=Paper=
{{Paper
|id=Vol-1347/paper25
|storemode=property
|title=Are you reading what I am reading? The impact of contrasting alphabetic scripts on reading English
|pdfUrl=https://ceur-ws.org/Vol-1347/paper25.pdf
|volume=Vol-1347
|dblpUrl=https://dblp.org/rec/conf/networds/IakovlevaPD15
}}
==Are you reading what I am reading? The impact of contrasting alphabetic scripts on reading English==
https://ceur-ws.org/Vol-1347/paper25.pdf
Are you reading what I am reading?
The impact of contrasting alphabetic scripts on reading English
Tatiana Iakovleva Anna E. Piasecki Ton Dijkstra
CNRS, France UWE Bristol Radboud University
59, rue Pouchet Coldharbour Lane Montessorilaan 3
75017 Paris Bristol BS16 1QY 6500 HE Nijmegen
tatiakovleva@ Anna.Piasecki@ T.Dijkstra@
yahoo.fr uwe.ac.uk donders.ru.nl
1 Introduction vergence and divergence in Russian and English
script coding for cognates and non-cognates.
This study examines the impact of the cross- Cognates are translation equivalents with signifi-
linguistic similarity of translation equivalents on cant cross-linguistic form overlap in phonology
word recognition by Russian-English bilinguals, and/or orthography (e.g., ‘marriage’ in English,
who are fluent in languages with two different ‘mariage’ in French). Cognates are generally
but partially overlapping writing systems. Cur- processed more quickly by bilinguals than
rent models for bilingual word recognition, like matched control words (for an overview of stud-
BIA+, hold that all words that are similar to the ies, see Dijkstra, Miwa et al., 2010). However, as
input letter string are activated and considered far as we know, cognate processing for the Rus-
for selection, irrespective of the language to sian-English language pair has not been exam-
which they belong (Dijkstra and Van Heuven, ined before.
2002). These activation models are consistent
with empirical data for bilinguals with totally 2 Predictions
different scripts, like Japanese and English (Mi-
wa et al., 2014). Little is known about the bilin- We are making the following predictions about
gual processing of Russian and English, but stud- English word recognition by Russian-English
ies indicate that the partially distinct character of bilinguals:
the Russian and English scripts does not prevent 1. In English word processing, Russian-English
co-activation (Jouravlev and Jared, 2014; Marian bilinguals will activate lexical candidates that are
and Spivey, 2003; Kaushanskaya and Marian, similar to the input word in both Russian and
2007). English (language non-selective lexical access).
Many Russian-English translation equiva- 2. English-Russian cognates will be recognized
lents are in part composed of shared letters that more quickly than English control words, due to
can potentially activate both Russian and English co-activation and convergence (cognate facilita-
word candidates. Often, these letters have am- tion effect, Dijkstra, Miwa et al., 2010; Lemhöfer
biguous phonemic mappings across the two lan- and Dijkstra, 2004).
guages. The degree of ambiguity is high espe- 3. Cognates with ambiguous orthography, i.e.
cially when shapes of block-letters and letters in shared letters mapping onto different phonemes
italics overlap across languages. For instance, a in the two languages, will be processed more
printed Russian letter ‘и’ does not look like any slowly than cognates with mismatching orthog-
letter of the English alphabet, but the shape of its raphy, due to decreased facilitation from the oth-
handwritten equivalent ‘u’ perfectly coincides er cognate member.
with the English hand-written grapheme. We The following two predictions are more
identified 5 overlapping pairs of printed English speculative and exploratory in nature.
block-letters and Russian letters in italics (g, r, 4. Response times to cognates with transparent
m, n, u). orthography, i.e. shared letters mapping onto the
Our study started from the assumption that same phonemes in the two languages, will be
even when a bilingual reads English words in about equal to those for cognates with mismatch-
printed font, letter shapes also activate handwrit- ing orthography, because transparent orthogra-
ten Russian letters with similar shapes in a bot- phy and shared phonology will lead to increased
tom-up way. We focused on the impact of con- lexical competition, but, at the same time, the
Copyright © by the paper’s authors. Copying permitted for private and academic purposes.
