TRIGGERS
IN L2 ACQUISITION: THE CASE OF SPANISH N-N COMPOUNDS*
clac 5/2001
Juana M. Liceras, Lourdes Díaz
University
of Ottawa & Instituto Universitario Ortega y Gasset, Universitat Pompeu
Fabra
jliceras@uottawa.ca
ABSTRACT
Spanish has
left-headed compounds which are not as productive as their left-headed
counterparts in other languages. This presence or absence of productivity has
been attributed to a binary parameter according to which N-N compounding, as
opposed to nominal constructions in which the head noun takes a complement as
in "the destruction of the city”, would
be the superset or marked option. Furthermore, the idiosyncratic nature of
Spanish N-N compounding has been attributed to the make-up of Spanish Nouns.
Specifically, it has been proposed that Spanish Nouns have a ‘word marker’
which triggers L1 acquisition of these constructions. Based on the results of
two picture tests intended to elicit actual command of N-N compounding strategies,
as well as word order patterns and gender marking patterns, we argue that: 1)
N-N compounding is not a marked construction; 2) adult L2 acquisition of
Spanish N-N compounds is triggered by head directionality (a processing
trigger) rather than by the ‘word marker’ (a representational trigger) which is
supposed to trigger L1 acquisition of these compounds.
1. Triggers
and learnability in L2 acquisition
Determination of what can act as a
trigger is important for acquisition theory because it will indirectly shed
light on the relationship between ‘nature’ (the innate principles of the
Language Acquisition Device) and ‘nurture’ (the contribution provided by the
linguistic environment). Defining triggers is also important for L2 acquisition
theory because it can serve as a point of reference for the identification of
actual L2 triggers.
Within the Chomskian framework,
formal analyses intended to account for language variation provide specific
proposals as to what may be the elements —abstract features (White 1992; Eubank
1996; Epstein, Flynn and Martohardjono 1996; Sprouse 1998) or overtly realized
morphemes (Zobl and Liceras 1994; Meisel 1997; Müller 1998; Vainikka &
Young-Scholten 1998) needed for projecting a grammar. It is these abstract or
explicit elements that are supposed to act as triggers for language
acquisition.
In this paper, we would like to
propose that acquisition of N-N compounds is related to two different types of
triggers: (a) a ‘representational trigger’ which stems from the properties of
words; and (b) a ‘processing trigger’ which is based on head directionality.
Based on two recent proposals intended to explain the different nature of N-N
compounding in English and Spanish, we discuss Spanish IL data elicited from
speakers of various language groups. We argue that it is mainly head
directionality that plays a role in adult L2 acquisition of Spanish N-N
compounds.
2. N-N
Compounding
N-N compounding is not a very productive
construction in Spanish and it is usually left headed, as illustrated in
examples (1) to (3).
(1) Hombre araña
Spider
man
(2) Perro policía
Police
dog
(3) Mujer pulpo
Octopus
woman
In fact, in
languages such as English, in V2 languages and in languages such as Chinese, Japanese
or Korean, N-N compounding is a highly productive strategy whereas Spanish-like
languages prefer a derivational strategy, as illustrated in (4), a ‘case
marking’ strategy —PP modifiers as in (5)—, or adjectival modification as in
(6).
(4) Manzano
Apple
tree
(5) Caja de herramientas
Tool box
(6) Vaca
lechera
Dairy
cow
This typological
difference led Snyder (1995) to propose a ‘Compounding Parameter’ which
differentiates languages according to whether their substantive categories
(Nouns, Adjectives, Verbs, Prepositions) are plus or minus affixal. This
implies that in languages such as English whose referential categories share
the [+affixal] property with morphemes, N-N compounding, as in the English
examples of (1) – (3), and complex predication of the type illustrated in (7)
and (8) are highly productive grammatical constructions.
(7)
Run up the bill
(8) Blow
one’s hair dry
However, in
languages such as Spanish, whose referential categories do not share the
[+affixal] property with morphemes, N-N compounding or complex predication are
NOT productive grammatical constructions. Thus, the property [+/- affixal] is
assumed to be the trigger for the acquisition of these constructions.
