

7

EA

87

85

.0

EB

33

87.9

G

I 45

93

.3

GN

16

75

.0

UI

50

92.0

M

I 36

91

.7

Tot

al

L2s

180

90.0

EA

87

EB

33

G

I 45

G

N

16

Ul

50

MI

37

Tot

al

L2s

181

EA

87

71

.3

EB

33

81.8

5.7

3 .O

4.4

6.3

2.0

2.8

21.8

12

.1

shou

lder

. G

I 45

82

.2

8.9

37.2

25

.0

47.7

37

.5

51.0

47

.2

43.5

95.4

10

0.0

97.8

87

.5

74.0

86

.5

88.4

beyo

nd

~

43.0

4.

6 50

.0

3.1

43.2

4.

5 56

.2

30.6

30

.5

16.7

39.5

5.

1

astr

ide

4.6

6.1

2.2

-

2.8

2.2

12.8

15

.6

2.3

6.3

6.1

5.5

6.8

4.6

3 .O

2.0

2.8

I .7

3.4

6.3

8.0

5.4

3.9

2.3

2.3

6.2

2.60

2.

3 4.

28

2.13

12

.2

11.4

6 8.

19

5.1

4.19

.62

2.84

18

.7

17.9

5 4.

0 7.

56

2.8

5.03

3.3

7.47

1.1

1.57

2.

2 1.

79

6.2

2.17

18

.0

15.3

0 8.

1 4.

36

7.7

4.90

4.6

6.1

2.40

8.

9 5.

26

4 4 4 4 4 4 3 3 4 4 4 4 2 2 2 2 2 2 3 3

,626

,3

70

,712

.0

22*

.085

,381

,892

.4

18

,001

* .1

09*

.284

*

.113

.456

,4

08

.337

,0

01'

,113

,086

.494

,1

53

(Con

tinu

ed)

Tabl

e 4

Dis

trib

utio

n of

Res

pons

es in

the

SC

T C

onte

xts

by E

A a

nd t

he L

2 Le

arne

rs, c

ontin

ued

on

top

SCT

C

onte

xt

LG

N

on

upon

of

on

to

over

ab

ove

from

OAR

UR

X

df

15

Dog

s m

ust

be

kept

---

a

leas

h.

16

The

hel

icop

ter

is fl

ying

---

the

apar

tmen

t bu

ildin

g.

17

Jane

’s

apar

tmen

t is

---

the

seco

nd

GN

UI

MI

Tot

al

L2s

EA

EB

GI

GN

u1 M

I T

otal

L2

s

EA

EB

GI

GN

UI

MI

Tot

al

L2s

EA

EB

GI

16

50

37

181 87

33

45

15

47

36

176 87

33

45

16

50

37

181 87

33

44

93.8

78

.0

91.9

84.0

92.0

90

.9

95.6

40

.0

59.6

80

.5

77.3

98.9

97

.0

100.

0 fl

oor,

. . .

GN

16

75

.0

6.3

4.0

5.4

7.2

67.8

15

.1

81.8

28

.7

68.9

28

.9

68.7

12

.5

62.0

24

.0

78.4

21

.6

71.3

22

.1

18.0

2.

7

8.8

6.9

1.1

6.1

3.0

4.4

6.0

2.1

38.3

2.

8 16

.7

2.3

20.5

3.4

3 .O

2.2

18.8

4.

0 10

.0

1.7

5.0

1.1

3.0

25.0

3.71

13

.95

6.64

17.3

4

.53

4.56

50

.33

35.0

0 11

.90

20.2

2

6.19

.1

5 18

.22

9.15

2.

18

6.33

. 00

.oo

22.7

4

3 3 3 3 2 2 2 2 2 2 3 2 3 3 2 3 1 1

P ,2

95

.003

* ,0

84

.001

*

,765

,1

02

.om

* .o

m*

.003

*

.ooo*

.103

,9

27

.ooo*

.0

27*

,337

,096

1 .O

oo

1 .O

oo

L .oo

o*

18

. . . d

irec

tly -

--

the

vari

ety

stor

e.

19

Jim

nai

led

the

plaq

ue -

-- t

he

boar

d.

20

He

stoo

ped

to

avoi

d hi

tting

his

he

ad -

-- th

e la

mp.

u1

49

MI

37

Tot

al

L2s

179

EA

87

E

B

33

GI

45

GN

16

U

I 16

M

I 37

T

otal

L2

s 18

0

EA

87

EB

33

G

I 45

G

N

16

UI

48

MI

37

Tot

al

L2s

179

EA

87

E

B

33

GI

45

GN

16

U

I 49

M

I 37

T

otal

L2

s 18

0

100.

0 97

.3

96.6

37.9

33

.3

51.1

50

.0

83.3

70

.3

60.3

86.2

4.

6 87

.9

64.4

25

.0

30.6

54

.0

53.9

29.9

30

.3

40 .O

18

.7

6.2

10.8

21.2

33.3

24

.2

22.2

12

.5

14.3

24

.3

20.0

64.4

72

.7

71.1

68

.7

57.1

48

.6

62.8

28.7

36

.4

6.7

18.7

4.

2 18

.9

agai

nst

8.0

9.1

15.5

31

.2

18.4

27

.0

18.9

~

.6

1.1

3.0

4.4

10.2

10

.8

6.7

2.3

2.2 .6

1.1

.oo

1 1.

000

2.7

2.78

2

.249

2.8

1.2

2.2

18.7

18

.4

16.2

10.6

1.1

12.5

6.

2

2.8

2.73

1.78

3.

08

12.9

5 23

.02

18.2

1

14.9

0

1.73

9.

45

7.93

33

.00

11.8

5

15.2

2

2 3 3 3 3 3 3 4 4 4 4 4 4

,255

,620

,3

79

.005

* ,0

00'

.om

*

.002

*

.786

.0

51

,094

.0

00*

.019

*

.004

*

3.0

2.06

3

.559

20

.0

19.3

4 3

.OW

* 43

.7

45.1

6 3

.OOO

* 51

.0

60.5

1 3

.OOO*

18

.9

24.7

3 3

.OW

*

27.2

43

.57

3 .0

00*

(Con

tinue

d)

Tabl

e 4

Dis

trib

utio

n OJ R

espo

nses

in t

he S

CT

Con

text

s by

EA

and

the

L2

Lear

ners

, co

ntin

ued

on

top

SCT

C

onte

xt

LG

N

on

upon

of

on

to

over

ab

ove

from

O

AR

U

R

X

df

p

21

She

sew

ed th

e E

A

87

60.9

bu

tton

--- t

he

EB

33

60.6

sl

eeve

. G

I 44

59

.1

GN

16

37

.5

UI

50

76.0

M

I 36

63

.9

Tota

l L2

s 17

9 63

.1

22

She

jum

ped

---

the

hors

e's

back

.

