2012. Vol.3, No.1, 49-56
Published Online January 2012 in SciRes (
Copyright © 2012 SciRes. 49
Linguistic Relativity Revisited: The Interaction between L1 and
L2 in Thinking, Learning, and Production
Hye K. Pae
School of Education, Univ ersity of Cincinnati, Cincin nati, Ohio, USA
Email: hye. pae
Received October 1st, 2011; revised November 5th, 2011; accepted December 9th, 2011
The linguistic relativity hypothesis (LRH; a.k.a., Whorfian hypothesis) is reconsidered with respect to
second language (L2) acquisition. With ebbs and flows over time, the notion of LRH went through dis-
missal and resurgence in linguistics, psychology, and anthropology. Empirical evidence gleaned from the
pseudo-linguistic domains, such as color categorization, time perception, spatial cognition, and number
recognition, supports the weak form of LRH. This article briefly reviews the conflicting views, discusses
empirical evidence, and expands the premise of LRH to L2 learning. Of interest is the interface of syntax
and semantics in English language learners’ (ELLs) ergative verb usage in which ELLs tend to overpas-
sivize English ergative verbs (e.g., appear, happen, break). The source of prevalent overpassivization er-
rors is discussed using the LRH framework.
Keywords: Linguistic Relativity Hypothesis; Language and Cognition; Second Language Acquisition
The relationship between cognition and language as well as
their reciprocal influence have been a long debated topic in psy-
chology and applied linguistics. The fundamental questions re-
lated to the nexus of cognition and language involve whether
we “think in language” and whether language shapes our thoug-
hts (Casasanto, 2008). These inquiries are related to linguistic
relativity in which linguistic differences yield differences in
speakers’ thoughts. The linguistic relativity principle1 (a.k.a., the
Whorfian Hypothesis, the Sapir-Whorf Hypothesis) posits that
language shapes the speaker’s thought and cognition (Gumperz
& Levinson, 1996; Lucy, 1996; Slobin, 20 03). According to this
hypothesis, different languages demonstrate a wide range of vari-
abilities in the speaker’s semantic categories and linguistic rep-
resentations. As a result, speakers of different languages percei-
ve the world differently and the conceptual system of the world
is constrained by the give n natural language (Gentner & Goldin-
Meadow, 2003).
Although it postulates a relationship between language and
cognition, linguistic relativity does not support unidirectiona-
lity or causality from cognition to language. In this regard, lin-
guistic relativity does not share commonality with linguistic de-
terminism which asserts that cognitive processes and thoughts
have a causal association with the structure of a language. A-
long this line, Casasanto (2008) claims that linguistic relativity
is different from linguistic determinism because linguistic rela-
tivity does not strictly determine the speaker’s thought.
Using the linguistic relativity hypothesis (LRH) as a theo-
retical framework, the purpose of this paper is to 1) review the
competing views on the linguistic relativity principle or the
Whorfian hypothesis, 2) discuss its significant role in percep-
tual domains, and 3) expand the linguistic relativity principle to
the area of second language (L2) or foreign language (FL) ac-
quisition. Specifically, the cognitive and perceptual aspects of
LRH will first be discussed. Secondly, the interaction between
language and cognition will be discussed with empirical evi-
dence. Next, the linguistic intricacy of English ergative verbs
related to overpassivization errors2 made by English language
learners (ELLs) will be explained using the spectrum of LRH.
Finally, the implications of LRH in cross-cultural language lea-
rning will be discussed.
Competing Views
The premise of LRH stemmed from the work of a linguist
named Sapir (1929; cited in Tohidian, 2009) who had studied
English in comparison to Native American languages and con-
cluded that differences in linguistic features across languages
yielded differences in the perception, understanding, and inter-
pretation of the world. This notion was followed and substanti-
ated by the work of Benjamin Whorf (1956; cited in Tohidian,
2009), resulting in a theory referred to as the Sapir-Whorf hy-
pothesis or as the Whorfian hypothesis3, which reflects the no-
teworthy influence of Whorf on theory-building. Later, scholars
broke down LRH into two versions according to the degree of
intensity: strong LRH and weak LRH (Hunt & Agnoli, 1991;
Tohidian, 2009). The strong form of LRH hypothesizes that
language dictates and controls thought and perception, while
the weak form of LRH posits that language affects thought.
Empirical evidence has rejected the strong form of LRH (Hunt
& Agnoli, 1991; Regier & Kay, 2009), by raising questions
about direct translation from one language to another, the pres-
ence or absence of a particular form in a language, and the qua-
2The term “error” is operationalized as any deviations from the prescribed
syntax, regardless of the nature of production or judgment (i.e., transitory
mistakes or persistent errors).
3In this paper, the term LRH and the Whorfian hypothesis are used in-
1According to Tohidian (2009), the label linguistic relativity is more com-
mon in these days.
lity of evidence used in Whorf’s assertion. The weak form of
LRH has also been criticized to be a vague and unprovable
hypothesis. However, Hunter and Agnoli (1991) have claimed
that LRH can be testable, quantifiable, and falsifiable with re-
spect to the influence of the lexical, syntactic, semantic, and
pragmatic aspects of language on cognition and thought.
Over the decades, LRH has gone through ebbs and flows in
the field. In the 1960s, the idea of linguistic relativity faded out
from the research framework, since the prevalent paradigm in
linguistics and anthropology stressed the universal nature of hu-
man language and cognition (Pinker, 1994). The linguistic uni-
versality hypothesis, motivated by Chomsky’s Universal Gra-
mmar (UG), posits that the universal repertoire of thought and
cognition precedes linguistic constraints each language entails.
UG overshadowed the trend in linguistics at the time and re-
sulted in a concomitant dismissal of LRH in psychological
inquiry. However, UG has limitations in the capacity of expla-
ining cross-linguistic variations, such as low intertranslationabi-
lity and loan-word use (Hunt & Agnoli, 1991). When she pub-
lished her memoire entitled Lost in Translation: A Life in a New
Language, Hoffman (1989) had a poignant penetration into the
intertwined nature of language and perception of the world,
because a thought expressible in one language does not al-
ways map into an equivalency in another. For example, the wo rd
serendipity” cannot be translated into a single Korean word,
because there is no corresponding counterpart word in Korean.
