5th International Symposium on Bilingualism, 20-23 March 2005, Barcelona

Written Picture Naming in Turkish-French Bilingual Children
Biran E. Mertan1 and Ilhan Raman2 1Eastern Mediterranean University, North Cyprus 2Middlesex University, UK

Abstract The current study explored the role of orthographic transparency on written picture naming
for 20 pictures in bilingual children whose first language was Turkish (L1) and second language was French (L2). The participants were 10-11 year old children and residents in France. A total of 14 children were selected from a sample of 103 and were matched on socio-economic background, age and scholastic attainment. They were allocated to two groups of seven depending on whether they received their early primary education either in Turkey or France. Each child was required to have been in the French education system for at least two years as well as attending extracurricular Turkish language classes in France. In addition, all pictures were rated to have name agreement, familiarity and similar complexity by independent judges. In a counterbalanced design, children were instructed to write the picture names down in French and Turkish. The study was conducted in the children’s schools. An interaction between Turkish and French languages at the lexico-semantic level in bilingual children was predicted in relation to their L2 spelling proficiency. This is because while the mappings between orthography and phonology in Turkish are totally transparent and consistent, in French they are relatively opaque and inconsistent. An analysis of the spelling error data show that children educated in Turkey in their early years made more errors in French than Turkish but this did not reach statistical significance. On the contrary, children who received their early education in France made three times more errors on average in Turkish than French. The results suggest that acquisition of spelling skills in second orthography is perhaps determined by the complexity of the first orthography children are exposed to. Children who learnt the transparent Turkish orthography first, made an equal number of errors in Turkish and French despite French being more opaque than Turkish. On the other hand, children who acquired the French orthography first, struggled with Turkish despite the simple mappings between its orthography and phonology. This raises a question in relation to transferring of phonological and orthographic skills between orthographies that have distinct mappings between orthography and phonology. The current results raise a further question whether orthographic and phonological overlap are necessary to establish shared lexical entries for cognates in attaining L1 and L2 fluency in the bilingual child.

Background
• Cognates in two different languages share the same etymology. • They are conceptually, orthographically and phonologically similar. • Empirical research has shown that cognates are processed much faster and more accurately than noncognates. • It is important to investigate how cognate objects affect bilingual spelling performance in children

Background cont’d
• Cross-linguistic effect of shared orthography, phonology and semantics is not clear • However, one needs to note previously reported facilitatory effects of crosslinguistic orthographic and semantic similarity on response latencies to target words, but inhibitory effects of phonological overlap as in Dutch-English (Dijkstra,Grainger & van-Heuven, 1999)

Turkish Orthography
• The modern Turkish orthography consists of a 29 letter Latin alphabet of eight vowels and 21 consonants. • The modern Turkish orthography was deliberately designed to embody the sounds in the spoken language in a totally transparent representation. • Print-to-sound translation rules for each of the letters in the Turkish alphabet are totally one-to-one, explicit and context independent, i.e. no sources of irregularity.

Turkish vs. French Writing Systems
• Although Turkish and French vary greatly on the orthographic transparency continuum, there are some shared characteristics:
– both scripts use the Latin alphabet – both written and read from left to right – shared letters/graphemes and phonemes

Turkish and French Alphabets
Turkish alphabet
abcçdefgğhıijklmnoöprsştuüvyz

French alphabet
abcdefghijklmnopqrstuvwxyz •Letters unique to each alphabet are in bold •23 letters are shared between two orthographies

Orthographic Transparency Continuum

Alphabetic Writing Systems

Turkish Transparent

Italian Spanish
Relatively transparent

English French Persian Mixed

Arabic Hebrew Opaque

Orthographic transparency continuum based on print-to-sound mappings

Questions
• Is there a relationship between orthography and acquisition of literacy in two languages? • How do the unique properties of a language shape fluency of spelling? • In bilingual, biscriptal Turkish-French readers, to what extent does the properties of L1 and L2 influence language processing?