EA

E

B

GI

GN

U

I M

I To

tal

L2s

1

32.2

27

.3

36.4

37

.5

4.0

25 .O

23.5

87

27.6

19

.5

39.1

33

36

.4

9.1

45.4

45

55

.6

24.4

16

43

.7

6.2

37.5

49

59

/.2

4.1

6.1

37

64.9

10

.8

5.4

180

53.9

5.

6 20

.6

23

The

dan

cer

EA

87

ra

ised

her

arm

E

B

33

high

--- h

er

GI

45

head

. G

N

16

UI

50

MI

37

Tot

al

L2s

181

35.6

64

.4

27.3

72

.7

31.1

68

.9

25.0

68

.7

50.0

46

.0

43.2

51

.3

37.6

59

.7

1

6.9

12.1

.9

7 2

,615

2.

3 2.

3 3.

31

3 ,3

46

12.5

12

.5

12.8

3 3

.005

* 8.

0 12

.0

23.1

0 3

.OOO*

8.

3 2.

8 2.

96

3 ,3

98

7.8

5.6

6.66

3

,083

8.0

5.7

3.0

6.1

3.33

4

,505

6.

7 13

.3

19.0

3 4

.001

* 12

.5

4.74

4

.315

10

.2

20.4

32

.18

4 .O

OO*

2.7

16.2

24

.98

4 .CO

O*

5.6

14.4

32

.26

4 .O

OO*

.42

1 ,5

15

.ll

1 .7

45

6.3

5.93

2

,052

4.

0 6.

94

2 ,0

31'

5.4

5.83

2

,054

2.8

2.68

2

,263

24

Whe

n yo

u st

and

---

a hi

ll, y

ou

can

see

in a

ll di

rect

ions

.

25

I put

the

dis

h --

- th

e ta

ble.

26

She

sat --

- a

pillo

w o

n th

e ch

air.

EA

E

B

GI

GN

U

l M

I To

tal

L2s EA

EB

G

I G

N

u1

M

I T

otal

L2

s

EA

EB

G

I G

N

UI

MI

Tot

al

L2s

87

33

45

16

50

38

182 87

33

45

16

50

38

182 87

33

45

15

50

38

181

41.4

12

.6

35.6

60

.6

12.1

21

.2

40.0

2.

2 53

.3

68.8

31

.2

58.0

4.

0 18

.0

52.6

7.

9 23

.7

53.8

5.

5 29

.7

85.0

6.

9 84

.8

75.6

2.

2 81

.2

6.2

88.0

8.

0 92

.1

5.3

84.6

4.

4

71.3

17

.2

72.7

15

.1

80.0

8.

9 86

.7

6.7

66.0

4.

0 71

.0

10.5

73.5

8.

8

4.6

6.1

2.2

2.0

7.9

3.9

9.2

1.1

6.1

4.13

4.

4 7.

19

5.97

4.

0 16

.0

19.5

6 2.

6 13

.2

11.9

9

3.8

7.1

13.4

4

5.7

2.3

6.1

9.1

4.93

2.

2 20

.0

13.5

1 12

.5

4.57

4.

0 3.

27

2.6

2.46

1.6

9.3

8.17

5.7

1.2

3.0

3.0

1.03

4.

4 4.

4 3.

69

6.7

4.86

6.

0 22

.0

20.9

6 10

.5

2.57

5.5

8.3

8.72

4 4 4 4 4 4 3 3 3 3 3 3 4 4 4 4 4 4

,388

,1

26

,201

.0

01*

.017

*

.009

*

,177

.004*

,206

,3

51

,483

.043

*

.905

,4

49

,302

.o

m*

,631

.069

(Con

tinu

ed)

Tabl

e 4

Dis

trib

utio

n of R

espo

nses

in

the

SCT

Con

text

s by

EA

and

the

L2

Lear

ners

, con

tinue

d

on

top

SCT

Con

text

L

G

N

on

upon

of

on

to

over

ab

ove

from

O

AR

U

R

X

df

p a

mem

ber

-

27

Are

you

---

the

EA

ball

team

this

EB

su

mm

er?

GI

GN

U

I M

I T

otal

L2

s

28

The

pro

fess

or

EA

w

as l

ectu

ring

EB

fa

r --

- ou

r G

I he

ads.

G

N

UI

MI

Tot

al

87

33

45

16

49

38

181 87

33

45

45

16

37

73.6

90

.9

80.0

18

.8

30.6

65

.8

60.2

36.8

63

.2

36.4

63

.6

33.3

64

.4

31.2

50

.0

28.0

48

.0

35.1

43

.2

of

4.6

3.0

- .6

18.4

3.

4 6.

1 11

.1

8.9

6.3

75.0

24

.5

44.9

2.

6 31

.6

11.6

27

.6

2.2

18.8

24

.0

21.6

4.62

4.

87

55.6

9 42

.07

24.2

4

26.1

0 . 00

2.04

16

.81

22.8

8 20

.64

3 .2

02

3 ,1

81

3 .0

00*

3 .oo

o*

3 .0

00*

3 .0

00*

I 1.

00

2 ,3

61

2 .0

00*

2 .0

00*

2 .0

00*

L2s

181

32.6

54

.1

13.3

12

.69

2 .0

02*

Lege

nd:

LG

Lan

guag

e gr

oup

OA

R

Oth

er a

ccep

tabl

e re

spon

ses

UR

U

nacc

epta

ble

resp

onse

s

Ijaz 433

Both contexts involved the componential combination 7 C-V-M/ 7 A. No difference in responses was found between EA and the ESL learners on SCTS (A mirror is- the wall) which involves the componential combi- nation 7 C-V-M/ T S. SCT5 differs from the typical meaning of on to a lesser degree than do SCT6 and 15 in that the difference resides solely in the absence of the component of VERTICALITY. SCT6 and 15 both have a component of ATTACHMENT. Moreover, in SCT6 and 15 con- tact does not occur with a flat surface as in SCT5 and in most contexts with a typical or central meaning of on. It is noteworthy that, of the contexts with a noncentral meaning of on, SCTS elicited the highest frequency of responses with on from EA (96.4 percent).

Of the two contexts with a nonvertical, motional meaning ( 7 C-V 7 M) of on (SCT20 and 21), only SCT20 (He stooped lo avoid hitting his head- the lamp) elicited a significant difference in responses between EA and the combined L2 learners. The concepts involved in the two contexts differed from each other in that SCT 20 had a component of COLLISION and SCT21 (She sewed the button-the sleeve) one of ATTACHMENT.