Korean takes several words in a phrase to express the English
word “serendipity”. Loan words used in many languages may
stem from this translation barrier or a lack of transparent equi-
valency from one language to another. Hunt and Agnoli (1991)
note that “language differentially favors some thought processes
over others, to the point t hat a thought that is easily expressed in
one language might virtually never be developed by speakers of
another language” (p. 378). However, balanced bilinguals seem
to maintain their abilities to think differently in different langua-
ges according to the circumstantial demand by going back and
forth from one language to another (Hunt & Agnoli, 1991).
In a similar vein, Casasanto (2008) has pointed out that the
anti-Whorfian school of thought is at fault in lumping two dif-
ferent questions together. According to him, the conflation of
two distinctively different questions “do we think in language?”
and does language shape thought?” created an artifact in the
argument structure and the logical flow of which the anti-Who-
rfian UG group claimed. Pinker (1994) asserts that the founda-
tional categories of reality are not “in” the world but are im-
posed by one’s culture, calling the direct relationship between
thought and language a “conventional absurdity” (p. 47). He
further calls the Whorfian hypothesis a myth. However, Garn-
ham and Oakhill (1994) argue that using the different number
of words for snow (the well-known example) used in Eskimo
and English to support or debunk the relationship between lan-
guage and cognition is questionable, because the difference in
the number of words in the two languages is resulted not from
the fundamental difference in thought but from the needs of the
environmental condition. They note that one group of English
speakers (i.e., skiers) uses a number of different words to refer
to snow, which is different from that used by typical English
speakers. Casasanto (2008) calls for a need of reframing of
Whorf’s inquiry into the relationship among language, concept,
and experience by discrediting Pinker’s (1994) assertion against
LRH. He argues that “language can shape the way people think
even if they do not think in language” (p. 65).
Effects of linguistic relativity have been reemphasized in the
1990s with empirical experiments in the domains of spatial co-
gnition (Levinson, 1996; Li & Gleitman, 2002), number (Gor-
don, 2004; Miller, Major, Shu, & Zang, 2000), color (Gilbert,
Regier, Kay, & Ivry, 2006; Kay & Kempton, 1984), and time
perception (Boroditsky, 2001; January & Kako, 2007). Current
research has focused on exploring paths in which language in-
fluences thought and on determining to what extent language
affects cognition (Lucy, 2003; Slobin, 20 03). This empirical evi-
dence supports the weak form of LRH. Casasanto (2008) endo-
rses a premise that language can shape the way we think, be-
cause speakers of different languages develop unique concep-
tual repertoires through cognitive processes over time. In addi-
tion, Regier and Kay (2009) also note that the perception of the
world is filtered through the semantic category of the native
language (first language; L1).
Language is an important vehicle for thought, in part, be-
cause language facilitates the understanding of others’ knowl-
edge and beliefs. Gentner and Goldin-Meadow (2003) empha-
size the role of language in the understanding of abstract, rela-
tional, information and the integration of spatial information in-
to a meaningful unit. Despite a gap between language and thou-
ght, structural differences in languages result in a different con-
strual of the world. Gentner and Goldin-Meadow (2003) have
compiled theoretically informed debates on the Whorfian hypo-
thesis, encompassing space, number, motion, gender, mind, the-
matic roles, and ontological discussion on objects and substan-
ces. These inquiries on around three categories of language uti-
lity: language as lens, language as tool kit, and language as ca-
tegory maker. The premise language as lens posits that the lan-
guage we speak shapes our perception of the world. The view
language as tool kit concerns as to whether the language we
speak expands our ability to represent and rationalize symbolic
systems and belief systems. The theme language as category
maker relates to whether the language we speak affects how we
make the category distinctions, such as spatial semantics and
relations, nonverbal classifications, and ontological distinctions
between objects and substances. Irrespective of the inquiry di-
fferences, consensus converges onto the fact that “language is a
powerful mediator of cognition when we speak and much of
our lives is spent in language-related activities” (Gentner &
Goldin-Meadow, 2003: p. 11).
Empirical Evidence of the Linguistic Relativity
Although it does not reflect a complete map of consciousness
or thought, language is at least a representational map which
varies across languages (Clark, 2003), because speakers selec-
tively choose different information, details, and interpretations
based on the representations of reality, depending on the lan-
guage spoken. A pile of research studies indicates that speakers
of different languages view and think about the world in rela-
tively different ways. A claim that language shapes the way we
think has been supported by a handful of research findings us-
ing pseudo-linguistic stimuli. Experimental evidence has rekin-
dled interests in LRH by examining the extent to which langua-
ge affects nonlinguistic, semantic, cognition in domains such as
space, color, number, and time (Boroditsky, 2001; Casasanto,
2005; Gilbert, Regier, Kay, & Ivry, 2006; Gordon, 2004; Lev-
inson, 1996; January & Kako, 2007; Kay & Kempton, 1984; Li
& Gleitman, 2002; Miller, Major, Shu, & Zang, 2000; Regier &
Copyright © 2012 SciRes.
Kay, 2009). Cultural variations can influence the speaker’s per-
ception and thought with respect to the lexical, syntactic, se-
mantic, and pragmatic aspects of language (Hunt & Agnoli, 1991).
As a way to control confounding factors related to a particular
language, attempts have been made by designing strictly pseu-
do-linguistic tests to assess the way in which speakers of dif-
ferent L1s perceive and conceptualize abstract conceptual ref-
erents, such as time and spatial orientation. As briefly indicated
earlier, Casasanto (2008) argues that Pinker’s (1994) assertion
of anti-Whorfian is logically fallacious and misleading, indi-
cating that his opposition to LRH is on the basis of ill-formed
argument structure.
