CONTENTS
3.
THE RESEARCH REVIEWED
3.1
KELLERMAN 1978
3.2
RICHARDS 1983
3.3
RUTHERFORD 1987
3.4
ZOBL 1989
3.5
SORACE 1991, 1993
3.6
HUBBARD 1994
3.7
PUTZER 1994
3.8
YIP 1994, 1995
3.9
HIRAKAWA 1995
3.10
OSHITA 1995
Probably the first scholar to use the term
ergative, derived from the ancient Greek word for work “ergon”, was A.
L. Dirr in 1928 (Tchekhoff 1978:12). He used “ergative” to describe languages
in which the subjects of intransitive verbs and the objects of transitives
are treated the same syntactically or morphologically. The term continues
to be employed in the field of language typology with rougly the same definition.
The linguistic phenomenon now often referred
to as ergative verbs is a fairly recent discovery, first identified by
Perlmutter (1978) in the Relational Grammar framework. Burzio (1981) integrated
what Perlmutter had named unaccusatives into Government & Binding theory
and relabelled them ergatives.1 The term ergative
reflects a certain resemblance in the syntactic behaviour of these verbs
to a construction in the above mentioned ergative languages, but is somewhat
inaccurate since the corresponding case in those
languages is called absolutive (Yip 1995). In the last two decades, studies
on the topic of ergative verbs have proliferated and an impressive array
of books and articles has been published on the matter. Burzio 1986, Belleti
and Rizzi 1981, Perlmutter and Postal 1984, Den Besten 1985, Grewendorf
1989, Levin and Rappaport Hovav 1995 to list just a few (Internet 1).
Their findings have been applied to SLA-research
in several publications like Zobl (1989) and Yip (1995), but not nearly
as much work has been devoted to the SLA-research than has been allocated
to the study of the general theoretical problems of ergatives. It is my
intention in this paper to summarize the acquisition-related work on ergative
verbs that has been done so far, to review the terminology used, to point
out differences and similarities in the approaches, and to look into the
data that has been gathered.
The first section of my paper will examine
the definition of ergative verbs and the terminology employed in the different
studies discussed in the next part. The second section will describe the
studies that have dealt with the acquisition of ergative verbs, covering
the hypotheses, database, theoretical framework and conclusions of each
study. In the last part, I will reach my own conclusion, bringing together
all the presented evidence into one picture and suggesting further avenues
of inquiry.
The ‘Unaccusative Hypothesis’ as suggested
by Perlmutter (1978) states that there are two classes of intransitive
verbs: unaccusatives and unergatives. Semantically, they can be differentiated
in that the subject of an unaccusative verb “does not actively initiate
or is not actively responsible for the action of the verb” (Internet 1)
but bears the semantic role of theme or patient that is usually associated
with the object. Most change of state verbs fall into that category. However,
unaccusatives also share syntactic features that set them apart from
unergatives. In the Relational Grammar framework the difference between
the two is formulated as: unaccusatives take an initial arc-2 but no arc-1
and unergatives take an initial arc-1 but no arc-2. In the final stratum
both take an arc-1 (Perlmutter & Postal 1984, Sorace 1993). Arc-1 and
2 refer to the grammatical relations of subject and direct object respectively.
Figures 1 to 3 illustrate this concept.
Figure 1: Simplified stratal diagram
for Something strange happened (Hubbard 1994)
happened something strange
Figure 2: example diagram for unergatives
sleep
the Martians
Figure 3: example diagram for unaccusatives
happened
something strange
(Perlmutter 1983, Perlmutter & Postal
1984).
In Government-Binding theory the members
of the two classes have the D-structure
syntactic configurations:
(1)unaccusative: [SNPe][VP V NP]
(2)unergative:[SNP[VPV]]
In other words, the essential assumption
is that some intransitive verbs have underlying direct objects while others
have underlying subjects. Or, to put it in terms of argument structure,
some intransitive verbs subcategorize for one internal argument and no
external argument, while others subcategorize for an external argument
without taking an internal one (Sorace 1993). The resulting different verb
classes can be schematized as follows:
__________________________________
|
|
Transitive
Intransitve
|
________________________
|
|
Ergative/Unaccusative
Unergative
The term Unaccusative emphasizes the unability
of ergative verbs to assign accusative case to the direct object, which
therefore has to move to the subject position in order not to violate Case
Filter (Yip 1995). Unergative verbs are sometimes also referred to
as simple intransitives (Yip 1995), or active intransitives (Zobl 1989).
