Analyzing the structure of code-switched written texts

As more written language data become available, the interest in written language mixing / codeswitching (LM/CS) is increasing (Sebba, Mahootian & Jonsson 2012; Sebba 2013). LM/CS in non-naturalistic (e.g., literary) texts raises issues related to gauging (1) the authenticity and representativity of a textual corpus, and deciding (2) whether categories/mechanisms of spoken LM/CS apply to written LM/CS.1 We focus on Guarani-Spanish LM/CS (Jopara) as represented in the Paraguayan novel Ramona Quebranto (RQ). We apply the framework of Muysken (1997; 2000; 2013), developed as a taxonomy of spoken LM/CS. Our contribution extends its applicability to written LM/CS. We show that Jopara has a mix of insertional and backflagging strategies, with infrequent alternations.