Of the contexts with spatial meaning features of over, only one, SCT 18 (The apartment is- the second floor, directly - the variety store), elicited a significant difference in responses between EA and the com- bined ESL learners. It involved a static, noncentral meaning of over. In the other SCT context with a static, noncontiguous meaning SCT4, (There is a picture-the sofa), a trend toward a significant difference (p0.076) in responses between EA and the combined ESL learners was obtained.

Both contexts with a figurative meaning (SCT27 and 28) elicited a significant difference in responses between EA and the combined L2 learners.

Our findings support Hypothesis 2 that ESL learners would approxi- mate native speakers more closely in the meaning they ascribe to typical or central instances of semantic categories than in the meaning ascribed to noncentral ones.

TRANSFER OF THE WEIGHT OF THE SEMANTIC DIMENSIONS OF L1 WORDS TO THE CORRESPONDING L2 WORDS

The meaning ascribed to the target wordslo by native German and Urdu speakers was, to a considerable degree, influenced by the meaning

434 Language Learning Vol. 36, No. 4

of corresponding lexical structures in their mother tongue. Both German speaking groups differed11 substantially from EA in the weighting of the most salient semantic dimension of on and over (figures 1 and 2). In the meaning ascribed to on, both German speaking groups underemphasized the component of + CONTACT and overemphasized + MOVEMENT. GI overemphasized + MOVEMENT to a greater degree than did GN. GI also overemphasized + VERTICALITY. The overemphasis of the component of +VERTICALITY in the meaning of on by GI was reflected in the semantic-relatedness test by the close relationship subjects in that group attributed to on and on top of. Of the L2 learner groups, GI was the only one that rated on and on top of as significantly more closely related in meaning than the EAs. On the SCT, GI overused on top of as compared to on on SCTl ( T C 7 V-M/ T S/ 7 E/ T EXTR), SCT24 ( T C T V-M/ 7 e/ 7 EXTR), SCT7 ( T C T V-M/ T S/ T E), and SCT9 ( T C 7 V 7 M/ T D/ T S/ T E). In the meaning ascribed to over, GI and GN both underemphasized +MOVEMENT, and GI strongly underemphasized + VERTICALITY and + CONTACT. The two German speaking groups differed to an approximately equal degree from EA in the weighting of the most salient semantic dimension of on and over, but GI showed stronger differences than did GN in the weighting of the words' less salient semantic dimensions. Generally speaking, subjects in GN revealed lower English proficiency in their linguistic usage than did those in GI.

The underemphasis of the component of +CONTACT and the overemphasis of that of + MOVEMENT in the meaning attributed to on by native German speakers appears to have been related to the fact that auf, the close German translation equivalent to on, can denote both contiguous and noncontiguous meanings. In its noncontiguous meaning, auf corresponds to motional meanings of English up.

The close relationship attributed to on and on top of by GI reflects transfer of the meaning of auf which has a stronger semantic component of +VERTICALITY than does on (Ijaz 1985). Indeed, the most typical meaning of aufmore closely approximates that of on top of than that of on. The overuse of on top of in certain contexts in which on and on top of were acceptable as responses seems to have been prompted by the assumption that on did not adequately express the weight of the component of + VERTICALITY involved. That is, the erroneous meaning attributed to on by GI became indirectly reflected in the overuse of on top of.

Ijaz 435

The overemphasis of the component of -VERTICALITY and the resultant underemphasis of -VERTICALITY in the meaning ascribed to on by native German speakers also became reflected in linguistic usage in the difficulty they experienced with contexts with nonvertical meanings of contiguity. This difficulty was also related to native-language constraints. Auf, the close German translation equivalent to on, cannot express nonvertical, spatial meanings, but in German such meanings are linguistically categorized differently than in English and expressed by a separate term, an. Significantly, a common erroneous usage by native German speakers in contexts with nonvertical meanings was at, the close translation equivalent of German an. The use of this term reflects native- language transfer.

The predominantly static meaning ascribed to over by native German speakers derived from transfer of the conceptual structures underlying uber, a German term that is inclusive of the meanings of over and above, but unlike over has a strongly static meaning (Ijaz 1985).

Like the native German speakers, the native Urdu speakers differed substantially from EA in the weighting of the most defining semantic dimension of on and over. In the meaning attributed to on UI underemphasized the component of +CONTACT and in that of over that of +MOVEMENT. On the semantic-relatedness test UI rated on and over as significantly more close’ly related to each other than did EA. On the SCT an error characteristic of UI involved the substitution of on for over (and above) and, in some instances, that of over for on.

As in the case of native German speakers, the erroneous meaning ascribed to on and over by native Urdu speakers seems to have been influenced by transfer of the meaning of the corresponding words in their mother tongue to the L2 words. In Urdu the distinction between +CONTACT and -CONTACT is not made linguistically salient by separate words. /Per/ and /ooper/, the closest Urdu translation equivalents to English on and over, largely overlap in meaning. The terms essentially differ from each other in that /ooper/ implies the existence of an elevated surface or level in space, whereas elevation is irrelevant to the meaning of /per/. As a result of native language constraints, native Urdu speakers appear to have found it difficult to constraints, native Urdu speakers appear to have found it difficult to distinguish linguistically between contiguous and noncontiguous meanings in English.

It becomes apparent that in the case of native German and Urdu

436 Language Learning Vol. 36, No. 4

speakers the semantic boundaries ascribed to on and over were influenced by the weight of the semantic dimensions of the corresponding words in their mother tongue. Our findings clearly support Hypothesis 3 .

ERRORS AS A REFLECTION OF THE LINGUISTIC CLASSIFICATION OF SEMANTIC CONCEPTS IN THE L1

The L2 learners in the different groups shared many erroneous lexical usages on the SCT, but some of the words used were more characteristic of subjects with a particular native language than of others. It was shown above that in contexts with nonvertical meanings of contiguity, native German speakers typically produced at as an erroneous response, a word whose usage was prompted by native language transfer. The contexts involved were:

SCT 6 I5 20 21

The keys are hanging---the hooks. Dogs must be kept---a leash. He stooped to avoid hitting his head---the lamp. She sewed the button---the sleeve.

In the same contexts, native Urdu speakers frequently produced with. On SCTl5 in and with were the most frequently used deviant terms. The nature of the erroneous terms typically produced by native German and Urdu speakers in those contexts reflects the classification of the concepts involved in the respondents’ native language.

Differences in the linguistic classification of a concept within a lan- guage or across languages derive from a different weighting of the semantic dimensions relevant to the meaning. For example, each of the four contexts above expresses a meaning of nonvertical contact between two entities. Apart from the components of + CONTACT and -VERTI- CALITY, each involves other semantic components as well. In German the semantic distinction between vertical and nonvertical contact is emphasized and made linguistically salient by a different categorization of the two types of meaning. Neither English nor Urdu shares this dis- tinction with German and, depending on the context, each language emphasizes different semantic components. The components emphasized become linguistically reflected in the alternative acceptable response terms in the different contexts.