Oshita (1995), following Levin & Rappaport Hovav (1995), draws a distinction
between unaccusatives and ergatives on the grounds of an existing or nonexisting
transitive alternant. Thus, in his account, verbs like sink, break and
open would be unaccusatives, while happen, appear and die are ergatives.
Other scholars, for example Sorace (1993), distinguish two subclasses,
paired and unpaired unaccusatives, where paired comprises all the ergatives
that have either a transitive or an unergative counterpart. However, the
terminology is still a matter of a heated ongoing academic debate, which
has led to outbursts like Pullum’s (1988:585): “unaccusative verbs [were]
renamed [by Burzio] - in a truly crackbrained piece of terminological revisionism
- ‘ergative verbs’.” Probably the most elegant formulation of the unaccusative
hypothesis’ basic claim comes from Baker (1983:1): “all seemingly intransitive
verbs are not created equal.” This also holds true in language acquisition
where several studies have shown that learners have a hard time with unaccusative/ergative
verbs. The following section reviews some of the SLA-related studies on
this topic, their findings and accounts of the phenomenon.
3.1 Kellerman
(1978)
In 1978 Eric Kellerman conducted a study
with 291 Dutch learners of English, using a grammaticality judgement task
for nine sample sentences. The sentences contained the English translations
for nine different meanings of the Dutch word breken, where breken could
actually be transfered and translated with the English verb break. Kellerman
then ranked the meanings from one -judged most acceptable and thus most
transferable- to nine, the one judged least acceptable. The ranking showed
remarkable regularity across the nine language proficiency groups (A2 =second
year pupils, KU2=ninth year pupils), except for the one sentence where
break was used as an unaccusative verb. Chart 1 shows the rankings of this
sentence: the cup broke.
Chart 1:
While the beginning learners generally accept
this sentence as grammatical, it drops to rank four and five among the
more advanced learners. Kellerman was quite puzzled by this finding and
surmised that advanced learners might perceive that use of break as somehow
more marked. This may be the case either because of its purported rarity
in the input or because of the subjects’ lack of agency. Kellerman also
reports that his subjects substituted either the passive or the transitve
form of break for the supposedly wrong one or they suggested the verb smash
whose Dutch equivalent is an unpaired unaccusative.
3.2 Richards
(1983)
In his study, Richards analysed data from
various preceding studies2 dealing with error
analysis for learners of English as a second language from multiple native
language backgrounds. He sorted out the errors that were common to the
different L1s including Japanese, Chinese, Burmese, French , Czech,
Polish, Tagalog, Maori, and the major Indian and West African languages.
Ergatives verbs show up in his crosslinguistic sample data in sentences
like:
(1) *The sentence is occurs ...
(2) *The telegraph is remain ...
(3) *He was died last year.
(4) *One day it was happened.
He argued that these nontarget-like forms
are caused by the learners’ interpretation of was as a past tense marker
and analogously is as a present tense marker.
3.3 Rutherford
(1987)
Rutherford discussed learners’ erroneous
use of intransitives as transitives exemplified in sentences like:
(1) *The shortage of fuels occurred the
need for economical enginge.
(2) *This construction will progress my
country. 3
He argued that these nonstandard constructions
originate from an overgeneralization of a zero-derivation rule used to
form causative verbs. This rule would derive intransitve causative verbs
from transitive ones, accounting for the alternation with verbs like open,
melt, sink etc. What he called zero-derivation closely resembles Keyser
& Roeper’s (1984) lexical rule for deriving ergative verbs.
3.4 Zobl (1989)
Helmut Zobl worked with a corpus of compositions,
written by 114 university-age students from varying L1 backgrounds ( 90
Japanese,10 Arabic,10 Spanish,1 Chinese,1 Turkish,1 Thai and 1 Indonesian)
enrolled in ESL-programs in Canada and the US. The data was pooled and
ergative, active intransitive and transitive constructions were extracted.
Then, all non-target like instances, where verbs occurred in structures
like
(1) be + Ven
(2) V + NP
were sorted out. Chart 2 illustrates the
results for (1):
Chart 2 :
No instances of (2) were found with active
intransitives and transitives, while for ergatives the target-like/nontarget-like
ratio was 67/13. Moreover, only two of the 17 speakers who produced nonstandard
uses of the auxiliary be with ergatives were among the ten who used be
incorrectly with the other verbs, indicating that the faulty selection
of be does not have the same origin in the two cases.
