Optionality in non-native grammars: L2 acquisition of German constructions with absent expletives
Transcript of Optionality in non-native grammars: L2 acquisition of German constructions with absent expletives
This is an Author's Accepted Manuscript of the article in Bilingualism: Language and
Cognition 8 (3), 2005, 177–193 available at: http://dx.doi.org/10.1017/S1366728905002245
The publisher should be contacted for permission to re-use or reprint the material in any
form.
Optionality in non-native
grammars: L2 acquisition
of German constructions
with absent expletives*
ALDONA SOPATA Adam Mickiewicz University, Poznań, Poland
This paper investigates the knowledge of constructions with absent expletives by advanced and high-proficiency non-
native speakers of German whose first language is Polish. German grammar is known to license null subjects due to the
strength of AGRP but not to identify them. Therefore only expletive subjects can be absent in German, except for Topic-
drop and, crucially, the expletive subjects have to be absent in certain cases due to the Projection Principle. The
knowledge of this phenomenon by second language (L2) learners has been investigated by two methods, elicited written
production task and grammaticality judgment tests. High-level non-native speakers of German differ significantly from
native speakers in both types of tasks. The differences are clearly not the result of transfer. The results reported here
reveal permanent optionality in L2 grammars suggesting a deficit in the grammatical representations of L2 learners.
1. Introduction Syntactic optionality in the grammars of proficient non-
native speakers raises questions concerning the nature of
adult second-language (L2) acquisition. A possible
explanation of this phenomenon concerns changes in the
language faculty affecting the way languages are acquired
in post-childhood. A broad range of hypotheses on the
fundamental nature of L2 knowledge has been developed. One of the approaches concerning the problem of the
role of Universal Grammar (UG) in L2 acquisition is the
hypothesis put forward by Vainikka and Young-Scholten
(1994, 1996), which claims full access to UG and transfer
from the L1 to contribute to L2 development. Vainikka and
Young-Scholten argue that interlanguages are initially
defined by L1 lexical categories. Functional projections
subsequently evolve, triggered by L2 input. According to
their hypothesis, the full success of L2 learners is possible
in all language areas. Full access to UG is also claimed by Schwartz (1998)
and Schwartz and Sprouse (1994, 1996, 2000). Syntactic
representations in the interlanguage are unimpaired and
accessible for the purposes of reparameterisation. The
initial state of the L2 acquisition process is said to be the
full L1 grammar. The differences between the native and
non-native grammars are claimed to be a result of mapping
* I would like to thank Jurgen¨ Meisel and other researchers in
Collaborative Research Center on Multilingualism in Hamburg for
their valuable discussion of the issues and problems involved in this
research. I am grateful to Lynn Eubank and two anonymous
reviewers for their valuable comments on the early versions of this
paper. All remaining errors are my own.
Address for correspondence Adam Mickiewicz University, ul. Pasieka 6/2, 61-657 Poznan, ́Poland E-mail: [email protected]
difficulties (Lardiere, 1998a, b, 2000) from syntactic
fea-tures to overt forms (viz. the Missing Surface
Inflection Hypothesis of Haznedar and Schwartz, 1997;
Prevost and White, 2000) or of L1 transfer. This
hypothesis predicts that the functional categories are
initially transferred from L1 and then parameters are
reset. L2 acquisition should result in the successful
acquisition of phenomena connected with functional
categories in L2 provided that such problems as L1
transfer or mapping difficulties are not involved. Another way of explaining the differences between
native and L2 grammars is attributing them to some
kind of impairment of the language faculty. Tsimpli and
Smith (1991), Tsimpli and Roussou (1991), and Smith
and Tsimpli (1995) argue that the features of functional
categories become connected with the morphophonolo-
gical surface paradigm in the L1 acquisition process,
and that this connection cannot be reset in L2
acquisition. Whereas unparameterised UG principles
constrain inter-language, parameterised UG principles
become unavail-able to second language learners. The impairment of the syntactic module of language
faculty is also claimed by Hawkins and Chan (1997).
According to their hypothesis, L2 acquisition is constrained
by UG, but functional categories are restricted to the L1
repertoire of interpretable features. L2 learners cannot reset
the features from the L1 value to the L2 value. The
checking of features is substituted in L2 by associ-ative
memory processes, the use of which in turn results in
optionality throughout the L2 acquisition process. Beck (1997, 1998) and Eubank, Bischof, Huffstutler,
Leek and West (1997) argue that lexical and functional
2 Aldona Sopata
categories are transferred initially from L1, but crucially
functional categories are transferred without the feature
values. The strength of values of V-features cannot be
specified in the course of L2 acquisition, which results
in permanent optionality. This hypothesis predicts
therefore that the interlanguage will be UG constrained
but the phenomena connected with the specification of
features (e.g. verb raising) will be optional even in the
interlanguage of very advanced L2 learners. These hypotheses can be summed up in the following
way. All researchers cited above claim that the
interlangu-age is UG-constrained, but to varying
degrees. Predictions differ in respect to the development
of functional cate-gories in L2 acquisition. On the one
hand, we have those hypotheses claiming the possibility
of specification of features (triggered directly by L2
input or reset from the transferred L1 value). By
contrast, other hypotheses argue for the impairment of
the syntactic module of language faculty. Values of
features are transferred from L1 and not reset or remain
unspecified. The failure of specification of features
results in the optionality in related syntactic phenomena. Empirical research can therefore show if the develop-
ment of phenomena connected with the specification of
features results in the successful acquisition of them or
remains optional to the advanced stages in SLA (second
language acquisition). The acquisition of constructions
with absent expletives in L2 German by L1 Polish speakers
serves as a good testing ground for this problem.
2. Linguistic background Expletives are semantically empty elements which play
only a syntactic role. Safir (1984, p. 203) defines them
in the following way: “An NP is expletive if it does not
count as an argument for the Theta Criterion”.1
The presence of expletives in a language was at first
claimed to depend on its status as a null subject language
or as a non-null subject language. Expletives are absent in
null subject languages, which allow the empty category
pro, while in non-null subject languages, which do not
allow pro, expletives are present. German is an interesting case in this respect.
Referential subjects cannot be null in German: (1) Er sagte, dass er tanzt.
he said that he dances
“He said that he danced.” 1 In German it is necessary to distinguish between quasi-arguments
(Chomsky 1981, p. 325; Grewendorf 1989, p. 149) (cf. i) and non-arguments – expletives.
i. Gestern regnete *(es) hier.
yesterday rained it here “Yesterday it rained here.”
(2) *Er sagte, dass tanzt.
he said that dances “He said that dances.
There are expletives in German (EX = EXPLETIVUM): (3) Es wird getanzt.
EX is danced
“It is danced.” In some cases, however, the expletives have to be absent: (4) Er sagte, dass (*es) getanzt wird.
he said that EX danced is
“He said that it is danced.” Rizzi (1986) proposes a modified version of pro. He
denies the close relation between the licensing of pro in
one language and its inflexional richness. The richness
of a inflexional paradigm is said to play a role only in
the identification of pro, but not in its licensing. Two
principles are proposed by Rizzi: (5) a. Licensing principle for pro (Rizzi 1986, p. 519)
pro is governed by X0 y.
b. Identification of pro (Rizzi 1986, p. 520)
Let X be the licensing head of an occurrence of
pro: then pro has the grammatical specification
of the features on X coindexed with it. Muller¨ and Rohrbacher (1989) investigate numerous
languages from various language families to see how the
principles of licensing and identification of pro function.
They distinguish four language groups: (6) a. languages which do not allow pro because INFL
is not a licensing head b. languages which allow pro but INFL is not richly
enough specified to supply an argument with the
necessary grammatical features c. languages in which INFL can determine only
some but not all features d. languages which make the reconstruction of all
grammatical specifications possible An example of the first group can be English, where INFL
is not a pro-licensing head. In German INFL is a pro-
licensing head, but it cannot specify phi-features (person,
number, gender), which results in the fact that only
expletives can be pro (Safir, 1984; Sternefeld, 1984;
Grewendorf, 1989). German therefore belongs to the
second group according to the classification of Muller¨ and
Rohrbacher. The third group can be represented by
Icelandic, in which the pro-licensing head can supply the
argument with some (but not all) phi-features. In Icelandic
not only expletives but also quasi-arguments can be empty
(e.g. with weather verbs). Languages in which pro is
formally licensed and INFL is specified richly enough to
allow for referential elements to be empty make up
the fourth group. Italian is a well-known example of this group of languages.