In a color perception study, Drivonikou and colleagues (2007)
have utilized a color identification task and a color detection
test to examine a hypothesis that linguistically coded color ca-
tegories (e.g., green, purple, pink, etc.) influence color discri-
mination. They found that language showed significantly stron-
ger effects of color discrimination on the right visual field than
the left visual field. This asymmetry was attributable to the
contralateral projection of visual fields onto cerebral hemi-
spheres and the specializd function of the left hemisphere for
language (Drivonikou, et al., 2007). In a subsequent study (Re-
gier & Kay, 2009), similar results were found in that arbitrary
color names influenced color perception in the right visual field
only and that color naming across languages reflected universal
tendency as well as local linguistic convention. Regier and Kay
(2009) have concluded that the view of linguistic relativists was
half right with respect to color perception and color naming.
Although the mechanism that an individual’s cognitive proc-
esses influence linguistic features or vice versa has not been fu-
lly mapped out, equating thinking with language is a way to di-
sentangle the relationship between language and thought. Ex-
perimental evidence that shows differences across different L1
speakers in the lexicalization of time can be attributable to cro-
ss-linguistic differences in the way time duration is represented
by speakers of different languages (Casasanto, 2008). A claim
that thought is mediated by language to some degree seems to
be in agreement (Casasanto, 2008), given that speakers of dif-
ferent languages think differently about time and spatial meta-
phors. This indicates that language reflects and shapes the or-
ganizational structure of temporal representations.
The relationship between time and perception has been in-
vestigated with respect to spatial orientation as well. For most
western-language speakers, time is semantically perceived as a
horizontally organized concept (Boroditsky, 2001). However,
this spatial orientation with time does not appear to be universal.
This was shown through a cross-language study. Boroditsky
(2001) has conducted three experiments to investigate whether
language affects the way the speaker thinks about time with
Mandarin and English speakers. Mandarin speakers tend to
think about time on a vertical plane even when they are think-
ing for English. For example, Mandarin speakers are faster to
confirm that March comes earlier than April when they see ve-
rtical object priming in a vertical array than horizontal priming.
The extent to which Mandarin-English bilinguals think about
time vertically is associated with age of initial L2 exposure and
acquisition. Boroditsky (2001) concluded that language serves
as an influential vehicle in shaping thought about abstract enti-
ties and that L1 plays a part in shaping habitual thoughts to
some degree. These results suggest that temporal spatial-time
cognition is cross-linguistically sensitive, which is evidenced
using spatial metaphors (e.g., “Pushing a meeting forward/ba-
ck,” “Traditions are handed down through generations,” “Her
birthday is coming up”; Casasanto, 2008). However, January
and Kato (2006) challenged Boroditsky’s (2001) study results
in terms of validity and its implications. They failed to replicate
the finding of Boroditsky’s (2001) study. It is still unclear as to
the source of the conflicting results. As Casasanto (2008) indi-
cates, designing a stimulus material and its validation are a big
challenge in research projects. No matter how carefully the in-
strument is constructed, any experimental design that relies on
comparable performance data between two groups may have
confounding effects on group differences and methodological
flaws are hardly avoidable (Casasanto, 2008). It is possible that
a comparative fallacy obscures systematicity in real group dif-
ference by underestimating or overestimating interlanguage re-
Psychological and cognitive aspects of categorical perception
have also received researchers’ attention. The categorization of
objects seems to be influenced by the linguistic feature. Ameel
and colleagues (2005) showed that French and Dutch bilingual
speakers, who came from Belgium and shared one cultural ba-
ckground, classified objects (e.g., bottles and dishes) into a ca-
tegorization of the shared features of the naming pattern. They
concluded that the classification of objects was dependent upon
the common features the objects share as well as the language-
specific factors. Color discrimination or perception was also
found to be language-specific. For instance, Roberson, Hanley,
and Pak (2009) have found that color discrimination thresholds
between color categories are different in English and Korean
speakers who use different color terms and different threshold
A difference of self-perception in the use of different lan-
guages was also found. Kemmelmeier and Cheng (2004) had
bilingual students from Hong Kong fill in independent and int-
erdependent self-construal scales in both English and Chinese.
They hypothesized that there was a significant difference in the
self-construal between the two-language groups. When they
described themselves in English, the bilingual students showed
a more independent self-construal, whereas their self-construal
was skewed toward a more interdependent scale when they des-
cribed in Chinese. This study suggests that langua ge can serve as
a cognitive cueing system that prompts to define self-perception
according to the language used at the moment in the face of spe-
cific situational demands. This finding is in a similar line with an
episodic description of multiple identities temporally formed when
people talk in different languages (Hunt & Agnoli, 1991), sug-
gesting that different cultures and different languages lead to
different speech acts affected by the language spoken.
Lee (1997) explains the Whorfian hypothesis in terms of the
role of language in teaching and thinking in order to improve
pedagogical practice by reflecting on the language-mind-ex-
perience relationship. The author places an emphasis on the
notion of “linguistic thinking,” “thought insofar as it is linguis-
tic,” or “language in cognition” (p. 432), because cognitive pro-
cesses are linguistic in nature. Language and thought are inex-
tricably interrelated entities. As language is closely intertwined
with conceptual activity, linguistic thinking plays an integral
part in communicative activities and meaning making processes
(Lee, 1997). Speakers become linguistically conditioned throu-
gh a consistent and continuous usage of a speech pattern, be-
cause persistent language use contributes to the organization of
an experiential reality in a certain way (Lee, 1997). However,
Casasanto (2008) warns that the fact that language affects thou-
Copyright © 2012 SciRes. 51
ght does not mean that people “think in language” or language
plays a privileged role via a special channel in shaping thought.
He also suggests that an awareness of cross-linguistic cognitive
differences demonstrated by diverse L1 speakers can be a step
toward an understanding of the boundaries of cultural diversity.
In addition, it can be a tool for investigating how the way of
thinking plays out in relation to linguistic activity as well as
knowledge acquisition and representation (Casasanto, 2008).
