Bilingual Resonance: Reactivating Text Elements Between L1 & L2
R. Brooke Lea (Macalester College), Paul van den Broek, Jazmin Cevasco (University of Minnesota), & Aaron Mitchel (The Pennsylvania State University)
Resonance Background
• Recent studies addressed the question of how readers access information from long term
memory (e.g., Albrecht & Myers, 1995; Lea et al., 1998; McKoon et al., 1996).
•For example, Albrecht & Myers (1995) showed that small contextual cues (e.g., “leather
couch”) could reinstate unsatisfied goals that had been backgrounded in a long passage.
•Their experimental strategy was to present readers with target sentences that would appear
inconsistent only if the participant remembered the unsatisfied goal.
•The results showed that readings times on those target sentences were significantly slower
when the contextual cue was presented just before the sentence.
•The cue reminded readers of the unsatisfied goal.
•The authors discussed their results in terms of a resonance process that activates
propositions in memory that overlap with propositions in working memory.
It was temporarily quiet on the battlefront. The young lieutenant in charge of the unit realized that
they were short of ammunition. Because he had word that the enemy would counterattack the next
morning, he would have to go to headquarters before sundown to request more ammunition for
the unit. He got in an old dirty jeep and sped toward headquarters. He hadn't gotten far when he
spotted a badly wounded soldier who clearly needed immediate attention. Tendría que ir al cuartel
general después de encontrar a alguien que atendiera al soldado. El teniente buscó un médico,
pero no lo halló. Mientras tanto, el soldado herido estaba sangrando mucho. El teniente rasgó en
pedazos la camisa del soldado en pedazos y envolvió firmemente la herida con ellos. Finalmente,
observó complacido que el soldado dejaba de sangrar. En ese momento, un médico llegó y relevó
al cansado teniente. El teniente regresó en el viejo y sucio jeep y encontró un lugar para tomar una
rápida siesta. No había dormido por varios días. Al día siguiente, probablemente habría mas
batalla.
Spanish/English
El frente de batalla estaba temporalmente tranquilo. El joven teniente a cargo de la unidad se dio
cuenta de que las municiones estaban escaseando. Dado que sabía que el enemigo contraatacaría a
la mañana siguiente, tendría que ir al cuartel general antes del anochecer para conseguir mas
municiones para la unidad. Se subió a un viejo y sucio jeep y se dirigió rápidamente hacia el
cuartel general. No había llegado muy lejos, cuando vio un soldado mal herido que claramente
necesitaba atención inmediata. He would have to drive to headquarters after he found someone to
tend to the soldier. The lieutenant looked for a medic but there was none around. Meanwhile, the
wounded soldier was bleeding badly. The lieutenant tore the soldier's shirt in strips and wrapped
it tightly around the wound. Finally, he was pleased to see the bleeding stop. At that point, a
medic arrived and took over from the tired lieutenant. The lieutenant got back in the old dirty jeep
and found a place to take a quick nap. He had not slept for several days. Tomorrow there would
probably be more fighting.
Bilingualism Background
• A central debate in bilingualism research concerns the nature of mental representations for
multiple languages (Francis, 1999).
• Considerable debate as to whether lexical activation is selective with respect to language -both lexicons activated simultaneously, and in parallel, or do bilingual speakers only
activate the target language?
• Recent research suggest that bilingual speakers activate both languages simultaneously
(e.g. Jared & Kroll, 2001).
• And evidence from the visual world paradigm that bilinguals, even in purely monolingual
contexts, cannot “turn off” the non-target language (Spivey & Marian, 1999; Marian &
Spivey, 2003).
• These findings are captured in the latest Bilingual Interactive Activation (BIA +) model of
language comprehension (Dijkstra & Van Heuven, 2002), which predicts simultaneous
activation of both languages, with a language node cuing the target language and allowing
for correct lexical selection (see Figure 1).
• While not much research exists on bilingual lexical access during reading, there is
evidence that lexical activation while reading is non-selective (Durgunoğlu, 1997).
•If bilinguals activate both languages in parallel, then when reading a passage that switches
between languages, do they make global inferences across languages?
Figure 1.
The BIA+ Model (adapted from Dijkstra & Van Heuven, 2002)
Task schemaSpecifies processing steps for task at hand
•
• Receives
continuous input from the identification
system
• Decision criteria determine when a response is
made based on relevant codes
Identification System
Language
nodes
L1/L2
Semantics
Lexical
Orthography
Lexical
Phonology
Sublexical
Orthography
Sublexical
Phonology
3000
• Four language combinations: English-English (E-E); EnglishSpanish (E-S); Spanish-English (S-E); Spanish-Spanish (S-S)
2955
2712
2378
2299
2500
Design. 2 (cue) X 2 (First Half Language) X 2 (Second Half
Language) first two factors within-, third between-subjects.