In English either on or from is acceptable in SCT6, and in Urdu either

rjaz 43 7

/per/ (on) or /say/ (with). That is, English either emphasizes the seman- tic component of CONTACT (on) or the source from which the keys are suspended Vrorn). Urdu either emphasizes CONTACT (/per/) or ATTACHMENT ( /say/) .

In SCT21 on, onto, and to are the MARS in English and /per/ and /say/ in Urdu. English on emphasizes the component of CONTACT and onto and to that of ATTACHMENT. Similarly, Urdu /per/ emphasizes CONTACT and /say/ ATTACHMENT.

In SCTlS the only MAR in English is on, but in Urdu either /say/ (with) or /meh/ (in) is acceptable. While English emphasizes the seman- tic component of CONTACT, Urdu stresses either that of ATTACH- MENT (/say/) or the concept of inclusion based on the dog’s collar ring (/rneh 1).

Finally, in SCT20 on, upon and against are the MARS in English, but in Urdu /say/ (with) is the only acceptable response. In this context, English emphasizes either the semantic component of CONTACT (on, upon) or COLLISION (against), Urdu emphasizes COLLISION (/say/).

The language-specific erroneous responses by native German and Urdu speakers in the above contexts show that the classification of the concepts in the subjects’ native language is transferred to the L2. This finding also supports Hypothesis 3 . Evidently, native language concep- tual patterns strongly influence the meaning ascribed to words in a sec- ond language.

LINGUISTIC STRATEGIES BY THE L2 LEARNERS

The differences in response patterns between native English speakers and ESL learners in the SCT contexts derived from a different distribution of alternative acceptable responses, the frequency of unacceptable responses, or both. Differences in the distribution of MARS were related to the frequency of errors, but not solely caused by them. Different linguistic strategies were noted in the lexical usage of the ESL learners in contexts with different meanings. The strategy employed in a given context appeared to be influenced by (a) the degree of resemblance of the given meaning to the typical meaning of on and over, respectively and (b) by a semantic equivalence hypothesis which assumes that conceptual patterns and linguistic/semantic coding practices in the L1 provide the essential criteria for those in the L2.

438 Language Learning Vol. 36, No. 4

Over- and Underuse of ON and OVER

In many contexts involving noncentral meaning features of on in which several MARs could be used, on was overused by the L2 learners and alternative appropriate responses underused. Each of the contexts in which on was significantly overused and alternative MARs underused by the combined ESL learners (SCT8, 14, 19 and 22) involved a motional, contiguous meaning with resultant contact from a vertical direction ( T C T V T M ) . On was associated more readily with each of these contexts than more specific terms such as upon and onto whose use was apparently avoided.12

Neither the German nor the Urdu language has a close translation equivalent to upon and onto. That is, these terms represent semantic distinctions specific to the English language. In each of the four contexts involved, German would use auf and Urdu /per/. In either case, the closest English translation equivalent to the word used in the subjects’ native language would be on. For the L2 learners the use of on in these contexts solely involved a transfer of meaning from the word appropriate in their native language to its English translation equivalent. By contrast, the use of upon or onto would have entailed delimiting the semantic boundaries of a word that did not have a close translation equivalent in their mother tongue, a linguistic strategy that might have been more difficult and possibly resulted in an error. The use of on in these contexts as compared to that of words of more specific meaning appears to have been the result of a simplification strategy.

The L2 learners consistently underused on in the following contexts:

SCT

I5 20 27

7 The books are---the shelf Dogs musf be kept---a leash. He stooped lo avoid hitting his head---the lamp Are you---the ball team this summer?

Three of these contexts involved noncentral meanings of on (SCT15:

TFIG), but one involved a meaning relatively typical of on (SCT7: T C 7 V-M/ T S/ 7 E). Significantly, in each of the four contexts, a high proportion of subjects in EA used on.13 The combined ESL learners produced a large number of errors (SCT7: 9.9 percent; SCTl5: 20.5 percent; SCT20: 27.2 percent; SCT27: 27.7 percent). On SCT7 and 27,

7 C-V-M/ T A; SCT20: T C-V 7 M/ 7 COL; SCT27: T C-V-M/

rjaz 439

89 and 92 percent, respectively, of all the erroneous responses produced by the second-language learners involved the use of in. In these contexts as well as in the other two, the vast majority of the erroneous responses by native German and Urdu speakers was the result of native-language transfer. On SCT27 in and /meh/ (in) would be appropriate in German and Urdu, respectively. In SCT20 an (at) is appropriate in German and /say/ (with) in Urdu. In SCT15 an would be the term used in German, and /say/ or /meh/ in Urdu. Finally, in SCT7, either auf or /per/ (on) would be appropriate in German and Urdu, respectively, or in or /meh/ (in). The relational meaning involved in each of the four contexts is either categorized differently or can be categorized differently linguisti- cally in languages other than English. That is, the linguistic categoriza- tion of the relevant meanings is highly language specific to English and considerable conceptual restructuring was required on the part of the language learners to acquire the respective meaning features of on.

It is noteworthy that in SCT20 against, an MAR used by 8 percent of EA subjects was significantly overused by the combined ESL learners. Of all the contexts in which on was an MAR, SCT20 was the only one in which a preposition other than on was significantly overused by the L2 learners. The overuse of against, like the erroneous use of at and with by native German and Urdu speakers, respectively, reflects the attempt by the subjects to emphasize the component of COLLISION relevant to the meaning of the sentence which, in languages other than English, appears to be made linguistically more salient.

The difficulty the second-language learners experienced with nontypi- cal meaning features of the terms also reflected in the use of over. Over was underused by all L2 learner groups in SCT18 (The apartment is-the second floor, directly-the variety store). The difference from EA was significant for UI as well as for the combined ESL learners. The context involved a static, noncontiguous meaning (-C 7 V-M): a meaning that was nontypical of over.

Failure to Transfer Noncentral Meaning Features of Translation Equiva- lents from the L1 to the L2

In some instances, nontypical meaning features of close translation equivalents to on in the subjects’ native language were not transferred to the L2, although such transfer would have resulted in the use of an appropriate response term in English. Such a linguistic strategy was par-

440 Language Learning Vol. 36, No. 4

ticularly characteristic of subjects in GI. In SCT25 (Zput the dish-the table), the only acceptable response in German would be auf, the close translation equivalent of English on. Transfer of the respective meaning feature ( 7 C T V 7 M/ T D/ T S/ 7 E) of auf to on would have resulted in an appropriate response. Instead, 20 percent of GI subjects produced an erroneous response; 17.8 percent used onto. The MARS were on and upon.