Piera (1995) provides an alternative
account of the differences between English-like and Spanish-like compounds. He
proposes that it is the actual make up of Spanish Nouns that is at the core of
the distinction. Following Harris (1991a, 1991b), he maintains that the
structure of Spanish Nouns, unlike the structure of English Nouns, has a word
marker (WM), as shown in (9a) - (9b):
(9) a. N[[ perr ] o ]
(9)
b. N [ dog ]
This [+/- WM]
property, an alternative to Snyder’s [+/-affixal] proposal, constitutes the
trigger for N-N compounding. According to Piera (1995), the double-bracketed
structure of perro in (9a) prevents
adjunction of an N to the left of any given N, as shown in (10a), whereas the
English counterpart in (10b) shows that it is not the case in English:
(10) a. N
[*policía [[ perr ] o ]
(10)
b. N [police [ dog ]
This explains why the right-headed equivalents of examples (1) –
(3) are ungrammatical, as shown in (11) - (13):
(11) *Araña hombre[i]
Spider man
(12) *Policía perro
Police dog
(13) *Pulpo mujer
Octopus woman
It is only the
head that carries gender and plural marking morphemes, as well as derivational
morphemes, as shown in (14) to (15)[ii]:
Plural Diminutive
(14) Hombre
araña hombreS araña hombrecito
araña
Spider man spider
men little spider man
(15) Perro
policía perroS policía perrito
policía
Police dog police
dogs little police dog
(16) Mujer pulpo mujereS pulpo mujercita pulpo
Octopus woman octopus women little octopus woman
3. Triggers
and learnability in L1 acquisition
3.1. Parameters
and the subset principle. The ‘compounding parameter’ as defined by
Snyder (1995)
respects the subset principle (Berwick 1985) in that English represents the
superset language because both the [+affixal] and the [-affixal] constructions
are productive options. On the other hand, languages such as Spanish represent
the subset option of the parameter because only the [-affixal] option is
productive. This implies that N-N compounding is the marked option. As already
stated, the parameter does not make any predictions in terms of head
directionality.
Based
on the above, we hypothesize that: 1) N-N compounding will be difficult to
acquire because it is a marked construction. Consequently production of N-N
compounds will be very limited at the early stages. 2) No predictions can be
made in terms of head directionality. 3) In terms of L1 transfer, if the L1
does not play a role, N-N compounding will be a difficult option for all
learners. However, if the superset option is transferred, there will be
differences between the French group (French pairs with Spanish) and the other
groups.
3.2. [+/- word marker]. Piera’s (1995)
analysis implies that the WM triggers the acquisition of N-N compounding and that
head directionality (adjunction to the left) will be linked to the presence
versus absence of this WM. Based on the claim that the WM is a trigger for
adult L2 acquisition of Spanish N-N compounds, we make the following
predictions: 1) There will be a limited production of Spanish N-N compounds
until learners are exposed to a considerable amount of positive input. 2) When
compounds are produced they will be left-headed (Spanish-like) because presence
of the proposed WM prevents adjunction to the left. 3) Gender marking of
compounds will be Spanish-like (only marked on the head of the compound).
If the WM does not trigger the
acquisition of Spanish N-N compounds, production may depend on the role of the
L1. Therefore we predict that: 1) Only subjects from English-like languages
will produce a large number of N-N compounds from the early stages. 2) Most of
the N-N compounds will be right headed. 3) Gender marking will not follow the
Spanish N-N pattern.
4. The
study
We
report here on a subset of the data that we are in the process of collecting as
part of a large research project being conducted in Ottawa (Canada) and
Barcelona (Spain). We discuss the results of two different tests administered
to 68 students whose first languages were Indo-European (French, English,
German, Russian, Polish, Danish, Swedish) and Non Indo-European (Chinese,
Japanese and Korean). The subjects were studying Spanish in Canada
(institutional setting) and Spain (natural setting).
These
two different tests are part of a Picture Task Experiment designed to elicit
information concerning the acquisition of ‘compounding strategies’ in Spanish[iii].
