Error: Some Problems of Definition, Identification, and

Download Report

Transcript Error: Some Problems of Definition, Identification, and

Interlanguage Syntax
Based on Second Language Syntax by Roger Hawkins &
Second Language Acquisition and Universal Grammar by White
Presenter: Shu-ing Shyu (徐淑瑛)
Acknowledgements: 李家慧,邵雅琪,謝寶玉,康郁敏,廖吟倫
Introduction
How people acquire the syntax of L2s:
Two topics:
• developmental problem-- Explain
how knowledge of syntax develops
over time
• logical problem of SLA-- Explain what
makes it possible for L2 speakers to
build mental grammars in the first
place, with impoverished input
Introduction
--Universal Grammar
--Chomsky (1995): minimalist program:
to characterize the mechanisms
made available by the brain for
building mental grammars for
specific languages
--principles and parameters
Aim of the UG based second
language acquisition
Learners build subconscious mental
grammars progressively (the
developmental problem), and that
they deploy the mechanisms of an
underlying UG to do so (the
logical problem).
What Is a Grammar?
Grammar: a set of instructions for
generating all the grammatical
sentences of a particular L.
Generative Grammar: a grammar which
is able to generate all and only the
grammatical sentences of a L.
Mental Grammar: Native speakers’
subconscious and internalized
grammar
Innate Grammar-Building: Evidence
• “Input-determined” view: output from what
is exposed, created by analogy,
ungrammatical Ss due to not encountered
• Problem:
• 5a. *What did she discover who had written?
• (wh-island)
• b. *Who did she discover what had written?
• Super raising
• c. Who did she discover had written what?
•  Syntactic knowledge is underdetermined
by the input.
1.4 Innate Grammar-Building: Evidence
• Problem of Input-determined view :
• Internalized grammatical knowledge:
• e.g. theta role relations between transitive Vs and
ditransitive Vs -• Ditransitive Vs are monosyllabic or stress on the first V
• Possession relation between indirect O and direct O (send,
give, serve, vs. explain, drove)
• She offered me a job. /*She explained me the problem.
• Underdetermination by the input  the
principles and parameters of UG are
biologically determined.
Innate Grammar: biologically determined
--Uniformity of success (L1)
--Acquisition is rapid.
--Acquisition is effortless.
-- Correction feedback is largely irrelevant to acquisition.
Universal Grammar in L1 Acquisition
• UG is proposed as part of an innate biologically
endowed language faculty (Chomsky 1965, 1981;
Pinker, 1984, 1994).
• S0 :child’s initial state// PLD: primary linguistic data
Ss: mother tongue
• The role of input
• The role of evidence (feedback)
• The role of time: critical period
• L1: Linguistic competence of native
speakers of a language can be
accounted for an abstract and
unconscious linguistic system.
• Native-speaker grammars: are
constrained by built in universal linguistic
principles, known as Universal Grammar
(UG).
Studying L2 Syntax
Assumption: the same innate mechanisms
underlying L2 grammar-building (L1 & L2)
• Non-native speaker grammars: refers to
interlanguage grammars.
• L2 learner language: is systematic & errors
produced by learners do not consist of
random mistakes but, rather, suggest rulegoverned behavior ( Adjemian 1976, Corder,
1967, Nemser 1971 and Selinker 1972).
• L2 learners, like native speakers, represent
the language that they are acquiring by
means of a complex linguistic system.
Full Access theories
• Full Access theories (Schwartz and Sprouse
(1994, 1996); Epstein et al (1996, 1998), Grondin &
White (1996): UG must constitute the initial state in
L2 acquisition.
• --initial state for the L2 L is the set of grammatical
representations determined by the L1, e.g. abstract
feature for ‘past tense’, instead of the surface
phonological realizations of morphemes –s, -ed
• --restructuring the initial-state G based on the L2
input
• --L1 transfer is relevant, but only once syntactic
representations have been sufficiently elaborated to
instantiate the property in question.
Full Access without Transfer
• --Full Access without Transfer: L1 is not
implicated in the interlanguage
representation, initially and subsequently.
(Epstein et al (1996), Flynn 1996).
•  Parameters are set to L2 values, on the
basis of UG interacting w/ L2 input, without a
prior stage of L1 settings.
•  L2 parameter values are attainable
•  no changes in interlg parameter settings
during the course of development
The Full Access and Full Transfer
Hypothesis
• (Schwartz and Sprouse (1994, 1996)): L1 G
constitutes the initial state but that there will
be subsequent restructuring in response to
properties of the L2 input.
• --Parameters are initially set at their L1
values but will subsequently be reset.