These inquiries and activities will facilitate the discovery of the
locus and structure of our mental representation through inves-
tigation into differences of cross-linguistic cognition and the
Whorfian hypothesis (Casasanto, 2008).
Linguistic Re lativit y H ypothesis and Sec o nd - and
Foreign-Language Acquisition
The idea of LRH was also investigated in bilingualism. Kou-
sta, Vinson, and Vigliocco (2008) examined the semantic effect
of grammatical gender intricacy (present in Italian but absent in
English) using an untimed picture naming task by fluent bilin-
guals and English native speakers. The Italian-English bilinguals
produced more gender-preserving errors with the task in Italian
than in English. However, the proportion of gender-preserving
errors for bilingual speakers did not show a significant difference
from those of either English monoli nguals or Italian monolinguals.
The authors interpreted these results as intra-speaker relativity in
semantic representations for Italian-Engli sh bilinguals, indicating
evidenc e for linguis tic re lativit y rath er than linguisti c determ inism.
Given the rekindled interest in linguistic relativity, English
ergative verb processing by ELLs provides an excellent exam-
ple as to how languages and thoughts are interlinked. Ergative
verbs in English are intransitive verbs that do not take objects
but convey passive meanings in sentences. While the transitive
verb is relatively straightforward in terms of its function and
usage, the intransitive verb bears several functional layers. In-
transitive verbs include ergative verbs and unergative verbs.
The unergative verb carries the subject’s volitional act (i.e., a-
gentive role; e.g., smile, sleep, walk) and is comparatively tran-
sparent at the semantic level, while the ergative verb entails the
subject’s unwilled or nonvolitional action, and the subject plays
a theta role (i.e., patienthood). The ergative verb involves an-
other level of layers, containing two subtypes of the ergative
verb. The first subtype of ergatives is the verb which has a tran-
sitive counterpart (e.g., break, sink, roll), and the other subtype
is the verb which does not have a transitive counterpart (e.g.,
appear, disappear, emerge).
Because verbs have the power to govern how a sentence con-
veys who did what to whom, it is crucial to assign the verb’s
role in a sentence according to the verb characteristics (Pinker,
1994). The importance of the distinction between ergative and
unergative intransitive verbs lies in the fact that L2 learners of
English tend to overpassivize ergative verbs at a significantly
greater rate than they do with unergative verbs (Balcom, 1997;
Oshita, 2000; Sorace & Shomura, 2001). Research shows that
this type of errors occurs among speakers of different L1 grou-
ps (Ju, 2000; Oshita, 2001; Yip, 1995; Zobl, 1989). Especially
when ergative verbs are used with inanimate nouns, the ergative
active voice becomes a big hurdle in L2 judgment and produc-
tion for many ELLs. Critical observations have been made on
the tendency of overpassivization of which L2 learners of Eng-
lish compose using ergative verbs. For example, L2 English
learners tend to produce incorrect passive sentences like “4the
battery was died yesterday,” instead of “the battery died yes-
terday.” The subject (the battery) is not capable of carrying out
the action, but rather undergoes the action expressed by the
verb (died). As such, when the subject is affected by the action
of the verb, yielding a passive meaning, the complexity is at-
tributable to ergative verbs (Cowan, 2008). Interesting is that
the prevalent overgeneralization of the passive voice is found in
advanced ELLs’ English production. Master (1991) has compi-
led a language sample showing this phenomenon. Master (1991)
notes that Asian students are more susceptible to this error type
than any other groups.
The correct forms of present third person singular and past
tense were failed to appear in Issic’s performance. [Chinese]
Contrastive analysis was not totally rejected; instead it was
emerged as a partial explanation of interlanguage. [Thai]
In markedness theory, when L1 is unmarked and L2 is
marked, transfer errors are persisted until the late stages of SLA.
Learning style preference and method are significantly inter-
acted when other variables are controlled. [Taiwanese]
His critical period hypothesis was argued that there is no
reason to assume that the language faculty atrophies with age.
The mathematical basis of the report is shown the following
results. [Japanese]
The content in this paper is used the right form and has three
parts. [Hmong] (Master, 1991: p. 19)
This overpassivization tendency observed in ELLs has re-
sulted in a number of research studies in the recent three dec-
ades (Balcom, 1997; Ju, 2000; Kondo, 2005, Oshita, 2001; Per-
lmutter, 1978; Yip, 1995; Zobl, 1989). This propensity has been
explained by numerous hypotheses, including the transitiviza-
tion hypothesis (Perlmutter, 1978; Yip, 1995), the post-verbal
NP movement hypothesis (Balcom, 1997), the unaccusative tra-
p hypothesis (Oshita, 2001), the relationship between argument
structure and its morphosyntatic instantiation (Kondo, 2005),
the locus of causality (Ju, 2000), and the split intransitivity
hierarchy (Sorace, 2000). Some researchers view the problem-
atic overpassivization as false mapping relations between lexi-
cal semantics and syntax, while others make connection to syn-
tax-bound optionality.
Despite many explanations generated, there is no study that
links the tendency to LRH. If persistent errors are found in a
specific grammatical point especially in advanced L2 English
learners, it can be speculated that a cognitive mechanism plays
a critical role in the process of that particular grammatical point
beyond significant L2 exposure and input. The following sec-
tion discusses how the overpassivization tendency can be ex-
plained through LRH with respect to subject agentivity and the
interface of syntax and semantics.
Korean English Language Learners’ Ergative
Verb Processing
A study by Pae, Schanding, and Kwon (2011), which inves-
tigated the Korean ELLs’ performance on a grammaticality ju-
dgment task of English ergative verbs, showed a dominant over-
passivization error pattern in Korean ELLs. The research que-
Copyright © 2012 SciRes.
ries were analyzed with respect to morphological markings (i.e.,
presence or absence of “be + p.p”; i.e., active vs. passive), the
hierarchy of agentivity (humans, concretes, or abstracts), noun
animacy (animate vs. inanimate), word characteristics (single-
words vs. multi-words), the presence or absence of counterparts
(paired vs. nonpaired), and the properties of the ergative verb
(state vs. motion). If the locus of the overpassivization relates to
cognitive complexity beyond L2 grammatical input, word fre-
quency would not affect the performance of Korean ELLs. As
expected, found was no significant association between word
frequency and the participants’ performance on ergative-sen-
tence judgment and between the sentence length and grammati-
cal judgment in terms of accuracy.