The use of onto may have been partly prompted by the subjects’ inabil- ity to define the semantic boundaries between upon and onto, closely related terms that overlap in several of their meaning features. But the high percentage of subjects with a German background who produced this erroneous usage clearly suggests native language interference. It is possible that because a noncentral meaning feature of German auf was involved, subjects were reluctant to transfer this meaning to its English translation equivalent. The use of onto may have derived from the sub- jects’ attempts to specify the component of DIRECTION relevant to the context. Subjects seem to have transferred the meaning of the German prepositional combination a u f . . . zu which expresses a meaning of in the direction of. The English translation equivalent is on . . . to. While the use of a u f . . . zu in the given context would be as inappropriate in German as onto is in English, it represents the only prepositional struc- ture in German that is both related to auf and has a strong semantic component of DIRECTION. Since neither English upon nor onto has a close translation equivaIent in German, the subjects apparently opted for the one of the two terms that had at least an approximate equivalent in their native language. It is noteworthy that GN, UI, and MI, whose overall English proficiency was somewhat lower than GI and who pro- duced substantially more erroneous responses, tended to use on in SCT25, thereby avoiding an incorrect response.

LANGUAGE SPECIFIC CLASSIFICATION AND CONCEPTUAL COMPLEXITY OF NONCENTRAL MEANINGS OF SEMANTIC CATEGORIES AND UNIVERSALITY OF THE CENTRAL MEANING OF ON

Overall, the ESL learners more closely approximated native English speakers in the meaning they ascribed to typical or central meaning features of on and over than in that of noncentral ones. On was either over- or underused by the ESL learners in contexts with noncentral meaning features of the word and over was underused in a context

Ijaz 441

involving one of its noncentral meaning features. Noncentral meaning features of on with a shared componential combination of TC 7 V 7 M were shown to be linguistically classified in a manner language specific to English. Other noncentral meaning features of on proved to be categorized differently linguistically across languages. In the case of a different linguistic categorization of meanings in the learners’ L1 than in the L2, strong effects of native-language transfer were observed in expressing such meanings in the L2. In the case of a similar linguistic categorization of noncentral meanings in the L1 as in the L2, but a greater differentiation of those meanings in the L2, either L2 words that specified noncentral meaning features were underused by the L2 learners or the respective meaning feature of the word in the L1 failed to be transferred to its L2 translation equivalent by ESL learners of high L2 proficiency. The difficulty experienced by the L2 learners with noncentral meaning features of on and over was apparently related to the language specific classification of noncentral meanings and, in the case of high proficiency L2 learners, their perceived language specificity to the L1.

To judge from the frequency of errors produced, of the SCT contexts SCT15, 20 and 27 presented the greatest difficulty for the ESL learners. The contexts not only involved noncentral meaning features of on, but the meanings were also linguistically categorized differently in English than in the subjects’ native language. Significantly, each of the three contexts involved a nonvertical meaning of on. It appears that the linguistic categorization of nonvertical, contiguous meanings with the same semantic category that expresses vertical, contiguous meanings is highly language specific to (idiomatic in) English, although some nonvertical meaning features may be categorized in a similar fashion in other languages: (T C-V-M in Urdu). It is possible that if meanings involve a different linguistic categorization in the L2 than in the learner’s L1, the labeling of such meanings in the L2 involves considerable conceptual restructuring and, for this reason, learning L2 words that express such meanings may be particularly difficult for language learners. Meanings that were categorized in a similar fashion in the L1 and in the L2 but differentiated more in the L2 also proved to be difficult for the ESL learners, but to a somewhat lesser degree than those that were categorized differently. In such instances, the use of the inclusive on, which had a close translation equivalent in the subjects’ L1,

442 Language Learning Vol. 36, No. 4

represented an acceptable alternative to the use of L2 words with language specific semantic boundaries.

In only two of the 28 contexts (SCT3: T C T V - M / T S and SCT13: -C 7 V T M/ 7 HBT) do the English, German and Urdu languages use single close translation equivalents (Table 5). The two contexts involve typical meanings of on and over, respectively. In the linguistic classification of the relational meanings involved in the other SCT contexts, the three languages differ from each other to varying degrees.

Contexts with typical meanings of on and over elicited the highest agreement in responses among native English speakers and second- language learners with different native languages with that of on to a slightly higher degree than that of over. In contexts with certain noncentral contiguous meanings, on was used as a substitution for more specific terms such as upon and onto by the language learners. In contexts with noncontiguous meanings, on was used as an erroneous substitution for over (and above) by second-language learners whose native language did not make a distinction between contiguous and noncontiguous meanings. Of all the six words investigated, on was the only one that had a close translation equivalent across the English, German, and Urdu languages. All these findings suggest that the term on incorporates a meaning that is conceptually more basic than other meanings in the semantic domain investigated, invariant crosslingually and universal. The central meaning of on appears to represent what Rosch et al. (1976) have referred to as the basic level concept of the semantic category investigated. Because of its universality, it was transferred more readily from the learners’ native language to the target language than noncentral meanings and native-language constraints on its transfer were minimal.

DISCUSSION

The meaning ascribed to English words by advanced ESL learners was shown to be influenced by linguistic and cognitive factors. Linguistic factors were related to the differential weighting of semantic dimensions in corresponding meanings in the L1 and the L2 and related differences in the linguistic classification of meanings. Cognitive factors were found to be related to the structure of conceptual categories, the greater varia- bility of noncentral members than of central ones, and the effects of

Zjaz 443

native-language transfer. Both factors were found to be closely related in that noncentral members of semantic categories were classified differ- ently across languages, whereas typical members were not. Linguistic factors differed with language learners of different native languages. Cognitive factors accounted for the degree of inherent conceptual com- plexity of meanings. Cognitive factors became defined by the dimensions relevant to a meaning, and linguistic factors by their weighting in the learners’ native language and the target language.

In the meaning ascribed to on and over as well as in the linguistic strategies employed in expressing relational meanings, the second- language learners essentially relied on a semantic equivalence hypothesis. This hypothesis facilitates the acquisition of lexical meanings in the L2 in that it reduces it to the relabeling of concepts already learned in the L1. It confounds and complicates vocabulary acquisition in the L2 by ignoring crosslingual differences in conceptual classification and differences in the semantic boundaries of seemingly corresponding words in the L1 and the L2.

