Effects of Syntactic

1
Effects of Syntactic Phonological Similarity in
Korean Center-embedding ConstructionsSun-Hee Lee
and Mineharu NakayamaThe Ohio State
University16th Annual CUNY Conference on Human
Sentence Processing, Cambridge, MA,
March 2003

Figure 1. Magnitude Estimation- All 6 Conditions
-

• 2.1. Predictions
• If Korean is like Japanese,
• the nominative -ka/-i sentences (1c-f)
  will be harder than the topic -nun sentences
  (1a, b) because of syntactic and phonological
  similarity.
• If there is an phonological similarity,
• ka-i/i-ka sentences (1d, e) will be
  harder than the ka-ka/i-i (1c, f) sentences.

4.0. Discussion 1. Korean shows effects of s
yntactic similarity. - Results replicate fi
ndings in Japanese. - Babyonyshev and Gibso
n (1999), Gibson (2000), Lewis and
Nakayama (2002), Uehara and Bradley
(2002), Vasishth (2002) 2. Korean shows effect
s of phonological similarity. - Results a
re different from Uehara and Bradleys (1996)
null result, which neither included topic
sentences nor controlled noun types as in this
study. - Vasishth (2002)

• 1.0. Research Questions
• Similarity-based interference (Lewis 1996)
• Evidence from Japanese (Lewis and Nakayama,
  2002)
  
• The serial position of syntactically
  similar NP arguments affects the difficulty of
  Japanese center-embeddings.
  
• Q1. Does Korean replicate this effect?
• Q2. Does Korean show phonological similarity
  interference contra Uehara and Bradley (1996) ?
  
• Korean is structurally similar to Japanese, but
  has two phonologically distinct nominative
  markers.
  
• -ka with nouns that end in vowels
• -i with nouns that end in
  consonants.
  
• e.g. Mary-ka vs. John-i

• 2.2. Methodology
• Non-cumulative moving window study implemented
  in
  
• Psyscope (Cohen, MacWhinney, Flatt, and
  Provost 1993)
  
• Magnitude estimation task (Bard et al 1996)
• 2.3. Design
• Three two-syllable nouns (i.e., two characters)
  and two verbs.
  
• NP1, NP2 proper nouns NP3 a common noun
• Controlled for familiarity of nouns
  plausibility of test
  
• sentences
• Total 80 Sentences 24 test sentences (1 type
  with 4 tokens)
  
• Subject 48 Korean native speakers (ages 19-36,
  mean28.2)

Figure 2. Magnitude Estimation -
Topic Sentences vs. Nominative Sentences -

• Due to the finding 1 ii above, effects of
  Japanese syntactic similarity discussed in Lewis
  and Nakayama (2002) were actually effects of
  syntactic and phonological (morphophonemic)
  similarity.
• - Lewis (2002) and Vasishth (2002) feature
  bundles
  
• 4. Evidence for similarity interference in
  working memory
  
• - See Gordon et al. (2001) and Gordon et al.
  (2002) among others.