Subject Agentivity
The animacy and agentivity of subject nouns are closely re-
lated to syntactic and semantic characteristics of verbs and fol-
lowing nouns as direct objects. Verb transitivity defines the
syntactic structure of sentences and the meaning of sentences
(Hinkel, 2002). Transitivity hinges on the parameter of nouns
and verbs, such as the subject noun capacity for agentivity, the
action capac ity of verb meaning, the degree of free will or voli-
tion that the subject noun entails, and the lexical characteristics
of direct objects. These features of nouns define the sentence
parameter in terms of transitivity. The degree of volitionality
that the subject shows as an agent can assign the object’s role as
a patient in the syntactic structure of a sentence (Delancey,
The linguistic features of English may affect ELLs’ capabil-
ity to process and use passive voice judgment and constructions
appropriately or inappropriately in English. In Korean, lexical
animacy of nouns is closely tied to the feasibility of noun agen-
tivity and verb transitivity. However, English usually does not
rely on noun animacy and its semantic constructs, such as agen-
tivity and patienthood, in the formation of a sentence. Let us
look at the sentence, “Ben broke the vase.Ben is the agent of
the action, and the vase is the recipient of the action. The re-
verse order in the sentence (i.e., The vase broke Ben) does not
yield acceptable syntax, but “the vase broke” is acceptable in
English in which the verb broke becomes an intransitive, and it
does not reflect the consideration of intentionality. However,
the Korean language imposes the expression of a passive mea-
ning on the sentence structure because the vase cannot be an
agent due to its inanimate nature (e.g., 병이 깨졌다; change
of status The vase was broken).
The degree of accountability for the action establishes the
gradient features of noun agentivity or patienthood. The agenti-
tivity of the subject’s role is clear in the example of the sen-
tence “I suffered from a cold,” where the subject “I” is clearly
not a causer of the verb action “suffer” but a recipient of the
verb action. Hence, an overpassivization error is likely to be
made by Korean ELLs as in “4I was suffe red from a cold.” The
patienthood trait becomes more salient than agentivity in the
sentence “The pen fell from the desk.” The subject, the pen, is
obviously the receiver of the verb (fall) action; therefore, the
passive form “4The pen was fallen from the desk” comes natu-
rally to many Korean ELLs over the active voice. This type of
sentence production and acceptability by English learners
points to a cognitive organization related to English ergative
verbs, transferred from L1 specificity.
English is liberal in taking subjects in sentences, regardless
of the subject characteristics, as seen in the prevalence of the
active voice with inanimate agents in English. In line with Ma-
ster’s (1991) notation on Asian students’ tendency to overpas-
sivize, Korean students are inclined to think that inanimate su-
bjects with active verbs are unacceptably anthropomorphic. En-
glish permits active voice with an inanimate subject when the
verb is an inherent aspect or function of that subject. For in-
stance, the intrinsic utility of a thermometer in a sentence “A
thermometer measures the temperature is a measuring instru-
ment (Master, 1991). According to the anthropomorphic princi-
ple, however, the subject should have the ownership of the
action in the sentence. Hence, animate nouns are to be used as
agents of actions in active voice, while inanimate nouns receive
actions that verbs express in passive-voice structures. The ther-
mometer itself cannot measure the temperature, because it is not
the doer of the measuring action and because the thermometer
is an instrument used for temperature measurement (Master,
1991). It should read “a thermometer is used to measure the
temperature” as a proper way of Korean expression. It can also
be expressed as “We measure the temperature with a thermo-
meter,” because the agent who operates the instrument should
serve as a subject in the sentence. This is compatible with the
notion of the hierarchy of inanimacy as well. For instance, “the
car needs gas” is permissible because the car has a higher ani-
macy value than gas on the hierarachy continuum (Hinkel,
2002). As Hinkel (2002) points out, however, a sentence “the
article discusses the government” may become confusing or
problematic for Korean ELLs because of the violation of the
sheer hierarchy. The “article” is lower than “government” on
the continuum of the animacy hierarchy. Hence, the passive
form (The government is discussed in the article) tends to be
produced or judged by Korean ELLs as an acceptable sentence
rather than the active sentence (the article discusses the gov-
It seems that the mechanism of language processing is or-
ganized by L1 linguistic structures and semantic representations.
Since the representation of syntax and semantics stored in the
speaker’s receptive repertoire is multifaceted, the organization
and retrieval of syntax may vary according to semantic knowl-
edge stored in long-term memory and the ability to make use of
the connections between syntax and semantics.
The Interface of Syntax and Semantic s
The syntactic structure be + p.p lexicalizes the semantic in-
formation in the change of the verb status. The ergative verbs
that tend to be overpassivized by Korean ELLs have semantic
attributes of both “change” and “recipienthood.” For example,
in the sentence the bottle broke, the verb broke places the mea-
ning of the verb in the change of status and, at the same time,
confines the subject to be the recipient of the action of the verb.
This goes against the normative way of Korean expression in
which the agentive doer executes the action for the change of
Since the verb largely defines how a sentence conveys who
did what to whom in what way in what context, the main verb
typically dictates the sentence structure and meaning. In Eng-
lish, the subject of the sentence is not always the “doer” of the
action of the verb as indicated earlier. The subject of the sen-
tence is the “doer” only when the verb defines as such (Pinker,
1994). For example, the verbs in the following sentences assign
4An ungrammatical sentence.
Copyright © 2012 SciRes. 53
different roles.
