Language against the Odds or rather not: a case study

1
Language against the Odds or rather not a case
study

• Mila Vulchanova team, Language Acquisition
  Language Processing Lab, NTNU

2
The classical picture of Savants

• Splinters of genius among a sea of difficulties
• Savant syndrome is a rare but extraordinary
  condition in which persons with serious mental
  disabilities, including autistic disorder, have
  some island of genius that stands in marked,
  incongruous contrast to overall handicap
• Treffert (2000)

3
The hierarchy

• Splinter skills
• Talented savants special skill is more prominent
  and in contrast to overall disability (within
  individual)
• Prodigious savant special skill oustanding and
  spectacular even in a non-impaired individual
  (cross-individual)

4
Recent research

• A more subtle and nuanced picture (Vital et al.
  2009, Happé Vital 2008, Howlin et al. 2009)
• Special abilities characterised by discrepancy at
  two levels
• Intra-individual
• Inter-individual
• Savant skill vs. exceptional cognitive skill
  (Howlin et al. 2009)
• Special ability positively associated with IQ
  (Vital et al. 2009)
• What is it about autism that predisposes to the
  development of talent?

5
Outline

• The case on-going research
• Issues
• diagnosing talented individuals
• the classical savant picture revised
• Comorbidity
• The source of language talent
• Tentative conclusions

6
People

• Valentin Vulchanov, NTNU
• Rik Eshuis, NTNU
• Matthias Weisgerber, Konstanz University
• Margarita Stankova, New Bulgarian University
• Hermundur Sigmundsson, Psychology NTNU

7
The case history

• EP, 10 years old, female
• L1 Bulgarian
• Lives in a relatively small village lives with
  father and grandmother (parents divorced)
• Has acquired German from watching (childrens
  programmes on) a German TV channel!?
• According to teachers has very good memory can
  memorise poems read in class on a single hearing

8
Profile

• Talent language learning
• Deficits dyslexia, autistic traits
  (Asperger/PDD)
• Dyslexia language talent an unusual
  combination
• The most commonly attested cases of talent
  witness a double dissociation between language
  and non-linguistic cognition (Smith Tsimpli
  1995) However the magnocellular theory (cf.
  Stein Talcott 1999) attributes dyslexia to a
  more encompassing deficit in the processing of
  fast transient stimuli
• The comorbidity among language, speech and
  reading disorders (Goulandris et al. 2000,
  Pennington Bishop 2009)
• Autistic traits language talent more common,
  however far from the classical idiot savant
  picture (IQ scores average)

9
German L1 or L2

• Preliminary testing (in Bulgaria) very advanced
  German, has enrolled in a school with intensive
  German instruction due to this
• Results from comprehensive testing in Trondheim
  by a German linguist extremely advanced level of
  German, close to L1 (comparison to peers
  acquiring German as L1)
• Recalls more German motion words than Bulgarian
• Child L2 acquisition? But unusual circumstances!

10
IQ scores

• HAWIK-R
• VIQ 109, PIQ 125, IQ 119
• Test z-score Cum.Prob.
• Information .009 .504 (within
  the range of average 50 of the norm group)
• Vocabulary .335 .631 (within
  the range of average 50 of the norm group
• Arithmetic .639 .739 (within
  the range of average 50 of the norm group)
  (but upper side)
• Comprehension .883 .811 (within
  the range of top 25 of the norm group)
• Similarities .927 .823
  (within the range of top 25 of the norm group)
• Picture Completion 1.683 .954 (within
  the range of top 5 of the norm group)
• Block Design 2.433 .993
  (within the range of top 1 of the norm group)
• Object Assembly 1.871 .970 (within
  the range of top 5 of the norm group)
• Picture Arrangement 1.206 .886 (within
  the range of top 25 of the norm group)
• Coding -.576 .282 (within the
  range of average 50 of the norm group
  (but lower side)

11
(No Transcript)
12
HAWIK-R results comments

• Excelling in Block design is the most common
  exceptional skill among individuals with ASD
  traits (Howlin et al. 2009)
• All scores are quite within the norm, and many
  are very high (BD, PCo, OAss)
• PCo, OAss point in the direction of attention to
  detail, local processing (the Weak Coherence
  hypothesis of Frith Happé), but also
  hyper-systematising (Baron-Cohen et al. 2009)
• Recent hypothesis by Happé Vital (2009) that
  attention to detail accounts for co-occurring
  talent in ASD individuals
• This of course does not explain the language
  talent, but maybe
• Block design, Object assembly and Coding are
  meant to check visual-motor coordination A
  disscoiation b/n BD OAss on the one hand and
  Coding on the other ? problem with localising and
  sequencing small targets, temopral
  processing/magnocellular account (Stein Talcott
  1999)