•  have changes in interlg parameter settings
during the course of development
Partial-access
• Partial-access (indirect) view, i.e. via
L1: Schachter (1988)
• L2 learners have full access to UG
principles but can only access those
parameters operative in their L1; they may
be able to reset L1 parameters by means of
general learning strategies
• L2 & L1 acquisition are the same in part;
adults fail to achieve full linguistic
competence
No-access view
• (the Fundamental Difference
Hypothesis): Clahsen and Muysken
(1986; Meisel 1991)
• L2 leraners no longer have access to the
principles and parameters of UG; genearl
learning strategies replace UG
• L2 ≠ L1 acquisition; adults fail to achieve
full linguistic competence
Acquisition of Functional
Categories
• IP, CP
• DP
Acquisition of Verbal Inflectional
Morphology
• Dulay & Burt (1973):
• present progressive & contractible copula be >
possessive ‘s & 3rd sg. present -s
• Dulay & Burt (1974):
• present progressive & article a/the, contractible
copula be > possessive ‘s & 3rd sg. present -s
• Bailey et al. (1974) :
• present progressive & contractible copula be,
plural –s > possessive ‘s & 3rd singular present -s
Acquisition of Verbal Inflectional
Morphology
• Andersen (1978)
• V-related morphemes:
• copula > aspect (±progressive: be V-ing) >
tense (±past) > S-V agreement (±3rd P sing.)
• (omitting: have, V-en, and ø article)
• N-related morphemes:
• the > a/ plural –s > possessive ‘s
VP and IP in the L2 acquisition of
English verbal morphology
• (1) Initial State: a stage without IP
•  copula be as an expletive V, which selects AP, NP, PP,
• --He cook/ He cooking (yesterday) /She old/She’s old (from
Stauble)
• -- She no is old.
• (2) the emergence of I
• Aux be selects only V-ing establishing Infl and IP
• Past tense (later than progressive aspect)
• Gueron & Hoekstra 1995: Non-local binding relation between a
Tense Op and I
• (3) 3rd person singular present tense
Acquisition stages
• (1) VP
• (2) IP represented, initially via the
minimally specified copula be
• a. I (be) selects VP complement (V-ing)
• b. Tense Op  past/present tense
• (3) Spec-Head agreement in IP (L1
influence)
• (based on Stauble, 1984)
Acquisition of Nominal Phrases
(the English articles the, a, Ø)
The distribution and interpretation of
English articles
The distribution and interpretation of
English articles
• Bickerton (1981) suggests that these
differences can be captured in terms of two
binary features:
• Whether the article and associated NP:
• refer to a specific entity
[+/- specific referent]
• are already known, from the previous
discourse or form context, to the person
[+/- hearer knowledge]
- HK
+ HK
•
•
•
•
•
•
•
•
•
•
•
(5)
(6)
(7)
(8)
A:
B:
A:
B:
A:
B:
A:
A:
B:
+ SR
(5) a/ 
(7) the
-SR
(6) a/ 
(8) the/a/  (generic)
How will you get a ticket for the England-France match?
I have a contact. I have  contacts.
What does she want to do when she’s married?
Have a baby/ Have  babies.
What does she want to do when she’s married?
Have a baby/ Have  babies.
What will they do with the baby when they go to Africa?
I saw a rabbit eating my carrots yesterday.
The rabbit can cause problems for the gardener.
A rabbit can cause problems for a gardener.
 Rabbits can cause problems for  gardeners.
Studies of the second language
acquisition of English articles
• Parrish (1987)
• --1 19 year-old Japanese speaker when
living in the US for 3 weeks when data
collection began
• --6 yrs classroom instruction in Japan 
beginning level
• --4 months of data collection
• --structured interactions (storytelling &
description of a place)
Parrish (1987)
Parrish (1987)
Results:
• The subject makes a lot of use of the zero
article: 228/436 (52.3%) even in +SR +HK
• The [+SR -HK] cell tells us that there are
some cases where the subject has
overgeneralized the: 9.4%
• -- overgeneralizing : [+SR, +HK]
(inaccuracy rate): 32.1%
Parrish (1987)
• The accuracy of each article in
obligatory contexts
Article
a
the
zero
Number
6/32
37/50
12/13
%
19
74
92
-- is used most in a non-target-like way.
A was never used in contexts where the
or  are required by NSs
Parrish (1987)
• The relative accuracy of each article 
• the is typically being used to mark NPs with
specific reference.
• a is used to mark the fact that the NP is not
known to the hearer.
• a emerges later than the.
• Ø is widely overgeneralized and appears to
function as a ‘default’ article.
& Perdue (1992: 61-88) Klein
• Klein & Perdue (1992: 61-88)
• Two Punjabi-speaking subjects
• In the UK for 13 and 20 months
• Had little instruction in English prior to that
period
• --one subject: no definite article (p. 68)
• --the other: “bare N is the most
frequently used referential device”
Huebner (1985)
• Huebner (1985)
• L2 English of a Hmong speaker from Laos.
• Longitudinal one-year study
• This subject was acquiring English in an
untutored setting in the US and undertook
a follow-up study 20 months later
• Date: every 3 weeks from free
conversation
Two studies of the second language
acquisition of English articles
• A contrast between da (a phonological
approximation to native the) and Ø – there is no
contrast between da and a.
• After 6 weeks, da flooded all contexts.