1) a. The teacher gave the student a book. [The subject is do-
ing the giving]
b. The student received a book from the teacher. [The sub-
ject is bein g gi ven to or the s ubject is doing the receiving]
2) a. He shattered the window. [The subject is doing the sha-
b. The window shattered. [The subject is being shattered]
The first set 1) of the sentences has two different words that
specify the roles of the doer and receiver. However, the second
set 2) of the sentences has the same lexicon used with animate
and inanimate subjects. Although the verb receive conveys the
meaning of “being given,” the verb in sentence 1b can be inter-
preted as the subject to be doing the action of receiving. The
direction of affectedness by the action of the verb is different
between the two sentences. In other words, the subject 2a is the
doer or causer of the action and the affected object is the win-
dow, whereas the subject of the latter 2b is the receiver of the
action, which is affected by the action of verb as an experiential
entity. The Korean verb equivalencies of the verbs used in the
first set are consistent in terms of the role of the head wo- rds
and the past markers. The equivalencies of the verbs used in the
second set have the same head word, but go through a varia-
tion that signals the action-receiving meaning or the change of
status (e.g.,-졌다). As seen here, the Korean language imposes
passive syntactic-semantics on the verb when a nonanimate
subject is used in the sentence. Hence, the deep structure of the
second set of sentences are as follows:
3) a. He shattered the window.
[S + Vtransitive + O] transitive
subject: doer object: experiential receiver
b. The window shattered.
[S + Vintransitive] ergative
subject: experiential receiver
c. The girl danced.
[S + Vintransitive] unergative
subject: doer
As seen above, the verb dance in 3c does not deviate from
the notion that the subject is a doer and does not create com-
plexities as the verb shatter does, as seen in 3b. It is speculated
that the organization and representation of L1 linguistic features
and semantics play a crucial role in the judgment and accep-
tance of ergative sentences.
Figure 1 displays a comparison of aggregated accuracy and
latency performance by the Korean ELLs. The data points rep-
resent logarithmic values for the purpose of placing the data on
a same continuum. The passive voice was an incorrect ergative
expression, whereas the active form was a correct counterpart.
The score indicates that the significant number of Korean lea-
rners of English accept incorrect passive forms of sentences as
correct expressions. This is consistent with the findings of pre-
vious research. Interestingly, the passive form took shorter time
in an acceptability judgment task than the active counterpart,
which means that the participants tended to accept the passive
voice without hesitation. This finding suggests that English er-
gative verbs are processed based on the meaning of patienthood
and the change of status, as it is already registered in L1.
To summarize, the findings of Pae, Schanding, and Kwon’s
(2011) study suggest that the prevalence of overpassivization is
language specific, indicating that there may be an interlanguage
rule that requires an additional processing for active voice to be
used with inanimate subjects. The ELLs’ confusion about erga-
tivity and the relationship between the instrumental subject and
the verb seems to stem from English-specific linguistic proper-
ties. This phenomenon has been observed in English production
and judgment by ELLs, but the same pattern has not been found
when English native speakers learn other languages, such as
Chinese, Japanese, and Spanish (Kondo, 2005; Shan & Yuan,
2008; Sorace & Shomura, 2001). The lexical and semantic fea-
tures of sentence constituents in ELLs’ L1 seem to have an impa-
ct on their grammaticality judgment, which is related to the impli -
cation of LRH. Korean ELLs have particular di sa dvantage s when
they process or produce ergative-verb sentences, because their L1
does not have syntactically and semantically derived active voi-
ces that intrinsically entail passive meanings, as English does.
Cognitive Mechanism beyond L2 Input
There is consensus on the positive correlation between L2
input and L2 proficiency level. Since L2 acquisition or learning
is a slow process of form-function mapping, explicit and im-
plicit instruction as well as real-life experiences with L2 will
lead to optimal learning that enables ELLs to perform success-
ful oral and written production (Ellis, Basturken, & Loewen,
2001). However, errors associated with ergative verbs do not
seem to support the pivotal role of L2 input in SLA. Ironically,
the overpassivization of ergative verbs are more likely to be
found as persistent errors in advanced ELLs than beginners
(Oshita, 2001; Masters, 1991). Then, there should be a cogni-
tive factor that is conditioned by L1 in the process of L2 Eng-
lish ergative verbs. Hunt and Agnoli (1991) have explained the
Whorfian hypothesis from a cognitive psychological perspec-
tive, and they concluded, supporting the weak version of LRH,
that language is a window to view the speaker’s thought.
Hoffman, Lau, and Johnson (1986) have also found a differ-
ence in the use of stereotypes by bilingual English-Chinese
speakers. The English-Chinese bilinguals were first asked to
see descriptions of people as to whether they were conformed
to either English or Chinese stereotypes. Later the bilinguals
were asked again whether certain behavioral descriptions which
had not been included in the original stimuli were congruent
with the characteristic of the target individual. A difference was
found according to the language used. When addressed in Eng-
lish, the participants used English stereotypes, but when asked
in Chinese, they used Chinese stereotypes. This finding indica-
tes an activation of unconsciously driven cognition to solve the
given task as well as the influence of language on different as-
pects of thought, suggesting different cognitive factors involved
according to the coding system used.
Krashen’s (1985) Input Hypothesis, one of his five hypothe-
Figure 1.
Korean English learners’ accuracy and latency performance on active
and passive forms.
Copyright © 2012 SciRes.
ses in second language acquisition (SLA), has been influential
in SLA as a viable theory. According to him, sufficient com-
prehensible input yields the presence of i + 1, where i repre-
sents previously acquired linguistic competence and knowledge,
and comprehensible input leads to grammatical fluency. How-
ever, Karshen’s hypotheses have been criticized due to a lack of
empirical evidence, obscure definitions, and unfalsifiable the-
ory (McLaughlin, 1987). It seems that Korean ELLs’ overpas-
sivization of ergative verbs goes beyond Krashen’s comprehen-
sible L2 input, as the error pattern is observed in advanced
The proposal that cognition and thought are conditioned by
language spoken has been long debated. Speakers who talk di-
fferently about the world also think about it differently because
language not only reflects the organization of our temporal lin-
guistic and nonlinguistic representations, but also shapes indi-
viduals’ unique conceptual repertoires (Casasanto, 2008). The
weak form of the Whorfian hypothesis has gained social scien-
tists’ recurring attention. Despite dismissal, LRH that the se-
mantic organization and structure of a language shape or con-
strain the ways in which the speaker concept ualizes, understa nds,
and interprets the world has been supported by a multitude of
empirical findings. The relationship between language and other
cognitive domains, such as spatial cognition, number, color,
and time perception, has been well documented (Boroditsky,
2001; Gordon, 2004; Kay & Kempton, 1984; Levinson, 1996).