The split of the null subject parameter into two
principles helps one to understand the empirical data. It
does not, however, define what the necessary conditions
are under which a language allows for null subjects. The
intuition that null subjects appear in languages with rich
agreement features is quite strong. The difficulty lies in
the formulation of conditions that are theoretically and
descriptively adequate. Speas (1994) proposes a solution
to this problem. Speas follows a proposal of Rohrbacher (1992), who
claims that in languages with strong agreement – that is,
with agreement that licenses null arguments – each
agreement morpheme has its own lexical entry; by con-
trast, languages that have weak agreement – that is,
agree-ment that does not license null arguments – the
morphemes do not have independent lexical entries. In
languages with weak agreement, verbs are listed in the
lexicon in verbal paradigms. Speas (1994, p. 185) calls
this Rohrbacher’s Generalisation: (7) a. Strong morphemes have individual lexical entries.
b. Weak morphemes do not have individual lexical
entries.
In languages with strong agreement, an agreement
morpheme heads the AGR projection. In languages with
weak agreement, the agreement morpheme is a part of
the inflexional paradigm. The descriptive generalisation
as follows (Speas, 1994, p. 186): (8) a. A language has null subjects if AGR is base
generated with a morpheme in it. b. A language cannot have null subjects if AGR is
base-generated on the verb. c. A language has null subjects if it has no AGR.
Speas (1994, p. 186) explains this generalisation by
using a principle of Economy, a sub-case of the general
principles of Economy of representation, which is stated
in the following way: (9) Project XP only if XP has content. A projection without content is a representation that
cannot be interpreted. A projection in which the head
and the specifier are empty (i.e. are without content)
would violate the principle. In non-null subject languages, in which the agreement
morphemes cannot head the AGR-projection, the overt
subjects have to license the AGR-projection. Therefore,
either an NP must move to Spec,IP or an expletive must be
inserted. In null subject languages, in which AGR is base
generated with a morpheme in it, AGR-projection has
content and there is no necessity for the specifier of AGR-
projection to be filled. In languages such as Japanese the
AGR-projection is not projected at all. Speas (1994,
Optionality in non-native grammars 3 p. 188) claims that the AGR-projection is necessary
only in languages which have some sort of residual
agreement. Languages like Japanese and Chinese do
have functional heads such as TENSE and ASPECT, but
they lack the head AGR. Speas’ proposal explains the licensing condition for
pro; it does not, however, eliminate the need for the
identification principle of pro. Speas (1994, p. 191)
states explicitly: There can be languages in which AGR has its own lexical
entry and hence suffices to license the AGR projection, but
whose AGR features are not strong enough to identify pro.
German is such a case. The fact that German allows dropped expletives shows that German licenses pro. Referential subjects, however, cannot be dropped in German except for cases of Topic-
drop2 (see Sigurdsson, 1993 for various strategies of
identification of pro). That referential subjects are required indicates that pro cannot be identified because the Agreement features are not richly enough specified.
Polish is a null subject language. Like Italian, Polish
allows referential subjects to be null: (10) Mowie´ po polsku.
speak Polish “I
speak Polish.” AGR is strong in Polish, so AGR affixes are sufficient to
license the AGR projection; therefore subjects may be null.
Moreover, AGR affixes in Polish are rich enough to assign
phi features, so referential subjects can be null. As in other
“genuine pro-drop languages” (Sigurdsson, 1993, p. 247),
Polish has no expletives. Given Speas’ analysis, this is not
surprising because the verbal paradigm in Polish is very
rich. It contains not only person- and number-features, but
also gender-features in the past forms. The effects of the strong or weak value of AGR in
languages are shown in Table 1. Languages with no
AGR (Chinese, Japanese) are omitted. German has null subjects only when the subject is
expletive, but not all expletives are null in German.
Speas (1994) proposes an explanation of the distribution
of overt vs. null expletives in German using the
Projection Principle. Overt expletives show up only
clause-initially in matrix clauses, but not claused
internally or in any embedded clause: (11) Es wurde gestern auf dem Schiff getanzt.
EX is yesterday on the ship danced
“There was dancing on the ship yesterday.” 2 Topic-drop is a phenomenon restricted to colloquial German. Null
subjects are identified here by discourse-identified operator. See Rizzi (1994) for detailed discussion.
4 Aldona Sopata
Table 1. The effects of strong or weak value of AGR
AGR – strong (morpheme generated
under AGR so pro licensed)
AGR – weak (morpheme
generated onV so pro not licensed) A. Phi features – rich (A1) (A2)
No expletive in Spec,AGR Referential or expletive subject required in Spec,AGR
Null referential subject optionally permitted in Spec,AGR
[This is Polish] B. Phi features – not rich (B1) (B2) No expletive in Spec,AGR Referential or expletive subject required in Spec,AGR Referential subject required in Spec,AGR
[This is standard German]
(12) Gestern wurde (*es) auf dem Schiff getanzt.
yesterday was EX on the ship danced “There was dancing on the ship yesterday.”
(13) Er sagte, dass (*es) getanzt wurde.
he said that EX danced was
“He said that it was danced.” In (11) we see that overt expletives show up in Spec,CP,
but they do not appear in Spec,IP (AGRP) in (12) and (13).
Spec,AGRP may be null when the subject is non-
referential because the agreement morphology in German
suffices to license the AGR projection. According to Speas
(1994, p. 191), examples (12) and (13) do not have a null
expletive. Rather, Spec,AGRP is here truly empty because
the head of AGRP is an affix which has its own lexical
entry and suffices to license AGRP. An expletive is necessary in Spec,CP if COMP does
not have a contentful item base-generated in it but only
the moved verb occupies C (e.g. example (12)). Verb
movement is not in itself sufficient to license a
projection and the functional projection must be
licensed by distinct content prior to verb movement. Considering the AGR strength parameter and the
Projection Principle, the situation in Polish and German
can be presented as follows: (14) Polish: No expletive in Spec,AGR
Null referential subject optionally
permitted in Spec,AGR (15) German: No expletive in Spec,AGR
Referential subject required in Spec,AGR REFERENTIAL OR EXPLETIVE SUBJECT REQUIRED IN SPEC,CP IF ONLY THE MOVED VERB OCCUPIES C
Children acquiring German as their L1 begin with a mental grammar allowing them to produce subjectless
constructions.3 In the course of L1 acquisition they
come to a developmental stage in which referential subjects are required and expletive subjects are absent in Spec,AGR. This development can be explained with the specification of the features of AGR.
Likewise, L2 learners have to specify the strength of
the AGR projection in the acquisition process. Learners
of German whose native language is Polish have to find
out that the AGR projection in German is strong and
therefore licenses null subjects. The AGR affixes are,
however, not sufficiently rich to assign phi features to
referential subjects, which results in the fact that only
expletives and not referential subjects can be null.4
3. Considerations for SLA The possibility of specifying features under a functional
head in the course of L2 acquisition has received
increasing attention in SLA research. As already
indicated, the present study investigates the possible
effects of specifying the strength feature under AGR in
the process of SLA. The various possibilities of AGR
specification and its effects in languages with rich and 3 See Weissenborn (1992) for detailed discussion.
4 There are other proposals of explaining the null subject phenomenon in
the literature (e.g. Haider, 1994; Vikner, 1995). Rohrbacher (1999, p.
246) argues, on the basis of Speas’ theory, that only the phono-logical
content and not the semantic content of a phrase is relevant for the
Principle of Economy of Projection. He defines “strong” agree-ment in
terms of minimal distinctive marking of [1st] and [2nd] person.
Eguzkitza and Kaiser (1999) and Kaiser (2002) deny the connection
between non-referential pro and the null subject phenomenon. They
connect them, however, with the V2 parameter. Adopting this view
would change the linguistic background of the study, but it would still
lead to similar conclusions in the aspect of L2 acquisition.