Zobl argued that the difficulty learners
clearly displayed with the ergative verbs stems from the particular semantic
and syntactic properties of these verbs. Namely that in English, they form
an exception to the rule in two ways. Firstly, they do not correspond to
the canonical mapping of thematic roles to grammatical relations in English,
which is subject-agent and object-theme. Secondly, they do not even obey
the rules for the exception to this canonical mapping, that is the passive,
which has to be marked morphologically in English. The data shows that
learners tend to overgeneralize passivization to ergative verbs and that
they sometimes map the deep structure of the ergative verbs [0 [V NP]]
directly onto surface structure. Not only do these errors confirm the adequacy
of models of grammar with multi-level linguistic representations and the
unaccusative hypothesis’ analysis of the ergative verbs’ deep-structure,
but they also support the claim that principles of universal grammar play
a role in second language acquisition and in the formation of learners’
interlanguage grammar.
Zobl suggests that learners first acquire
a lexical NP-movement rule that allows them to externalize the internal
argument of ergative verbs prior to syntax. Once they become aware of the
discrepancy between ergative verbs and the fundamental characteristics
of English mentioned above, they become uncertain as to the correctness
of their earlier analysis which in turn gives rise to the indeterminacy
in their interlanguage grammar.
3.5
Sorace (1991, 1993)
Sorace’s paper in 1991 compared grammaticality
judgements on the selection of the perfective auxiliary “be” or “have”
in French and Italian, made by French learners of Italian and Italian learners
of French. She found that the intuitions of subjects with French L1 were
consistenly more accurate than those of Italian L1. She attributed this
to the fact that auxiliary selection in Italian, which is closely linked
to unaccusativity (Burzio 1986), is much more consistent and learnable
than the French system where this link has been eroded over time.
In her 1993 study, Antonella Sorace investigated
the linguistic competence of “near-native” speakers of Italian with English
and French L1. She let three groups, English, French L1 speakers of Italian
plus a control group of native speakers, pass grammaticality judgements
on samples of Italian auxiliary selection. Her data showed that not only
were the “near-native” speakers different from the Italian control group,
but there was also a significant and regular dissimilarity between the
French and English learners. The English subjects’ judgements displayed
indeterminacy, whereas the French manifested consistent difference. Sorace
pointed out that this result was predictable because the French auxiliary
system is only partially similar to the Italian in that only a core group
of ergative verbs take auxiliary be, whereas in Italian all ergatives
select be. In English there is simply no alternation between perfective
auxiliaries; unaccusativity is primarily semantic. Therefore, the English
learners’ competence remains incomplete where the French learners’ competence
is divergent.
3.6 Hubbard
(1994)
Hubbard found that ungrammatical sentences
like
(1) *This problem is existed for many years.
(2) *Something strange was happened before
I could open the door.
which earlier research had sometimes attributed
to an incomplete mastery of the English tense /auxiliary system (see Richards
1983), occurred in the vast majority (over 90%) with unaccusative verbs.
He argued that the similarities between passives and unaccusatives, for
example in morphology (like in Italian or French) or in syntax where they
share the same underlying structures, led learners to believe “that English
unaccusative verbs (or some subset of them) receive passive morphology.”
(Hubbard 1994:56)
3.7 Putzer
(1994)
In an error analysis conducted on a large
sample of data from advanced Italian learners of German, Putzer discovered
that the frequency of incorrect case assignment (nominative/accusative)
was much higher in passive sentences than in active constructions. When
he noticed that in those active constructions where errors did in fact
occur the subject usually bore the thematic role of theme, he drew
the conclusion that it must be the noncorrespondence of these particular
active sentences and the passive with the canonical alignment subject-agent,
object-theme that induced learners to make errors. He also found that some
learners introduced the “typical German syntactic dummy pronoun for a missing
agent es” (Putzer 1994:256, translation by the author) in order to solve
their linguistic dilemma.
3.8 Yip
(1994, 1995)
In her study (1994), Virginia Yip examined
the effect of Consciousness - Raising on the acquisition of ergative verbs.
