A cross-linguistic investigation of the onset effect in reading aloud: No need to mope about the MOPE
The global population speaks approximately 7,000 different languages and approximately half of the world’s population is bilingual. The unique orthographic characteristics of different languages can help us to gain insight into the seemingly simple process of transforming written to spoken words (i.e., reading aloud). In this thesis five language contexts were investigated: Persian, English, Dutch-English, Russian-English, and Spanish.
- Kalinka Timmer
- 26 September 2013
- Full text in Leiden University Repository
With both behavioral and electroencephalography (EEG) measures the on-line time course of reading (aloud) is investigated in first, second, and cross-language situation. The behavioral results of the studies presented in this thesis revealed that target words (e.g., CARPET) were read aloud faster when preceded by phonologically onset-related primes (e.g., kettle) but not orthographically onset-related primes (e.g., circus). The phonological priming effect was present in the native language, the second language, and from second to first language (e.g., phone – FIETS). However, the effect is not present in Persian where words are printed without vowels or when reading is silent.
The event-related potentials (ERPs) revealed both orthographic and phonological activation peak around 150 milliseconds after target presentation for first, second, and cross-language situations. Thus, even if we have an accent in a second language, the phonological code is activated similarly to a native speaker. In addition, though not visible in the behavioral responses, ERPs revealed early activation of orthography and phonology for Persian word reading (with unmarked vowels) and for silent reading. This suggests that reading starts with same processes – whether silent or aloud or in different scripts.
To conclude, this thesis favors computational models assuming early conversion of orthography-to-phonology. However, with approximately half of the world’s population being bilingual and the results presented in this thesis, it is important that computational models of reading processes extend their horizon.