Optionality in non-native grammars 5 Table 2. The various possibilities of AGR specification
AGR – strong (morpheme
generated under AGR, so
pro licensed)
AGR – weak (morpheme
generated on V, so pro
not licensed)
AGR – unspecified
(morpheme generated either on
V or licensed) A. Phi features – rich (A1) (A2) (A3) No expletive in Spec,AGR Referential or expletive subject Expletive optionally permitted required in Spec,AGR in Spec,AGR Null referential subject Null referential subject optionally permitted in optionally permitted in Spec,AGR Spec,AGR REFERENTIAL OR EXPLETIVE REFERENTIAL OR EXPLETIVE SUBJECT REQUIRED IN SPEC,CP SUBJECT REQUIRED IN SPEC,CP IF ONLY THE MOVED VERB IF ONLY THE MOVED VERB OCCUPIES C OCCUPIES C [This is Polish except for the V2 effect with subject in
Spec,CP]
B. Phi features – not rich (B1) (B2) (B3) No expletive in Spec,AGR Referential or expletive subject Expletive optionally permitted required in Spec,AGR in Spec,AGR Referential subject required in Referential subject required in Spec,AGR Spec,AGR REFERENTIAL OR EXPLETIVE REFERENTIAL OR EXPLETIVE SUBJECT REQUIRED IN SPEC,CP SUBJECT REQUIRED IN SPEC,CP IF ONLY THE MOVED VERB IF ONLY THE MOVED VERB OCCUPIES C OCCUPIES C [This is standard German]
not rich phi features are shown in Table 2. It contains
not only the cases that occur in mature languages, but
also a possibility proposed by Beck (1997, 1998) in
regard to verb raising. Beck (1998, p. 317) argues that
an impairment to the strength feature under the
functional head would effectively cause verb raising to
become optional because the strength feature either
requires or prohibits verb raising in mature languages.
The same situation is possible in the case of morpheme
generation. If AGR is unspecified, a morpheme is
generated either on the verb or under AGR. This would
cause the related phenomena to become optional. In light of the discussion presented above, it is now
possible to consider different scenarios of L2 acquisition
of German constructions with absent expletives by
Polish native speakers. Adopting the full- access
hypothesis of Schwartz and Sprouse, we should expect
full transfer from the L1 by early learners. This means a
transfer of the strength feature under AGR from L1
Polish. As the value of this feature in Polish is strong,
this value should be transferred into early interlanguage
and cause the requirement of no expletives in Spec,AGR
(i.e. no expletives sentence-internally). Crucially, this
value should stay the same during the whole course of
L2 acquisition, as the subsequent restructuring of the
grammar occurs, according to Schwartz and Sprouse,
only to the extent to which positive data are available
from L2. Consequently, according to this hypothesis, L2
learners of German with Polish as a native language will
have the strong value of AGR, a value which prohibits
the appearance of expletives in Spec,AGR at each stage
of L2 acquisition. The full-access proposal made by Vainikka and Young-
Scholten differs from that of Schwartz and Sprouse to
some extent, but it makes the same predictions about the
acquisition of constructions with absent expletives at an
advanced stage of the acquisition of German by Polish L2
learners. After the very early stage of only lexical
projections in interlanguage, L2 learners will first develop
underspecified functional projections which will later
become fully specified. According to this view, at the
underspecified FP stage the (obviously underspecified)
value of AGR will cause the optional use of expletives in
Spec,AGR (Table 2, column 3). However, the more
advanced learners at the specified FP stage will have the
strong value of AGR (Table 2, column 1). Therefore,
6 Aldona Sopata
Table 3. Predictions based on the various SLA approaches
Schwartz & Sprouse from the beginning to the end state of L2 acquisition
AGR – strong
⇓ no expletive in Spec,AGR
Vainikka & Young-Scholten early stage* advanced stage
AGR – unspecified AGR – strong
⇓ ⇓ expletive optionally permitted in Spec,AGR no expletive in Spec,AGR
Beck from the beginning* to the end state of L2 acquisition
AGR – unspecified
⇓ expletive optionally permitted in Spec,AGR
optional insertion of expletives in Spec,AGR should
diminish over the course of L2 acquisition and finally
result in the successful acquisition of the phenomena
connected with the appropriate value of AGR. In contrast to hypotheses claiming full access to UG,
there are approaches that argue for an impaired access of
adult L2 learners to UG. With regard to the specification of
features of FP in the course of L2 acquisition the most
explicit proposal has been made by Beck (1997, 1998). She
predicts that the specification of the strength feature under
a functional head is impossible in L2 acquisition, and that
an impairment of this feature would effectively cause
optionality in the phenomena controlled by it. In our case,
then, the hypothesis would predict an unspecified value of
AGR during the whole course of L2 acquisition, causing
the optional appearance of expletives in Spec,AGR
throughout. Predictions based on these approaches are
summarised in Table 3.
4. Hypotheses The purpose of the study is to investigate the
interlanguage of Polish L2 learners of German regarding
their use of expletives claused internally. This can prove
or disprove the predictions made by the hypotheses
regarding the role of UG in L2 acquisition. Hypotheses
for the study are thus presented as follows: (16) Expletives do not appear in Spec,AGR (i.e. clause
internally) in the interlanguage of German L2
learners. (17) Expletives appear optionally in Spec,AGR (i.e.
clause internally) in the interlanguage of German
L2 learners.
5. The study 5.1. Subjects A total of 176 subjects participated in the study. There was
a control group of 25 native speakers of German and four
groups of Polish learners of German as L2. They learned
German primarily in the classroom/university setting. The subjects were divided into groups according to their
general language proficiency. The first group was made up
of final year pupils of two high schools. They had 3–6
German language lessons per week. The dominant teaching
method was the communicative one. Contact with the
target language in the group was relatively weak. The second, third and fourth groups were respectively
first, second and third year students at the German
Philology Department and at the German Department of the
Teacher Training College at Adam Mickiewicz University
in Poznan, Poland. Each year these students participate in
an intensive language course: They have 12 hours of
language classes per week, and most of their theoretical
lessons (e.g. Descriptive Grammar of German, German
Literature, German Culture) are also conducted in German.
Contact with the target language is quite intensive. Various
learning materials, intended to develop all language skills,
are used. Most of the subjects, especially participants of
groups 3 and 4, have a longer stay in Germany behind
them. They can be considered as quite highly motivated
with a great deal of interest in the Q1 German language.
Tables 4-7 provide general information about the
subjects, their exposure to the target language and their
knowledge of other languages.
5.2. Materials and procedures It is desirable to validate accounts of L2 competence by
using diverse evidence. Grammaticality judgments are
Table 4. General information about subjects
Number of Age
Group persons Sex mean from–to
Optionality in non-native grammars 7 Table 7. Knowledge of other languages in groups 3 and 4 Group 3 Group 4
1 49 female – 37 17.9 17–18
male – 12
2 37 female – 28 20.7 19–25 male – 9
3 30 female – 26 21.8 21–26 male – 4
4 35 female – 29 23 21–26 male – 6
Table 5. Information about the exposure to
target language
English – 24 people
(worse than German)
English – 1 person
(better than German)
Russian – 3 people
(worse than German)
French – 2 people
(worse than German)
English – 31 people
(worse than German)
Russian – 15 people
(worse than German)
French – 3 people
(worse than German)
Spanish – 1 person
(worse than German)
Italian – 1 person
(worse than German)
Mean time of Age at language exposure the onset Group Level (years) (years)
intermediate
1 Zertifikat Deutsch 6.9 9–15 als Fremdsprache
upper-intermediate
2 Deutsche 7.5 9–16 Mittelstufen-prufung¨
advanced
3 Kleines Deutsches 7.9 11–16 Sprachdiplom
4 very advanced 8.3 9–20
Table 6. Knowledge of other language in groups 1 and 2 Group 1 Group 2
grammaticality judgments and elicited written production tasks.
The elicited written production task consisted of various
reformulation tasks. Crucially, the reformulation of
sentences required the use of a sentence structure where the
expletive subjects had to be absent. In order to get intuitive
answers from the participants, the tasks contained many
sentences not connected with the problem under
investigation. They functioned as distractions. The tasks
were explained in both Polish and German. The first task was a completion task. The participants
were asked to complete sentences according to the
following sentences: (18) Es wird an die Oma gedacht.
Ex is of the grandmother thought “Grandmother is being thought of.”
Er sagte, dass.................... he said that
Target form: Er sagte, dass an die Oma he said that of the grandmother
English – 10 people
(worse than German)
English – 4 people
(better than German)
Russian – 31 people
(worse than German)
Russian – 4 people
(better than German)
English – 27 people
(worse than German)
English – 1 person
(better than German)
Russian – 5 people
(worse than German)
French – 4 people
(worse than German)
gedacht wird. thought is “He said that grandmother is being thought of.”
