The Maghrebi Language Identification Corpus (MAGLIC) is a new resource for language recognition comprising over 600 conversational telephone speech recordings in four North African languages: three Maghrebi Arabic dialects (Algerian, Libyan and Tunisian) plus North African French. Calls from consented subjects were recorded on a custom telephony platform in Tunisia, and call metadata was collected including recording time, anonymized phone number and speaker sex. Segments from collected calls were selected for use as development and test data in the NIST 2022 Language Recognition Evaluation, where results indicate that systems show confusability among the three Maghrebi Arabic dialects, despite an inter-annotator agreement study indicating that humans can distinguish these varieties with high reliability. The MAGLIC corpus will be published in the LDC Catalog, making it broadly available for language recognition and other language-related research, education and technology development.