2000
15411409
1500
1526
1330
1000
500
0
Eng/Eng
Eng/Span
cue
Span/Eng
Span/Span
no cue
Results: Experiment 1
(See Figures 2 - 4)
• Main effect for cue: F (1, 39) = 9.30, p = .004
• Significant cue-effect in all language combination conditions.
• No interactions.
• No relationship between L2 proficiency and size of cue effect.
• Resonance clearly cross the bi-lingual divide.
• Proficiency at this level is not related to the cue effect.
Figure 3
Experiment 1: Cue Effects
(Cue Effect = Cue – No Cue)
577.71
600
500
412.72
400
300
200
206.25
132.18
0
Eng/Eng
Eng/Span
Span/Eng
Span/Span
Figure 4
Experiment 1: Relationship Between L2
Proficiency and Cue Effects
3500
• Exp 2 subjects were significantly less proficient in
Spanish; average CBM score in Experiment 2 was 99
compared to 128 words in Experiment 1, (t (88) =
5.17, p < .001).
• Subjects took a vocab test of the 24 cue words; only
subjects who scored 24 out of 24 on this test were
included in the analyses.
• Pattern of means was similar to Experiment 1
(Figure 5).
• Significant main effect for cue: F (1, 47) = 10.51, p =
.002.
•Language-Congruence X Cue Interaction (Figure 8)
approaching significance, F (1, 47) = 1.90, p = .177.
• Significant cue effect in all four language
combination conditions.
• L2 proficiency and cue effect DID SHOW a
significant positive correlation (Figure 7).
• Hints at a connection between L2 proficiency and the
effectiveness of contextual cues.
Experiments 1 & 2
r = .081
• Combining Experiments 1 & 2:
• Main Effect for Language Congruence, F >10
(Single Language texts slower than Mixed texts)
• Main Effect for Cue, F >20 (Cue slower than
No Cue).
• Marginal Language Congruence X Cue
interaction, p = .098, (Cue effect was larger in
Single-Language texts than in Mixed texts;
Figure 8.)
• No significant differences between Exps 1& 2,
except for correlation between L2 proficiency and cue
effect in Spanish texts.
r = -.128
3135
2912
2581
3000
2546
2500
2000
1581
1500
1513
1315
1442
1000
500
0
Eng/Eng
Eng/Span
Span/Eng
cue
Span/Span
no cue
r = .541
p = .003
Figure 6
Experiment 2: Cue Effects
(Cue Effect = Cue – No Cue)
588.95
600
500
400
316.18
300
198.25
200
139.24
r = .450
100
p = .016
0
Eng/Eng
Eng/Span
Span/Eng
Span/Span
Figure 8
Conclusions
Cue Effects for Single-Language vs. Mixed-Language Texts
(Cue Effect = Cue – No Cue)
Bottom-up reactivation processes like
Resonance appear not to be affected in
bilingual contexts.
 Reactivation effects in the mixed-language
passages show that contextual cues need not
share orthographic, phonemic, or languagespecific features to effectively reinstate
backgrounded concepts.
 Such results suggest that Resonance can
operate at an abstract, conceptual level; no
overlap other than meaning appeared in the
mixed-language texts.
 However, the marginal language-congruence
X cue interaction may indicate that the
additional overlapping features in the samelanguage texts provided stronger cues.
 These reactivation effects do not require a
very high level of proficiency in L2 (found in
students of 2nd semester college Spanish).
 The cue-effect is not related to language
proficiency at relatively high levels of L2
ability; at lower L2 skill levels, proficiency is
associated with a larger cue effect (this may be
due in part to the task).
 These results provide further evidence for
parallel, non-selective access models of
bilingual processing.

500
398
426
400
References
Albrecht, J.E., & Myers, J.L. (1995). The role of context in accessing distant information
during reading. Journal of Experimental Psychology: Learning, Memory, &
Cognition,21, 1459-1468.
Dijkstra, A., & Van Heuven, W. J. B. (2002). The architecture of the bilingual word
recognition system: From identification to decision. Bilingualism: Language
and Cognition, 5, 175-197.
Durgunoğlu, A. Y. (1997). Bilingual reading: Its components, development, and other
issues. In A. De Groot & J. Kroll (eds.), Tutorials in bilingualism.
Psycholinguistic perspectives, pp 255-276. Mahwah, NJ: Lawrence Erlbaum.
