TL;DR
This paper investigates whether prosodic phrasing can help second language learners discover unfamiliar word order rules in a new language. They found that non-aligned prosodic phrasing during training, which contrasts with syntactic structure, surprisingly aids learners in identifying clause-internal word order differences, but only when syntactic variability during testing is low.
What Problem Does This Paper Address?
Learning the grammatical rules of a new language is a complex process. One significant challenge is mastering word order, especially when it differs from one's native language. While native speakers often use prosodic cues (like intonation and pauses) to disambiguate sentences, it's less clear how effectively second language learners can leverage these cues to learn new grammatical structures. Specifically, this paper focuses on how prosodic phrasing might help learners discover word order alternations within clauses, a common feature in many languages but not all. The authors aim to determine if specific types of prosodic phrasing during learning can facilitate the acquisition of these novel word order patterns.
Methods
The researchers conducted two experiments using a novel language learning paradigm. Participants were native English speakers with no prior knowledge of German.
Experiment 1: * Training: Participants were trained on simple and complex German sentences. The complex sentences featured a subordinate clause with a verb-final word order, contrasting with the verb-second order in main clauses. Crucially, the prosodic phrasing of these complex sentences was manipulated: * Aligned Prosody: Prosodic boundaries matched the syntactic structure (e.g., a pause after the main clause verb). * Non-aligned Prosody: Prosodic boundaries did not match the syntactic structure (e.g., a pause after the subordinate conjunction "dass," creating a unit with the main clause). * Testing: Participants were presented with sentences that were either grammatically correct or contained errors in word order or lexical items. They had to indicate if the sentence correctly described a presented picture. * Analysis: Mixed-effects logistic regression models were used to analyze the proportion of correct responses, considering participant group (aligned vs. non-aligned training), sentence condition, and experiment phase.
Experiment 2: * Modification: This experiment was similar to Experiment 1 but introduced more syntactic variability during the test phase by including additional complex sentence structures (e.g., scrambling). * Goal: To assess if increased syntactic complexity during testing impacts the effectiveness of prosodic cues learned during training.
Datasets
Not explicitly discussed as pre-existing datasets. The study created its own set of German sentences for training and testing, tailored to manipulate prosodic phrasing and introduce specific grammatical errors (word order and lexical). The sentences were designed to be understandable to English speakers through cognates and simple imagery.
Key Insights and Ideas
- Non-aligned Prosody as a "Highlight": The study suggests that prosodic phrasing that contrasts with the syntactic structure (non-aligned) can be more effective for learning clause-internal word order differences than phrasing that aligns with it. This non-aligned phrasing might create a more direct comparison between the main clause and the subordinate clause, making the word order difference more salient.
- The Importance of Low Syntactic Variability: The effectiveness of prosodic cues appears to be highly dependent on the testing environment. When the test phase involves a high degree of syntactic variability, the benefits of prosodic training diminish significantly.
- Transfer of Native Language Knowledge: Participants struggled with incorrect word order in German subordinate clauses because they often mirrored English word order, leading them to incorrectly judge these as correct. This highlights the strong influence of native language grammar.
- Cognates as Reliable Cues: Participants were generally good at identifying incorrect lexical items, likely due to the use of German words that were close cognates with English.
Key Results
- Experiment 1:
- Simple Sentences: The non-aligned group performed better overall than the aligned group. There was a marginal interaction between sentence condition and experiment phase.
- Complex Sentences: The non-aligned group again performed better overall. A significant interaction between participant group and experiment phase was observed, with the non-aligned group showing significant learning in the second half of the experiment, while the aligned group did not.
- Experiment 2:
- Simple Sentences: Participants performed differently across sentence conditions, but there were no significant differences based on participant group or experiment phase.
- Complex Sentences: No significant effects of sentence condition, participant group, or experiment phase were found. The increased syntactic variability in the test phase seemed to negate any benefits of prosodic training.
Limitations and Future Work
- Limitations:
- Participants performed poorly on incorrect word order trials, suggesting a strong interference from native English word order.
- The study used a limited set of German sentences and a specific set of cognates.
- The "novel language" was German, which, while unfamiliar to the participants, is a real language with complex grammar.
- Future Work:
- Investigate this effect with a larger number of participants and a wider range of L1 language backgrounds.
- Explore the interaction between prosodic cues and other linguistic features in language acquisition.
- Further research into how learners overcome native language interference when encountering different word order rules.
Summary updated