Transcript CML-2007

When ‘Colorless green ideas’
meet working memory span
Olga V. Fedorova
Ekaterina V. Pechenkova
Lomonosov Moscow State University
[email protected]
CML-07 workshop ‘Science of Cognition: Bridging Gaps
between Cognitive Linguistics and Experimental Psychology’
July 31, 2007
Memory is…
encoding, storage and retrieval of information
Human memory stores
?
Which
way
does
language
affect
memory?
 motion
 emotion
 images
 logic and
linguistic
structures
?
Which
way does
memory
affect
speech?
Classical cognitive approach:
multiple memory systems
REHEARSAL
SHORT-TERM
MEMORY
LONG-TERM
MEMORY
?
Span: 7±2 chunks
Capacity limits are not known
Duration: maintains
information for 18-30
sec (without rehearsal)
Maintains information for the
lifetime (potentially unlimited
duration)
Recoding information
into verbal format
Declarative and procedural
formats, including schemata and
semantic networks
Working memory
CENTRAL
EXECUTIVE
VISUOSPATIAL
SCETCHPAD
EPISODIC
BUFFER
PHONOLOGICAL
LOOP
LONG-TERM MEMORY
(2001)
Alan
Baddeley
Reading span test =
PROCESSING
COMPONENT
Patricia
Carpenter
+
STORAGE
COMPONENT
Meredith
Daneman
Daneman, M., & Carpenter, P. (1980).
Individual differences in working memory and reading.
Journal of Verbal Learning and Verbal Behavior, 19, 450-466.
Reading span in Russian
Elaborated by Olga Fedorova and colleagues in 2001
Processing: reading sentences
Storage: maintaining the final words of these sentences
70 sentences, 12-15 words each
NB! Russian sentences appear to be much longer if we
count the number of syllables, since the portion of
one or two syllable words in Russian is significantly
less than in English
Visual or auditory presentation
Reading span test
(Russian version)
Тест на определение
объема оперативной
памяти
аудиовариант
(audio)
Данный эксперимент тестирует объем Вашей
оперативной памяти; информация в ней хранится
всего несколько секунд.
Вам необходимо будет запоминать и потом
воспроизводить последние слова предложений,
которые Вы будете слышать, точно в той же форме и в
том же порядке, в которой они были произнесены.
This is the test of your working memory. The working
memory stores information for just a few seconds.
You will be listening to sentences, and you will have to
memorize the final words of these sentences. Then you
will have to recall these words keeping their form and
the order of appearance.
Количество предложений будет неуклонно
возрастать, тем самым Ваше задание будет
усложняться.
Если Вы не можете вспомнить последнее слово
какого-то предложения, переходите к следующему.
The number of the sentences in a series will be
increased step by step, so the difficulty of the task will
grow.
If you cannot recall some word, please, skip it and go
on to the next one.
Эксперимент начинается!
Ready? Press the spacebar…
нажмите пробел!
Слушайте предложение 1
In September, every child, including myself,
will go to school.
Listen to the sentence 1
Слушайте предложение 2
Mike sat down on the bench next to his
home and started to look at the people
passing.
Listen to the sentence 2
Пожалуйста, повторите последние
слова этих двух предложений
Please, recall the final words of these
two sentences
?
Школу
School
Прохожих
Passing
Эксперимент начинается!
Ready? Press the spacebar…
нажмите пробел!
Слушайте предложение 1
I like ice-cream very much, but afterwards I
have an ache in my throat.
Listen to the sentence 1
Слушайте предложение 2
Tomorrow all members of our family will go
to the mountains for skiing.
Listen to the sentence 2
Слушайте предложение 3
In order to get to the restaurant from our
house, you should turn to the right and
then again to the right.
Listen to the sentence 3
Слушайте предложение 4
I haven't seen Anton since then because he
has moved from our district.
Listen to the sentence 4
Слушайте предложение 5
As her birthday present, the girl was given
a big fancy doll.
Listen to the sentence 5
Пожалуйста, повторите последние
слова этих пяти предложений
Please, recall the final words of these
five sentences
?
Горло
Throat
Лыжах
Skiing
Направо
Right
Района
District
Куклу
Doll
Russian version vs. English
Russian
English
Proportion of the Low-Span
participants (span<4)
64,5%
50%
Auditory presentation
advantage (vs. visual)
greater
weaker
Amount of the word form
errors
more
less
Short final words advantage
(vs. long)
yes
yes
What does the test
really measure?
