Reaction time for gender concord

1
Reaction time for gender concord

• Christer Johansson Janne Cecilie von Koss
  Torkildsen
• University of Bergen
  University of Oslo
• Computational linguistics
  Language, logic information

2
Experiments on gender agreement

• What we have done
• Reaction time experiments
• What we will do
• Event-related potentials (ERPs)

3
The use for that which does not exist

• Detection of missing forms a measure of
  system acquisition.
• Our measures can be used to distinguish between
  groups, but not individuals ? not a diagnostic
  criterion

4
Gender processing and Scandinavian languages

• Gender agreement one of few morphologically
  complex phenomena left in Scandinavian languages
• Few experimental studies of gender processing for
  Scandinavian languages that we know of.
  (Especially for Event Related Potentials).

5
Adjective Noun concord

• There are no good neuter forms of certain
  adjectives in Norwegian and Swedish.
• Implications for Scandinavian gender classes.
• Two empirical studies of reaction times
• Concord
• Incongruency
• Existing word / Novel word

6
Examples
7
Some Scandinavian adjectives without neuter form

• ett latt barn, ett rätt lejon
• A lazy child, an afraid lion
• det lata barnet, det rädda lejonet
• Definite adjectives have a-ending (ok!).
• ett sött barn, ett solitt resultat, ett rött
  djur, ett matt lejon.
• Forms with very similar phonological structure
  are accepted.

8
Pettersson (1989)

• kry, glad, god, ked, led, blid, frod, pryd, snöd,
  strid, vid, vred, flat, kåt, lat, rät, såt(a),
  fadd, nödd, rädd, stadd, disträ, kavat, brydd,
  gladd, X-id,
• A large number of words
• Low frequency (såt?) to high frequency (rädd,
  lat).
• Systematic
• Similar syllabic structure
• Similar semantics

9
Inter-language contrasts

• Norwegian en slu man / et slutt barn
• Swedish en slug unge. ett slugt barn. (!).
• Danish en glad man / et gladt menneske

10
Semantics of missing adjectives
11
Natural semantic clustering

• Non-gender languages
• In some cases of aphasia/agnosia, the patient
  selectively loses the ability to name either
  living or non-living entities.
• Warrington, E. K., and T. Shallice. (1984)
  Category specific semantic impairments. Brain
  107829-853.
• Classes may form without linguistic expression.
• Possibly classes are formed by how words are
  described, which puts focus on adjectives.

12
Problematic adjectives

• ... denote properties that are
• Hard to visualize (draw a picture of afraid)
• Hard to verify (How do we know that someone is
  afraid?)
• Internal states not directly observable

13
However

• common neuter
• en trött man, ett trött barn
• A tired man a tired child
• en slug man ett slugt barn
• A sly man a sly child

14
Grammatical gender and semantic features

• Common assumption grammatical gender is assigned
  arbitrarily, i.e. house is masculine in
  Russian, feminine in Romance languages and neuter
  gender in German and Scandinavian.

15
Gender semantics for nouns

• Neuter Gender are mostly non-living
• Neuter mass / one of many, low identifiability
• children, mosquitoes etc.
• Common high individuality.
• a specific nail en spik, any one nail ett spik
• the mosquito that bites you en mygga.
• any one mosquito ett mygg
• a chair seats one individual en stol
• a table is a shared space ett bord.

16
Reaction Time Studies
17
Design

• Subjects 25 students and staff Lund and Oslo.
• Swedish 19 subjects (6 too slow responses)
• Norwegian 21 subjects (4 too slow).
• 240 NPs determiner adjective head noun
• Presented in 4 blocks without repetition.
• Pause between blocks
• Forced choice YES NO
• Decision and reaction time were recorded.