Test #1 was intended to find out whether learners of Spanish were aware of the
specific characteristics and the low degree of productivity of the N-N
compounding strategy in Spanish. We provided pictures of beings and objects
(see Appendix) which can normally be labeled as N-N compounds as in examples
(1) to (3) above. We also provided pictures which cannot be labeled as N-N
compounds, as indicated in (4) to (6) above. In Test #2 we provided a set of
pictures of possible N-N compounds paired as masculine and feminine (see
Appendix), so that there was an ‘octopus man’ and an ‘octopus woman’ or a
‘pirate ship’ (masculine) contrasted with a ‘pirate bicycle’ (feminine). Each
picture was only shown once and subjects were given enough time to name them.
In order to avoid non-production of
N-N compounds due to lack of actual vocabulary, students were shown pictures
(with their names) of the isolated Nouns which could form the N-N compounds
(see Appendix).
Students were classified as
Beginners, Intermediate or Advanced depending on their performance on the
placement test ‘SGEL’[iv].
All examples in (17) - (21) were
considered acceptable, but only cases such as (17) were counted as N-N
compounds.
(17) barco pirata
ship pirate
(18) barco de piratas
ship of pirates
(19) llavero
key-’er’ (key chain)
(20) vaca de leche
cow of milk (dairy cow)
(21) lleva llaves (deverbal)
hold keys (key holder)
Examples such as
(22) - (25) were considered unacceptable because they were either ungrammatical
—(22)— or did not reflect any possible compounding strategy in Spanish.
(22) *pirata barco
pirate ship
(23) perro largo
long
dog (wiener dog
(24)
pulpo
octopus
(25) sofá con cama
sofa with bed
6. Results
Results are presented in Tables 1 to 5. Table 1 shows that all
groups but the B-Spain group produce a substantial amount of N-N compounds. Therefore,
the hypothesis that compounding as a general strategy is marked is not
confirmed. The fact that the B-Canada group produces a high number of
English-like N-N compounds confirms the hypothesis that the WM does not act as
a trigger for these learners.
TABLE 1. Test #1. Total production of N-N
compounds
|
Level / L1
Setting
|
B
Spain
8x20
|
B-E
Canada
8x10
|
I-E
Canada
8x5
|
I-F
Canada
8x10
|
A-E
Canada
8x7
|
A-F
Canada
8x8
|
A-V2
Spain
8x4
|
A-F
Spain
8x4
|
|
Spanish
N-N
|
(36/160)
0.22
|
(21/80)
0.26
|
(15/40)
0.37
|
(43/80)
0.53
|
(34/56)
0.60
|
(36/64)
0.56
|
(26/32)
0.81
|
(22/32)
0.68
|
|
Non-Spanish
N-N
|
(11/160)
0.06
|
(37/80)
0.46
|
(10/40)
0.25
|
(11/80)
0.13
|
(6/56)
0.10
|
(6/64)
0.09
|
(1/31)
0.03
|
(2/32)
0.06
|
|
Total of
N-N
|
(47/160)
0.29
|
(58/80)
0.72
|
(25/40)
0.62
|
(54/80)
0.67
|
(40/56)
0.71
|
(42/64)
0.65
|
(27/32)
0.84
|
(24/32)
0.75
|
B= beginners; I = intermediate; A = advanced
E = L1 English; F = L1 French; V2 = L1 German & Danish
The two Beginner groups provide
opposite trends in terms of their choices of N-N versus N-PP constructions. In
fact, the figures seem to indicate that the B-Spain group performs as well as
the advanced groups. However, as we will show below, these results are
misleading because this group does not favor the N-N strategy but rather a
different type of strategy.
These
data also indicate that Spanish compounding becomes a preferred strategy at the
intermediate and advanced levels, which we take as an indication that learning
has taken effect via positive input.
Table 2 contains the results for
test #2 in relation to head directionality. In this table we have split the
B-Spain group into two: the Indo-European B-Spain group and the Non
Indo-European B-Spain group. The results are consistent with the ones presented
in Table 1 in that, it would seem that neither the Indo-European nor the Non
Indo-European group have problems with word order when it comes to choosing the
N-N compounding strategy. Namely, they produce few N-N Spanish compounds but
when they do, these compounds are left headed as in Spanish. This does not
happen in the case of the B-Canada group, which we take as evidence that the
latter group has not received enough input to realize that N-N compounding as
such is not productive in Spanish.
|
Level / L1
Setting
#Items/Subj.