Pseudo-linguistic experiments and cross-language studies in re-
lation to LRH are essential not only to determine how language
affects cognition but also to understand the cognitive conse-
quences of and L1 influences on bilingualism. The cross-lingui-
stic and L1 semantic effects on L2 learning support the canon
of LRH. Lingui stic feature s affecting cog nitive processe s and me -
chanisms are also reconceptualized with respect to L2 learning.
Irrespective of the passive meaning of the sentence, English
ergative verbs refuse to appear in consortium with direct obje-
cts and to use them in passive forms because they are essentia-
lly intransitive verbs. When English sentences and verbs have a
low degree of transparency between syntactic and semantics,
the opaqueness of ergative verbs creates confusion for Korean
ELLs; as a result, they experience interference from their L1
whose language has consistency between the verb and its se-
mantic properties. Although overgeneralization is a common
interlanguage strategy that ELLs utilize when they face difficult
or confusing structures of English (Nassaju & Fotos, 2011), the
fact that the overpassivization error is made via sophisticated
semantic processing seems to go beyond overgeneralization mi-
stakes. As the reaction time indicates, Korean ELLs tend to take
more time and produce more errors, in a sentence acceptability
task, with ergative active voice (i.e., correct form) than with er-
gative passive voice (i.e., incorrect form; Pae, Schanding, & Kwon,
2011). Ergative passive forms are processed more automatically
without delay than ergative active forms. This suggests that
conscious control conditioned by L1 appears to be executed in
the face of the deep structure of grammatical complexity.
Comprehensible input has been emphasized as a critical con-
cept for ELLs with and without learning difficulties. According
to Krashen’s (1984) Input Hypothesis, L2 input that is compre-
hensible leads to an understanding of the essence of syntactic
and semantic intricacies of L2. The input hypothesis places va-
lues on relevant background, and posits that knowledge and a-
ppropriate contextual information are crucial. All evidence cen-
ters on the premise that language is an influential tool in shap-
ing thought (Casasanto, 2008; Gentner & Gol din-Meadow, 2003 )
and that L1 is a mediator in L2 acquisition or learning (Cowan,
2008; Koust, Vinson, & Vigliocco, 2008; Master, 1991). This
suggests that L2 input alone is not sufficient enough to acquire
optimal proficiency of L2.
The implications of ergative verb processing by ELLs have
to do with the Whorfian hypothesis. When L1 has a drastically
different linguistic structure from that of L2, the linguistic in-
tricacy of L1 may affect L2 acquisition and learning. In addi-
tion to this underlying intrinsic mechanism, it becomes a big
obstacle for ELLs to internalize the complexity of the English
language, when not only must they know the nature of the sub-
ject (i.e., agentive or instrumental), but also the features of the
verb, such as intransitive, stative, ergative/unaccusative, and
unergative, as well as the argumen t stru c t ure and discourse flow
(Master, 1991). The pedagogical recommendation is to address
the interaction between language and cognition as well as its
influence on L2 learning beyond L2 instructional and compre-
hensible input.
To conclude, if advanced ELLs of different L1 groups per-
sistently make error s in the use of English ergative verbs, it may
be an indication that there is a linguistic element that hinders a
mastery of L2. This article echoes a quote by Wilhelm von
Hamboldt, German educator, linguist, and philosopher: “Lan-
guage is the formative organ of thought. Intellectual activity,
entirely mental, entirely internal, and to some extent passing
without trace, becomes through sound, externalized in speech
and perceptible to the senses. Thought and language are there-
fore … inseparable from each other.”
Ameel, E., Storms, G., Malt, B. C., & Sloman, S. A. (2005). How bi-
linguals solve the naming problem. Journal of Memory and Lan-
guage, 53, 60-80. doi:10.1016/j.jml.2005.02.004
Balcom, P. (1997). Why is this happened? Passive morphology and
unaccusativity. Second Language Research, 13, 1-9.
Boroditsky, L. (2001). Does language shape thought? Mandarin and
English speakers’ conceptions of time. Cognitive Psychology, 43,
1-22. doi:10.1006/cogp.2001.0748
Casasanto, D. (2008). Who’s afraid of the big bad Whorf? Crosslin-
guistic differences in temporal language and thought. Language
learning, 58, 63-79. doi:10.1111/j.1467-9922.2008.00462.x
Chaika, E. (1989). Language the social mirror. New York: Newbury
House Publishers.
Clark, E. V. (2003). Languages and representations. In D. Gentner, & S.
Goldin-Meadow (Eds.), Language in mind: Advances in the study of
language and thought (pp. 17-24). Cambridge, MA: The MIT Press.
Cowan, R. (2008). The teacher’s grammar of English: A course book
and reference guide. New York: Cambridge University Press.
Delancey, S. (1990). Ergativity and the cognitive model of event struc-
ture in Lhasa Tibetan, Cognitive Linguistics, 1, 289-321.
Drivonikou, G. V., Kay, P., Regier, T., Ivry, R. B., Gilbert, A. L.,
Franklin, A., & Davies, I. R. L. (2007). Further evidence that Whor-
fian effects are stronger in the right visual field than the left. PNAS,
104, 1097-1102. doi:10.1073/pnas.0610132104
Ellis, R., Basturkmen, H., & Loewen, S. (2001). Learner uptake in
communicative ESL lessons. Language learning, 51, 281-318.
Garnham, A., & Oakhill, J. (1994). Thinking and reasoning. Oxford:
Copyright © 2012 SciRes. 55
Copyright © 2012 SciRes.