Her subjects were ten students, with various native languages4,
enrolled in an advanced ESL-class at university. Yip used a grammaticality
judgement test on ergative verbs as a pre-test, then the students participated
in a Consciousness - Raising class, and two weeks later the same grammaticality
judgement test was administered as a post-test. The subjects tended to
score better after the instruction except for the verb cook which seemed
recalcitrant. Yip also noted a particular success for the correction of
the passivization of unpaired ergatives like happened. But she conceded
that because of the small range of her experiment “the results are at best
suggestive.”(Yip 1994:136)
In her book (1995) on interlanguage and
learnability of native speakers of Chinese Yip also discussed the learnability
problems of ergative verbs. Her argument is based on the results of a pilot
study with 20 Chinese advanced learners of English who had to complete
a production task and a grammaticality judgement task involving: grammatical
passives, grammatical ergatives, ungrammatical passivized ergatives, ungrammatical
pseudo-passives, grammatical auxiliaries, and ungrammatical auxiliaries.
To a large extent her findings replicate those of Zobl’s (1989). Learners
do not overgeneralize the ergative construction, but produce and accept
passivized ergatives as grammatical. This can be seen as evidence for the
psychological reality of the similar D-structure analysis of unaccusatives
and passives as having an underlying direct object which moves into subject
position. Of course the difference is, that the passive morphology signals
the unusual mapping of the underlying direct object to subject position.
It is important to note that ergatives do exist in Chinese and their passivization
is clearly ungrammatical, just as it is in English.
In terms of learnability, Yip pointed out
an essential distinction between paired and unpaired unaccusatives. She
argued that the latter should be easier to acquire than the former, because
once the learner realizes that the input gives no instances of them being
passivized the Uniqueness Principle will operate and the incorrect forms
will drop out of the interlanguage. However, the transitive counterparts
of the paired unaccusatives can occur freely with the passive and the semantic
difference between the two is very subtle. In that case, the Uniqueness
Principle might even block the acquisition of the ergatives. Yip carefully
suggested that learners probably first acquire the more learnable unpaired
ergatives, which then in turn might serve as a model for the paired ergatives.
3.9 Hirakawa
(1995)
Makiko Hirakawa’s study investigated the
acquisition of English unaccusative constructions by native speakers of
Japanese. Her data contained the results of a production task and a grammaticality
judgement task administered to 22 Japanese intermediate learners of English
and a control group of 14 native speakers. In the production task, passivized
unaccusatives showed up very rarely (13 times out of 132 possible contents)
and only with paired unaccusatives. The overall results of the grammaticality
judgement task showed that learners’ intuitions were less accurate but
not deviant from those of native speakers. However, Hirakawa noted that
ungrammatical passives were sometimes accepted for unpaired unaccusatives
which have a transitive alternant in Japanese. In general her findings
do not suggest that learner overgeneralize passive to unaccusative verbs,
“although a few verbs in both tasks caused some difficulty to some
learners” (Hirakawa 1995). Hirakawa attributed those errors to incomplete
lexical knowledge and L1 transfer.
3.10 Oshita
(1995)
Hiroyuki Oshita studied a large sample of
data taken from the Longman Learner Corpus, which contains written English
produced by learners from various L1 backgrounds. He limited his research
to Spanish, Italian, Korean and Japanese as L1s and a selection of ten
unpaired unaccusative verbs5. Following theoretical
considerations Oshita proposed the ten plausible surface structures for
unpaired unaccusatives listed below:
Preverbal NP: i. NP - V
ii. NP -
be - Ven
Postverbal NP without “passive” structure:
iii. there - V - NP
iv. it - V - NP
v. 0 - V - NP
Postverbal NP with “passive structure”:
vi. there - be - Ven - NP
vii. it - be - Ven - NP
viii. 0 - be - Ven - NP
Transitive: ix. NP1 - V NP2
Special: x. there be + NP + V
Structure i. represents the correct standard surface structure of unaccusatives. Structure iii. is the only acceptable construction of unaccusative verbs with an expletive subject. All the other structures, while being clearly ungrammatical, are all structurally possible in the interlanguage of learners of English. Oshita then classified his token sentences into the ten slot grid above. Chart 3 illustrates the result:
Chart 3:
It is evident that in the vast majority
(over 90% in fact) unaccusatives appear in the correct structure i. For
the sake of visibility chart 4 shows the same results without structure
i. and split for the four L1s.
Chart 4:
On the basis of his data, Oshita criticized
the existing explanations for the occurrence of passivized unaccusatives.