|
B-Non/Ind
Spain
16x9
|
B-Ind
Spain
16x11
|
B-E
Canada
16x10
|
I-E
Canada
16x5
|
I-F
Canada
16x10
|
A-E
Canada
16x7
|
A-F
Canada
16x8
|
A-V2
Spain
16x4
|
A-F
Spain
16x4
|
|
Spanish
N-N
|
(24/132)
0.18
|
(19/166)
0.11
|
(54/160)
0.33
|
(28/79)
0.35
|
(75/160)
0.46
|
(53/109)
0.48
|
(73/126)
0.57
|
(58/64)
0.90
|
(41/62)
0.66
|
|
NonSpanishN-N (w.o)
|
(7/132)
0.05
|
(14/166)
0.08
|
(71/160)
0.44
|
(22/79)
0.27
|
(20/160)
0.12
|
(23/109)
0.21
|
(20/126)
0.15
|
(3/64)
0.04
|
(16/62)
0.25
|
|
Total N-N
|
(31/132)
0.23
|
(33/166)
0.19
|
(125/160)
0.78
|
(50/79)
0.63
|
(95/160)
0.59
|
(76/109)
0.69
|
(93/126)
0.73
|
(61/64)
0.95
|
(57/62)
0.91)
|
B=
beginners; I = intermediate; A = advanced
E
= L1 English; F = L1 French; V2 = L1 German & Danish; Ind = Indo-European;
Non/Ind = Non Indo European
The results in Table
2 also show that the Advanced French/Spain group seems to have more problems
with word order than the other Advanced groups, even than the
Intermediate French/Canada group. However, the results are due to the uneven
distribution of the right- headed constructions, since most of them were
produced by one of the subjects. Thus, for some learners, directionality seems
to be a problem up to the advanced levels.
Table 3 shows the choice of native
options —N-N as in (17) or N-PP as in (18)—. These results are also consistent
with the previous ones in that choice of Spanish-like N-N compounds increases
with level. Also, the Indo-European/Spain group produces less N-N compounds
than the other groups but more N-PP constructions. This may be due to the fact that
they have been in contact with a richer input than the B-Canada group but,
having L1s that are typologically more distant from Spanish than the L1s of the
Indo-European group, they are not yet aware of the fact that the pictures in
this test favor actual N-N choices in native Spanish.
TABLE 3. Test #2. ‘Compounding
strategies’: Native options
|
Level L1
Setting
#Items/Subj.
|
B-Non/Ind
Spain
16x9
|
B-Ind
Spain
16x11
|
B-E
Canada
16x10
|
I-E
Canada
16x5
|
I-F
Canada
16x10
|
A-E
Canada
16x7
|
A-F
Canada
16x8
|
A-V2
Spain
16x4
|
A-F
Spain
16x4
|
|
Spanish
N-N
|
(24/132)
0.18
|
(19/166)
0.11
|
(54/160)
0.33
|
(28/79)
0.35
|
(75/160)
0.46
|
(53/109)
0.48
|
(73/126)
0.57
|
(58/64)
0.90
|
(41/62)
0.66
|
|
Spanish
N-PP
|
(1/132)
0.007
|
(19/166)
0.11
|
(6/160)
0.03
|
(4/79)
0.05
|
(16/160)
0.10
|
(6/109)
0.05
|
(18/126)
0.14
|
0
|
0
|
|
Total
Native-like
|
(25/132)
0.18
|
(36/166)
0.21
|
(60/160)
0.37
|
(32/79)
0.40
|
(91/160)
0.56
|
(59/109)
0.54
|
(91/126)
0.72
|
(58/64)
0.90
|
(41/62)
0.66
|
B= beginners; I = intermediate; A = advanced
E = L1 English; F = L1 French; V2 = L1 German & Danish; Ind =
Indo-European; Non/Ind = Non Indo-European
The choice of non-Spanish options is
illustrated in Table 4. It shows that, within levels, the F-Canada students
perform better than the E-Canada students. This was predicted on the basis of
L1 transfer.
