LIDES/CHILDES

1
LIDES/CHILDES

• the taxonomic phase of syntactic code-mixing
  research

2
The Language Interaction Data Exchange System
(LIDES)

• The main aims of LIDES are to
• maximise the use of available data
• allow data sets to be compared
• answer questions that cannot be answered on the
  basis of one data set
• improve the quality of research and of the data
  in this field through mutual scrutiny
• promote advances in language interaction research
  and research methodology, analogous to the
  advance in first language acquisition research
  resulting from the CHILDES project (see Sokolov
  Snow 1994)
• help ensure that data gathered using public funds
  is made available to the (scientific) community

3
The LIPPS Group

• is an international group of researchers whose
  aim is to develop transcription and coding
  standards for bilingual data, and to set up a
  computerised database of code-switching data.
• is composed of
• the founding members Penelope Gardner-Chloros
  (Birkbeck College London), Roeland van Hout
  (Tilburg University), Melissa Moyer (Universitat
  Autònoma de Barcelona) and Mark Sebba (University
  of Lancaster)
• Members Ruthanna Barnett, Eva Codó, Montse
  Forcadell, Maria Carme Torras, Ad Backus, Lous
  Boumans, and other members who are engaged in
  research in the field of language interaction and
  use and/or contributed data.
• the advisory committee, formed by Pieter Muysken,
  François Grosjean, Maria Teresa Turell, Marianne
  Starren, Jeannine Treffers-Daller, Jacomine
  Nortier and Margaret Deuchar

4
CHILDES

• the database of language acquisition data
  transcripts (English USA UK, Celtic, East
  Asian, Germanic, Romance, Slavic, and other
  languages bilingual acquisition, language
  disorders, narratives), digitised audio and video
  files
• CHAT (Codes for the Human Analysis of
  Transcripts), or
• Conversation Analysis (CA) Transcription, or the
• Berkley Transcription System (BTS) for Signed
  Langauge
• CLAN (Child Language Analysis Programmes)
• language acquisition teaching material and an
  online bibliography (also available in Endnote)

5
The Transcription and Coding Systems

• The LIDES transcription and coding format
• (The LIDES Coding Manual IJB 4/2)
• has been adapted from CHAT (McWhinney 2000) for
  the use on language interaction data.
• It consists of
• headers
• the main speaker and
• dependent tiers

6
Headers

• _at_Participants DOR Dorit Adult, LIL Lily Sibling,
  TRU Trude Sibling
• _at_Age of DOR 74
• _at_Coder Eva Eppler
• _at_Education of DOR gymnasium
• _at_Filename jenn1.cha
• _at_Language English_at_2, German_at_4
• _at_Language of LIL, MEL, DOR, TRU, EVA English,
  German
• _at_Age of Onset of DOR 20
• _at_SES of LIL, MEL, DOR, TRU MMC
• _at_Sex of LIL, MEL, DOR, TRU female
• _at_Begin
• _at_Date 18-JAN-1993
• _at_Situation playing a game of cards
• _at_Comment LIL is telling the story of her car
  accident

7
LIDES Basic

• DOR ich_at_4 lt hab(e)_at_4 geglaubt_at_4 the_at_2 lorry_at_2
  is_at_u gt xxx .
• LIL no_at_2 lt the_at_2 lorry_at_2 did-'nt_at_2 do_at_2 /.
• DOR der_at_4 wind_at_4 hat_at_4 (e)s_at_4 aufgeblasen_at_4
  und_at_4 the_at_2 lorry_at_2 hat_at_4/.
• glo the wind has it open blown and the lorry ha
  s
• tra the wind blew it open and the lorry

8
CHILDES (Brown Corpus)

• _at_Participants CHI Eve Target_Child, MOT Sue
  Mother,
• _at_ID enbrownCHI16.0Target_Child
• _at_Age of CHI 16.0
• _at_Date 15-OCT-1962
• CHI more cookie .
• mor qnmoreadvmore ncookie .
• spa IMP
• int distinctive, loud
• trn qnmore ncookie .
• MOT how about another graham cracker ?
• mor advwhhow ptlaboutprepaboutadvabout
  detanother ngraham ncracker ?
• trn advwhhow prepabout detanother ngraham
  ncracker ?
• CHI more juice ?
• mor qnmoreadvmore njuice ?
• trn qnmore njuice ?