Authors of the test:
working memory for reading
Further common view:
storage capacity that
determines linguistic skills
Alan Baddeley:
effectiveness of the central
executive and episodic buffer
Kyllonen & Christal:
general intelligence
Reading span test
is a dual task
Dual task paradigm
Common resource (structural or energetic)
Yes
No
A tradeoff between a primary
and a secondary task
No interference
Reading span test
is a dual task
+
PROCESSING
STORAGE
COMPONENT
COMPONENT
Which
No component is
Reading sentences one is
Memorizing final
explicitly
set
the
words
as a primary
primary task
task? extracting maintaining
THERE IS A TRADEOFF,
BUT…
What does the test
really measure?
Our hypothesis:
human ability of
organizing one’s
memory
and of using memory
aids and semantic
schemata
Study I:
semantically related final words
Experimental condition
И мне стало грустно, когда я подумал, что они не
пойдут завтракать в шикарный ресторан.
Содрав с маленьких сухощавых рук перчатки,
Нина последний раз в жизни ела моллюски.
Увидев маленькое кафе, он юркнул туда, чтобы
передохнуть, и заказал еще одну чашку кофе.
Вино выдерживали слишком долго, и оно совсем
испортилось и потеряло свой цвет и вкус.
Study I:
semantically related final words
Experimental condition
And I felt sad when I thought that they wouldn’t go for
a breakfast to a fancy restaurant.
Nina teared off her gloves from her small lean hands
and for the last time in her life was eating shellfish.
When he got a sight of a small café, he dropped in to
have some rest and ordered another cup of coffee.
The vine was aged for too long, so it finally was spoilt
and has lost its color and taste.
Study I:
semantically related final words
Control condition – standard test
Мы дошли до двери, и я видел, как она быстро
переступила порог и пошла по вестибюлю.
Надо было достойно вознаградить усилия того,
кто достиг совершенства в своем ремесле.
Он сидел на скамье и показался мне темной,
бесформенной грудой в образе человека.
Граф закрыл окно и задернул шторы: сквозняк и
электрический свет привлекают молнию.
Study I:
semantically related final words
Control condition – standard test
We came up to a door, and I saw her quickly stepping
over the doorstep and going along the vestibule.
An appropriate reward was needed for the effort of
one who reached perfection in his profession.
He was sitting on the bench and seemed to me as a
dark shapeless pile in appearance of a human.
The earl closed the window and drew the draperies
since the draught and the electric light attract the
lighting.
Study I: Results
40 participants,
visual presentation, reading aloud
Scores: 5 percent for each group reported correctly
Semantically
related final
words
Standard test
58%
45%
High span
Low span
Study I: Conclusion
Accessibility of one’s
semantic schemata and memory aids
is a factor that influences
the working memory span
 Short final words are better
recalled than the long ones
 Artificial articulation difficulties
decreases the test score
 Semantically related words that
could be linked to each other by
means of visual memory aids are
recalled better than a typical
random word set
 Any other mental load decreases
the results
A related subcomponent
of Baddeley’s model
Russian reading span data
supports Baddeley’s model
Phonological loop
Visuospatial
scetchpad
Central
executive
Working memory span correlates
with text comprehension
EXAMPLE:
Comprehension of the text with large reference
distances in people with high and low ‘reading span’
Daneman, M., & Carpenter, P. (1980).
Individual differences in working memory and reading.
Journal of Verbal Learning and Verbal Behavior, 19, 450-466.
Reference distance 2
Reference distance 6
It was midnight and the jungle was very
still. Suddenly the cry of a wolf pierced
the air. This anguished note was followed
by a flurry of activity. All the beasts of
the jungle recognized that an urgent
meeting had been summoned by the lion,
their king. Representatives from each
species made rapid preparations to get
to the river clearing. This was where all
such emergency assemblies were held.
The elephant and tiger were the first to
arrive. Next came the gorilla, panther and
snake. They were followed by the owl and
the crocodile. The proceedings were
delayed because the leopard had not
shown up yet. There was much
speculation as to the reasons for the
midnight alarm. Finally he arrived and the
meeting could commence.