13
Dissociations/asymmetries

• Between verbal IQ and PIQ in some areas, with
  peaks in PIQ
• BD, PCo, OAss all (supposedly) tap spatial
  cogniton/abilities
• An observed dissociation between HAWIK-R PIQ (and
  those tasks specifically) and a problem with
  spatial language (literal interpretation of
  prepositions, verbs, prefixes), inability to
  describe dynamic spatial events (objects falling
  down, moving in space) when telling stories
  (primarily resorts to gestures/sound imitation)

14
Dyslexia profile

• Many of the behavioural symptoms (BDA)
• General
• Vision, reading spelling
• Hearing and speech
• Writing and motor skills
• Math and time management
• Memory and cognition
• Behaviour, health, development and personality

15
Symptoms

• General
• Talented in art, drama, music, story-telling,
  special interest in certain technical inventions
  (e.g., visit to Science Centre, Trondheim)
• Gets lost easily or loses track of time
• Difficulty sustaining attention
• Learns through hands-on experience,
  demonstrations, observation, visual aids

16
Symptoms

• Vision, reading and spelling
• Complains of dizziness, headaches while reading
• Reading and writing shows omissions, reversals in
  letters, poor syllabification
• Seems to have difficulty with vision, yet eye
  exams do not reveal a problem
• Extremely keen sighted and observant
• (relatively) Poor comprehension
• Spells phonetically and inconsistently

17
Symptoms

• Hearing and speech
• Extended hearing, hears things not apparent to
  others, easily distracted by sounds
• Observe that this is consistent with some
  autistic traits (Baron-Cohen et al. 2009 on
  sensor hyper-sensitivity)

18
Symptoms

• Writing and motor skills
• Trouble with writing, handwriting varies,
  illegible
• Clumsy, uncoordinated, poor at ball games (Stein
  Talcott 1999, Sigmundsson team, on-going
  research)
• Often confuses left/right, poor sense of
  direction (also proprioception, own body in
  surroundings)

19
Symptoms

• Math and time management
• Difficulty telling time (cannot read the clock),
  managing time (no sense of time)
• Math tasks knows answers but cannot do it on
  paper
• Can do arithmetic, but fails on word problems
  (cf. performance on HAWIK-R arithmetic)
• However, telling time may have a different
  underlying mechanism, e.g., transferring visual
  information (spatial) to digital/numerical!!!

20
Symptoms

• Behaviour, health, development, personality
• Extremely disorderly
• Late in crawling, walking, problems tying
  shoe-laces (Stein Talcott 1999)
• Sensitive to foods
• Light-sleeper, wetting
• Mistakes and symptoms increase dramatically with
  confusion, emotional stress

21
RAN-test

• RAN (Denckla Rudel 1975) results compared to
  controls clearly show a dyslexic profile specific
  for that age (9-10 years)

22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
Reading

• Text
• z-vlaue 2.395
• Cum. prob. .992
• In the slowest 1 of the population (20 normal
  controls)
• Compared to controls
• Extremely slow
• Many mistakes clitics, complex words
  (polysyllabic), prosody

28
Further IQ testing

• KBIT-2 test compared to 20 controls
• Test z-score
  Cum.Prob.
• Verbal knowledge .474
  .68
• Riddles .435
  .67
• Matrix .187
  .57
• Verbal IQ .586
  .72

29
Working memory

• Language vs. sound recall ceiling performance
• Read vs. hear (words) recall tasks ceiling
  performance
• Read vs. hear (words) recall combined with
  production (increased difficulty)
• Read write down words recalled 6
• Hear repeat words on first trial only one
  word/ second trial 6 / third trial 9
• Rubinsteins (1970) memory task (hear repeat
  words) significantly better than average for
  peers (2nd 3rd trial 9 words out of 10)
• Excellent visual memory (for pictures of
  objects), but also words
• Memory results contradict the general dyslexic
  profile (e.g., poor working memory)

30
(No Transcript)
31
(No Transcript)
32
Dissociation

• A general conclusion on the basis of (L1/L2)
  language talent and dyslexia an
  asymmetry/dissociation between oral (a strength)
  and written (a weakness) modality
• Crossing of modalities is a particular problem
• Possible explanation in terms of the
  phonological loop of working memory as the best
  candidate for the language acquisition device
  (Gathercole 1999, Perani 2005)
• Support good at rhyming