• Week 21, drop da from [-SR -HK]
• Week 27, drop da from [+SR -HK]
• 20 months later, a had begun to appear in the
[+SR -HK]
Two studies of the second language
acquisition of English articles
• Prrish’s and Huebner’s subjects show a similar
pattern:
• da/the used frequently and predominantly in [+SR +HK],
but also used in [-HK] contexts
• a used much less frequently, but restricted to [-HK]
• All three studies (+Andersen, 1978) suggest that
L2 learners of English acquire the properties
encoded by the English article system
incrementally
• Andersen, 1978: the > a/ plural –s > possessive ‘s
Andersen (1978), Parrish (1987), & Huebner
(1985)
(bare NP)
Specificity in the NP (marked by the/da)
Hearer knowledge in the NP (marked by a/ Ø)
Possessive ‘s
The structure of English
determiner phrases
• Abney (1987)
• John refuses to leave
• John’s refusal to leave
• Possessive ‘s is a morpheme realizing the
category determiner, or D
The structure of English
determiner phrases
• Abney’s proposal is that the article the, a, Ø,
pronouns like my, your, her, etc., and
demonstratives like this, that, etc., belongs to
the class D.
• NPs are the projection of determiner-less Ns
The structure of English
determiner phrases
• [Tense OPi …[IP… Ii ….]]
• [D OPi …[DP… Di ….]]
• If determiner is not co-indexed with the DOperator its interpretation will be ‘unknown
referent’ and a/ Ø will be selected.
[DOPi [IP I Saw Di/Dj rabbit in the garden yesterday]]
• The old man’s friend
• The friend of the old man
• It’s assumed that ‘s assigns genitive Case to its
specifier so DP raise to specifier of the topmost DP to
receive Case.
Grammar-building in the SLA of DPs
(Andersen (1978), Parrish (1987), & Huebner (1985))
(bare NP)
 NP (lexical projection)
Specificity in the NP
(marked by the/da)
 D (head-complement
local selection)
Hearer knowledge in the
NP (marked by a/ Ø)
 non-local D-OP relation
Possessive ‘s
 Spec-Head relation
Grammar-building in the SLA of DPs
(Andersen (1978), Parrish (1987), & Huebner (1985))
VP
I(be) selects VP complement
(NP/AP/ V-ing)
 VP (lexical projection)
 I (head-complement local
selection)
Tense Op (past)
 non-local Tense OP
relation
S-V agreement
 Spec-Head non-local
relation ( (L1 influence)
Initial state in SLA
• the point from which L2 learners start to build
grammars
• Minimal Trees Hypothesis: Vainikka & YoungSholten 1994; 1996a, 1996b)
• ‘only lexical categories are present at the earliest
stage of …L2 acquisition, and that during acquisition
functional projections develop in succession.’ (1996a:
7)
• initial transfer from the L1 of the properties of lexical
categories
• functional categories are not L1 influenced but
developed only in response to positive evidence from
the L2
Initial state in SLA
• The Valueless Features H. (Eubank
(1993/1994, 1994a, 1996)
• --All the categories instantiated in the L1 are
initially transferred into the mental G for the L2, but
the specifications chose for the F categories by
the L1 are neutralized (not specified)
• --optionality
• --the lg-specific properties of VP are transferred in
the early stages
Initial state in SLA
• --Full Access theories (Schwartz and
Sprouse (1994, 1996); Epstein et al (1996,
1998), Grondin & White (1996):
• UG must constitute the initial state in L2 acquisition.
• initial state for the L2 L is the set of grammatical
representations determined by the L1, e.g. abstract
feature for ‘past tense’, instead of the surface
phonological realizations of morphemes –s, -ed
• restructuring the initial-state G based on the L2
input
• L1 transfer is relevant, but only once syntactic
representations have been sufficiently elaborated to
instantiate the property in question.
Initial state in SLA
• Modulated structure building (Hawkins) 2001)
• combination of minimal tree and full access/transfer theories
• initial L2 G: lexical projections in principle, & their structural
properties are determined in principle by L1 (minimal trees)
• restructuring towards the L2 may be rapid, depending on the
evidence available and the nature of the transferred property in
question (full transfer/full access theory)
• functional Ps are established later than lexical Ps (minimal trees),
the rapidity of establishment depending on the evidence available
• ‘modulated’ structure building structure building is influenced by
properties of the L1 at the relevant point in the construction of a G,
and not before
Robertson (2000)
• Optionality: the variable use and
nonuse of a particular surface feature,
with no difference in meaning between
the two forms.
Robertson (2000)
• 3 principles:
• (1) ‘determiner drop’, analogous to ‘pro-drop’, an NP
with def or indef reference need not be overtly
marked for [+/- definiteness] if it is included in the
scope of the D of a preceding NP.
• (2) ‘recoverablity’ principle: an NP need not be overtly
marked for [+/- definiteness] if the information
encoded in this feature is recoverable from the
context.
• (3) ‘Lexical transfer principle’: apply Chinese
demonstratives and numeral yige