The many faces of distribution: tracing the development of linguistic

advertisement
The many faces of distribution:
tracing the development of
linguistic structures in learner
writing
Maisa Martin
University of Jyväskylä, Finland
Development of linguistic skills




Three dimensions of development: fluency,
accuracy and complexity.
All difficult to operationalize but complexity
particularly so.
The topic: many faces of complexity – in
search for measures.
Related concepts:


distribution
context
Stucture of the presentation





The Cefling Project and the SLATE network
DEMfad Model
Grammatical complexity measures
Problems and solutions (?)
Discussion
CEFLING: The linguistic basis of the
Common European Framework levels:
Combining second language acquisition
and language testing research
SLATE



Second Language Acquisition and Testing in
Europe
An informal network to bring together
researchers from SLA and Testing
Research around the Common European
Framework of Reference (CEFR) levels and
their relationship to linguistic development
Cefling Rationale
CEFR very influential in Finland:
- * school curricula
- * adult education curricula
- * National Certificates of Proficiency
- * citizenship requirements
Research Question

What combinations of linguistic features
characterise learners’ performance at
the proficiency levels defined in the
Common Framework and its Finnish
adaptations?
Subjects and Languages






Written performances of adults taking the
National Certificate of Proficiency
examination (3 texts per participant)
Similar texts on similar tasks from young
learners (grades 7 – 9, ages 13-16)
L2 Finnish – L1 varies
L2 English – L1 Finnish or Swedish
L1 control groups
Tasks: formal and informal messages,
argumentative texts, narratives
Rating of the Data




functional can-do scales, no reference to
language
three – four trained raters per writing sample
inclusion for main data: complete interrater
agreement or two in agreement, one + or – one
level
Problems with certain levels
 not enough A1 for adults and C1 and C2 for
young writers
Coding of the Data



CHILDES (.chat format, CLAN tools)
Basic coding for all data, structural features
as needed (several for Finnish, fewer for
English at the moment)
Automatic analyses for English only
DEMfad MODEL
(Franceschina et al. 2006)
D
D = Domain
E = Emergence
f
E
a
M
M = Mastery
f = frequency
d
a = accuracy
d = distribution
DEMfad MODEL






Domain = linguistic structure or vocabulary
area in focus
Emergence = first occurrence (chunk or not)
Mastery = 80 – 90 % occurrence in obligatory
contexts
Frequency per 1000 words (~ Fluency)
Accuracy (%, criteria defined for each D)
Distribution ~ Complexity?
Complexity measures




Wolfe-Quintero et al. 1998
Larsen-Freeman 2006
Verspoor et al. 2004
Banerjee et al. 2007
Wolfe-Quintero et al.


Clauses, T-units, sentences
Reduced clauses, dependent clauses,
passives, passive sentences, adverbial
clauses, adjective clauses, nominal phrases,
prepositional phrases, preposed adjectives,
pronouns, articles, connectors, transitional
connectors, subordinating connectors,
coordinating connectors.
Larsen-Freeman et al.


grammatical complexity = average number of
clauses per t-unit
many other measures as well, but listed as
measures of fluency, accuracy, or lexical
complexity
Verspoor et al.







Development is defined as greater fluency (more words
per entry, fewer Dutch words, longer sentences, and
more complexity) and/or accuracy (fewer misspellings,
fewer grammatical errors, and so on).
i.e. do not separate grammatical complexity from other
features (choice below mine):
the number of conjunctions used to connect clauses,
the use of tenses
the word order
determiners
prepositions
Banerjee et al.








T-units
Main clauses
Dependent clauses: relative
Dependent clauses: adverbial
Dependent clauses: non-finite
Fragments
’Double’ embedded clauses
Ellipsis
Problems

Overlap between frequency, accuracy, and
complexity: is this a problem?







No, if overall development is the aim
Yes, if more detailed profile is sought
Low predictive value
Concentration on syntax – morphology can be
complex as well
Is complexity processability?
Is complexity computational?
Is complexity desirable?
Complexity as processability

Processability Theory




Processing capacity a feasible factor in the growth
of complexity +
Clear framework with spelled-out methodology +
Addresses limited area of the growth of
complexity (syntax) Dependent on one grammatical theory (LexicalFunctional Grammar) -
Example 1: Computational
complexity. Error % by word type
30
25
20
i:i
i:e
15
i:e:C
si
10
5
0
A1
A2
B1
B2
C1
C2
TOTAL
Is complexity desirable?
Example 2


Subordinate clauses per T-unit
What if


Subordination is a simple matter of starting the
clause with a conjunction – with adults the idea of
subordination exists > why complex?
The culture values short and clear sentences and
lengthy and complicated expressions are not
considered good style?
Other approaches


Construction Grammar: growth as extension
of constructions (semantic, lexical, structural)
Growth as distribution: number of contexts of
a structure


”Close distribution” (mandatory arguments,
Paavola 2008)
Growth at several levels: layers of
morphology and syntax (Nieminen 2007 in
L1)
Conclusion

Wolfe-Quintero et al. 1998:
”the goal of - - grammatical complexity studies
should be to correlate individual measures for
a wide variety of structures - - but that is
largely uncharted territory.”
References








Banerjee J, Franceschina F, Smith AM. 2007 Documenting features of written language
production typical at different IELTS band score levels. In: IELTS Research Reports 7. London:
British Council.
Franceschina, F. 2007 Aspects of the development of number marking in L2 English. Paper
presented at AFinLA Conference, Kouvola, Finland, 9-10 November 2007.
http://www.jyu.fi/hum/laitokset/kielet/cefling/en/pub
Larsen-Freeman, D. 2006 The Emergence of Complexity, Fluency,and Accuracy in the Oral and
Written Production of Five Chinese Learners of English. Applied Linguistics 27/4: 590–619.
Martin, M. 2007 Emergence, mastery and distribution. Can acquisition criteria be combined?
Paper presented at EUROSLA 17 Conference, 11-14 September 2007, Newcastle-upon-Tyne,
UK. http://www.jyu.fi/hum/laitokset/kielet/cefling/en/pub
Nieminen, L. A complex case: a morphosyntactic approach to complexity in early child language
2007. Jyväskylä Studies in Humanities.
Paavola, V. 2008 Haluatko menna muunkansa kalastaman? Verbiketjujen kehkeytyminen suomi
toisena kielenä -oppijoiden kielessä. MA thesis. University of Jyväskylä.
Verspoor, M.H., K. de Bot & W.M. Lowie, “Dynamic systems theory and variation: a case study in
L2 writing.” In H. Aertsen, M. Hannay & R. Lyall, Words in their places: a Festschrift forJ. Lachlan
Mackenzie. Amsterdam: VU, 2004. pp. 407-421.
Wolfe-Quintero, K. & Inagaki, S. & Kim, H.-E. 1998 Second language development in writing:
measures of fluency, accuracy, and complexity. Technical report #17. Second Language Teaching
& Curriculum Center. University of Hawai’i at Mãnoa.
Download