TABLE
4. Test #2. ‘Compounding strategies’: Non-native options
|
Level / L1
Setting
|
B-Non/Ind
Spain
16x9
|
B-Ind
Spain
16x11
|
B-E
Canada
16x10
|
I-E
Canada
16x5
|
I-F
Canada
16x10
|
A-E
Canada
16x7
|
A-F
Canada
16x8
|
A-V2
Spain
16x4
|
A-F
Spain
16x4
|
|
N-N
word order
|
(7/132)
0.05
|
(14/166)
0.08
|
(71/160)
0.44
|
(22/79)
0.27
|
(20/160)
0.12
|
(23/109)
0.21
|
(20/126)
0.15
|
(3/64)
0.04
|
(16/62)
0.25
|
|
N-P-N
|
(26/132)
0.19
|
(19/166)
0.11
|
(16/160)
0.10
|
(2/79)
0.02
|
0
|
0
|
0
|
0
|
0
|
|
Gender
|
(2/132)
0.015
|
(10/166)
0.06
|
(9/160)
0.05
|
(7/79)
0.08
|
(6/160)
0.03
|
(5/109)
0.04
|
(7/126)
0.05
|
(1/64)
0.01
|
0
|
|
Other
|
(74/132)
0.56
|
(85/166)
0.51
|
(3/160)
0.01
|
(16/79)
0.20
|
(43/160)
0.26
|
(22/109)
0.20
|
(7/126)
0.05
|
(2/64)
0.03
|
(5/62)
0.08
|
|
Total non-
native
|
(109/132)
0.82
|
(128/166)
0.77
|
(99/160)
0.61
|
(47/79)
0.59
|
(69/160)
0.43
|
(50/109)
0.45
|
(34/126)
0.26
|
(6/64)
0.09
|
(21/62)
0.33
|
B= beginners; I = intermediate; A = advanced
E = L1 English; F = L1 French; V2 = L1 German & Danish; Ind = Indo-European;
Non/Ind = Non Indo-European
It also shows
that the choice of the N-P-N strategy —as in (26) - (28)— is only used at the
early stages and it is mostly favored by the non Indo-European speakers.
(26) araña de hombre
spider
of man
(27) carta de bomba
letter of bomb
(28) perro de largo
dog
of long
Thus,
directionality (word order) constitutes a problem for all beginners because
this is a right-headed option. Therefore, these data suggest that transfer of
L1 word order guides the production of N-N compounds at the early stage. We
interpret this as evidence that directionality is the key to adult L2
acquisition of Spanish N-N compounds.
In terms of Gender marking, subjects
produced instances of both Spanish and non-Spanish options, as shown in (29) to
(32).
(29)mujer pulpo
octopus woman
(30)hombre
araña
spider
man
(31) *araño
hombre
man
spider (masculine)
(32)*mujer
pulpa
octopus
(feminine) mujer
The results in
table 4 indicate that even though there are very few non-Spanish options they
do not disappear with level. Furthermore, these results have to be interpreted
differently. Namely, since all Spanish nouns have intrinsic gender —hombre
‘man’ (masculine); mujer ‘woman’ (feminine); araña ‘spider’ (masculine and
femenine) - the production of instances of non-Spanish gender marking with the
total items included in the test does not provide an account of the
rule-governed behavior which would indicate knowledge of Spanish gender marking
for compounds. In fact, we should only count the six items with differentiated o/a markings associated with
masculine or feminine respectively. Therefore in terms of actual choices, the
percentages on ‘gender’ in table 4 should be multiplied by three, which
represents a rather high percentage of non-Spanish gender marking for the
advanced levels. We take this as an indication that adult L2 learners do not
access the WM to acquire Spanish N-N compounds.
In terms of setting our data show that
compounding strategies are linked to setting as follows: N-N is favored in the
institutional setting (Canada) while N-P-N is favored in Spain, mainly by the
Non Indo-European group. In fact, we believe that this strategy, which looks
like an interlanguage strategy (or construction) in terms of its surface
realization, corresponds to two different mental representations. In the case
of the Indo-European group, learners produce an N (the modifier) and then a PP
which happens to contain the head and results in a non-interpretable Spanish
sequence, as indicated in (33):
(33) N[N
araña PP [de hombre]]
spider / of man
In the case of the
Non Indo-European group, learners transfer the PP right headed option from
their L1s, so that the modifier is followed by the preposition forming a PP,
and the head follows as in (34):
(34) N[
PP[araña de] Nhombre]]
spider of / man
Thus, in both
cases, these constructions indicate that learners are struggling with head
directionality, which we take as evidence for the triggering role of word order
in the acquisition of Spanish N-N compounds.