The second task was a transformation task. Students
were asked to transform active sentences into passive
sentences. The word given in the second line exhorted
the use of the target structure. postulated to be combined with other measures on the
same or comparable items and subjects (Chaudron, 1983,
p. 369; Birdsong, 1989, pp. 116–118; Mandell, 1999). By
using a multiplicity of tasks, it might be possible to tease
apart knowledge from the task effects (Klein and
Martohardjono, 1999, p. 16). Two methods were used in
the study in order to ensure the validity of the results:
(19) Immer reist man mit viel always travels PRONOUN (person) with a lot of
Gepack¨.
luggage “People always travel with a lot of luggage.”
Immer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . always
8 Aldona Sopata
Target form: Immer wird mit viel Gepack¨ always is with a lot of luggage gereist.
travelled “People always travel with a lot of luggage.”
The third task consisted of connecting sentences with
the appropriate conjunction from a given set, e.g. dass
“that”, weil “because”. To fulfil this task, the given
sentences had to be reformulated. (20) Die Jugend jubelt. Es wird ihm geholfen.
the youth cheers EX is him helped “The youth cheers. The people will help him.” Target form: Die Jugend jubelt, weil ihm
the youth cheers because him
geholfen wird. hepled is “The youth cheers because the people will help him.”
The fourth task was a typical reformulation task.
Students were asked to reformulate given sentences in
such a way that they would begin with the given words. (21) Jetzt geht aber schnell ins Bett!
now go-PLURAL but fast to bed “But now go to bed fast.”
Jetzt wird.................... now is Target form: Jetzt wird schnell ins Bett gegangen.
now is fast to bed went “Now go to
bed fast.” Each task contained two sentences involving the use of
ex-pletives and three or four sentences not connected
with the investigated problem. The time was limited to
30 minutes. Students were encouraged to ask for
clarification of lexical problems. Many potential (lexical
and formal) problems were clarified at the outset of the
session in the mother tongue of the participants. The second method used in the study was a test of
grammaticality judgments. Our understanding of
various factors involved in the creation of linguistic
intuitions of L2 learners is very limited. The learning
environment that fosters the development of
metalinguistic knowledge is connected with additional
difficulties. As Sorace (1996, p. 385) puts it: It can be a more complex task, however, to decide about the kind
of norm consulted by learners in the process of producing a
judgment . . . It is difficult to tell whether subjects reveal what
they think or what they think they should think. The elicitation of
immediate judgment responses under well-defined time
constraints may provide a partial solution to this problem. To ensure the elicitation of intuitive judgments in the study
the time was limited to 15 minutes. So the second method
used in the study – grammaticality judgments can be seen
as indicators of internalised grammars of interlanguage. The test was constructed in such a way which should
exclude the potential influence of performance and some
task dependent factors. The tasks were clearly and
explicitly formulated in Polish and German and were
explained orally at the beginning of each session. The
target sentences were mixed up with other sentences not
involving the problem under investigation. Out of 32
sentences, 16 items included the phenomena relevant to
this study. Eight items were connected with referential
subjects, eight items were connected with absent explet-
ives. The test contained sentence pairs with and without
subjects. The sentences were mixed in a way that the
following sentences never belonged to the same group.
This reduced the possibility that the students would be
aware of the knowledge in which they were being tested.
The examples were sentences not connected with the
investigated problem. The students were asked to judge the sentences in regard
to their correctness and to indicate how sure they were
about their judgments. They had to therefore choose from
seven blanks. Three of them could be marked if the
students judged the sentences to be correct. The first blank
was to be chosen when the subjects were absolutely sure
about their judgment, the second when the subjects were
quite sure, and the third when the subjects were quite
unsure about their judgments. They could choose from
amongst three similar possibilities if they judged the
sentences as incorrect. The seventh blank would be marked
if the subjects did not know the answer at all. The
participants were asked to correct those sentences which
they judged to be incorrect. The sentences used in the test of grammatical
judgment can be divided into the following four groups: (i) Grammatical sentences with a preposed phrase
(22) Auf der Autobahn wird sehr schnell gefahren.
on the highway is very fast gone “One can go
fast on the highway.” (ii) Ungrammatical sentences with a preposed phrase
(23) *Auf der Autobahn wird es sehr schnell gefahren. on the highway is EX very fast gone
(iii) Grammatical sentences with a preposed finite or
non-finite clause
(24) Um den Nachbarn nicht aufzuwecken, wird to the neighbour not wakeup is
geschlichen. stolen “You should steal, so as not to wake up
the neighbour.” (iv) Ungrammatical sentences with a preposed finite or
non-finite clause
Table 8. Test of referential subjects Number of subjects Number of subjects with correct with incorrect
Group answers (%) answers (%)
1 23 (47%) 26 (53%) 2 34 (92%) 3 (8%)
3 29 (97%) 1 (3%)
4 35 (100%) 0
C(ontrol) 25 (100%) 0
(25)*Um den Nachbarn nicht aufzuwecken, wird es to the neighbour not wake up is EX geschlichen.
stolen Students were encouraged to ask questions about any formal or lexical problems.
The goal of the study is to investigate the acquisition of
constructions with absent expletives by Polish learners of
German. To get a clear picture of the acquisition process of
this phenomenon, the groups were tested to ascertain,
whether they manage the use of referential subjects in
German, too. The sentences used to test this phenomenon
were pairs of four grammatical and four ungrammatical
items. There were items with subjects preposed by a phrase
(example (26)) and by a clause (example (27)). (26) Nach der Schule kommst du nach Hause.
after the school come you home “After the school you come home.”
(27) Katja glaubt, dass sie klug ist.
Katja thinks that she wise is
“Katja thinks that she is wise.”
Optionality in non-native grammars 9 In ungrammatical counterparts of the sentences the
subject was absent. Those students who did not use the
referential subjects correctly were filtered out. As a
result of this procedure, the possibility that the false use
of absent expletive subjects will be due to the fact that
the null subject parameter was not set yet can be
excluded. The problems of the use of referential and
expletive subjects can therefore be treated separately.
6. Results The first step in the experiment was conducted to filter
out those students who it seemed had not yet mastered
the use of referential subjects in L2 German. The results
of relevant grammaticality judgments are presented in
Table 8. As already mentioned, only the answers of
those participants who use referential subjects correctly
were considered relevant to the further investigation.
They were regarded as such if they correctly judged all
grammatical sentences as correct, correctly judged all
ungrammatical items as incorrect and in addition
corrected them in an appropriate manner. The acquisition of constructions with absent
expletives was investigated by two methods. The results
of the first method, the elicited written production task,
are given in Tables 9 and 10. The use of expletives by
the participants differs in respect to the kind of
sentences. Therefore, the results are divided into two
tables, depending on the kind of sentences involved. The exact number of participants who correctly used
constructions with absent expletive subjects in the created
contexts is given in the tables 9 and 10. However, the
tasks offer only a high possibility for using the
investigated structures, they cannot really exhort their use.