9
Analysis Programmes

• a very small selection of the existing ones

10
CHECK

• checks the overall and detailed structure of
  newly transcribed CHAT files
• depfile (depfile.cut) compare the syntax of the
  files with that of the programme
• 00depad.file specific to data sets

11
FREQ

• computes the frequencies of the words in a file
  or files.
• freq s_at__ tDOR Jen1.cha
• establish the distribution of words from each
  language as produced by speaker DOR in file
  Jen1.cha

12
COMBO

• carries out string searches that match patterns
  of letters, words, or groups of words and symbols
  in the data files
• combo s\_at_2\_at_4 find all code-switches from
  English to German
• combo s\_at_2\_at_4\_at_4\_at_2 _at_ u tDOR
• find all code-switches in both directions in all
  the files entered for speaker DOR
• combo sbecause\_at_2\_at_4 w2 w2
• find all utterances containing because followed
  by a German words and list the two utterances
  preceding and following the hit

13
MLU MLT

• MLU
• automatically calculates the number of morphemes
  per utterance and the number of utterances for
  individual speakers.
• MLT
• computes the mean number of words per utterance.

14
FLO

• creates a simplified version of a main CHAT/LIDES
  line
• DOR ich lt hab(e) geglaubt the lorry is gt
  xxx .
• LIL no lt the lorry did-'nt do /.

15
The syntax-pragmatics interface

• because German main clause word order

16

• ... we do not yet know enough about the relation
  between frequency distributions of specific
  grammatical patterns in monolingual speech data
  and properties of the grammar to handle frequency
  in bilingual data.
• (Muysken 1995 185)

17

• ... it is possible that foreign conjunctions do
  not trigger verb-final in Dutch and German matrix
  clauses simply because they are used in functions
  that require main clause order. ... It is hardly
  feasible to examine this hypothesis in relation
  to the published examples because these are for
  the most part presented out of context. (Boumans
  1998 121)

18
Restrictive (Rutherford 1970)

• The main and the subordinate clause form one
  complex proposition
• with a presupposed main clause
• and
• an asserted subordinate clause.

19
Non-restrictive (Rutherford 1970)epistemic
(Sweetser 1990)disjunctive (Quirk et. al. 1973)

• The because-clause provides a reason why a
  speaker knows or has asserted a proposition i.e.
  an elaboration of a proposition, rather than a
  cause or reason within one complex proposition.

20
Discourse (Schiffrin 1987, Schleppegrell 1991)

• Continues, expands or elaborates the ideational
  content of the text, i.e. and interpretive link.
• In discourse this interpretative link may be
  interactional or textual rather than semantic.

21
Conclusion 1

• The quantitative figures on this construction
  type demonstrate that,
• contrary to what Muyskens (2000 29) claims,
• we do know enough about the relation between
  frequency distributions
• of at least one specific grammatical pattern
• in monolingual speech data and properties of the
  grammar to handle frequency in bilingual data.

22
Conclusion 2

• The results verify Boumans (1998 121) hypothesis
  that
• ... foreign conjunctions do not trigger
  verb-final in Dutch and German clauses because
  they are used in functions that require main
  clause order.
• These functions are epistemic and discourse in
  both languages.

23

• This result also demonstrates that competent
  code-switchers identify exceptional cases of
  categorial equivalence at the syntax-pragmatics
  interface in order to facilitate switching.

24
the taxonomic phase

• Muyskens (2000) Bilingual Speech aims to tie
  together of a set of intermediary results of the
  taxonomic phase of syntactic code-mixing
  research. He identifies three primitives

25

• the facilitating influence of categorical
  equivalence and
• equivalent surface word orders
• the facilitating role of peripherality
• restrictions on function words, both as selected
  and as selective elements

26
The most important findings

• to emerge from LIDES based research in general
  my own work in particular is that is that
• although it is based on more data on more
  language combinations
• my grammatical analysis is conducted in a
  lexically based dependency grammar rather than a
  hierarchical phrase-structure theory

27

• it supports two of the primitives identified
  by Muysken
• the facilitating influence of categorical word
  order equivalence, and
• the facilitating role of peripherality in the
  clause