Effects of number of translations in a translation recognition task ICP Berlin Germany 11 juillet 2008

Jean-Marc Lavaur & Jannika LaxénUniversity of Montpellier III, FRANCE

International Congress of psychology. Berlin, Germany. July 2008.

Language proficiency

Context of use

Translation situations

Language proximity

Word characteristics

Words characteristics

Intralingual factorsIntralingual factors

word frequency, concreteness, homography

Interlingual factors Interlingual factors

number-of-translations, dominance of the translations

Words with one translationWords with one translationWords with one meaning and one translation

For example: ARBRE TREETREE

Words with many meanings but only one translationFor example: DRÔLE FUNNYFUNNY

Words with more than one translationWords with more than one translationWords with one meaning but many translations

For example: BATEAU BOAT SHIPBOAT SHIP

Words with many meanings and many translationsFor example: FEMME WOMAN WIFEWOMAN WIFE

Lexical level

Semantic level

FatherPère

Concrete wordsConcrete words

Lexical level

Semantic level

Love Amour

Abstract wordsAbstract words

Detect semantic factors that influence

relations between translation equivalents and thus, in part, determine the performance in out-of-context translation.

Impact of three interlingual factorsImpact of three interlingual factors number number of translations (experiments 1 & 2) dominancedominance of these translations (experiments

1 & 2) semantic similaritysemantic similarity between the multiple

translations (experiment 2)

Hypothetical representation of semantical nodes shared by the translation equivalents for words with one or more-than-one translations.

More-than-one translations

FEMME WOMAN WIFE

One translation

ARBRE TREE dominant non-dominant

Hypothetical representation of semantical nodes shared by the translation equivalents for words with more than one translation equivalents of which one dominant (D+) and one non dominant (D-).

Semantically similar

BATEAU BOAT SHIP

Semantically dissimilar

TEMPS TIME WEATHERdominant non dominant dominant non dominant

Factors/ hypothesesFactors/ hypotheses

Number-of-translationsOne translation < More-than-One translation

Dominance of the translationsDominant translation < non-dominant translation

Participants 32 native speakers of French students of

English at the University of Montpellier (France)

MaterialMaterial

28Nail - Clou

28Nail - Ongle

28Moon-Lune

L2-L1

28Femme- Wife

28Femme-Woman

28Lait-Milk

L1-L2

More-than-One translation pairs

Dominant Dominant Non-Non-dominantdominant

One translation pairs

Translation Equivalents

*lunemoon

yes

*shoelune

no

450

500

550

600

650

700

L1L2 L2L1

one translationdominant translationnon dominant translation

Experiment 1

Factors / hypothesesFactors / hypothesesNumber-of-translations

One translation < More-than-One translation Semantic similarity of translations

Semantically similar < Semantically dissimilar

Participants 24 native speakers of French, students of

English at the University of Montpellier (France)

Material

10Nail – OngleNail - Clou

10Husband- Mari

Husband -Epoux

20Moon-Lune

L2-L1

10Argent – MoneyArgent - Silver

10Maison – House

Maison - Home

20Lait-Milk

L1-L2

“More-than-One” translation pairs

Similar Similar DissimilarDissimilar

“One translation” pairs

Translation Equivalents

612

683

450

500

550

600

650

700

Experiment 2

Similar Dissimilar

Semantic similarity effect

500

550

600

650

700

L1L2 L2L1

One translation two translations (similar)two translations (dissimilar)

450

500

550

600

650

700

L1L2 L2L1

one transaltiondominant translationnon dominant translation

Our results confirm our hypotheseswords with one translation are processed

faster than words with more-than-one translations.

dominant translations are also processed faster than non-dominant translations.

words with two semantically similar translations are processed faster than words with two semantically distant translations.

DiscussionDiscussion

Despite the complexity of relations between translation equivalents, the bilingual semantic memory remains efficient

The more the translation equivalents share features (hypothetically) at the semantic level the faster the translation recognition

(One translation< Dominant translation< Non-dominant translation)

The similar results in both directions of translation can be taken as an argument in favour of the hypothesis that both translation directions use the semantic route of translation.

Merci, Kiitos, Tack, Gracias, Grazie etc...