Oshita showed that the passivized unaccusatives cannot be regarded as overgeneralized
adjectival passives (as suggested by Hubbard 1994) because the majority
of examples gleaned from his research corpus “express an action, not a
state as in the case of adjectives” Oshita (1995:17). Oshita also rejected
the view (voiced by Yip 1994 & 1995) that learners re-analyse unaccusatives
as transitives because if that were the case, he argued, “transitivized”
unaccusatives should appear at least as often as “passivized” unaccusatives.
However, the ratio of the relevant structures ii. and ix. in his research
corpus is 35 to 13. Furthermore, Oshita disagrees with Zobl’s (1989) analysis
that learners associate the lack of a logical semantic subject with the
passive morphology, since the structure postverbal NP plus “passive” is
attested only twice in his corpus. He argued that if Zobl’s interpretation
were correct, constructions like vi.- viii. where the subject is replaced
by an expletive should occur more frequently. The possibility that passivized
unaccusatives could be attributed to direct transfer of similar L1 structures
to the L2 can also be discounted. Yip 1995 already reported that Chinese
does not offer a transferable grammatical structure and yet passivized
unaccusatives occur in Chinese interlanguage. Oshita’s work confirmed these
observations in that Japanese and Korean learners used passivized unaccusative
as exemplified in structure ii. even more frequently that Italian and Spanish
speakers in whose languages transferable structures do exist.
Moreover, Oshita interpreted the statistically
significant crosslinguistic differences he observed for Italian/Spanish
and Korean/Japanese L1 in structures ii. and iv/v. respectively as evidence
for the psychological reality of “null expletives”. Disregarding the correct
construction with the expletive there in iii., he found that speakers of
Italian and Spanish, in whose L1s overt and covert (null) expletives exist,
show a strong preference for the expletive constructions iv. and v. The
contrary is true for the speakers of Korean and Japanese, which in Oshita’s
opinion presents convincing evidence for the nonexistence of expletives
in those languages.
After the discussion of the previous explanations
for learners’ difficulties with unaccusatives, Oshita proposed a developmental
hypothesis which is intended to account not only for learners’ errors but
also explains why the majority of tokens occurs in the correct structure
NP V. Following Zubizarreta (1987) and Levin & Rappaport Hovav (1995)
he posits two levels of representation in the lexicon, lexico-semantic
representation and argument structure. The former encodes a verb’s semantic
characteristics while the latter contains the verb’s syntactic configuration.
The two levels are connected by systematic regularities expressed by linking
rules. Levin & Rappaport Hovav (1995), for example, suggest four linking
rules (LRs): Directed Change LR, Existence LR, Default LR, and Immediate
Cause LR for the projection of A-structure from the semantic representational
level. The details need not concern us here. Oshita proposed that learners,
when they first acquire unaccusative verbs, wrongly adopt a single LR which
allows them
to externalize the single argument of any
intransitive verb. In this manner Oshita’s account is similar to Zobl’s
(1989). In contrast to Zobl (1989), who suggested that the restructuring
of the learners’ grammar is triggered by the acquisition of the passive
construction and the identification of the canonical mapping structure
of Enlish, Oshita holds the opinion that the realization of the semantic
intricacies associated with unaccusative verbs induce advanced learners
to modify their earlier analysis. The processes which are involved in this
re-analysis lead the learners to produce more complex and often nontarget-like
surface structures.
The studies reviewed in this paper clearly
demonstrate two points. Firstly the Unaccusative Hypothesis is right. Secondly
some unaccusative verbs present a certain difficulty to learners. The following
section is intended to explain the deliberate vagueness of the second point.
While the phenomenon has been recognized
by researchers and teachers (Parker M. G. 1997, Melles G. B. 1997, see
appendix) all over the world, it is not very well documented. The research
indicates that the acquisition of unaccusatives poses problems for learners
from many different language backgrounds (e.g. Richards 1983, Zobl 1989,
Putzer 1994, Yip 1994, Oshita 1995). However, it is not clear
how much of this learnability problem is universal and how much is due
to L1 interference. Sorace (1991, 1993) and Oshita (1995), for example,
observed marked acquisition differences depending on the native language
background.
Furthermore, it should be noted that the
data samples presented in the studies discussed in this paper are quite
heterogenous in nature. Basically, there are three different kinds of data:
grammaticality judgement data (Kellerman 1978, Sorace 1991 & 1993,
Yip 1994 & 1995, Hirakawa 1995), elicited written production data (Hirakawa
1995) , and corpus written production data (Zobl 1989, Putzer 1994, Oshita
1995). The disparity of the data, some of which was collected in connection
with research questions totally unrelated
to the acquisition of unaccusatives (Kellerman 1978, Richards 1983, Putzer
1994), makes generalizations somewhat difficult.