In Vito Pirrelli, Claudia Marzi, Marcello Ferro (eds.): Word Structure and Word Usage. Proceedings of the NetWordS Final
Conference, Pisa, March 30-April 1, 2015, published at http://ceur-ws.org
112
transparency will lead to increased semantic co- indicated that bilinguals considered not only
activation of cognates in the two languages. block-letters but also corresponding handwritten
5. English control words with mismatching or- graphemes when rating the visual similarity be-
thography will be processed more quickly than tween words.
words with ambiguous orthography, because less In total, 37 Russian-English participants (10
interference from the Russian alphabet is ex- male vs. 27 female; age: 19-60 years) took part
pected in the first case. in the study. At the moment of testing, all par-
ticipants were residing in English-speaking coun-
3 Method tries: 11 participants in Bristol, UK, 21 partici-
pants in Sheffield, UK, and 5 participants in New
To test these hypotheses, we first constructed a
Zealand. After the experiment, all participants
large database of Russian-English cognates with
rated their proficiency in English on a scale from
three, four, five or six letters in length. To our
1 (the lowest) to 6 (the highest). Average ratings
knowledge, no such database is currently avail-
for reading, writing, speaking, and listening var-
able to the community of researchers. Next, 75
ied between 4.4 and 5. Except for two partici-
English cognates were selected as test words in a
pants, ages of L2 acquisition (AoA) ranged be-
lexical decision task. Orthographic coding was
tween 6 and 19 years. Length of residence in an
performed on English cognate words written in
English-speaking country varied between 3
lower-case block letters in Arial font. The result-
months and 21 years (mean = 33 years, SD = 11
ing items were allocated to three categories: 1)
years).
Cognates with Ambiguous Orthography
Participants performed an English lexical de-
(CAO=Minus condition), composed of letters
cision task, in which they pressed a “yes” or a
that have different phonological mappings in
“no” button depending on whether a presented
English and Russian (e.g. ‘guru’ might be read as
word was English or not. They were asked to
/digi/ if a Russian monolingual was asked to read
press a button as quickly and accurately as possi-
this string of letters); 2) Cognates with Transpar-
ble. The items were presented in a pseudo-
ent Orthography (CTO=Positive condition),
randomized order to each participant. The ex-
composed of letters that largely share their or-
periment was programmed in E-Prime. Reaction
thographic-phonological mappings with letters of
times (RTs) and accuracy of responses were
the Russian alphabet (e.g. in ‘koala’ the only
measured. Only correct responses to real words
mismatch with the Russian alphabet is the graph-
were included in the analyses of reaction times.
eme ‘l’); 3) Cognates with Mismatching Orthog-
raphy (CMO=Base condition), composed mostly 4 Results
of letters that do not exist in the Russian alphabet
(e.g. ‘filter’). The cognate types were matched First, all responses faster than 300 ms and slower
across conditions (CAO/CTO/CMO) in word than 3 s were removed from the data set, because
length, frequency, and degree of cross-linguistic they were not considered as valid measurements.
orthographic overlap between Russian and Eng- Next, the data from 9 participants were excluded
lish alphabets. Three groups of control words from analysis, because they had a response accu-
were then selected that matched the cognates of racy below 70%. We removed 5 cognates, 8 con-
each type with respect to these three dimensions. trol words, and 14 non-words from the items,
Finally, each cognate and non-cognate was because these items had an accuracy below 70 %
matched with a pseudo-word generated with the or had extremely slow responses. For the remain-
help of the Wuggy-software (crr.ugent.be). ing 28 participants, after removing these items,
Next, 20 Russian-English bilinguals were cognate and control word conditions were still
asked to rate the visual similarity between the matched with respect to length and frequency (as
English cognates and their Russian translation shown by non-significant t-tests). None of the
equivalents. They also rated the semantic simi- remaining responses were further apart than 2.5
larity of all selected item pairs. Rating results SDs from the participant mean in each condition.
showed that bilinguals mostly considered ortho- The mean RT for non-words was 892 ms. Table
graphic congruence (as opposed to incongru- 1 presents the mean RTs for words in each cog-
ence) between the orthography of Russian and nate and control word condition, as well as their
English translation equivalents and gave higher accuracy.
ratings to English words that have shared orthog-
raphy with the Russian alphabet. Ratings also
113
Condition RT dif- fect is also observed in cognates with (partially)
Cognates Controls mismatching orthography, the cognate effect
Type ference
661 (82.2) 727 (112.7) may in part be ascribed to the phonological and
Base 66 semantic overlap in these cognates. Thus, the
.97 .95
711 (105.7) 734 (106.4) orthographic input representation quickly leads
Minus 23 to an activation of sublexical and lexical phono-
.94 .93
656 (89.01) 730 (113.1) logical representations (cf. Peeters et al., 2013).
Plus 74 In line with prediction 3, the cognate facilita-
.97 .92
tion effect is modulated by the degree of shared
Table 1. Mean reaction times and accuracies for transparent overlap between Russian and English
word categories (standard deviations between alphabets. Cognates with transparent orthogra-
parentheses). phy were processed faster than cognates with
ambiguous grapheme to phoneme mappings.
