Carrol, G.
(2015)
Found in translation: a psycholinguistic investigation of idiom processing in native and non-native speakers.
PhD thesis, University of Nottingham.
Abstract
Idioms, as highly familiar word combinations, are processed quickly by native speakers, but are problematic for non-native speakers even at high levels of proficiency. In this thesis I explore the representation of idioms in the monolingual and bilingual lexicons. In a series of studies I investigate how native and non-native speakers of English process English idioms and idioms translated from another language. In Study 1 I used a lexical decision task to test how much an expected word is primed following the first part of an idiom, e.g. on the edge of your… seat. English native speakers and Chinese-English bilinguals were tested using English idioms and translations of Chinese idioms (e.g. draw a snake and add… feet). In Study 2 I presented the same materials in short passages to allow for more natural presentation and used eye-tracking to investigate the reading patterns for all items. I also compared figurative and literal uses of the same items to see how easily non-native speakers were able to process non-compositional meaning in the L2. In Study 3 I used the same methodology (eye-tracking of idioms used in short sentence contexts) with a higher proficiency group (Swedish-English bilinguals), with much shorter, less predictable idioms (e.g. break the ice/bryta isen) and included a set of idioms that exist in both L1 and L2. All three studies point to the same conclusion: that even in an unfamiliar translated form, the expected lexical combination was facilitated (idioms showed faster processing than control phrases), but only the highest proficiency participants also showed evidence that they were able to process the figurative meanings without disruption. Congruent items show no additional advantage, hence it is clearly L1 knowledge of what words ‘go together’ that drives the effect in translation.
In Study 4 I extended this by contrasting idioms with other types of formulaic phrase: literal binomials (king and queen) and collocations (abject poverty). All types showed faster reading compared to equally plausible control phrases. I then used formulaic component words in separated contexts to see whether any lexical priming effects are observed when the formulaic frame is compromised. Only idioms showed evidence of a formulaic advantage in this condition, while binomials showed evidence of semantic priming and collocations showed evidence of disruption. Importantly, different factors relevant to each formulaic type show an effect on how they are processed, e.g. idioms were driven by predictability, while binomials were driven more by the strength of semantic association between component words.
The results overall provide a valuable new perspective on how formulaic units are represented in the mental lexicon. The fact that faster processing is seen for translated forms shows that idioms are not processed as unanalysed whole units, since L1 influence must be contingent on the individual words activating translation equivalent forms. This also shows that non-native speakers do not show fundamentally different processing in their L2 than native speakers, and ‘known’ word combinations are processed quickly regardless of the language of presentation. Compared to idioms, other formulaic types also show fast processing in canonical forms, but are more variable in whether or not the component words also show lexical priming in non-formulaic contexts. Formulaicity therefore exists at multiple levels of representation, encompassing lexical, structural and conceptual properties of word combinations.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