Stimuli
• The picture names were cognates in Turkish and French with similar pronunciation and spelling

Turkish kamyon

French camion

Object Name French abat-jour automobile cinéma camion casquette parachute piano chapeau téléphone gendarme ampoule cigarette collier cassette manteau douche radio valise télévision autobus

Object Name Turkish abajur otomobil sinema kamyon kasket paraşüt piyano şapka telefon jandarma ampul sigara kolye kaset manto duş radyo valiz televizyon otobüs

English translation lampshade car cinema lorry cap parachute piano hat telephone gendarme light-bulb cigarette necklace cassette coat shower radio luggage television bus

Predictions
• Cross-linguistic orthographic and semantic similarity on response latencies to target words, is known to have facilitatory effects whereas phonological overlap yields inhibitory effects (Dijkstra,Grainger & van-Heuven, 1999) • This effect was found in Dutch-English, two Indo-European languages • It is important to examine the role of shared phonology, orthography and semantics in spelling by using cognates in two orthographies with similar orthography and phonology • Turkish is a Turkic language while French belongs to the Indo-European cluster

Predictions
• Children were predicted to conduct more spelling errors in French overall due to its unpredictable nature between print and sound. • Moreover, we predicted errors to reflect children's early language instruction in L1: those who received their L1 training in Turkey were predicted to make more spelling errors in French compared to Turkish. Similarly, children who received their L1 training in France were predicted to make more spelling errors in Turkish compared to French. • Moreover, an interaction between Turkish and French languages at the lexico-semantic level in bilingual children was predicted in relation to their L2 spelling proficiency. • This is because while mappings between orthography and phonology in Turkish are totally transparent and consistent, in French they are relatively opaque and inconsistent.

Method
• Design
– In a counterbalanced design, children were instructed to write cognate picture names down in French and Turkish. The critical variable was whether they received their early primary education in Turkey or France.

• Participants
– The participants were 10-11 year old children and residents in France. A total of 14 children were selected from a sample of 103 and were matched on socio-economic background, age and scholastic attainment. Each child was required to have been in the French education system for at least two years as well as attending extracurricular Turkish language classes in France.

• Materials/ Stimuli/ Procedure
– The stimulus picture names were cognates in Turkish and French with similar pronunciation and spelling in L1 and L2 (e.g., light bulb is ampul in Turkish and ampoule in French). In addition, all pictures were rated to have name agreement, familiarity and similar complexity by independent judges. The study was conducted in the children’s schools.

Graph 1: Spelling errors in Turkish and French according to primary education
9 8 8 7 7 6

5 5 4 3 3

Mean

2 1 Turkish education French Education

TR_ERRSP FR_ERRSP

GP

Line graph showing the interaction between L1 & L2 errors
8

7

6

Estimated Marginal Means

5

4

ERROR
3 1 Turkish Errors 2 2 French Errors 2.00 French Educ

1.00 Turkish Educ

GROUP

Results
• There was a significant difference between the two education groups on Turkish errors t(12) = -2.5 p<0.025 as well as French errors t(12) = 4.8 p<0.0001. • Children with French education made more spelling errors in Turkish than children with Turkish education. • Formal analysis also showed a significant interaction between the disproportionate errors two education groups make in the two languages F(1,1) = 25.2 p<0.0001 as demonstrated by Graph 1 and Figure 1.

Models of Bilingual Language Processing
• Current models of bilingual language processing differ in their assumptions about the organisation of the lexical system in L1 and L2. • The effects of language we observed here are compatible with models that assume separate lexica for L1 and L2

Kroll & Stewart Revised Hierarchical Model (1999)
lexical links L1 conceptual links conceptual links concepts L2

pictures

Figure 8

Discussion
• The results suggest that acquisition of spelling skills in the second orthography is perhaps determined by the complexity of the first orthography. • Children who learnt the transparent Turkish orthography first, made an equal number of errors in Turkish and French despite French being more opaque than Turkish. • On the other hand, children who acquired the French orthography first, struggled with Turkish despite the simple mappings between its orthography and phonology.

Future Directions
• This raises a question in relation to transferring of phonological and orthographic skills between orthographies that have distinct mappings between orthography and phonology. • The current results raise a further question whether orthographic and phonological overlap are necessary to establish shared lexical entries for cognates in attaining L1 and L2 fluency in the bilingual child. • The results suggest that pedagogical practices could involve the use of cognates particularly when there is a considerable amount of phonological and orthographic overlap.