Gentner, D., & Goldin-Meadow, S. (2003). Whither Whorf. In D.
Gentner, & S. Goldin-Meadow (Eds.), Language in mind: Advances
in the study of language and thought (pp. 3-14). Cambridge, MA:
The MIT Press.
Gilbert, A. L., Regier, T., Kay, P., & Ivry, R. B. (2006). Whorf hy-
pothesis is supported in the right visual field but not the left. Pro-
ceedings of the National Academy of Sciences of the USA, 103, 489-
494. doi:10.1073/pnas.0509868103
Gumperz, J. J., & Levinson, S. C. (Eds.) (1996). Rethinking linguistic
relativity. New York: Cambridge University Press.
Hinkel, E. (2002). Why English passive is difficult to teach (and learn).
In E. Hinkel, & S. Fotos, (Eds.), New perspectives on grammar
teaching in second language classrooms (pp. 233-259). New York;
Lawrence Erlbaum Associates
Hoffman, E. (1989). Lost in translation: A life in a new language. New
York: Dutton.
Hoffman, C., Lau, I., & Johnson, D. R. (1986). The linguistic relativity
of person cognition. Journal of Personality and Social Psychology,
51, 1097-1105. doi:10.1037/0022-3514.51.6.1097
Hunt, E., & Agnoli, F. (1991). The Whorfian hypothesis: A cognitive
psychology perspective. Psychological Revi ew, 98, 377-389.
January, D., & Kako, E. (2006). Re-evaluating evidence for linguistic
relativity: Reply to Boroditsky (2001). Cognition, 104, 417-426.
Ju, M. K. (2000). Overpassivization errors by second language learners:
The effect of conceptualizable agents in discourse. Sea Service
Leadership Association, 22, 85-111.
Kay, P., & Kempton, W. (1984). What is the Sapir-Whorf hypothesis?
American Anthropologist, 86, 65-79.
Kemmelmeier, M., Cheng, B. (2004). Language and self-construal
priming: A replication and extension in a Hong Kong sample. Jour-
nal of Cross-Cultural Psychol ogy , 35, 705-712.
Kondo, T. (2005). Overpassivization in second language acquisition.
International Review of App li e d L i n gu i st i c s, 43, 129-161.
Kousta, S-T., Vinson, D. P., & Vigliocco, G. (2008). Investigating
linguistic relativity through bilingualism: The case of grammatical
gender. Journal of E xperimental Psychology, 34, 843-858.
Krashen, S.D. (1985). The input hypothesis: Issues and implications,
New York: Longman.
Lee, P. (1997). Language in thinking and learning: Pedagogy and the
New Whorfian framework. Harvard Educational Review, 67, 430-
Levinson, S. (1996). Frames of reference and Molyneux’s question:
Crosslinguistic evidence. In P. Bloom, & M. Peterson (Eds.), Lan-
guage and Space (pp. 109-169). Cambridge, MA: MIT Press.
Li, P., & Gleitman, L. (2002). Turning the tables: Spatial language and
spatial cognition. Cognition, 83, 265-294.
Lucy, J. A. (1992). Grammatical categories and cognition: a case study
of the linguistic relativity hypothesis. New York: Cambridge Univer-
sity Press.
Master, P. (1991). Active verbs with inanimate subjects in scientific
prose. English for S p ecific Purpose, 10, 15-33.
McLaughlin, B. (1987). Theories of second-language learning. London:
Edward Arnold.
Nassaji, H., & Fotos, S. (2011). Teaching grammar in second language
classrooms: Integrating form-focused instruction in communicative
context. New York: Routeledge.
Oshita, H. (2001). The unaccusative trap in second language acquisition.
Studies in Second Language Acquisition, 23, 279-304.
Park, K-S., & Lakshmanan, U. (2007). The unaccusative-unergative
distinction in resultatives: Evidence from Korean L2 learners of Eng-
lish. Proceedings of the 2nd Conference on Generative Appro ach es to
Language Acquisition North America (GALANA). 328-338., document #1573.
Pae, H. K., Schanding, B., & Kwon, Y.-J. (2011). Overpassivization of
ergatives by adult English language learners. Paper presented at the
American Association for A pplied Linguistics, Chicago, Illinois.
Perlmutter, D. M. (1978). Impersonal passives and the unaccusative
hypothesis. Proceedings of the Berkeley Linguistics Society, 4, 157-
Pinker, S. (1994). The language instinct. New York, NY: William Mor-
row and Company.
Regier, T., & Kay, P. (2009). Language, thought, and color: Whorf was
half right. Trends in Cogniti v e S c i ences, 13, 439-446.
Roberson, D., Hanley, J. R., & Pak, H. (2009). Thresholds for color
discrimination in English and Korean speakers. Cognition, 112,
482-487. doi:10.1016/j.cognition.2009.06.008
Shan, C.-K., & Yuan, B. (2008). “What is happened” in L2 English
does not happen in L2 Chine se, ERUOSLA Yearbook, 8, 164-190.
Sorace, A., & Shomura, Y. (2001). Lexical constraints on the acquisi-
tion of split intransitivity: Evidence from L2 Japanese. Studies in
Second Language Acquisiti o n , 2 3, 247-278.
Sorace, A. (2000). Gradients in auxiliary selection with intransitive
verbs. Language, 76, 859- 890. doi:10.2307/417202
Tohidian, I. (2009). Examining linguistic relativity hypothesis as one of
the main views on the relationship between language and thought.
Journal of Psycholinguistic R e s ear ch , 38, 65-74.
Yip, V. (1990). Interlanguage ergative constructions and learnability.
CUHK Papers in Linguistics (pp.45-68). Hong Kong: Chinese Uni-
Yip, V. (1995). Interlanguage and learnability: From Chinese to Eng-
lish. Amsterdam: John Benjamins.
Zobl, H. (1989). Canonical typological structures and ergativity in
English L2 acquisition. In S. M. Gass, & Schacter (Eds.), Linguistic
perspectives on second language acquisition (pp. 203-221). New
York: Cambridge University Press.