33
The two languages

• As expected, problems and success! persist in
  both languages
• Poor spelling, (relatively) poor reading
  comprehension in both Bulgarian and German
• Good at auditory language
• A somewhat special register profile speaks a
  literary variety of the L1 (in contrast to her
  background)
• A special expressive language profile uses
  stereotyped and highly idiomatic language not
  typical of her age (but no metaphors!)
• Mechanism Tomasello (2002) holoprhases are
  typical of the initial/early stages of LA (before
  generalizations based on analogy)

34
Grammar

• Mistakes at the level of advanced grammar
• Bulgarian clitic doubling, verb argument
  structure (subject/direct object), Information
  Structure phenomena
• Potential explanation in line with deficit and
  LA clitics are small (function) words and are
  usually omitted in early stages of acquisition
  (lack phonological weight and go unnoticed)?
  the very same items are omitted in EPs texts
  written under dictation, errors even in copying
  clitics, also present difficulty in reading
• German cases (e.g., similar to clitics in Bg)

35
Further problems

• Prosody when reading both at word level and
  larger chunks related to dyslexia (cf. The
  golabl acoustic pattern perception hypothesis of
  Foxton, Talcott, Witton, Brace, McIntyre
  Griffiths 2003)

36
Some puzzles

• Literal interpretation of language (related to
  autistic traits?)
• Proverbs and sayings, metaphors
• Verbal prefixes originally derived from spatial
  prepositions (poor spatial cognition, e.g.,
  orientation problems)
• Translates literally from German

37
More puzzles

• Learned German from exposure to TV impossible
  according to current research (Zimmerman et al.
  2007, also p.c.)
• Possible explanation (in line with factors in
  LA)
• Massive exposure never tires of watching her
  favourite programmes
• Exploits relative strengths of the dyslexic
  profile good hearing, attention to detail,
  learning from demonstration, hands-on
• Good auditory skills
• Good working memory
• Biological factors brain plasticity (started
  acquiring German at age 4)

38
Implications for L2 learning

• Good news L2 learning still possible for some
  dyslexic individuals
• Provided
• The strengths of the dyslexic profile are used
  good hearing, the relative oral (auditory)
  language strength, hands-on learning profile
• However
• Reading/writing still suffer especially when
  orthographies differ (e.g., Norwegian (relatively
  transparent) ? English (non-transparent)

39
Language and autistic traits

• CAST score 22, above clinical cut-off point
• Observation-based not as pronounced in social
  behaviour, as in perseverance patterns (obsession
  with topics etc.) and attention to detail
• Communication has acquired certain stereotypical
  reactions, such as theatrical laughter, feigns
  emotion

40
Language talent and autistic traits

• Profile here consistent with proposal in Happé
  Vital (2008) - attention to detail/local
  processing style responsible for the language
  talent
• Evidence acquires language in a special way,
  based on fragments, language use resembles a
  puzzle, e.g., putting (already stored) bits and
  pieces together
• Snyder (2009) privileged access to lower level,
  less-processed information
• Baron-Cohen et al. (2009) hyper-systematising,
  hyper-attention to detail (which is the result of
  sensory hypersensitivity), cf. E.s peaks (BD,
  OAss), also Picture Arrangement good scores

41
Language and autistic traits

• Highly specialised and advanced vocabulary,
  somewhat reminiscent of WS children (e.g.,
  exotic/extinct animals)
• Uses very elaborate and advanced stereotypical
  phrases (e.g., More than you can imagine!),
  .e.g, fragments of conversation

42
Language talent and autism

• Further support in linguistic theories of
  language acquisition, e.g., Tomasellos
  holophrastic stage (pre-analogy stage)
• All linguistic items are acquried as lexical
  items based on the association of form and meaning

43
Some of the puzzles explained

• Literal interpretation compositionality in
  semantics the meaning of a composite expression
  is a function of the meanings of its
  sub-components
• E. processes language in a piece-meal fashion,
  but only when form (an acoustic string) can be
  associated with meaning leading to literal
  interpretation the hyper-compositionalist!

44
Contrast with poor syllabification

• Syllables are not meaningful, thereby cannot be
  parsed (only 4 correct out of 19 words)
• (Un)expectedly good at word-chains (a bit slower
  compared to controls, due to temporal load) since
  this involves parsing at the level of meaningful
  units

45
The full picture Fragments on the road to system?

• Event perception Attention to detail, tells
  isolated events, not a coherent story
• Communication fragments of communicative
  gestures (laughter, simulating (real) emotion)
• Fragments of language (constructions)
  stereotypical phrases that replace expression of
  own thoughts spontaneously

46
(No Transcript)