18
10 main classes

• NPROB problematic form NEUTER context
• (et latt barn / a lazy child)
• CPROB problematic form COMMON context
• (en latt unge / a lazy kid)
• CYES concord common gender (en fin unge)
• NYES concord neuter gender (et fint barn)
  
• CNO incongruent common context (en fint unge)
• NNO incongruent neuter context (et fin barn)
  
• NONS 4 categorier gender x congruent

19
Balanced Factors

• Phonological and Semantical Complexity
• Head words common neuter
• Kvinna fruntimmer, unge barn, katt lejon,
  varelse djur
• Woman wench kid child cat
  lion, being - animal
• Adjectives
• Lat sömnig, rädd skrämd, rigid styv, gravid
  lycklig
• Lazy sleepy, afraid scared, rigid stiff,
  pregnant happy
• Semantic as well as phonological complexity
  matched.
• Training
• Nouns flicka äpple girl apple
• Adjectives liten söt small sweet

20
Expectation

• Principle Conflicting cues take time
• NPROB correct form does the word exist?
  
• CPROB wrong form does the word exist?
• CNO/NNO clear incongruency
• CYES/NYES no conflict

21
Reaction times (Swedish)

• Normalized (1.3 30 slower than individual
  median).
• Highly significant main differences. NPROB very
  slow.

N/C Incongruent
NPROB
Novel words
CPROB
N/C Concord
22
Reaction times (Swedish)

• NPROB most problematic. Slowest. Very different.
• Neuter slower than common gender. (Frequency
  effect)
• Novel words are slower for congruency, but faster
  for incongruency.

23
Conclusion (Swedish)

• Reaction times faster for clear decidability and
  high frequencycommon gender.
• We have separated the problematic forms by their
  reaction times for lexical decision.

24
Norwegian

• Very similar, but different words
• et krytt/latt/slutt/vantrott/statt
  (1)
• et sultent/snilt/slitent/forkomment/fint(2)
• barn/dyr/esel/kvinnfolk
• en kry/lat/slu/vantro/sta
• unge/person/katt/kvinne

25
Reaction times (Norwegian)
nno
nprob
nyes
cno
cprob
cyes
26
Reaction times Expected only
Problematic forms accepted in common gender form
with neuter context. Lot of hesitation 55
slower, and only 3/4 affirmative responses.
Congruency effect no slower
than yes. Neuter slower than common.
nno kry (yes)
nprob (no)
nno (no)
cno (no)
nyes (yes)
cprob (no)
cyes (yes)
27
Reaction time (Norwegian)

• NPROB similar to normal incongruency
• Problematic latt, rätt, etc treated as common
  gender???
• Neuter slower than common gender. (Frequency
  effect)
• Just like in Swedish!!
• Incongruent slower than congruent.
• CPROB CNO. Problematic forms treated as
  incongruent with common gender!
• Recall Both cases have the same forms
  latt, krytt, slutt, vantrott, statt
• et _ barn (nprob) en _ unge
  (cprob)!!

28
Conclusion 1 (Norwegian)

• Different from Swedish.
• The problematic forms are reacted to as if they
  are incongruent with both genders.
• novel words?

29
Norwegian novel words
Norwegian strong effect of concord. Swedish
strong effect of gender (frequency).
cno
nno
nyes
cyes
30
Conclusion

• Swedish Problematic forms cause significantly
  slower response times.
• Norwegian Common gender is the preferred default
  (according to dictionaries), but accepting the
  common gender form for the problematic forms
  create significant hesitation.

31
Future research event-related potentials (ERPs)
32
(No Transcript)
33

• The scientist criticized Maxs of proof (Neville
  et al. 1991)

LAN
N400
P600
34
ERPs and syntactic processing

• ELAN
• Early left anterior negativity between 100 and
  300 ms after stimulus onset
• Reflects first-pass parsing processes,
  especially phrase structure violations
• LAN
• Left anterior negativity between 300 and 500 ms
• Reflects morpho-syntactic processing (i.e.
  agreement violations)
• P600
• Positive component peaking around 600 ms,
  centroparietal
• Reflects controlled processes of syntactic
  reanalysis and repair (i.e. garden-path sentences)

35
ERPs and lexical/semantic processing

• N400
• Negative component peaking between 350 and
  450 ms
• Reflects lexical integration, including
  morphological processes
• Centro-parietal scalp distribution

36
Language comprehension processes

• Initial structure building ((E)LAN)
• Lexical/semantic integration (N400)
• Late syntactic integration (P600)