Some answers are therefore irrelevant. Thus, four cases
can be
Table 9. The use of expletives in the elicited production task in sentences with a preposed phrase Correct use Correct use Correct use Correct use Correct use
Correct use in in 75% in 66% in 50% in 33% in 25% Correct use in Not 100% contexts contexts contexts contexts contexts contexts 0% contexts revealing
Group 4/4 3/3 2/2 1/1 3/4 2/3 2/4 1/2 1/3 1/4 0/4 0/3 0/2 0/1 0/0
1 4 4 3 0 0 1 0 0 0 0 11 1 1 1 0 0 0 9 2 13 4 0 0 5 1 1 1 1 0
17 5 0 6 1 1 3 1 3 16 3 1 0 1 1 0 0 0 0
20 3 1 2 2 1 0 0 4 10 7 1 0 1 1 0 0 0 0
18 9 4 2 2 0 0 0 C 25 0 0 0 0 0 0 0 0 0
25 0 0 0 0 0 0 0
10 Aldona Sopata Table 10. The use of expletives in the elicited production tasks in sentences with a preposed finite or non-finite clause Correct use Correct use Correct use Correct use Correct use
Correct use in in 75% in 66% in 50% in 33% in 25% Correct use in Not 100% contexts contexts contexts contexts contexts contexts 0% contexts revealing
Group 4/4 3/3 2/2 1/1 3/4 2/3 2/4 1/2 1/3 1/4 0/4 0/3 0/2 0/1 0/0
1 0 0 0 0 0 1 8 2 2 3
0 0 2 1 1 2 15 2 2 0 1 0 0 1 0 10 5 6 2
1 1 0 1 2 6 23 0 3 0 0 0 0 0 1 13 3 1 1
0 4 0 1 0 6 18 0 4 0 0 0 0 1 1 24 5 0 0
0 0 0 2 0 4 29 0 C 25 0 0 0 0 0 0 0 0 0
25 0 0 0 0 0 0 0
Table 11. The use of expletives in the elicited production
task in sentences with a preposed phrase
100% 75% 66% 50% 33% 25% 0% Group correct correct correct correct correct correct correct
1 78% 7% 7% 7% – – – 2 51% 15% – 18% 3% 3% 9% 3 69% 10% 3% 7% 7% 3% – 4 51% 26% 11% 6% 6% – – C 100% – – – – – –
represented by the 100%-correct usage of constructions
with absent expletives: the participant used four relevant
structures and in all cases the expletive was absent; the
participant used three, two or one relevant structures and
in all of them used the null subjects correctly. The situation
of the 100%-correct usage of construction with absent
expletices also occurred (a) if a participant wrote three
sentences using the construction with absynt expletives
targeted by the task and (b) if all three of his sentences were
correct with respekt to absent expletives and (c) if he did
not write the fourth relevant sentence at all or solved the
fourth task in another way not using the construction with
absent expletive. His result would then have been – correct
use in 100% contexts – 3/3, i.e., he used three relevant
structures, and all three sentences were correct with respekt
to absent expletives. The situation is similar in the case of
the correct usage in 50% and 0%. The exact numbers are
first given in the tables and then they are added in the row
below. Cases where the participants did not use any relevant
construction are regarded as not revealing. Tables 11 and 12
present the results in percentages.
Table 12. The use of expletives in the elicited production
task in sentences with a preposed finite or non–finite
clause
100% 75% 66% 50% 33% 25% 0% Group correct correct correct correct correct correct correct 1 – – 9% 5% 5% 9% 71% 2 3% 3% – 3% 6% 18% 68% 3 – 14% – 3% – 21% 62% 4 – – – 6% – 11% 83% C 100% – – – – – –
The results are also show in Figures 1 and 2.
The ANOVA-test and post-hoc Scheffe´ tests were
conducted. The one-way-ANOVA indicated a significant
difference between the groups in respect to the simple
clauses (F = 5,023455; p < 0,000847).5 The post-hoc
Scheffe´ test revealed a significant difference between the
second and the control group (p < 0,001606). In respect to the compound sentences, the one-way-
ANOVA indicated a significant difference between the
groups (F = 102,8596; p < 0,0000). The differences
between the experimental groups were proved to be
insignificant according to the post-hoc Scheffe´ test.
Crucially, the differences between each experimental
group and the control group are significant:
group 1/control group: p < 9,75E-28; group 2/control group: p < 0; group 3/control group: p < 9,85E-29;
5 The ANOVA-test was calculated for the results: 100% and 75%
correct uses of expletives.
120
100
80
%
60
40
20
0 Group 1 2 3 4 control
100% correct
75% correct 66% correct 50% correct 33% correct 25% correct
0% correct
Optionality in non-native grammars 11 Table 13. Grammaticality judgments about sentences with a preposed phrase Both judged
Group Correct Incorrect as correct Not revealing
1 3 17 5 21 2 12 23 19 14 3 17 22 8 11 4 18 27 10 15 C 50 0 0 0
Table 14. Grammaticality judgments about
sentences with a preposed finite or non–finite clause Figure 1. The use of expletives in the elicited production task in items with a preposed phrase.
Both judged
Group Correct Incorrect as correct Not revealing
120
100
80
%
60
40
20
0
Group 1 2 3 4 control
100% correct
75% correct
66% correct
50% correct
33% correct 25% correct
0% correct
1 1 23 1 21 2 8 28 7 25 3 11 25 4 18 4 19 29 9 13 C 50 0 0 0
Table 15. Grammaticality judgments about sentences with a preposed phrase Both judged Group Correct Incorrect as correct
1 12% 68% 20% 2 22% 43% 35% 3 36% 47% 17%
Figure 2. The use of expletives in the elicited production task in items with a preposed finite or non-finite clause.
group 4/control group: p < 0.
The use of expletives in Spec,AGR by L2 German
learners was also investigated by a second method, a
grammaticality judgment test. For purposes of scoring,
responses were regarded as “correct” (a) if subjects
indicated that they were absolutely or quite sure about their
judgments, (b) if the grammatical item was judged as
“correct” and (c) if the ungrammatical item (of the pair)
was judged as “incorrect”, and (d) if the ungrammatical
item was adequately corrected. Responses were regarded as
“incorrect” (a) if subjects indicated that they were
absolutely or quite sure about their judgments, (b) if the
grammatical item was judged as “incorrect” and (c) if the
ungrammatical item (of the pair) was judged as “correct”,
and (d) if the item judged as “incorrect” was adequately
corrected. Judgments, both those absolutely sure and
those quite sure, where both sentences (of the pair), the
grammatical and the ungrammatical, were
4 33% 49% 18% C 100% 0 0
judged as being correct, are listed separately. All other
answers are regarded as unrevealing. These are (a) the
unsure judgments, (b) judgments in which one sentence
of the pair was not judged at all or (c) judgments in
which the sentence judged as incorrect was not
corrected and therefore it is impossible to decide about
the source of the judgment. Tables 13 and 14 show the exact number of
judgments. The division according to the kind of
sentences involved has been kept. Tables 15 and 16
indicate the results as percentages. The number of all
revealing judgments is regarded as 100%. The results
are also presented in Figures 3 and 4. The one-way-ANOVA shows that the differences
between the groups in judging simple clauses are
significant (F = 23,69487; p < 1,44E-13). The post-hoc
Scheffe´ test reveals that the differences between the
experimental groups are not significant. The differences
12 Aldona Sopata
Table 16. Grammaticality judgments about
sentences with a preposed finite or non-finite clause Both judged Group Correct Incorrect as correct
1 4% 92% 4% 2 19% 65% 16% 3 27.5% 62.5% 10% 4 33% 51% 16% C 100% 0 0
Table 17. Grammaticality judgments task – results
according to the number of subjects with
particular responses
Table 18. Grammaticality judgments task – results
according to the number of subjects with
particular responses Number of Number of Number of subjects with subjects with subjects with correct incorrect optional Group responses responses responses
1 5% 68% 26% 2 15% 24% 61% 3 25% 25% 50% 4 16% 28% 56% C 100% – –
Number of Number of Number of Number of subjects with subjects with subjects with subjects with correct incorrect optional unrevealing Group responses responses responses responses
1 1 13 5 4 2 5 8 20 1 3 7 7 14 1 4 5 9 18 3 C 25 – – –
100 %
90 %
80 %
70 %
60 %
50 % 40 %
30 %
20 %
10 % 0%
gro
up 1
gro
up 2
gro
up 3
gro
up 4
co
ntr
ol g
rou
p
both examples
judged as correct
wrong
correct
between each experimental group and the control group are, however, significant:
group 1/control group: p < 1,69E-07;
group 2/control group: p < 6,6E-11;
group 3/control group: p < 2,76E-07;
group 4/control group: p < 2,37E-07. In the case of the compound clauses the difference between
the groups are again significant (F = 30,18382; p < 8,27E-
16). The post-hoc Scheffe´ test shows that the differences
between the experimental groups are not significant. But
the differences between each experimental group and the
control group are again significant:
group 1/control group: p < 3,7E-11;
group 2/control group: p < 9,7E-10;
group 3/control group: p < 4,54E-09;
group 4/control group: p < 1,93E-09.
To show the status of AGR in individual grammars of
the subjects the results of the second task are presented
below in another manner. Table 17 presents results
according to the number of subjects with particular
responses. The division according to the kind of sentences
has been ignored here. The first column presents the
number of subjects whose revealing judgments were all
correct. The second column shows the number of subjects
Figure 3. Grammaticality judgments about sentences with a preposed phrase. whose revealing judgments were all incorrect. The third
column shows the number of subjects whose revealing
judgments contained some kind of inconsistency; this
means, either both sentences of a pair were judged in
some cases as correct or some sentence pairs were
judged homogenously correct and some homogenously
incorrect. Finally, the fourth column gives the number of
subjects whose judgments were all unrevealing. Table
18 presents the results as a percentage. The number of
all subjects with revealing answers is regarded as 100%.
