Online Processing

1
Online Processing

• 37-924-01
• Dr. Sharon Armon-Lotem

2

• On-line experimental techniques tap into
  automatic unconscious processes involved in
  language comprehension and production and
  minimize participants reliance on explicit or
  metalinguistic knowledge.

3
Presentation by Vicky Chondrogianni Theo
Marinis
4

• There are two basic types of time-sensitive
  measures available to examine language
  processing behavioral measures (e.g.
  comprehension response times and production
  latencies) and physiological measures (e.g.
  event-related brain potentials (ERPs) and
  eye-movements).
• (Clahsen 2008, p. 3)

5
Behavioral methods

• Eye tracking
• Priming experiments measuring reaction time
  (children are usually slower than adults)
• Cross modal priming
• Monitoring task (self paced reading/listening)
  (Taxler 2005 from Clahsen 2008)

6
Eye tracking

• Linguistic abilities are assessed by tracking and
  recording eye movements in response to
  predetermined verbal and visual stimuli
• Eye-tracking in language processing studies
  allows researchers to track and record
  participants' eye movements when they
• Read a sentence
• Look at the pictures on the computer screen as
  they listen to sentences that describe these
  pictures

7

• Experimental eye tracking data is obtained to
  investigate
• understanding of spoken language
• cognitive processes related to spoken language
• ability to process and interpret metaphor and
  figurative language
• body language and lip reading
• turn taking in conversations
• audio-visual integration
• reading behavior
• tracking-task performance
• scene exploration strategies
• (http//www.tobii.com/eye-tracking-research/global
  /research/linguistics/)

8
Preferential looking, Head-turn method

• Children (and adults) tend to look at pictures
  corresponding to a sentence they hear. This can
  be used to test word comprehension as well as
  sentence comprehension

9
Shes kissing the keys/ball
10
(No Transcript)
11
Garden-Path sentences

1. Since Jay always walks a mile seems like a short
   distance to him.
2. The horse raced past the barn fell
3. As the woman edited the magazine about fishing
   amused all the reporters
4. As the woman sailed the magazine about fishing
   amused all the reporters

12

• magazine about fishing amused all the reporters
• noun post noun verb post verb
  

A regression is any eye movement that begins at
the right-most point the reader has fixated and
leaves the currently fixated region to the left.
This definition is therefore only concerned with
disruption occurring during initial processing.
First-pass time is the sum of the fixations
occurring within a region before the first
fixation outside the region. If the eye fixates a
point beyond the end of a region before fixating
the region for the first time, then the
first-pass time for that region is zero. (This
measure is equivalent to the gaze-duration
measure e.g., Rayner Duffy, 1986, when the
region is a single word.) Total time is the sum
of all fixaions in a region.
(Pickering Traxler, 1998)
13

• The regressions and total-time data demonstrate
  that
• readers misanalysed both types of ambiguous
  sentence
• sentences with implausible object analyses were
  harder to process during the critical noun phrase
• sentences with plausible object analyses were
  harder to process during the syntactically
  disambiguating verb phrase.
• The regressions data demonstrate further that
  readers incrementally interpreted the sentences,
  because plausibility effects emerged before the
  point of syntactic disambiguation.
• Readers must have initially treated the magazine
  about fishing as the object of the subordinate
  verb (with magazine as the head noun).

14

• Clackson, K. H. Clahsen 2011. Online processing
  of cataphoric pronouns by children and adults
  Evidence from eye-movements during listening. In
  Danis, N., Mesh, K. H. Sung (eds.), Proceedings
  of BUCLD 35. Vol.1, Cascadilla Press Somerville,
  MA, pp. 119-1 MORIA 17/1

15
Word monitoring Tasks (Tyler Marslen-Wilson,
1981)

• Monitoring for the word hand in auditory
  stimulus.
• a. John had to go back home. He had fallen out of
  the swing and had hurt his hand on the ground.
• b. John had to sit on the shop. He had lived out
  of the kitchen and had enjoyed his hand in the
  mud
• c. The on sit top to had John. He lived had and
  kitchen the out his of had enjoyed hand mud in
  the

16

• Participants 5, 7,10 and adults
• Reaction time was measured
• Findings all showed the same gradation
• The gap was smaller for 5s limited processing