A closer look at the raw data used in the
study by Oshita (1995) reveals that his findings are not as conclusive
as he suggests. The raw data which is illustrated in chart 4 was taken
and the ratio of tokens per structure (i.-x.) to the total number of tokens
per language was calculated. Chart 5 shows the results. In comparison to
chart 4 one sees that a general pattern is much harder to determine. Even
if the results for Korean, which are not really comparable because of the
small size of the sample6, are left aside,
no striking crosslinguistic pattern emerges.
Chart 5:
Another fact that Oshita’s study does not
take into consideration is possible lexical variation.
The two charts above show the share that
the ten unaccusative verbs used by Oshita (1995) have of the total number
of tokens (Chart 6) and the share they have of the nontarget-like structures
(Chart 7). Especially interesting are the verbs arrive (in dark blue),
whose share decreases from roughly a third to about one ninth, and
happen (in light green), whose share increases from a quarter to roughly
a third. Like the Hirakawa study (1995) this suggests, that for learners
not all ergative verbs are created equal. As for Oshita’s (1995) developmental
hypothesis that advanced learners experience more problems with unaccusatives,
it is hard to judge solely on the basis of Kellerman’s (1978) study, which
was not designed for this purpose.
What is needed to shed more light onto the
learnability problems posed by unaccusative verbs is a study which combines
longitudinal and latitudinal aspects. In other words it should examine
a crosssection of learners from several L1 backgrounds where each L1 section
also comprises learners at several stages of competence. The most practical
design for such a study would probably be modelled after Hirakawa (1995),
using elicited production tasks and grammaticality judgement tasks. Lexical,
L1, and developmental influences would have to be carefully disentangled
to yield a more complete picture of the complex problem of the acquisition
of unaccusative verbs.
Baker, M. (1983). Objects, themes, and lexical rules in
Italian. In Levin, L., Rappaport Hovav, M. &
ENDNOTES:
Zaenen, A. eds. (1983). Papers
in lexical-functional grammar. Bloomington: Indiana University
Linguistics Club.
Belleti, A. & Rizzi, L. (1981). On the syntax
of `ne': Some theoretical implications. The Linguistic Review
1: 117-155.
Besten, H. den. (1985). Some remarks on the ergative
hypothesis. In Abraham, W. (1985). Erklaerende
Syntax des Deutschen. Tuebingen,
p. 53-74.
Burzio, L. (1981). Intransitive verbs and Italian
auxiliaries. Doctoral Dissertation, MIT, Cambridge Mass.
Revised as Burzio (1986).
Burzio, L. (1986). Italian syntax: A government-binding
approach. Dordrecht: Reidel.
Dixon, R. M. W. (1994). Ergativity. Cambridge:
Cambridge University Press.
Fellbaum, C. (1986). On the middle construction in
English. Bloomington, IN: Indiana University
Linguistics Club.
Gass, S. & Schachter, J. (1989). Linguistic perspectives
on second language acquisition. Cambridge:
Cambridge University Press.
Grewendorf, G. (1989). Ergativity in German. Dordrecht:
Foris.
Hubbard, P. L. (1994). Non-transformational theories
of grammar. In Odlin, T. (1994). Perspectives on
Pedagogical Grammar.
Cambridge: Cambridge University Press.
Internet
1. The Lexicon of Linguistics. http://wwwots.let.ruu.nl/~Hans.Leidekker/lexicon/ll.html
Internet
2. Linguist List 7.1355: Ergatives in language acquisition.
http://www.sfs.nphil.uni-tuebingen.de/linguist/issues/7/7-1355.html
Juffs, A. (1996). Learnability and the lexicon: Theories
and second language acquisition research.
Philadelphia, PA: John Benjamins
Publishing Co.
Kellerman, E. (1978). Giving learners a break: Native
language intuitions as a source of predictions about
transferability. Working Papers
on Bilingualism. vol. 15: 59-92
Keyser, S. & Roeper, T. (1984). On the middle and
ergative constructions in English. Linguistic Inquiry
15: 381-416.
Levin, B & Rappaport Hovav, M. (1995). Unaccusativity:
At the syntax-lexical semantics interface.
Cambridge, Mass.: MIT Press.