7. Discussion The data gained in the study show that the hypothesis
stated in (16) above – expletives do not appear in
Spec,AGR (i.e. clause internally) in the interlanguage of
German L2 learners – must be rejected. The data are
consistent with the hypothesis stated in (17) – expletives
appear optionally in Spec,AGR (i.e. clause internally) in
the interlanguage of German L2 learners. Both of the
research methods used reveal that Polish L2 learners of
German do use expletives in Spec,AGR (i.e. clause
internally) in their interlanguage.
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0% co
ntr
ol g
rou
p
gro
up 1
gro
up 2
gro
up 3
gro
up 4
both examples
judged as correct wrong correct
Optionality in non-native grammars 13 preposed finite or non-finite clause as was the case in the
production task. The grammaticality judgments are more
homogenous.7 The fact that the answers of the highly
proficient group are not much better than the answers of
the less proficient groups is quite striking. In all groups in
the case of both sentence types, there is only a small
minority of correct judgments. The overwhelming majority
of judgments in all experimental groups indicates that the
subjects do accept expletives in Spec,AGR in L2 German.
As regards items with a preposed phrase (see Table 15), the
highest, though still quite small, number of correct
judgments is in the third group (36%). About half of
judgments in all groups is incorrect; this means that the
ungrammatical sentence was judged as correct and the
correct sentence as incorrect and Figure 4. Grammaticality judgments about sentences with a preposed finite or non-finite clause.
The use of constructions with absent expletives in the
elicited written production tasks differs depending on
the items with a preposed phrase and with a preposed
finite or non-finite clause.6 The use of constructions
with absent expletives in sentences with a preposed
phrase is relatively good, but it is still not native-like
even in the most advanced group. About half of the
subjects from groups 2 and 4 do not use expletives
claused internally correctly. In groups 1 and 3 is this
even a majority of the subjects. But there are still many
participants who do permit expletive to appear in
Spec,AGR and in the case of the most advanced group
nearly half of the subjects do it (see Table 11). The situation is quite different in the case of
sentences with a preposed finite or non-finite clause. In
all groups, except for the control group, the majority of
participants (always over 60%) use the expletive
subjects incorrectly in all structures which demand the
use of the constructions with absent expletive subject.
Other subjects use expletives optionally, e.g. in some
created contexts. Only 3% of subjects from group 2 did
not insert expletives in Spec,AGR as the specification of
the relevant feature in German demands (see Table 12). The results of group 1 should be treated, however, with
some caution, because the number of participants was
reduced due to the pre-test concerning the use of referential
subjects. 53% of the participants of the group were
excluded from further consideration because their answers
could not prove whether they had acquired the use of
referential subjects or not. The remaining number, 23
participants, is relatively small. Numerous answers of the
subjects from this group are unrevealing, too. The results gained by the second method indicate a
much smaller discrepancy between the appearance of
expletives in sentences with a preposed phrase and a
6 An explanation for this difference will be discussed in the next section.
it was adequately corrected. In other judgments
(between 17% in the third and 35% in the second group)
both sentences are accepted, which indicates clearly a
non-target-like optionality in the use of expletives
claused internally. The grammaticality judgments about the items with a
preposed finite or non-finite clause confirm the picture
(see Table 16). The number of correct judgments here is
even smaller. In the case of the very advanced group 4
this is only one-third. In all groups judgments which
show that the subjects accept the use of expletives in
Spec,AGR clearly dominate. The results of the task summarised according to the
number of subjects with particular responses (see Table
18) show that over half of the participants from groups
2, 3 and 4 accept the optional appearance of expletives
in Spec,AGR in their interlanguage. The data demonstrate therefore quite clearly that the
highly proficient L2 learners of German accept and use
expletives clause internally in their interlanguage. As
already mentioned, the participants of the study learned
German in the classroom and then in the university
setting. But they have participated in a very intensive
language course for many years. Most of the
participants of group 3 or 4 have stayed in Germany for
a longer period of time. Despite many years of exposure
to L2 and many years of instruction during which the
target constructions are explicitly taught several times,
the students overuse expletive subjects and set them in
structures in which their absence is required.
8. Role of metalinguistic knowledge The architecture of learners’ linguistic knowledge cannot
be seen as homogeneous. Bialystok and Ryan (1985)
7 Different evidential sources have to be compared with some caution, as
different tasks may not tap the same source of linguistic knowledge or may be connected with different ways of access to and retrieval of that knowledge. This will be discussed in the next section.
14 Aldona Sopata
postulate different ways of the linguistic knowledge
structuring: unanalysed knowledge, which is used
routinely and cannot be manipulated intentionally; the
speaker is not aware of its structure, and analysed
linguistic knowledge, which is accessible, explicit and
can be manipulated in the course of solving different
cognitive problems (Birdsong, 1989, pp. 54f.). Access to linguistic knowledge can proceed with
different level of cognitive control. Cognitive control is
defined by Bialystok and Ryan (1985) as the operation of a
mechanism that can focus on and retrieve information.
Meaning appears to be the most relevant aspect of a
linguistic message to a child or to an L2 learner:
If a language problem requires some attention to form, . . .
then the child must deliberately or intentionally focus on form
in order to supplement, derive, or override the meaning. . . .
There is a tendency, therefore, for increased values along the
control dimension to be correlated with an increase in the
proportion of attention that must be directed in order to
succeed on the task. (Bialystok and Ryan, 1985, pp. 213f.)
Various test tasks make differential demands on cognitive
control, linguistic knowledge and coordination of them.
Such differences can be observed in the tasks used in the
present study too. According to Bialystok and Ryan (1985,
pp. 234–238) and Birdsong (1989, pp. 58f.), judgments of
grammaticality demand only moderate analysed
knowledge. Transformational tasks, on the other hand,
demand high cognitive control, as the meaning in these
tasks is largely unaltered while structure is changed. The
Bialystok and Ryan model is quite convincing and
consistent with many findings reported in the literature (for
relevant discussion see Birdsong, 1989, pp. 51–62). The control dimension and the higher levels of
analysed knowledge can be modified by instruction.
This fact has to be crucial for the task-specificity in the
learning environment. The explicit metalinguistic rules
obtained via learning will play a larger role in the
transformational tasks. In the judgments of
grammaticality the internalised linguistic knowledge
will be accessed in a much higher degree. This is, in fact, what we see in the results of the present
study. The results of the elicited written production tasks,
which are transformational tasks, indicate a larger role of
the cognitive control and the explicit metalinguistic rules. It
is not surprising then, that the use of structures with absent
expletives is quite good in the case of the items with a
preposed phrasal material. This is how the canonical
examples of the absence of expletives look like in various
German as L2 courses. Sentences in which absent
expletives are preposed with phrasal material are then more
likely to be the area where the learned metalinguistic rules
play a role, as the use of the cognitive control by the task is
very high. The use of these rules seems to be not so clear
for the learners in the items with
preposed non-finite or finite clauses, as the results on
these items show. Crucially, however, the grammaticality judgments do
not show such discrepancy in the findings. The
judgments on the items with preposed phrasal material
and preposed non-finite or finite clauses are judged in a
similar vain. This confirms that the metalinguistic rules
obtained via general-purpose-learning do not play a
large role in this task. The grammaticality judgments
show a quite homogeneous picture of the internalised
grammars of the learners.
