new SLI corpus

Sat Jul 28 05:32:47 UTC 2001

Dear Info-CHILDES,

  I would like to announce the addition to CHILDES of a new corpus from Gina
Conti-Ramsden, Ludovica Serratrice, Kate Joseph, and Rachel Hick, all from
the University of Manchester.  I am referring to this corpus as conti3.sit
and conti3.zip, since it is the third corpus on specific language impairment
(SLI) contributed by Gina Conti-Ramsden and her co-workers at Manchester and
we don't want to confuse this Manchester SLI corpus with the Manchester
corpus on normally-developing language.  This one is in /clinical whereas
the other is in /english on the server.
  This corpus includes longitudinal data across 18 months from four English
SLI children.  It also includes a completely disambiguated %mor line for all
of the utterances in all of the files.  This is the first SLI corpus with a
full %mor line and could be an excellent source of information on
grammatical patterns in SLI.  I am appending the readme file. I would love
to send it as a nice Word attachment, but given the rash of viruses on the
Internet, perhaps it is best to continue to keep attachments off of
info-childes.

--Brian MacWhinney

MANCHESTER SLI

Address for correspondence: Gina Conti-Ramsden

Gina Conti-Ramsden 
Human Communication and Deafness
School of Education
University of Manchester
Oxford Road
Manchester M13 9PL
gina.conti-ramsden at man.ac.uk

Ludovica Serratrice
Human Communication and Deafness
School of Education
University of Manchester
Oxford Road
Manchester M13 9PL
Serratrice at man.ac.uk

Kate Joseph
Human Communication and Deafness
School of Education
University of Manchester
Oxford Road
Manchester M13 9PL
k.joseph at man.ac.uk

Rachel Hick
Human Communication and Deafness
School of Education
University of Manchester
Oxford Road
Manchester M13 9PL
r.f.hick at stud.man.ac.uk

This corpus includes longitudinal data from four monolingual British
children with Specific Language Impairment (SLI) ranging from 2;6 to 4;0 at
the beginning of the study and from  3;11 and 5;0 at the end of the study.

Participants

The four children in this study, one girl and three boys, were recruited
through a number of speech and language therapists in the North West of
England. The therapists were initially informed by letter of the criteria
for participation including the following: age range between 2;6 and 4;0,
early stages of multiword speech, no history of hearing problems, good
degree of intelligibility, non-verbal abilities within the normal range, no
obvious autistic tendencies, poor language abilities including poor
receptive abilities. Parents of children meeting these criteria were
subsequently contacted and visited at home by two researchers. In the
initial screening visit one of the investigators explained the aims and
methods of the study to the parents and helped them fill in an anglicised
copy of the MacArthur Communicative Development Inventory Words and
Sentences (Fenson, Dale, Reznick, Thal, Bates, Hartung, Pethick, Reilly,
1993) to assess the child¹s lexical skills and his/her productive use of
multiword utterances. The parents were also asked to complete a
questionnaire including information on parental education and occupation,
number of siblings, pregnancy and birth history, family history of speech
and language impairment, learning difficulties and mental retardation, and
child¹s previous ear infections and hearing problems. The Autistic Screening
Questionnaire (Berument, Rutter, Lord, Pickles & Bailey, 1999) was also
administered to ensure that the children did not have any obvious autistic
tendencies. During the initial visit a twenty minute spontaneous speech
sample was also obtained while one of the investigators played with the
child and administered the receptive component of the Reynell developmental
language scales (Edwards, Fletcher, Garman, Hughes, Letts & Sinka, 1997). In
order to be enrolled in the study the children had to present receptive as
well as expressive difficulties and only children below the 16th centile
were therefore included. A measure of non-verbal ability was also elicited
using the Leiter International Performance Scale (Leiter, 1969) to ensure
that the children were within the normal range.
Once the children were identified as potential participants in the study the
parents were required to sign a consent form in which they agreed to take
part in the study with their children for a period of up to 18 months.

Table 1 provides information on children¹s age, sex, birth order, MLU in
words at the start of the study and test scores measuring receptive language
skills and non-verbal ability.