17
Priming tasks - Lexical decision task
Kazanina 2006, University of Ottawa
18

• The cortical representations of the prime and
  target are interconnected or overlap in some way
  such that activating the representation of the
  prime automatically activates the representation
  of the target word.
• (Forster 1999, p. 6)

19
Cross-Modal Priming

• Auditory prime, visual target

Kazanina 2006, University of Ottawa
20
Phonological priming (Marslen-Wilson Zwiserlood
1985)
Kazanina 2006, University of Ottawa
21
Phonological priming (Marslen-Wilson Zwiserlood
1985)
Kazanina 2006, University of Ottawa
22
Coreference (McKee, Nicol McDaniel, 1993(

• Alive or not alive?
• The reindeer knows that the alligator with the
  gigantic teeth is looking at himself in an old
  shiny mirror.
• The reindeer knows that the alligator with the
  gigantic teeth is looking at him in an old shiny
  mirror.

23
Traces in relative clauses antecedent
reactivation (Roberts et al 2007 from Clahsen
2008) - The effect of memory span

• John saw the peacock to which the small penguin
  gave the nice birthday present __ in the garden
  last weekend.
• John saw the peacock to which the small penguin
  gave the nice birthday present __ in the garden
  last weekend.
• (PEACOCK, CARROT)

24
High span children and adults mean reaction time


25

• High memory span shorted RT to identical target
  (than unrelated) in the gap position (than
  control for the related only)
• Low memory span no antecedent reactivation, but
  no difference in comprehension

26
Monitoring task (self paced reading/listening)

• Participants listen and press a button for the
  next word/phrase. RT is measured as well as
  comprehension/judgment at the end.

27
(No Transcript)
28
(No Transcript)
29
Presentation by Vicky Chondrogianni Theo
Marinis
30
Self-paced reading (Taxler 2005 from Clahsen 2008)

• When Sue tripped the girl fell over and the vase
  was broken
• When Sue tripped the table fell over and the vase
  was broken
• When Sue fell the policeman stopped and helped
  her up

31
Presentation by Vicky Chondrogianni Theo
Marinis
32
Presentation by Vicky Chondrogianni Theo
Marinis
33
Presentations

• Word monitoring Montgomery, J. W., Leonard, L.
  B. (1998). Real-Time Inflectional Processing by
  Children with Specific Language Impairment
  Effects of Phonetic Substance. J of Speech
  Language and Hearing Research, 41(6), 1432-1443.
  HALA 24/1

34

• Montgomery, J. W., Leonard, L. B. (2006).
  Effects of Acoustic Manipulation on the Real-Time
  Inflectional Processing of Children With Specific
  Language Impairment. J of Speech Language and
  Hearing Research, 49(6), 1238-1256

35

• Participants 16 with TLD (Mean age 811) and 16
  with SLI (Mean age 90)
• Word recognition RT Task with enhanced cues for
  low and high substance markers
• Grammaticality Judgment Task for similar items
  (with and without enhancement)

36
Acoustic Enhancement

37
Grammaticality Judgment

• CAgtSLI
• Morpheme type effect
• Acoustic enhancement effect for both group with
  more impact on the low substance item
• Specific morpheme effect only for SLI

38
Word recognition
SLIgtCA LowgtHigh INFltSTEM (but not for SLI on low
substance morphemes) No effect of enhancement
39

• Enhancement help in the off-line task but not in
  the on-line task
• The task demand (fast response) mask the
  enhancement

40

• Cross modal picture priming Marinis, T. and van
  der Lely, H. (2007) On-line processing of
  wh-questions in children with G-SLI and typically
  developing children. International Journal of
  Language Communication Disorders, 42 (5). pp.
  557-582 SANDY 24/1

41

• Self Paced Listening Vicky Chondrogianni,
  Theodoros Marinis, and Susan Edwards. 2010.
  On-line Processing of Articles and Clitic
  Pronouns by Greek Children with SLI. In Franich,
  K., Iserman, K, Keil, L. (Eds.). Proceedings of
  the 34th Annual Boston University onference on
  Language Development, Volume 1, 78-89.