The word data were analyzed by means of a This finding can be explained by assuming that
repeated-measures Analysis of Variance (ANO- Russian words are co-activated with English
VA), using cognate type (3, MO vs. AO vs. TO) words to the extent that they match the English
and cognate status (2, cognate vs. control) as letter input, irrespective of whether this matching
within-subject factors. This analysis resulted in is in terms of block letters or handwritten visual
main effects of Cognate Status (F (1, 27) = similarity. In other words, it is purely a bottom-
94.11, p<.001), Item Type (F (2, 54) = 9.89, up (signal-driven) effect.
p<.001), and an interaction of Cognate Status
with Item Type (F (2, 54) = 10.22, p<.001).
Next, we did planned comparisons to test the
Cognate Minus (CMO) and Cognate Plus (CTO)
conditions against the Cognate Base (CMO)
condition. Significant differences were found
between the RTs between the Cognate Base con-
dition and the Cognate Minus condition (t(27)=-
5.0, p<.001 two-tailed) but not between the Cog-
nate Base and the Cognate Plus condition
(t(27)=.60, p=.55). There was a significant dif-
ference between the Cognate Base condition and Figure 1. Localist connectionist illustration of
the Control Base condition (t(27)=-6.54, p<.001). cognate representation and processing, adapted
Finally, no significant differences arose between from Dijkstra, Miwa et al. (2010).
the different control conditions (Control Base vs.
Control Minus, t(27)=-.67, p=.51; Control Base The finding that cognates with mismatching
vs. Control Plus t(27)=-.36, p=.72). orthography and shared orthography with trans-
parent grapheme-to-phoneme mappings are re-
5 Discussion sponded to about equally fast, is in line with pre-
diction 4, which is based on the representation
Russian-English bilinguals performed an English for cognates that has been proposed by Dijkstra,
lexical decision task with purely English control Miwa et al. (2010). As Figure 1 indicates, both
words and English-Russian cognates 1) with form representations of cognates are assumed to
mismatching orthography or 2) shared orthogra- be activated based on the input and they spread
phy with a) transparent or b) ambiguous map- activation to convergent semantic representa-
pings on phonemes in Russian and English. tions. The co-activation of form representations
Responses to cognates were faster than to results in lexical competition and interference
English controls (see Table 1). This cognate fa- (Dijkstra, Hilberink-Schulpen et al., 2010),
cilitation effect is in line with prediction 1 that whereas the convergence on semantics results in
lexical candidates in both Russian and English facilitation. As a result, the RT difference be-
are activated during Russian-English bilingual tween cognates with mismatching orthography
word recognition. and shared transparent orthography may be rela-
It also confirms prediction 2 that language tively small, due to a cancelling out of the effects
non-selective lexical access takes place in Rus- of increased lexical form competition and in-
sian-English word recognition. Because the ef- creased semantic co-activation.
114
Finally, in contrast to prediction 5, English and Dijkstra (under revision). The present study
control words with mismatching orthography provides confirmation for these models from a
were not processed more quickly than control completely independent perspective, that of
words with ambiguous orthography. Apparently, cross-linguistic similarity effects in scripts.
mismatching orthography in general did not re- To conclude, we presented evidence in favor
sult in any systematic interference on word proc- of language non-selective lexical access in Rus-
essing speed. Said differently, the noise intro- sian-English bilinguals, showing an English-
duced by spuriously activated word candidates Russian cognate facilitation effect, the size of
from Russian with overlapping letters in the oth- which depended on whether there was overlap in
er control conditions did not systematically affect orthography or not, and on whether this overlap
the lexical decision to the English target word, was ambiguous or transparent relative to phonol-
although it may have affected the participants’ ogy. These effects were shown to be lexical in
general decision-making strategies in the ex- nature, because mismatching orthography in con-
periment. In terms of interactive activation mod- trol target words with translations that are com-
els, the increase in noise could be cancelled out pletely different in form did not show any evi-
by a somewhat higher reliance on semantic codes dence of differential processing.
or global lexical activation (Grainger and Jacobs,
1996) for making the lexical decision. Acknowledgments
In all, the obtained patterns of results are in
This research was made possible with support
support of interactive activation models for bi-
from NetWordS, the «European Network on
lingual word recognition, such as the BIA+ mod-
word structure in the languages in Europe» (re-
el (Dijkstra and Van Heuven, 2002) when the
search grant n° 09-RNP-089). The authors are
assumption is made that cognates are represented
also deeply indebted to the EPSRC's RefNet re-
in terms of overlapping but lexically competing
search network that enabled us to collect a large
form representations and largely shared semantic
part of our data. We also wanted to thank the
representations in the two languages (Dijkstra,
anonymous reviewers for their helpful comments
Miwa et al., 2010), see Figure 1. Even the
on an earlier version of this paper.