9. Implications for the SLA theories German licenses null expletive subjects because AGR in
German is strong (Speas, 1994, pp. 180f.; Rohrbacher,
1999, p. 246). During the course of acquisition, one
might therefore expect the strong setting for AGR to
come about. Because the AGR features are not specified
richly enough to supply the referential subjects with phi-
features, referential subjects cannot be null. This leads to
the situation in which learners of German are exposed to
input in which referential subjects are always overt and
expletives are absent in Spec,AGR. Due to the
Projection Principle, referential or expletive subjects are
required in Spec,CP in German (Speas, 1994, pp.191f.). The predictions based on the Schwartz and Sprouse
(1994, 1996) proposal (see Table 3) were that learners from
the beginning to the end state of L2 acquisition would have
in our case the strong value of AGR and therefore use no
expletive in Spec,AGR (i.e. clause internally). The findings
reported above plainly disconfirm this acquisition scenario
as the L2 learners of German do permit expletives clause
internally in the vast majority of cases. The value of AGR
cannot be therefore strong in their interlanguage. The acquisition scenario based on the approach of
Vainikka and Young-Scholten (see Table 3) predicted
that the learners would have the unspecified value of
AGR at the early stage of L2 acquisition and would then
permit expletives to appear optionally in Spec,AGR. But
at the advanced stage of the acquisition course the value
of AGR would be specified as strong and L2 learners
would use expletive clause internally. The results
reported in this study are consistent with this view to the
extent that the less advanced learners will permit
expletives optionally in Spec,AGR but the results are
inconsistent with the predictions that the more advanced
learners will not use expletives claused internally. Finally, Beck’s (1997, 1998) proposal (see Table 3)
predicted that L2 learners at any stage of development
would permit optional appearance of expletives in
Spec,AGR because the value of AGR would stay
unspecified to the end state of L2 acquisition. The results
reported in previous sections confirm this approach as the
majority of learners in all groups including the most
advanced one clearly permit expletives in Spec,AGR.
This indicates that the value of AGR-projection cannot
be strong in their interlanguage. The overuse of expletives in L2 German by Polish
learners can, therefore, be explained only when we assume
that the value of the relevant feature was not transferred
from the mother tongue to the target language. Rather, the
relevant value is not specified at all causing the optionality
in the use of expletives clause internally.8
References Beck, M.-L. (1997). Regular verbs, past tense and frequency:
Tracking down a potential source of native speaker/non-
native speaker competence differences. Second Language
Research, 13, 93–115. Beck, M.-L. (1998). L2 acquisition and obligatory head
movement: English speaking learners of German and the
Local Impairment Hypothesis. Studies in Second
Language Acquisition, 20, 311–348. Bialystok, E. & Ryan, E. B. (1985). A metacognitive framework
for the development of first and second language skills. In D.
L. Forrest-Pressley, G. E. MacKinnon & T. G. Waller (eds.),
Metacognition, cognition and human performance, pp. 207–
252. Orlando, FL: Academic Press. Birdsong, D. (1989). Metalinguistic performance and inter-
linguistic competence. Berlin: Springer. Carroll, S. (1989). Language acquisition studies and a feasible
theory of grammar. Canadian Journal of Linguistics, 34,
399–418. Chaudron, C. (1983). Research on metalinguistic judgments:
A review of theory, methods and results. Language
Learning, 33, 343–377. Chomsky, N. (1981). Lectures on Government and Binding.
Dordrecht: Foris. Eguzkitza, A. & Kaiser G. A. (1999). Postverbal subjects in
Romance and German: Some notes on the Unaccusative
Hypothesis. Lingua, 109, 183–194. Eubank, L., Bischoff, J., Huffstutler, A., Leek, P. & West, W.
(1997). Tom eats slowly cooked eggs: Thematic verb raising
in L2 knowledge. Language Acquisition, 6, 171–200. Grewendorf, G. (1989). Ergativity in German. Dordrecht:
Foris. Haider, H. (1994). Deutsche Syntax generativ. Tubingen:¨ Narr.
Hawkins, R. & Chan, C. Y. (1997). The partial availability of
Universal Grammar in second language acquisition: The
8 The overuse of expletives by L2 learners of German points to the
possible involvement of the process of general-purpose-learning in the
L2 development. General-purpose-learning is claimed to require more
frequent exposure to input, possibly over a longer period of time and it
needs salient and unambiguous input data. Intra- and interindividual
variation is to be expected (see Carroll, 1989; Meisel, 1995, 1997).
Because the input data are not very salient and frequent, problems with
the learning of constructions with absent expletive subjects in L2
German can be expected if we assume that the features of functional
projections remain unspecified and the general-purpose-learning is
involved here. The interaction of the various factors in SLA is however
an area which awaits further research.
Optionality in non-native grammars 15
“failed functional features hypothesis”. Second Language Research, 13, 187–226.
Haznedar, B. & Schwartz, B. (1997). Are there optional infinit-
ives in child L2 acquisition? In E. Hughes, M. Hughes & A.
Greenhill (eds.), Proceedings of the 21st Annual Boston
University Conference on Language Development, pp. 257–
268. Somerville, MA: Cascadilla Press. Kaiser, G. (2002). Syntaktische Variation und generative
Syntaxtheorie. In E. Stark & U. Wandruszka (eds.), Syntaxtheorien: Modelle, Methoden, Motive, pp. 257–
272. Tubingen:¨ Narr. Klein, E. & Martohardjono, G. (1999). Investigating second
language grammars: Some conceptual and
methodological issues in generative SLA research. In E.
Klein, & G. Martohardjono, (eds.), The development of
second language grammars: a generative approach, pp.
3–36. Amsterdam: John Benjamins. Lardiere, D. (1998a). Case and tense in the “fossilized”
steady-state. Second Language Research, 14, 1–26. Lardiere, D. (1998b). Dissociating syntax from morphology in
a divergent L2 end-state grammar. Second Language
Research, 14, 359–375. Lardiere, D. (2000). Mapping features to forms in second
language acquisition. In J. Archibald (ed.), Second
language acquisition and linguistic theory, pp. 102–129.
Oxford: Blackwell Publishers. Mandell, P. (1999). On the reliability of grammaticality
judgement tests in second language acquisition research. Second Language Research, 15, 73–99.
Meisel, J. M. (1995). Parameters in acquisition. In P. Fletcher
& B. MacWhinney (eds.), The handbook of child
language, pp. 123–156. Oxford: Blackwell. Meisel, J. M. (1997). The acquisition of the syntax of negation in
French and German: Contrasting first and second language
development. Second Language Research, 13, 374–385. Muller,¨ G. & Rohrbacher, B. (1989). Eine Geschichte ohne
Subjekt. Zur Entwicklung der pro-Theorie, Linguistische
Berichte, 119, 3–52. Prevost, P. & White, L. (2000). Missing surface inflection or
impairment in second language acquisition? Evidence
from tense and agreement. Second Language Research,
16, 103– 133. Rizzi, L. (1986). Null objects in Italian and the theory of pro.
Linguistic Inquiry, 17, 501–557. Rizzi, L. (1994). Early null subjects and root null subjects. In T.
Hoekstra & B. Schwartz (eds.), Language acquisition studies
in generative grammar, pp. 151–176. Amsterdam & Philadelphia: John Benjamins.
Rohrbacher, B. (1992). English AUXˆNEG, Mainland Scandinavian NEGˆAUX and the theory of V to I raising. Proceedings of the 22nd Western Conference on
Linguistics, pp. 307-321. Rohrbacher, B. (1999). Morphology-driven syntax: A theory of
V to I raising and pro-drop. Amsterdam & Philadelphia:
John Benjamins. Safir, K. (1984). Missing subjects in German. In I. Toman (ed.),
Studies in German grammar, pp. 193–230. Dordrecht:
Foris. Schwartz, B. (1998). On two hypotheses of “transfer” in L2A:
Minimal trees and absolute L1 influence. In S. Flynn,
16 Aldona Sopata
G. Martohardjono & W. O’Neil (eds.), The generative study
of second language acquisition, pp. 35–59. Mahwah, NJ: Lawrence Erlbaum.
Schwartz, B. & Sprouse, R. (1994). Word order and nominative case in nonnative language acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In T. Hoekstra & B. Schwartz (eds.), Language acquisition studies in generative grammar, pp. 317–368. Amsterdam & Philadelphia: John Benjamins.
Schwartz, B. & Sprouse, R. (1996). L2 cognitive states and the Full Transfer/Full Access model. Second Language Research, 12, 40–72.
Schwartz, B. & Sprouse, R. (2000). When syntactic theories evolve: Consequences for L2 acquisition research. In J. Archibald (ed.), Second language acquisition and linguistic theory, pp. 156–187. Oxford: Blackwell Publishers.
Sigurdsson, H. A. (1993). Argument-drop in Old Icelandic. Lingua, 89, 247–280.
Smith, N. & Tsimpli, I.-M. (1995). The mind of a Savant: Language learning and modularity. Oxford: Blackwell. Speas, M. (1994). Null arguments in a theory of economy of projection. In E. Benedicto & J. Runner (eds.), Functional
projections, University of Massachussets Occasional Papers in Linguistics 17
pp. 179–208.