42

• The study examines whether Greek children with
  SLI and a group of age-matched typically-developin
  g (TD) children are sensitive to the omission of
  articles and clitic pronouns when they listen to
  sentences in real-time.
• Following Tsimpli Stavrakakis (1999)
  Interpretability Hypothesis a difference is
  expected between indefinite article vs. clitic
  pronouns and definite article

43

• 20 with TLD (Mean age 70) 13 with SLI (mean
  age 69)
• SPL with E-prime
• The grammatical version was recorded and the
  clitic/article was spliced out to generate the
  ungrammatical ones

44
(No Transcript)
45
(No Transcript)
46

• (5) Definite article subject position
• Yesterday a dolphin was playing in the sea with
  the other animals. Late / in the afternoon /
  (the) dolphin / chased / the fish.
• (6) Definite article object position
• Yesterday a kangaroo was playing with a green
  ball. The kangaroo / kicked / (the) ball / on the
  pitch / yesterday afternoon.
• (7) Indefinite article object position
• Yesterday a naughty fox chased some other
  animal. The fox / chased / (a) donkey / in the
  woods / yesterday at midday.
• (8) Accusative direct object clitic pronoun
• The lion wanted to eat the deer. The deer / got
  very scared / when / the lion / (it) bit / in the
  jungle / on the rocks.

47
Definite article in subject position
Both groups showed a main effect of
Grammaticality (TD children F1 (1, 26) 58.074,
p lt .001 F2 (1,7) 26.375, p .001 children
with SLI F1 (1, 12) 5.626, p lt .05 F2 (1,7)
5.397, p .053).
48
Definite article in object position
TD children showed a main effect of
Grammaticality in both analyses per subjects and
per items (F1(1,26) 247.376, p lt.001 F2(1,7)
21.804, p lt. 01). Children with SLI showed a main
effect of Grammaticality only in the analysis per
participants (F1 (1,12) 6.423, p lt.05 F2 (1,7)
2.292, p gt.1).
49
Indefinite article in object position

• Both groups showed a main effect of
  Grammaticality (TD group F1 (1,26) 105.969, p
  lt.001 F2 (1,7) 47.920, plt.001 SLI group F1
  (1,12) 62.471, plt.001 F2(1,7) 81.401,
  plt.001).

50
Clitic pronoun condition
TD children showed a significant main effect of
Grammaticality (F1 (1,26) 12.189, p lt.01 F2
(1,9) 8.627, p lt.01). In contrast, children
with SLI showed no main effect of Grammaticality
for either the participant or item the analysis
(F1 (1,12) 1.402, p gt .1 F2 (1,9) 1.243, p gt
.1).
51
Four major findings

• Children with TLD are sensitive to grammaticality
• Children with SLI are sensitive to omission of
  indefinite article but not to clitic omission as
  predicted by the Interpretability Hypothesis as
  well as the Surface Hypothesis
• Children with SLI are sensitive to definite
  article omission which is not predicted why?

52

• Frequency
• Form-function mapping (consistency)
• Therapy effect

53
Comparing across populations

• SLI vs. L2
• Children vs. adults

54

• Picture matching Marinis, T. and Chondrogianni,
  V. (2011) Comprehension of reflexives and
  pronouns in sequential bilingual children do
  they pattern similarly to L1 children, L2 adults,
  or children with specific language impairment?
  Journal of Neurolinguistics, 24 (2). pp. 202-212
  IRENA 24/1

55

• Chondrogianni, V Marinis, T. (2012). Production
  and processing asymmetries in the acquisition of
  tense morphology by sequential bilingual
  children. Bilingualism Language Cognition, 15,
  5-21

56

• 28 monolingual TLD and 39 L2ers with L1 Turkish
  (6-9)
• TEGI Production (Here is a teacher. Tell me
  what she does).
• Word monitoring task for grammatical inflections
  (Mary really likes to bake. Every day she bake(s)
  cakes and sometimes cookies and muffins)
• Comparison with SLI from Leonard Montgomery

57
Accuracy in the production of tense morphemes

• Main effect of group and morpheme
• -S lt -ed for L2
• Moderate correlation between LoE and s

58
RT for tensed morphemes

• Main effect for
• group (L2gtL1)
• morpheme type (non-tensedgttensed)
• grammaticality (ungrammaticalgtgrammatical)

59
RT on non-tensed morphemes

• No interaction gt the two groups were equally good
  at detecting ungrammaticality

60

• Two groups by scores on TEGI below and above the
  criterion score.
• For s sig dif for age and LoE
• For ed sig dif for LoE
• No effect for group for RT

61
Word recognition
INFltSTEM (but not for SLI on low substance
morphemes) no grammaticality effect