The weight of the semantic dimensions of words in the native language was transferred to the corresponding words in the L2. In lexical usage, the L2 learners consistently favored lexical/semantic structures that had close equivalents in their native language. In instances in which the L2 made finer semantic distinctions in a semantic field than did the L1, many L2 learners avoided the use of L2 words expressing distinctions language specific to the L2 in favor of a word with a more generalized meaning which had a near-equivalent in the L1. This linguistic strategy became most apparent in the extensive overuse of on and underuse of upon and onto in contexts in which both on and related terms were appropriate as responses. This finding supports the observations by a number of other researchers (Mackey and Savard 1967; Kellerman 1977; Levenston and Blum 1977; Blum and Levenston 1978a, b; Harley and Swain 1978; Harley 1982) that L2 learners appear to favor the use of terms of wide coverage. In instances in which the L1 made a distinction that was not made in the L2, the L2 learners sought to make that distinc- tion linguistically salient in the L2 as well. The large number of errone- ous responses with at by native German speakers in contexts with non- vertical, contiguous meanings provided a characteristic example. In instances in which the L1 and the L2 made different linguistic distinc- tions between concepts within a semantic domain, the L2 learners experi- enced considerable difficulty in delimiting the semantic boundaries

Tabl

e 5

SCT

Con

text

s, T

heir

Mea

ning

s an

d A

ccep

tabl

e R

espo

nse

Term

s in

the

Eng

lish,

Ger

man

and

Urd

u La

ngua

ges*

RES

PON

SE T

ER

MS

Com

pone

ntia

l SC

T

Con

text

C

ombi

natio

n En

glis

h G

erm

an

Urd

u 3 2 7 26

12 I 24

17 5 6 15

14

25 8 9

Ther

e is

a ba

sket

---

the

floo

r.

Two

wat

ches

are

---

the

tabl

e.

The

boo

ks a

re ---

the

shel

f.

She

sat --

- a

pillo

w o

n th

e ch

air.

The

chi

ldre

n w

ere

sitti

ng -

-- th

e se

e-sa

w.

The

cat

is -

-- t

he r

efri

gera

tor.

Whe

n yo

u st

and

--- t

he h

ill, y

ou

can

see

in a

ll di

rect

ions

.

Jane

’s a

part

men

t is

---

the

sec

ond

floo

r.

The

pic

ture

is -

-- th

e w

all.

The

key

s ar

e ha

ngin

g --

- th

e ho

oks.

Dog

s m

ust

be k

ept --

- th

e le

ash.

He

plac

ed h

is h

and

---

her

shou

lder

.

I put

the

dis

h --

- th

e ta

ble.

Mar

y cl

imbs

---

the

foo

tsto

ol to

re

ach

the

shel

f.

Som

e w

orke

rs c

limbe

d --

- th

e ro

of.

TC

TV

-M/?

S

7c

Tv-M

/ 7

% 7~

7

c Tv

-M/ 7 S/

7~

7 c 7

V-M

I 7 S/

TE

+ c

7 V-M

I 7

E/ 7

AST

R

7 c 7

V-M

I 7

S/ 7

E/

TEX

TR

7 c 7

V-M

/ 7 E/

TEX

TR

7 c 7 V-

M/

7 FI

G

7 C

-V-M

/ 7 S

7 C-

V-M

/ 7

A

7 C-V

-MI

?:A

Tc 7

v 7

~/

TD

/ 7~

7c

TvTM

/TD

/ 7s

/ 7~

7c

7~

7~

1

~D

/?

s/

~E

7c

7 v T

M/ 7

D/ 7

s/ T

E

on

on

on

on, u

pon,

with

on, u

pon,

ast

ride

on, o

n to

p of

, upo

n

on, o

n to

p of

, upo

n

on

on

on, f

rom

on

on, u

pon

on,

upon

onto

, on,

upo

n, o

n to

p of

, up

on

onto

, on,

upo

n,

up o

n, u

p to

, on

top

of

auf

auf

auf,

in

auf

auf

auf

auf

auf,

in

an

an

an

auf

auf

auf

auf

/per

/

/per

/, /o

oper

/

/per

/, /m

eh

/per

/, /o

oper

/

/per

/

/per

/, /o

oper

/

/per

/, /

oope

r/

/per

/

/per

/

/per

/, /s

ay/

/say

/, /m

eh/

/per

/

/per

/, /o

oper

/

/per

/, /o

oper

/

/per

/, /

oope

r/

22

19

21

20

18 4 16

13

10

I1

23

21

28

She

jum

ped

--- t

he h

orse

’s b

ack.

Jim

nai

led

the

plaq

ue ---

the

boar

d.

She

sew

ed th

e bu

tton

---

the

slee

ve.

He

stoo

ped

to a

void

hitt

ing

his

head

--- th

e la

mp.

Jane

’s a

part

men

t . .

. is

dire

ctly

---

the

vari

ety

stor

e.

The

pic

ture

is -

-- th

e so

fa.

The

hel

icop

ter

is f

lyin

g --

- th

e ap

artm

ent

build

ing.

The

dog

jum

ped

---

the

fenc

e.

The

kite

ros

e hi

gher

and

hi

gher

, --- th

e ho

uses

. . .

an

d --

- th

e tr

ees

The

dan

cer

rais

ed h

er a

rm

high

---

her

hea

d.

Are

you

---

the

bal

l te

am t

his

sum

mer

?

The

pro

fess

or w

as l

ectu

ring

far

---

our

head

s.

TC

T v

TM/

TD/ T

E

TC

TV

TM

/Ta

Tc-v

T

MI

TA

7

C-v

7 MI

7 CO

L

-C T

V-M

-C T

V-M

/ T

POST

-CT

VT

M

-C T

v~

M/ 7

HB

T

-C 7

v 7

MI 7 VB

T

-C 7

v TM

/ ~V

BT

-C 7 v 7

MI 7 VB

T

7 C

-V-M

/ 7

FIG

-C T

vTh.

11 T

FIG

onto

, on

, upo

n

on, o

nto,

to

on, o

nto,

to

on, a

gain

st, u

pon

abov

e, o

ver

abov

e, o

ver

over

, ab

ove

over

over

, abo

ve

over

, abo

ve

abov

e, o

ver

on

abov

e, o

ver

*Clo

sest

Eng

lish

Tra

nsla

tion

Equ

ival

ents

of

Ger

man

and

Urd

u R

espo

nse

Ter

ms

Ger

man

E

nglis

h U

rdu

Eng

lish

auf

on

/per

/ on

an

at

/o

oper

/ ov

er

in

in

/meh

/ in

~- -

~

~

/say

/ w

ith

auf

auf,

an

an

an

uber

uber

uber

uber

uber

uber

uber

in

uber

hinw

eg

...

/per

/

/per

/

/per

/, l

oope

r/

/say

/

/oop

er/

/oop

er/

/oop

er/

/oop

er/

/oop

er/

/oop

er/

/oop

er/

/meh

/

- /o

oper

/

446 Language Learning Vol. 36, No. 4

between the terms expressing those concepts. The difficulty of native Urdu speakers in delimiting the semantic boundaries between on and over serves as an example.

