A SOM-Based Analysis of Early Prosodic Acquisition of English by Brazilian Learners: Preliminary Results
Autor: | Guilherme A. Barreto, Ana Cristina C. Silva, Ana Cristina Pelosi Silva de Macedo |
---|---|
Rok vydání: | 2011 |
Předmět: |
Self-organizing map
business.industry Computer science Speech recognition First language SIGNAL (programming language) computer.software_genre Linear predictive coding Categorization Stress (linguistics) U-matrix Artificial intelligence business computer Natural language processing Word (computer architecture) |
Zdroj: | Advances in Self-Organizing Maps ISBN: 9783642215650 WSOM |
DOI: | 10.1007/978-3-642-21566-7_27 |
Popis: | In this paper the SOM is used in an exploratory analysis of transfer phenomena from first language (L1) to the second language (L2) related to word/lexical stress. The basic hypothesis tested is whether the parameterization of the speech signal of the learner's utterances by standard signal processing techniques, such as Linear Predictive Coding (LPC), used to encode the input of the network results in efficient categorization of speakers by the SOM. Preliminary results indicates that the combination LPC+SOM is indeed able to produce well-defined clusters of speakers that possess similarities regarding the transfer of stress patterns among Brazilian students in learning English as a foreign language. |
Databáze: | OpenAIRE |
Externí odkaz: |