Sorace, A. (1996). The use of acceptability judgments in SLA research. In W. C. Ritchie & T. K. Bhatia (eds.), Handbook of second language acquisition, pp. 375–409. San Diego, CA: Academic Press.
Appendix A. Elicited written production task. I Setzen Sie die folgenden Sätze in indirekte Rede mit
Konjunktiv! .
Proszę˛ wyrazić następujące zdania w mowie zależnej z trybem przypuszczajacym!
1. Der Vater sah sein Unglück kommen.
Er sagte, dass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Es wird an die Oma gedacht.
Er sagte, dass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Geh nach Hause!
Er sagte, dass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Es wird Deutsch gesprochen werden.
Er sagte, dass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Bernd sollte daran denken.
Er sagte, dass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
II Formen Sie folgende aktivischen Sätze in passivische
Sätze um! Proszę przekształcić zdania w stronie czynnej na zdania w stronie biernej.
1. Immer reist man mit viel Gepäck. Immer . . . . . . . . . . . 2. Das Gericht entzieht ihm den Führerschein. Der Führerschein.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. Das Auto soll den Fußgänger erfasst haben. Der Fußgänger . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4. Der Koch hat die Suppe kosten müssen. Die Suppe. ..
Sternefeld, W. (1984). On Case and Binding. In I. Toman (ed.),
Studies in German grammar, pp. 231–285. Dordrecht:
Foris. Tsimpli, I.-M. & Roussou, A. (1991). Parameter resetting in L2?
University College London Working Papers in
Linguistics, 3, 149–169. Tsimpli, I.-M. & Smith, N. (1991). Second language learning:
Evidence from a polyglot Savant. University College
London Working Papers in Linguistics, 3, 171–184. Vainikka, A. & Young-Scholten, M. (1994). Direct access to
X_ theory: Evidence from Korean and Turkish adults
learning German. In T. Hoekstra & B. Schwartz, (eds.),
Language acquisition studies in generative grammar, pp.
265–316. Amsterdam & Philadelphia: John Benjamins. Vainikka, A. & Young-Scholten, M. (1996). Gradual develop-
ment of L2 phrase structure. Second Language Research,
12, 7–39. Vikner, S. (1995). Verb movement and expletive subjects in the
Germanic languages. New York: Oxford University Press. Weissenborn, J. (1992). Implications of parametric variation for
adult second language acquisition: An investigation of the
pro-drop parameter. In J. Weissenborn, H. Goodluck & T.
Roeper, (eds.), Theoretical issues in language acquisi-tion,
pp. 269–299. Hillsdale, NJ: Lawrence Erlbaum. Received September 10, 2003 Revision received January 30, 2005 Accepted March 11, 2005 5. In dem Kaufhaus hat man auch am Sonntag verkauft. In
dem Kaufhaus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III Verbinden Sie die Sätze, indem Sie eine von angegebenen Konjunktionen benutzen. Bitte ändern Sie die Reihenfolge der Sätze nicht. Konjunktionen, die Ihnen zur Verfügung stehen: weil, trotzdem, obwohl, als
dass, dass. .
Proszę połączyć zdania używając jednego z podanych spójników. Proszę nie zmieniać kolejności zdań. Spójniki do dyspozycji: weil, trotzdem, obwohl, als, dass, dass. 1. Man hat es nicht gemacht. Man hat es machen sollen. 2. Die Jugend jubelt. Es wird ihm geholfen werden. 3. Das war ziemlich merkwürdig. Es wird lange her
geschrieben worden sein. 4. Er war in einer Schwimmhalle und konnte nichts
sehen. Es wurde eben geschwommen. 5. Die Aufgabe ist zu schwierig. Man könnte sie auf
Anhieb lösen. 6. Ulrike möchte Jörg nicht heiraten. Es ist klar. IV Formen Sie die Sätze um. Proszę przekształcić´ zdania. 1. Jörg hat ausziehen müssen.
Jorg¨ ist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Jetzt geht aber schnell ins Bett.
Jetzt wird . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3. Die Preise sind letztens sehr gestiegen.
Die Preise wurden . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Es ist mir gleichgültig, was du darüber denkst.
Was du darüber denkst, ist . . . . . . . . . . . . . . . . . . . . . . . . 5. Mit dem Boot müssen wir die ganze Insel umfahren. Die
ganze Insel muss . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6. Man singt während der Busfahrt gemeinsam. Während der
Busfahrt wird . . . . . . . . . . . . . . . . . . . . . . .
7. Es nervt mich, dass du immer zu spät kommst. Dass du immer zu spät kommst, . . . . . . . . . . . . . . . . . . .
Appendix B. Grammaticality judgments test. Beurteilen Sie die Korrektheit der Sätze! Proszę określić poprawność zdań. Beispiel I/Przykład I – Der Satz: Ich spreche gut Deutsch. Der
Satz wird als korrekt beurteilt und die Person ist ihrer
Meinung völlig sicher, also ein Kreuz in der ersten Rubrik.
Zdanie zostało uznane za prawidłowe i dana osoba jest pewna
swego zdania, a wiec krzyżyk w pierwszej rubryce. Beispiel II/Przykład II – Der Satz: Ich verstehen nicht.
Der Satz wird als nicht korrekt beurteilt und die Person
ist ihrer Meinung ziemlich sicher also ein Kreuz in der
fünften Rubrik. Zdanie zostało uznane za nieprawidłowe i dana osoba
. jest dość pewna swego zdania, a więc krzyżyk w piątej rubryce. Wenn Sie einen Satz nicht korrekt empfinden, schreiben Sie bitte, eine korrigierte Entsprechung des Satzes.
. Jeśli uważają Państwo jakieś zdanie za błędne, proszę napisać poprawiony odpowiednik tego zdania. Na przykład: Ich verstehen nicht. – poprawiony
odpowiednik – Ich verstehe nicht. 1. Mein Freund ist ein starke Raucher . . . . . . . . . . . . . . . 2. Nach der Schule komme nach Hause . . . . . . . . . . . . . .
Optionality in non-native grammars 17
3. Katja glaubt, dass klug ist . . . . . . . . . . . . . . . . . . . . . . . . 4. Morgen wird es erklärt, dass er kommt . . . . . . . . . . . . 5. Alle meine Entchen sind hier . . . . . . . . . . . . . . . . . . . . . 6. Verstehen Deutsch gut . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. Auf der Autobahn wird es sehr schnell gefahren . . . . 8. Manfred fragt, ob er die Prüfung bestanden hat . . . . . 9. Katja glaubt, dass klug ist . . . . . . . . . . . . . . . . . . . . . . . .
10. Morgen wird regnen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Dass der Sommer angefangen hat, ist deutlich geworden .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12. Wenn der Weg unbekannt ist, wird es gefragt . . . . . . 13. Er setzte sich zwischen seinen Schülern . . . . . . . . . . . 14. Während des Krieges wird gelitten . . . . . . . . . . . . . . . . 15. Dass der Sommer angefangen hat, ist es deutlich
geworden . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16. Morgen wird erklärt, dass er kommt . . . . . . . . . . . . . . 17. Morgen wird es regnen . . . . . . . . . . . . . . . . . . . . . . . . . . 18. Wenn der Weg unbekannt ist, wird gefragt . . . . . . . . . 19. Nach der Schule kommst du nach Hause . . . . . . . . . . 20. Er denkt immer an sich . . . . . . . . . . . . . . . . . . . . . . . . . . 21. Während des Krieges wird es gelitten . . . . . . . . . . . . . 22. Manfred fragt, ob die Prüfung bestanden hat . . . . . . . 23. Morgen wird er die Arbeit beendet haben . . . . . . . . . . 24. Um den Nachbarn nicht aufzuwecken, wird es
geschlichen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25. Wir verstehen Deutsch gut . . . . . . . . . . . . . . . . . . . . . . . 26. Der Alkohol begann, seine Gesundheit zu untergraben . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27. Auf der Autobahn wird sehr schnell gefahren . . . . . . 28. Dass Marie gewonnen hat, freut mich sehr . . . . . . . . . 29. Um den Nachbarn nicht aufzuwecken, wird geschlichen. .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30. Der Student hat die Arbeit geschafft . . . . . . . . . . . . . . 31. Wir beauftragen ihn, die Post abgeholt zu haben . . . . 32. Dass Marie gewonnen hat, freut es mich sehr . . . . . .