Francis, W. S. (1999). Cognitive integration of language and memory in bilinguals:
Semantic representation. Psychological bulletin, 125(2), 193-222.
Jared, D. & Kroll, J. F. (2001). Do bilinguals activate phonological representations in one
or both of their languages when naming words? Journal of Memory and
Language, 44, 2-31.
Lea, R. B., Mason, R.A., Albrecht, J.E., Birch, S.L., & Myers, J.L., (1998). Who knows what
about whom: What role does common ground play in accessing distant
information? Journal of Memory & Language, 39, 70-84.
Marian, V., & Spivey, M. (2003). Bilingual and monolingual processing of competing
lexical items. Applied Psycholinguistics, 24(2), 173-193.
McKoon, G., Gerrig, R.J., & Greene, S.B. (1996). Pronoun resolution without pronouns:
Some consequences of memory-based text processing. Journal of Experimental
Psychology: Learning, Memory, & Cognition, 22, 919-932.
Spivey, M. J., & Marian, V. (1999). Cross talk between native and second languages:
Partial activation of an irrelevant lexicon. Psychological Science, 10(3),
281-284
Experiment 2: Relationship Between L2
Proficiency and Cue Effects
Results: Experiment 2
(See Figures 5 - 7)
100
• Within each of the four combinations, half of the passages were
“cue” and half were “no-cue”
Why no L2 proficiency – cue-effect correlation?
 Perhaps lack of proficiency range
 Lower proficiency subjects in Experiment 2
 Forty-nine students from intermediate-level Spanish
classes at Macalester – in second semester of college
Spanish
 Subjects were proficient enough to understand the
stories, but less proficient than those used in
Experiment 1.

Milliseconds
English/Spanish
Figure 7
Experiment 2: Target Sentence Reading Times
Experiment 1: Target Sentence Reading Times
Subjects. 41 students from upper-level Spanish classes at
Macalester College participated for $10. They were all native
speakers of English (L1), and highly proficient in Spanish (L2).
Materials. Passages from Albrecht and Myers (1995) were
adapted by translating them into Spanish (see Table 1). The
passages each contained:
Introduction: with an unsatisfied goal, and the contextual cue
(e.g., “old, dirty jeep”)
• Intervening Episode
• Conclusion: contextual cue is present or absent
• Target Sentences: coherent, but not globally coherent if the
reader reactivates the backgrounded goal.
Target sentence reading times slower if readers are reminded of
the global anomaly.
• 24 experimental passages and 14 fillers.
• Spanish proficiency measured with the CBM reading test.
Figure 5
Milliseconds
It was temporarily quiet on the battlefront. The young lieutenant in charge of the unit realized that
they were short of ammunition. Because he had word that the enemy would counterattack the next
morning, he would have to go to headquarters before sundown to request more ammunition for
the unit. He got in an old dirty jeep and sped toward headquarters. He hadn't gotten far when he
spotted a badly wounded soldier who clearly needed immediate attention. He would have to drive
to headquarters after he found someone to tend to the soldier. The lieutenant looked for a medic
but there was none around. Meanwhile, the wounded soldier was bleeding badly. The lieutenant
tore the soldier's shirt in strips and wrapped it tightly around the wound. Finally, he was pleased to
see the bleeding stop. At that point, a medic arrived and took over from the tired lieutenant. The
lieutenant got back in the old dirty jeep and found a place to take a quick nap. He had not slept for
several days. Tomorrow there would probably be more fighting.
Experiment 2: Less Proficient Readers
Figure 2
Milliseconds
English/English
Experiment 1
Milliseconds
Memory-based text processing research has focused on how distal text concepts are
reactivated by an automatic resonance process. In a typical experiment, an adjective-noun
pair such as “leather couch” is presented as a contextual cue early in a passage, and then
repeated later in the passage. Concepts associated with the first instance of the cue are
reactivated upon presentation of the second instance of the cue. Researchers agree that
feature overlap is fundamental to triggering resonance, but little is known about how overlap
is defined psychologically. We gave English-Spanish bilinguals passages that were partly in
English and partly in Spanish, and crossed L1 and L2 with the beginning and end of each
passage to determine the effect that cue encoding in one language has on the translated
counterpart encoded in the other language. The results showed robust resonance effects
both within and between languages, and a modest L2 proficiency effect.
Table 1
Example Texts Used in Experiments 1 & 2
Milliseconds
Abstract
307
266
300
200
100
0
Single Language
Exp. 1
Mixed Languages
Exp. 2
Descargar

Slide 1