somewhat counter-intuitive prediction 4 can find
a reasonable explanation in terms of such mod-
els. Prediction 5 was not confirmed, but the actu-
References
ally obtained result can be interpreted in terms of Eriko Ando, Kazunaga Matsuki, Heather Sheri-
slightly shifted lexical decision criteria. dan, and Debra Jared. 2015. The locus of Ka-
This study confirms the presence of language takana-English masked phonological priming
non-selective lexical access in visual word rec- effects. Bilingualism: Language and Cogni-
ognition by different script-bilinguals, in line tion, 18: 101-117.
with, e.g., for Korean-English Kim and Davis
Ton Dijkstra, Béryl Hilberink-Schulpen, and
(2003) and for Japanese-English Hoshino and
Walter J. B. Van Heuven. 2010. Repetition
Kroll (2008), Miwa et al. (2014), and Ando et al.
and masked form priming within and between
(2015). Moreover, it bridges research on shared
languages using word and nonword neigh-
scripts and different scripts by considering the
bors. Bilingualism: Language and Cognition,
partially overlapping Latin and Cyrillic scripts of
13: 341-357.
English and Russian. It is innovative in showing
that cross-linguistic effects depend on the degree Ton Dijkstra, Koji Miwa, Bianca Brummelhuis,
of overlap in scripts depending on the exact Maya Sappelli, and Harald R. Baayen. 2010.
characteristics of the words involved. How cross-language similarity and task de-
The study also provides indirect support for mands affect cognate recognition. Journal of
various types of models that assume co- Memory and Language, 62: 284-301.
activation of word candidates that are ortho-
graphically similar to the input letter string. The Ton Dijkstra and Walter J. B. Van Heuven. 2002.
set of such candidates is often referred to as the The architecture of the bilingual word recog-
neighbourhood (Grainger and Dijkstra, 1992). nition system: From identification to decision.
Van Heuven et al. (1998) have shown that the Bilingualism: Language and Cognition, 5:
number of neighbours within and between lan- 175-197.
guages affects bilingual word recognition. This
result has recently been confirmed by Mulder
115
Jonathan Grainger and Arthur M. Jacobs. 1996. David Peeters, Ton Dijkstra, and Jonathan
Orthographic processing in visual word rec- Grainger. 2013. The representation and proc-
ognition: A multiple read-out model. Psycho- essing of identical cognates by late bilinguals:
logical Review, 103: 518-565. RT and ERP effects. Journal of Memory and
Language, 68: 315-332.
Jonathan Grainger and Ton Dijkstra. 1992. On
the representation and use of language infor- Walter J. B. Van Heuven, Ton Dijkstra, and Jon-
mation in bilinguals. In R. J. Harris (ed.), athan Grainger. 1998. Orthographic neigh-
Cognitive Processing in Bilinguals, 207-220. borhood effects in bilingual word recognition.
Amsterdam: Elsevier. Journal of Memory and Language, 39: 458-
483.
Noriko Hoshino and Judith F. Kroll. 2008. Cog-
nate effects in picture naming: Does cross-
language activation survive a change of
script? Cognition, 106: 501-511.
Olessia Jouravlev and Debra Jared. 2014. Read-
ing Russian–English homographs in sentence
contexts: Evidence from ERPs. Bilingualism:
Language and Cognition, 17: 153-168.
Margarita Kaushanskaya and Viorica Marian.
2007. Bilingual Language Processing and In-
terference in Bilinguals: Evidence From Eye
Tracking and Picture Naming. Language
Learning, 57: 119-163.
Jeesun Kim and Chris Davis. 2003. Task effects
in masked cross-script translation and phono-
logical priming. Journal of Memory and Lan-
guage, 49: 484-499.
Kristin Lemhöfer and Ton Dijkstra. 2004. Rec-
ognizing cognates and interlingual homo-
graphs: Effects of code similarity in language
specific and generalized lexical decision.
Memory and Cognition, 32: 533-550.
Viorica Marian and Michael Spivey. 2003. Bi-
lingual and monolingual processing of com-
peting lexical items. Applied Psycholinguis-
tics, 24: 173-193.
Koji Miwa, Ton Dijkstra, Patrick Bolger, and
Harald R. Baayen. 2014. Reading English
with Japanese in mind: Effects of frequency,
phonology, and meaning in different-script
bilinguals. Bilingualism: Language and Cog-
nition, 17: 445-463.
Kimberley Mulder and Ton Dijkstra (under revi-
sion). Revisiting the neighbourhood: The pro-
cessing of cross-language hermits and neigh-
bours in different tasks. Bilingualism: Lan-
guage & Cognition.
116