The second-language learners experienced the greatest difficulty with idioms: concepts that were categorized in a highly language-specific fash- ion in the L2. The numerous errors attributable to native-language trans- fer in contexts involving such meanings suggest that in such instances native-language conceptual constraints are particularly rigid. It is possi- ble that the conceptual restructuring required in such instances is closely akin to the acquisition of an entirely new concept. The finding by Zobl (1980), in the area of grammatical structures, that L2 learners find it more difficult to learn new concepts than to restructure existing ones also appears to hold true for lexical acquisition.

Among the meanings that presented considerable difficulty for the L2 learners were figurative meanings. This finding is in keeping with that of Kellerman (1979). Figurative meanings constitute a category of idioms and, like them, represent meanings that are often linguistically catego- rized in a language specific fashion. In any language, they also represent highly noncentral meaning features of the semantic category labeled by the word by which they are expressed in that they involve a nonconcrete level of abstraction.

In one instance, the semantic equivalence hypothesis failed to be applied, apparently as a result of interfering cognitive factors. This instance involved a nontypical meaning feature of a word in the subjects’ native language which had a direct translation equivalent in the target language. Instead of transferring the meaning feature of the L1 word to the corresponding L2 word, the L2 learners produced an erroneous term.

A similar strategy was observed by Kellerman (1978, 1980). Kellerman (1978) noted that of the polysemous senses of Dutch breken, central ones were transferred more readily to an L2 than noncentral ones. Kellerman (1978) has argued that this strategy is related to the perceived nontrans- ferability of noncentral meanings of semantic categories by L2 learners.14 Such an explanation makes sense in view of our finding that noncentral meanings of semantic categories are classified in a language specific fashion; however, the perceived nontransferability of noncentral mean- ing features of polysemous words in the L1 contrasts with the effects of native-language transfer observed in many erroneous L2 usages by the second-language learners in trying to express noncentral meanings of on and over in the L2 that are linguistically categorized differently in the L1

Ijaz 44 7

than in the L2. It appears that in instances in which a noncentral meaning of a semantic category is made linguistically salient in the L1, transfer of that meaning readily and often inappropriately occurs. But in contexts in which a noncentral meaning is not made linguistically salient in the L1, the linguistic label for the concept is perceived to be language specific to the L1. That is, the language specific status of noncentral instances of semantic categories is emphasized either by linguistic or cognitive means.

The second-language learners consistently applied the semantic equiv- alence hypothesis with correct linguistic results in contexts with typical meanings of on and over. Such meanings proved to be the only ones that were linguistically classified by single translation equivalents. In contexts involving such meanings, the strategy of the semantic equivalence hypothesis provided an excellent match for the crosslingual linguistic condition.

The combined second-language learners differed significantly from adult native English speakers in their similarity rating of 80 percent of the word pairs on the semantic-relatedness test. The differences in the semantic boundaries ascribed to the six terms between EA and the L2 learners were substantial, although the extent of the differences did not become apparent from the actual erroneous usages produced by the L2 learners on the SCT. Most second-language learners did not attribute an entirely inappropriate meaning to the words. The many correct uses of on and over, in particular, show that the essence of their meaning had been acquired. But some of the words’ polysemous meanings, particu- larly of on, and subtle semantic differences between related terms had failed to be acquired. That is, the essence of the words’ meanings had been mastered, but their semantic boundaries had not. Despite the fact that all L2 learners who participated in our study were very advanced ESL learners, the meaning ascribed to the words by them remained strongly shaped by native language conceptual patterns. This finding is in keeping with similar ones by d’Anglejan and Tucker (1973) and Strick (1980).

All these findings suggest that the linguistic usage of advanced second- language learners is influenced by their mother tongue in a much more subtle fashion than the frequency of erroneous usages clearly traceable to their native language might suggest. Native language conceptual patterns appear to be powerful determinants of the meaning ascribed to L2 words, and they seem to be very rigid and difficult to permeate.

448 Language Learning VoI. 36, No. 4

CONCLUSION

The meaning words have for second-language learners was shown to

1. contextual constraints and the number, weight and perspicuity of

2. differences in the weighting of those dimensions in the learner’s L1

3 . the effects of native-language transfer, and 4. cognitive factors related to the structure of semantic categories. Word usage in a second language was shown to be strongly influenced

by a semantic equivalence hypothesis which presumes that conceptual patterns and linguistic coding practices in the L1 provide the essential criteria for those in the L2; however, in one instance the use of this strategy was found to be constrained by cognitive factors.

The findings from our study suggest that the principle of prototypical- ity forms the basis of the structure of semantic categories in all lan- guages, as well as influencing the degree of difficulty involved in acquir- ing word meanings. Our finding that noncentral semantic concepts constitute language specific combinations of differentially weighted semantic dimensions may, in part, account for the different linguistic classification of semantic concepts across languages,

be influenced by a variety of factors.

the relevant semantic dimensions,

and the L2,

NOTES

‘A word’s polysemous senses will henceforth be referred to as “meaning features”. 2A specifically weighted semantic dimension within a given word meaning will hence-

forth be referred to as a semantic component. 3A detailed description of the possible semantic relationships that may exist between

corresponding words across two languages may be found in Ijaz (1984). 4The term “semantic category” is used to refer to a set of related word meanings, either

the different meaning features of lexical items or the meanings of related words in a word field.

5Most acceptable response terms other than MARS were produced by only one subject in EA, in a few instances by two. On only one item was an identical acceptable response other than an MAR given by three respondents in EA; however, on six items four respondents in EA used an identical acceptable response term. The use of an acceptable response term by four subjects in EA as a criterion for an MAR also seemed to be meaningful in terms of the data obtained from the second-language learners. On four of the six items on which an identical acceptable response term was given by four subjects in EA the response term in

Ijaz 449

question was significantly over- or underused by one of the language learner groups or by the combined ESL learners.

6The configurations obtained for the terms upon, onto, on top of and above may be found in Ijaz (1985).

’In the scaling procedure of the semantic relations of each term the notation of the dimensions was determined by their degree of salience, with dimension I being the most salient and dimension 111 the least salient. That is, the nature of the dimensions in the subject weight space obtained for the different terms is not necessarily identical. The labeling of the dimensions was based on three criteria: (1) the nature of the most salient semantic dimension of each word as became apparent in our pilot study; (2) the nature of the word pairs whose relationship was rated significantly differently by the L2 learner groups than by EA and (3) the nature of group specific erroneous lexical usages on the SCT involving the different words.

*The nature of these differences is discussed in greater detail in Linguistic Sfrategies by the L2 Learners below.

9Because of the small sample of GN (n = 16), the percentages of that group have to be treated with caution.

IOBecause of space limitations, the present analysis of the results will focus primarily on the terms on and over. A more detailed description of the results comprising all the six terms may be found in Ijaz (1985).

