The K. family

[Paul Fletcher]

I have only just become aware of the existence of FUNKNET, and I hope
subscribers will bear with me if, somewhat late in the day,  I add a
lengthy postscript to the recent discussion about the K. family. As a
linguist who has worked with Vargha-Khadem's group in London on some of the
data, I feel it's incumbent on me to make some comment. (I should point out
however that the views expressed below are my own and does not represent
any kind of London group view). While I can apologise for trying your
patience, I can make no apologies for concentrating on the data to make
some points which I think bear on the issues.

1. I think it's important to emphasise the differences between the K.
family and the vast majority of subjects designated 'SLI' in the literature
on grammatical impairment over the 25 years since Morehead & Ingram's
initial foray into this field in 1973. First, the sex distribution in the
K. family is very different to that of subjects designated SLI. A glance at
the pedigree for the K. family (first published in Hurst et. al.,
DevMed&ChNeurol. 32, 347-55) shows that of the 30 affected members across
three generations, half are female. This is in contrast to the
well-attested sex ratio of approx. 3:1, boys to girls, in the SLI
population as appearing in the literature. Second, while most of the SLI
subjects in publihsed studies are in the 4-9 yr age-range, the K. family
ranges from elementary school age individuals to a woman in her seventies.
In the light of these two factors alone it would make sense to consider the
K. family separately and not lump them under the general label 'SLI'. From
this I think it follows that we need to be cautious about comparing the
results from this family with non-familially-related, and on average much
younger, Greek and Japanese language-impaired subjects.

2. Re the 'articulation problems' of the affected members of the K. family:
I can assure subscribers to FUNKNET that these are real, and not an
artifact of the testing situation. Let me elaborate. In a paper I presented
at the Evolution of Human Language conference in Edinburgh in April,  I
looked in some detail at a subset of the K. family -- a second generation
affected female, her two affected sons, and an unaffected daughter and
granddaughter. (I showed video clips of these individuals at the
conference). I was particularly interested in vowel pronunciations, (for
reasons that I hope will become apparent). In the to my knowledge only
published (and hence refereed) paper on the phonology of the K. family
(Fee, E. (1995), Clinical Linguistics & Phonetics 9, 189-209),  the author
acknowledges, but decides to ignore, any vowel errors on the part of
affected family members. Because vowel errors are relatively rare in what
are termed phonological disorders in SLI kids, and also because they have
been descibed in the literature on developmental apraxia of speech (DAS),
this seemed an omission which needed correction.

The second and third generation affected members showed vowel problems, in
addition to the consonantal problems documented by Fee; the unaffected
third and fourth generation members did not. In the folowing brief
illustrative extract from one of the 3G affected males (aged 22 yrs.), the
vowels in caps. are mispronounced (including both occurrences of 'I'):

My nAme is ------ and todAy's dAte is the EIghteenth of the second
nInety-six. I am a chEf  and I can tell you how to mAke a Victoria sponge
mix. RIght what to do first......

About 25% of vowels here are errors. 9/10 occur in long vowel targets, of
which all are diphthongs. Further, they are all diphthongs whose offset is
a close front vowel. (Typically the onset is produced as the representative
of the vowel).  This feature of the affected family members I looked at
(and I acknowledge that so far it is only one branch of the family) links
with the DAS literature (see for example Morley 1957, 'Development and
disorders of speech in childhood', Pollock & Hall 1991, ClinLing&Phon  5,
207-24),  where similar diphthongal errors are described. Interestingly,
Pollock & Hall also summarise a constellation of other problems evinced by
those described as developmentally apraxic:
-- family history of speech and language problems
-- persistence of speech problems and their intractibility to remediation
-- non-verbal oral apraxia
-- 'soft' neurological signs
-- language disorder
-- word-finding difficulties
-- academic learning problems
This list of symptoms bears a remarkable similarity to the profile of
performance patiently uncovered by Vargha-Khadem et al. in the K. family.
The purpose of the comparison is not to label the affected members of the
K. family as developmentally apraxic ( a notoriously slippery label), but
to emphasise that as a group (and with the variation you would expect
within a group -- morphological performance also varies from affected
individual to affected individual) they display a constellation of symptoms
which is similar, at first blush, to a well-attested profile in other
impaired individuals. Some of the elements of this profile (in DAS) have
been attributed to motor planning problems. If this is correct, it may also
apply to some of the elements of the K. family profile. And it seems to me
to be premature, at the very least, to attribute an ontological primacy to
any single one of these symptoms. A necessary condition for ever
understanding what is going on here is a precise and exhaustive description
of the phenotype, and it's far too early to foreclose on that.

3. Of course you can't find out everything about the language of impaired
individuals by using standardised tests. But you can find out something.  I
take it that the excitement caused by Gopnik's earlier reports on affected
members of the K. family, among the linguistic community, had to do with
the fact that they appeared to identify a selective grammatical deficit.
If the fault-lines of a deficit in SLI individuals demarcate a 'natural
class' of categories from the grammar, this could be taken as support for
the Chomskyan view of how the LAD is structured and the role that it plays
in acquisition. For this assumption of a selective 'representational'
deficit to hold,  I take it that there should be evidence available from
both comprehension and production on the relevant deficits. In the
Edinburgh paper I also reported some findings form a comprehension test,
TROG (Test for the reception of Grammar, standardised in the UK by Dorothy
Bishop on 4 - 12 yr. olds), which provides some indirect evidence on
whether the clear evidence of a production-side morphological deficit is
paralleled in comprehension.  TROG is a sentence/picture matching task,
addressing a range of morphosyntactic areas: gender and number in pronouns,
plural, comparative, prepositions, and grammatical relations via
active/passive sentences and complex sentences with embedded relative
clauses.  The Vargha-Khadem group applied TROG to the K. family., with the
following outcomes:
a) There is a significant performance difference between affected and
unaffected family members, but affected members still perform at close to
90% on the task, indicating a receptive grammatical ability roughly
comparable to the average 10 yr old (the average 10 yr old Brit, that is).
b) Failing items on the TROG,  for the affected members of the family,  are
not randomly distributed, but involve those items on the test which involve
embedded relative clauses.
Verb inflectional contrasts, the crux of the issue for Gopnik, are not
included in the TROG, so we have no direct evidence on these items. However
passive verb morphology IS a part of the test, in the section in which
subjects have to match four reversible passives with appropriate pictures,
and affected family members who have output tense problems perform well on
this section. It is difficult to believe that a receptive command of the
verb forms signifying passive would cooccur with a lack of command of past
tense.  But unless specific production limitations are identified in
comprehension also, it is difficult to argue a competence-based deficit.

At which point I rest my case, fearful of trying the patience of FUNKNET
subscribers any further. I should point out that Myrna and I have known one
another a long time, and I for one am grateful to her for playing the role
that she has in bringing the issues relating to the family in particular
and to SLI in general under the spotlight. It will be obvious that we
disagree pretty fundamentally on tthe interpretation of the data.  I am
however with her unequivocally when she says 'the only way to answer this
question is to look at what they do and how they do it and why they do
it....'.