37
Related ERP findings

• Gender incongruencies in sentences
  LAN-P600 pattern (German, Dutch, Spanish)
• Gender incongruencies in noun phrases LAN
  and N400 (Spanish)
• Morphological aspects of lexical integration
  affect size of N400 (Dutch, Spanish)
• LAN for subject-verb disagreement was elicited
  for both real and novel words, whereas P600 is
  only present for real words (German)

38
Which components do we expect?

• Problematic adjectives --- a case of
  gender disagreement?
• ? LAN-P600 pattern-------------------------------
  ------------------
• Problematic neuter gender adjectives ---
  treated as nonsense-words?
• ? LAN, but no P600

39
Design issues

• Simple tests to establish a starting point.
• Congruency vs incongruency
• Real vs novel words
• Processing profile for the problematic cases
• Full phrase vs. adjectivenoun pair?
• -- et latt barn vs. latt barn

40
Future Experiments?

• Priming studies/picture/word interference
• High and low individuality as a priming
  factor

41
Thank you for listeninghttp//ling.uib.no/BREDT

• Johansson, C. 2003. How is grammatical gender
  processed?, proc. of CLIN2002.

42
Conclusion

• The observed hesitation to the problematic forms,
  and for Norwegian also for the prescribed default
  forms, indicate that there is a factor involved
  that is not observed directly in the word forms.
• Results suggest that internal states, with the
  noun features animate and/or individualized, is
  crucial for the formation of Scandinavian gender
  classes.
• superior time resolution compared to other
  functional brain imaging techniques

43
Exposures in each condition

• NPROB 16 (4 adjective 4 head nouns)
• CPROB 16 (15 for Swedish glad/glatt homonymi).
• CYES 40 (10 4)
• NYES 20 (5 4)
• CNO 20 (5 4)
• NNO 40 (10 4)
• NONS 4 x 4 x 4 64
• Fillers 24 (Training).
• 19 Swedish 21 Norwegian. 4500 to 5000 exposures

44
Conclusion

• Problematic cases
• SWEDISH NORWEGIAN
• N C matrix N
  C
• N 33 23 N 29 11
• C 13 -4 C 49 -1

45
Novel words
Swedish novel words Effect for gender, not
concord.
cno
nno
Norwegian novel words Effect for concord (not
gender).
nyes
cyes
46
Finding sentient beings

• A small experiment
• Time how long it will take to name 10 sentient
  beings in each gender category.
• Highly significant
• 7 out of 7 took much longer to finder neuter
  examples (in Norwegian).
• average 2 minutes slower in neuter.

47
Why doesnt German have similarly missing
adjectives?

• German is similar to Scandinavian
• 3 gender classes.
• Neuter gender similarly correlated with
  non-living.
• However
• In German the neuter gender is the default
  gender
• many exotic animal are neuter gender Das
  Dromedar, Kamel, Krokodil, Opossum, Zebra
• Many domestic animals Das Pferd, Schwein
• Diminutives Kind, Mädchen, Fräulein.

48
Why doesnt German have similar missing
adjectives?

• Less difference in production time (preliminary
  n3)
• Indicates that sentient beings are more readily
  available in all German genders.
• Therefore, German is much less likely to have
  missing adjective forms that depend on the
  internal state factor that seem so crucial to
  the Scandinavian problem.

49
What about English?

• English speakers
• Produce 10 sentient beings in less than half of
  the time it takes Scandinavians to produce the
  common gender list. (Gender does not facilitate
  word finding).
• There is a specific kind of aphasia/agnosia that
  selectively affects naming of either living or
  non-living nouns.
• a linguistic distinction between sentient and
  non-sentient beings seem to be universal. It
  has been argued that the distinction can be
  acquired by how such nouns are described gt back
  to the adjectives.

50
Future Research Event-related potentials (ERPs)

• superior time resolution compared to other
  functional brain imaging techniques
• electroencephalogram (EEG) recording of neurally
  generated voltage changes at various scalp sites
  with respect to a reference location.
• event-related potentials (ERPs) voltage changes
  time-locked to events

51
(No Transcript)