2.0. Test Sentences Markers topic
nun, nominative ka, -i, accusative -lul
1a.  NP1-nun NP2-ka NP3-lul V V (nun-ka type)
     ??-? ??-? ??-? ?????
????. (Korean) Euncwu-nun Youngay-ka k
yoswu-lul chacawasstako kiekhayssta.
Euncwu-top Youngay-nom professor-acc
visited remembered
   Euncwu remembered that Youngay had visited
the professor.   1b.  NP1-nun NP2-i NP3-lul V
V (nun-i type)      Euncwu-nun Huisen-i 
kyoswu-lul  chacawasstako kiekhayssta.
Euncwu-top Huisen-nom professor-acc
visited remembered    Euncwu rememb
ered that Huisen had visited the professor.
  1c.  NP1-ka NP2-ka NP3-lul V V (ka-ka type)
   Euncwu-ka   Youngay-ka kyoswu-lul
chacawasstako kiekhayssta. Euncwu-nom Youn
gay-nom professor-acc visited
remembered Euncwu remembered that Youngay h
ad
Figure 3. Magnitude Estimation
- 2 Types of Nominative Sentences -
References Bard, E. G., D. Robertson, and A. Sora
ce (1996) Magnitude estimation of linguistic
acceptability. Language 72.1, 32-68.
Cohen, J. D., B. MacWhinney, M. Flatt, and J.
Provost (1993) PsyScope A New graphic
Interactive Environment for Designing Psychology
Experiments. Behavioral Research Methods,
Instruments Computers 25.2, 257-271.
Babyonyshev, M. and E. Gibson (1999) The
complexity of nested structures in Japanese.
Language 75.3, 423-450. Gibson, E. (2000) Depende
ncy locality theory A distance-based theory of
linguistic complexity. In Marantz et al. (eds.),
Image, Language, Brain Papers from the first
mind articulation project symposium. MIT Press.
Gordon, P., C. Hendrick, and M. Johnson (2001)
Memory interference during language processing.
Journal of Experimental Psychology Learning,
Memory Cognition 27.6, 1411-1423.
Lewis, R. L. (1996) Interference in short-term
memory The magical number two (or three) in
sentence processing. Journal of Psycholinguistic
Research 25.1, 93-115. Lewis, R.L. (2002) Parsin
g as cue-based memory retrievalToward
computational models of the moment processes of
sentence comprehension. (Talk presented at the
Institue for Research in University of
Pennsylvania.) Lewis, R. L. and M. Nakayama (20
02) Syntactic similarity effects of embeddings
in Japanese. In Nakayama (ed.), Sentence
Processing in East Asian Languages. 85-110. 
CSLI. Uehara, K. and D. Bradley (1996) The eff
ect of -ga sequences on processing Japanese
multiply center-embedded sentences. In Park and
Kim (eds.) ,Language, Information and
Computation. Kyung Hee University.
Vasishth, S. (2002) Working Memory in Sentence
Comprehension Processing Hindi
center-embeddings. Ph.D. dissertation. Ohio State
University.

• 3.0. Results
• 1 Effects of Syntactic and Phonological
  Similarity
  
• The nominative sentences (1c-f) were
  significantly harder than the topic sentences
  (1a, b).
  
• (F1(1,47)33.209, pF2(1,46)38.285, p
• The nominative sentences with ka-ka/i-i (1c, f)
  were significantly harder than the topic
  sentences with nun-ka/nun-i (1a, b).
  
• (F1(1,47)34.481, pp
  
• The nominative sentences with ka-i/i-ka (1d, e)
  were also significantly harder than the topic
  sentences with nun-ka/nun-i (1a, b).
  
• (F1(1,47)27.359, pp
  
• 2 Effects of Phonological Similarity
• Among the nominative sentences, the same
  phonological sequences ka-ka/i-i (1c, f) were
  significantly harder than ka-i/i-ka (1d, e).
  
• (F1(1, 47)14.259, pF2(1,46)4.554, p

1d.  NP1-ka NP2-i NP3-lul V V (ka-i type)
   Euncwu-ka   Huisen-i kyoswu-lul
chacawasstako kiekhayssta. Euncwu-nom Hui
sen-nom professor-acc visited
remembered Euncwu remembered that Huisen h
ad visited the professor.   1e.  NP1-i NP2-ka
NP3-lul V V (i-ka type)      Huisen-i  Eu
ncwu-ka  kyoswu-lul  chacawasstako kiekhayssta.
Huisen-nom Euncwu-nom professor-acc v
isited remembered
Huisen remembered that Euncwu had visited
the professor.   1f.  NP1-i NP2-i NP3-lul V
V (i-i type)      Huisen-i   Swuceng-i k
yoswu-lul  chacawasstako kiekhayssta.
Huisen-nom Swuceng-i professor-acc
visited remembered
Huisen remembered that Swuceng had visited
the professor.