6. Conclusions
In this paper we have investigated
three main hypotheses: 1) that N-N Compounding is a marked strategy; 2) that
the ‘word marker’ of Spanish Nouns acts as a trigger for the acquisition of
Spanish N-N compounds; 3) that the ‘word marker’ does not act as a trigger. We
have not found confirmation that N-N compounding is a marked option because all
learners but the B-Spain group produce many instances of N-N compounds. We
attribute the lower production of N-N compounds by the B-Spain group to the
fact that they have been exposed to enough Spanish input to realize that N-N
compounding is not salient in Spanish. However, they have not had enough
exposure to incorporate Spanish N-N compounds into their interlanguage.
With respect to the triggering
effect of the WM, it could be argued that it does not show at the early stage
but that it does at the later stages because there is a large production of
left-headed Spanish compounds at the advanced level. However, based on the
pattern shown by the N-P-N strategy and Gender marking we would like to argue
that it is not the WM but directionality (changing from a right headed to a
left headed word order) that triggers the acquisition of Spanish N-N compounds.
Namely, while there is a correlation between the increase in the production of
N-N compounds and the disappearance of the N-P-N strategy, the status of Gender
marking is not equally affected by the level of competence. We take this as an
indication that adult L2 learners do not have access to the proposed WM. In
other words, this marker (realized or abstract) does not seem to constitute a
trigger for adult acquisition of N-N compounds.
7. References
BERWICK, R. 1985. The acquisition of syntactic knowledge.
Cambridge, Mass.: The MIT Press.
EPSTEIN, S., FLYNN, S & MARTOHARDJONO, G. 1996. Second language
acquisition: Theoretical and experimental issues in contemporary research. Behavioral and Brain Sciences 19,
677-758.
EUBANK, L. 1996. Negation in
early German-English interlanguage: more valueless features in the L2 initial
state. Second Language Research 12,
17-106.
HARRIS, J. W. 1991a. The exponence of gender in Spanish. Linguistic Inquiry 22: 27-62.
HARRIS, J. W. 1991b. The
form classes of Spanish substantives. In Booij, G. & van Marle, J. (eds), Yearbook of morphology 4. Dordrecht:
Kluwer , 65-68.
LARDIERE, D. 1998. Parameter-resetting
in morphology: evidence from compounding. In Beck, M-L. (ed.), Morphology and its interfaces in
second-language knowledge. Amsterdam: John Benjamins, 283-305.
MEISEL, J. 1997. The acquisition of the syntax of negation in French
and German: contrasting first and second language development. Second Language Research 13, 227-63.
MÜLLER, N. 1998. UG access without parameter setting: a longitudinal
study of (L1 Italian) German as a second language. In Beck, M-L. (ed.), Morphology and its interfaces in
second-language knowledge. Amsterdam: John Benjamins, 115-163.
PIERA, C. 1995. On compounding in English and Spanish. In Campos, H.
& Kempchinsky, P. (eds.), Evolution
and revolution in linguistic theory. Washington: Georgetown University
Press, 302-315.
SNYDER, W. 1995. Language acquisition and language
variation: the role of morphology. Ph. D. Dissertation. Massachusetts Institute
of Technology.
SPROUSE, R. 1998. Some notes on the relationship between
inflectional morphology and parameter setting in first and second language
acquisition. Morphology and its interfaces in second-language acquisition, ed.
M.-L. Beck, Amsterdam: John Benjamins, 41-67.
VAINIKKA, A. & YOUNG-SCHOLTEN, M. 1998. Morphosyntactic triggers in
adult SLA. In Beck, M-L. (ed.), Morphology
and its interfaces in second-language knowledge. Amsterdam: John Benjamins,
89-113.
WHITE, L. 1991. The verb movement parameter in second language
acquisition. Language Acquisition 1,
337-60.
ZOBL, H. & LICERAS, J. M. 1994. Functional
categories and acquisition orders. Language
Learning 44, 159-80.
FOOTNOTES