Sitting with Richie, Archie, Walter and the
rest of my gang in the Grill yesterday, I
began to feel uneasy. Robbie had put a
dime in the juke box. It was blaring one
of the latest "Rock and Roll" favorites. I
was studying, in horror, the reactions of
my friends to the music. I was especially
perturbed by the expression on my best
friend's face. Wayne looked intense and
was pounding the table furiously to the
beat. Now, I like most of the things other
teenage boys like. I like girls with soft
blonde hair, girls with dark curly hair, in
fact all girls. I like milkshakes, football
games and beach parties. I like denim
jeans, fancy T-shirts and sneakers. It is
not that I dislike rock music but I think it
is supposed to be fun and not taken too
seriously. And here he was, "all shook up"
and serious over the crazy music.
Question: Who finally arrived?
Question: Who was "all shook up" and
serious over the music?
% the percentage of correct
responses
Comprehension in subjects with
low vs. high reading span
100
90
80
70
60
50
40
30
20
10
0
Reading
span
2
3
4
5
small(2-3)
me dium (4-5)
large (6-7)
Reference distance, sentences
Why is the correlation?
 There is no correlation of linguistic skills with standard
single task working memory test like word or digit span
(Turner&Engle, 1989).
 The correlation is there even if we substitute the
sentences by arithmetic expressions and/or final words by
final figures (Turner&Engle, 1989).
 Modifications of the processing task (i.e. different types of
judgements about the sentence) give correlations at the
range 0.52-0.63 (Waters & Caplan, 1996)
 Test-retest reliability is only moderate: correlations at the
range 0,41-0,54 (Kuntsi et al., 2001)
Study II: Is the source of
interference syntactically
specific?
Processing component task:
informational visual search verbal with high
attention demands and low syntactic demands
Material: Russian grammatical, but nonsense
sentences (like one invented by L.V. Shcherba, 1920th)
with real Russian words embedded
Глокая куздра штеко будланула
бокра и курдячит бокрёнка…
Глокая куздра штеко будланула
бокра и курдячит бокрёнка…
by L.V. Shcherba, 1920th
Compare to
Colorless green ideas sleep furiously
by Noam Chomsky (1957)
ИНСТРУКЦИЯ
Просим Вас:
 прочитать предложение про себя один раз и
запомнить из него все русские слова;
 когда предложение исчезнет с экрана, назвать все
русские слова из этого предложения в той же
форме и в том же порядке, в которой они были в
предложении.
TEST INSTRUCTIONS
Please, read each sentence silently once, and
memorize all Russian words that would be there;
on disappearance of the sentence, recall all Russian
words keeping their word form and order
Предложение будет появляться на экране на
несколько секунд.
Количество русских слов в одном предложении
будет постепенно возрастать.
Each sentence will appear on the screen for a few
second. The number of the Russian words in the
sentence will be increased step by step.
Йоу, биктя ю рекетя быстро шлемзит
худамп хлеб лопатик.
Yoou, biktya yu reketya, bystro [rapidly]
shlemzit khudamp khleb [the bread]
lopatik
Нажмите пробел!
Брундая куздра, живущая ф трукате
хуранта, громко куздряет рядом
проходящих бяко людей.
Brundaya kuzdra, zhivuzhchaya [living] f
trukate khuranta, gromko [loudly]
kuzdryaet ryadom [near]
prokhodyazhchikh [passing by] byako
lyudey [people].
Нажмите пробел!
Study II: Results
32 participants,
visual presentation
Scores: 5 percent for each group reported correctly
Visual search
Standard test
54%
High span
74%
High span
There is a difference (ANOVA F=25.39, p<0.001)
There is a medium correlation
(Spearman’s rho=0.36, p=0.045
Study II: Conclusion
each subcomponent of
the Daneman-Carpenter task
requires both general and specific
processing resources
Conclusions
 Russian version of the standard reading span
test fits the classic model of the working
memory structure proposed by Alan Baddeley
 The test apparently provides a measure of
individual verbal processing abilities rather than
individual working memory capacity as such
Thank you twas brillig,
and the slithy toves for your did gyre
and gimble kind attention in the
wabe!
Спасибо воркалось.
Хливкие шорьки пырялись по наве, и
хрюкотали за внимание зелюки
как мюмзики в мове!