“The “differences” discussed with respect to the multidimensional scaling results are not statistically significant because inferential statistics were not applied and no technique for making statistical comparisons is currently known.

l2With the exception of UI in SCTl (7 C 7 V-M/ 7 S/ 7 E/ 7 EXTR), on top of was not significantly underused by any L2 learner group nor by the combined ESL learners in any SCT context.

131n SCT27 only 73.6 percent of native English speakers used on and a large variety of other responses were produced, many of which were neither prepositions nor related to each other: “a member of,” “joining,” “coaching,” “the captain of.” Their use was appar- ently prompted by a strong degree of ambiguity inherent in the context and the lack of sufficient semantic constraints to exclude all but a few closely related relational terms as a response. 14Kellerman’s argument of a differential transferability of meanings from the LI to the L2 was further supported by his (1980) finding that the transferability of nontypical meaning features of the Dutch word “oog” (eye) was related to their degree of similarity to the word’s primary sense as organ of sight on one hand, and their subjective frequency in the world of discourse on the other.

REFERENCES

Abberton, Evelyn. 1968. Some persistent English vocabulary problems for speakers of Serbo-Croatian. English Language Teaching Journal 22: 167-173.

Anglin, J. M. 1977. Word, Object and Conceptual Development. New York, New York: Norton.

Bates, E. and B. MacWhinney. 1981. Second-language acquisition from a functionalist perspective: pragmatic, semantic, and perceptual strategies. In Native Language and Foreign Language Acquisition, ed. H. Winitz, pp. 190-214. New York, New York: The New Academy of Sciences.

450 Language Learning Vol. 36, No. 4

Blum, S. and E. A. Levenston. 1978a. Universals of lexical simplification. Language Learning 28:399-415.

Blum, S. and E. A. Levenston. 1978b. Lexical simplification in second-language acquisi- tion. Studies in Second Language Acquisition 2:43-64.

Coleman, L. and P. Kay. 1981. Prototype semantics: the English word lie. Journal of the Linguistic Society of America 57:26-44.

d’Anglejan, A. and G. R. Tucker. 1973. Communicating across cultures. Journal of Cross- cultural Psychology 4:121-130.

DuSkova, L. 1970. On sources of errors in foreign language learning. International Review of Applied Linguistics in Language Teaching 7: 1 1-3 1.

Hakuta. K. A. 1975. Becoming Bilingual at Age Five: The Story of Uguisu. Honors B.A. Thesis, Harvard University, Cambridge, Massachusetts.

Harley, B. 1982. Age-Related Differences in the Acquisition of the French Verb System by Anglophone Students in French Immersion Programs. Unpublished Ph.D. Dissertation, University of Toronto, Toronto, Ontario, Canada.

Harley, B. and M. Swain. 1978. Form and Function in a Second Language: A Close Look at the Verb System. Paper presented at the Fifth International Congress of Applied Linguistics. August, Montreal, Quebec, Canada.

Ijaz, I. Helene. 1984. The acquisition of lexical meanings in a second language. In Multi- culturalism in Canada, eds. R. J . Samuda, J . W. Berry and M. Laferriere, pp. 229-237. Toronto, Ontario, Canada: Allyn and Bacon.

Ijaz, I . Helene. 1985. Native Language and Cognitive Constraints on the Meaning Ascribed to Select Spatial Prepositions by Advanced Adult Second-Language Learners of English. Unpublished Ph.D. Dissertation, University of Toronto, Toronto, Ontario, Canada.

Kellerman, E. 1978. Giving learners a break: native language intuitions as a source of predictions about transferability, Working Papers on Bilingualism 15:60-92.

Kellerman, E. 1979. The problem with difficulty. Interlanguage Studies Bulletin 4:27-48. Kellerman, E. 1980. Oeil pour oeil. Encrages Numero special: 54-63. Kellerman, E. 1983. Now you see it, now you don’t. In Language Transfer in Language

Learning, eds. Susan M. Gass and Larry Selinker, pp. 112-134. Rowley, Massachusetts: Newbury House Publishers.

Leech, G. 1974. Semantics. Harmondsworth, UK: Penguin Levenston, E. A. and S. Blum. 1977. Aspects of lexical simplification in the speech and

writing of advanced adult learners. In Actes du Cinquiime Coloque de Linguistique Appliquee de Neuchatel, ed. E. Roulet, pp. 51-71. Geneva, Switzerland: Droz.

Mackey, W. F. and J . G. Savard, 1967. The indices of coverage. International Review of Applied Linguistics in Language Teaching 7:71-121.

Mervis, C. B. 1980. Category structure and the development of categorization. In Theoreti- cal Issues in Reading Comprehension: Perspectives from Cognitive Psychology, Linguis- tics, Artificial Intelligence and Education, eds. R. S. Spiro, B. C. Bruce, and W. F. Brewer, Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Mervis, C. B., J . Catlin, and E. Rosch. 1975. Development of the structure of color categories. Developmental Psychology 1 1 :54-60.

Mougeon, R. M., M. M. Canale, and S. Carroll. 1979. Acquisition of English prepositions by monolingual and bilingual (FrenchIEnglish) Ontarian students. In Studies in First- and Second-Language Acquisition, eds. F. R. Eckman and A. J. Hastings, pp. 194-205. Rowley, Massachusetts: Newbury House.

Neimark, E. D. 1974. Natural language concepts: additional evidence. Child Development 45508-51 1 .

Ijaz 45 I

Oiler, John W., Jr. and N. Inal. 1971. A cloze test of English prepositions. TESOL Quarterly 5 : 3 15-326.

Rosch, E. H. 1973. On the internal structure of perceptual and semantic categories. In Cognitive Development and the Acquisition of Language, ed. T. E. Moore, pp. 11 1-144. New York, New York: Academic Press.

Rosch, E. H . and C. B. Mervis. 1975. Family resemblances: studies in the internal struc- ture categories. Cognitive Psychology 7:573-605.

Rosch, E. H. , C . B. Mervis, E. Gray, D. Johnson, and P. Boyes-Braem. 1976. Basic objects in natural categories. Cognitive Psychology 8:382-439.

Rosner, S. R. and D. S. Hayes. 1977. A developmental study of category item production. Child Development 48:1062-1065.

Saltz, E., E. Soller, and I. E. Siegel. 1972. The development of natural language concepts. Child Development 43:1191-1202.

Schachter, J . 1974. An error in error analysis. Language Learning 24:205-214. Strick, G. J . 1980. A hypothesis for semantic development in a second language. Language

Young, F. W. and R. Lewyskyj. 1981. ALSCAL 4 User’s Guide. Second ed. Chapel Hill,

Zobl, Helmut. 1980. The formal and developmental selectivity of L1 influence on L2

Learning 30: 155-176.

North Carolina: University of North Carolina.

acquisition. Language Learning 30:43-57.