Table 1: Manchester SLI Children

Child    Sex    Birth order    Age    MLUw    Reynell receptive
(centile)    Non-verbal I.Q.
Bonnie    F    1    4;0    2.85    1    90
Dan    M    2    2;6    1.06    10    129
Harry    M    1    3;4    1.78    13    120
Nathan    M    2    2;10    1.23    1    117

At the start of the study all the children were attending mainstream nursery
schools and were either receiving speech and language therapy or had
received it in the previous six months. By the end of the data collection
period two of the children, Bonnie and Harry, were enrolled in mainstream
primary schools and Bonnie had the support of a teaching assistant.

Data collection

Data were collected for a period of sixteen months at fortnightly intervals
in the children¹s homes with breaks due to illness and family commitments.
One of the children originally recruited in the study was excluded after
three months due to difficulties in keeping up with the recording schedule.
Bonnie replaced the child who was withdrawn four months into the project
therefore data collection only lasted twelve months for this child.
Each session lasted for approximately an hour and it involved the child and
the mother playing together in a quiet room. Older siblings were
occasionally present during the recordings with Harry and Nathan, and
Bonnie¹s newborn brother was also in the room in most of the recordings. The
investigator was not normally present during the recordings with Dan and
Harry, except when the mother had to leave the room or at the beginning and
at the end of a session in order to deal with the equipment. In the case of
Bonnie and Nathan the investigator spent longer periods of time with the
children while the mother was otherwise engaged, especially from session
sixteen onwards in Bonnie¹s recordings. The activities the children engaged
in during the recordings ranged from looking at picture books to playing
with toy trains, drawing, playing with Lego. The investigators also provided
a set of toys including miniature Playmobil people and a variety of assembly
sets such as a café, a log cabin with fishing boats, a farmyard, a
playground and a furnished house. The new toys were quite successful in
engaging the children¹s attention and in generating a considerable amount of
speech with their mothers and the investigator. In the last three months of
the study the investigators supplied three different story books with a
clear sequential narrative in the attempt to elicit more complex structures
and longer mother-child exchanges.

All sessions were audio-taped by a portable Sony digital minidisc recorder
MZ-R35 and an ATR97 ominidirectional boundary microphone placed on a flat
surface near the recorder. At monthly intervals the sessions were also
videotaped using a Panasonic VHS-C camera installed on a tripod.

In addition to collecting spontaneous data of mother-child interaction a
number of psychometric tests were also administered to assess children¹s
receptive and expressive lexical and grammatical abilities: the British
Picture Vocabulary Scales (Dunn, Dunn, Whetton & Pintillie, 1982), the
Expressive Vocabulary Test (Williams, 1997), the Clinical Evaluation of
Language Fundamentals-Preschool (Wiig, Secord & Semel, 1992), the Children¹s
Test of Non-word Repetition (Gathercole & Baddeley, 1996). All tests were
conducted in a quiet room by one of the female investigators, mothers were
not normally present unless the child was particularly distressed or
unwilling to co-operate. All of the tests were repeated once after a minimal
interval of six months.

Table 2. Manchester SLI Test Scores

    Bonnie    Dan    Harry    Nathan
Test    Age    Centile    Age    Centile    Age    Centile     Age
Centile
BPVS    4;1    30    3;1    58    3;9    34    3;4    55
CELF    4;3    7    3;9    53    3;10    45    3;4    47
Linguistic concepts        1        75        75        63
Basic concepts        1        75        37        63
Sentence structure        5        50        75        95
Receptive lang        1        70        66        82
Recalling sentences        5        16        25        16
Formulating labels        5        75        50        37
Word Structure        1        25        9        5
Expressive lang        2        34        21        16
CN-REP    4;10    11-15    3;11    50    4;7    25 - 50    4;0    25
EVT    4;3    21    3;3    47    3;10    61    3;4    68