Newmeyer, F. J. (1986). The politics of linguistics.
Chicago, Ill.: University of Chicago Press.
Oshita, L. (1995). L2 acquisition of English unaccusative
verbs: Structural, crosslinguistic, and
developmental aspects.
Unpublished Manuscript. University of Southern California.
Perlmutter, D. M. (1978). Impersonal Passives and the
Unaccusative Hypothesis. In Proceedings of the
fourth annual meeting of
the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society.
Perlmutter, D. M. (1983). Studies in relational grammar
1. Chicago, Ill.: University of Chicago Press.
Perlmutter, D. M. & Postal, P. M. (1984). The 1-advancement
exclusiveness law. In Perlmutter, D. M. &
Rosen, C. eds. (1984). Studies
in relational grammar 2. Chicago, Ill.: University of Chicago Press.
Perlmutter, D. M. & Rosen, C. (1984). Studies
in relational grammar 2. Chicago, Ill.: University of
Chicago Press.
Pullum, G. K.(1988).Topic ... comment: Citation etiquette
beyond thunderdome. Natural Language and
Linguistic Theory. 6:579-588.
Pullum, G. K. (1991). The great eskimo vocabulary
hoax: And other irreverent essays on the study of
language. Chicago: University
of Chicago Press.
Putzer, O. (1994). Fehleranalyse und Sprachvergleich:
Linguistische Methoden im
Fremdsprachenunterricht am
Beispiel Italienisch - Deutsch. Ismaning: Hueber.
Richards, J. C. (1983). A noncontrastive approach to
error analysis. In Wallace Robinett, B. & Schachter,
J. (1983). Second language
learning: Contrastive analysis, error analysis, and related aspects.
Ann
Arbor: University of Michigan
Press.
Rutherford, W. (1988). Second Language Grammar: Learning
and teaching. London: Longman.
Salkie, R. (1990). The Chomsky update: Linguistics
& politics. London: Unwin Human Ltd.
Scarcella, Robin. (1984). Cohesion in the writing
development of native and non-native English speakers.
Ph. D. diss. USC.
Schachter, J. & Rutherford, W. (1983). Discourse
function and language transfer. In Wallace Robinett, B.
& Schachter, J. (1983).
Second language learning: Contrastive analysis, error analysis, and
related
aspects. Ann Arbor: University
of Michigan Press.
Sorace, A. (1991). Unaccusativity and auxiliary choice
in non-native grammars of Italian and French:
assymetries and predictable
indeterminacy. Paper presented at the 16th Boston University Conference
on Language Development. Boston:
Unpublished Manuscript.
Sorace, A. (1993). Incomplete vs. divergent representations
of unaccusativity in nonnative grammars of
Italian. Second Language Research
9, 1: 22-47.
Tchekhoff, C. (1978). Aux fondements de la syntaxe:
L’ergatif. Paris: Presses Universitaires de France.
Wallace Robinett, B. & Schachter, J. (1983). Second
language learning: Contrastive analysis, error
analysis, and related aspects.
Ann Arbor: University of Michigan Press.
Yip, V. (1994). Grammatical consciousness-raising and
learnability. In Odlin, T. (1994). Perspectives on
Pedagogical Grammar.
Cambridge: Cambridge University Press.
Yip, V. (1995). Interlanguage and learnability: From
Chinese to English. Amsterdam: John Benjamins
Publishing Co.
Zobl, H. (1989). Canonical typological structures and
ergativity in English L2 acquisition. In Gass, S. &
Schachter, J. eds. Linguistic
perspectives on second language acquisition. pp. 203-221. Cambridge:
Cambridge University Press.
Zubizarreta, M. L. (1987). Levels of representation
in the lexicon and in the syntax. Dordrecht: Foris
Publications.
1 For an exhaustive account of the unaccusative
hypothesis’ history see Pullum 1991: pp.147-158.
2 The major sources mentioned by Richards
are: French 1949, Duskova 1969, Arabski 1968, Estacia 1964, Richards 1968,
Bhaskar 1962, Grelier n.d., Aguas 1964.
3 The two sentences were originally reported
by Scarcella 1983.
4 The L1s were: Spanish (1), Hebrew (1),
Korean (2), Chinese (2), Indonesian (2), German (1), and Greek(1).
5 The ten verbs were: appear, arise, arrive,
die, disappear, exist, fall, happen, occur, and rise.
6 The size of the Korean sample is about
one fifth of each of the other language samples.