TL;DR
This study investigates the lexical knowledge of adult Hebrew heritage speakers who are dominant in American English. It reveals that while their Hebrew vocabulary is significantly weaker than their English, they demonstrate creative strategies like code-switching and calquing to bridge lexical gaps, suggesting a unique linguistic blend that still allows for effective communication.
What Problem Does This Paper Address?
The field of heritage language (HL) bilingualism has seen significant growth, but research has largely focused on specific language pairs (e.g., Spanish, Korean, Russian in the US). This leaves a gap in understanding the generalizability of findings across different linguistic contexts. Specifically, Modern Hebrew, a language with a unique Semitic root-based morphology distinct from Germanic languages like English, has been understudied. This paper aims to fill this gap by examining the lexical abilities of adult Hebrew heritage speakers who are dominant in American English, exploring how their HL proficiency compares to their dominant language and how they navigate lexical challenges.
Methods
The researchers employed a mixed-methods approach:
- Multilingual Naming Test (MINT): Participants were asked to name pictures in both their heritage language (Hebrew) and their dominant language (English). This provided a quantitative measure of their lexical accuracy in each language.
- Narrative Elicitation Task: Participants were asked to tell a story in Hebrew based on a picture book. This allowed for a qualitative analysis of their vocabulary breadth, use of code-switching (inserting English words or phrases into Hebrew), and calquing (translating English phrases directly into Hebrew, sometimes creating non-standard constructions).
- Coding Schemata: Non-target responses from the MINT were categorized using a detailed system to understand the types of lexical errors or creative strategies employed. Narrative data was analyzed for code-switching and calquing instances.
Datasets
The study utilized two primary data sources:
- Multilingual Naming Test (MINT): This is a standardized test used to assess vocabulary in bilingual individuals. The study used existing MINT data for both Hebrew and English.
- Narrative Data: This was collected from the 40 participants in the study, creating a novel dataset of Hebrew heritage language narratives.
Key Insights and Ideas
- Dominance Matters: As predicted, participants showed significantly higher accuracy in their dominant language (English) compared to their heritage language (Hebrew) on the MINT. This highlights the impact of the dominant societal language on HL proficiency.
- Creative Lexical Gaps: When faced with lexical gaps in Hebrew, participants didn't just say "I don't know." They employed creative strategies like:
- Code-switching: Seamlessly inserting English words or phrases, suggesting English serves as a "framing language" for thought and expression.
- Calquing: Directly translating English phrases, sometimes resulting in novel but understandable Hebrew constructions.
- Hebrew Morphology is Robust: Despite lexical challenges, participants demonstrated a strong grasp of Hebrew's unique root-pattern word formation and morphology, suggesting that core grammatical structures are well-preserved.
- Universal Trends in Non-Target Responses: The types of non-target responses (e.g., semantic associations, hypernyms) mirrored patterns found in heritage speakers of other languages, suggesting some universal tendencies in how bilinguals handle lexical retrieval.
- "Child-like" Language Persists: The study found instances of participants using older or "child-like" Hebrew vocabulary, likely preserved from early childhood exposure, indicating a unique blend of historical and contemporary language use.
Key Results
- MINT Accuracy: Average accuracy on the English MINT was 92% (SD=5%), while on the Hebrew MINT it was 56% (SD=16%). This difference was statistically significant.
- Non-Target Response Types: The most common non-target response type was "Unknown" (903 instances), followed by "Semantic" (210 instances) and "Morpho-Phonological" (36 instances).
- Semantic Subtypes: Within semantic responses, "Hypernym" (58 instances) and "Associative" (53 instances) were most frequent.
- Cross-Linguistic Influence: Borrowing of English terms was common, with 6 participants producing "ziper" for "zipper."
- Narrative Analysis: Participants produced narratives with an average of 379 lexical units and 134 unique lexical units. The average number of code-switching instances was 3.5, and calques were 3. The range of code-switching instances was significantly wider than calques.
- Correlation: A significant negative correlation was found between the number of code-switching instances and Hebrew MINT accuracy, suggesting that more code-switching might indicate greater lexical gaps.
Limitations and Future Work
- Production-Focused: The study primarily examined lexical production. Future research could incorporate receptive measures to provide a more complete picture of HL lexical abilities.
- Narrative in SL: The narrative task was only conducted in Hebrew. Including a narrative in English would allow for a more direct within-participant comparison of verbosity and language use.
- Background Factors: The study did not deeply investigate the influence of background factors like input quantity/quality, motivation, or social context on HL proficiency. Future work could explore these variables.
- Morphosyntax: While non-target responses hinted at strong morphosyntactic knowledge, this study focused on the lexicon. Further research could delve into morphosyntactic abilities.
- Child HSs: The current study focused on adults. Extending this research to child Hebrew heritage speakers would help understand the stability of these patterns over time.
Summary updated