Transcription and coding

The spontaneous data were orthographically transcribed in CHAT format by the
three trained researchers who originally collected the samples. Pauses,
hesitations, interruptions, overlaps and retracings were also coded as
accurately as possible. All names, except the investigators¹, were changed
to preserve anonymity.
A complete morphological tagging was added to the main line by using the MOR
and POST programmes. Although the automatic tagging was fairly accurate and
reliable, additional manual checking of all the data was conducted to ensure
that ambiguous cases were resolved appropriately.
A number of changes to the coding system were introduced to include new
codes and minor amendments to previously existing ones:

1. New postcodes were introduced for tagging self-repetitions [+ srp] and
imitations [+ imi] of an immediately preceding utterance and non-lexical
responses [+ nlr]. The first two codes were used only for utterances
containing a minimum of two words and for those utterances that were an
exact repetition of the immediately preceding utterance used by the speaker
or identical to an immediately preceding utterance used by another speaker.
The [+ nlr] code was used whenever the child attempted a verbal response but
it was of a non-lexical nature, i.e. a vocalisation, although not a
recognisable word, that was obviously a response to a previous maternal
turn.

2. The MOR coding does not include a distinction between past tense forms
and perfective participle forms for regular verbs or for irregular verbs in
which the two forms are homophonous. An automatic search for past tense
forms therefore may also include a number of forms which are in fact
participial forms, e.g. in MOR does not make a distinction between ³I walked
for hours² and ³I have walked for hours², the form ³walked² is a past tense
in the former sentence but a participle in the latter, while the coding is
v|walk-PAST in both cases. To avoid the inclusion of participle forms in the
count for past tense forms an additional code was introduced as illustrated
below: 

walked  = v|walk-PERF  instead of   v|walk-PAST       when participle
put        = v|put&PERF    insead of    v|put&PAST        when participle

3. The distinction between adjectives and participles is often a very
difficult one to make and one which has implications in the coding of the BE
verb they appear with. The decision was made to treat the following words as
adjectives in constructions containing a BE verb unless there were clear
indications that the construction was really a passive, i.e. in the presence
of a by-phrase: allowed, bored, bothered, broken, called, closed, done,
excited, fed up, finished, fixed, gone, locked, meant, mixed up, mended,
shattered (= tired), shut, stuck, supposed.

*CHI:    this toy was broken.
%mor:    det|this n|toy v|be&3S adj|broken.

Publications using these data should cite:

Serratrice, L., Joseph, K.L. & Conti-Ramsden, G. (in press). The acquisition
of past tense in pre-school children with Specific Language Impairment and
unaffected controls: regular and irregular forms. In K.Lindner and Z. Penner
(eds.), Special Issue of Linguistics.

References

Berument, S.K., Rutter, M., Lord, C., Pickles A., & Bailey, A. (1999).
Autistic Screening Questionnaire: Diagnostic Validity. British Journal of
Psychiatry, 175, 444-451.

Dunn, L.M., Dunn, L.M., Whetton, C. & Pintillie, D. (1982). The British
Picture Vocabulary Scale. Windsor: NFER.

Edwards, S., Fletcher, P., Garman, M., Hughes, A., Letts, C. & Sinka, I.
(1997). The Reynell Developmental Language Scale III: the University of
Reading Edition. Windsor: NFER.

Fenson, L., Dale, P.S., Reznick, J.S., Thal, D., Bates, E., Hartung, J.,
Pethick, S., Reilly, J. (1993). The MacArthur Communicative Development
Inventories: user¹s guide and technical manual. San Diego, CA: Singular
Publishing Group. 

Leiter, R.G. (1969). The Leiter International Performance Scale. Chicago,
IL: Stoelting. 

Williams, K. (1997). The Expressive Vocabulary Test. Circle Pines, MN:
American Guidance Services.

Wiig, E.H., Secord, W & Semel, E. (1992). Clinical Evaluation of Language
Fundamentals-Preschool. New York, NY: The Psychological Corporation.
Gathercole, S.E. & Baddeley, A.D. (1996) The Children¹s Test of Nonword
Repetition. London: The Psychological Corporation. 