This is a detailed analysis of noun phrase errors in written English taken from a corpus of 7500 structures collected from the examination papers of 30 undergraduate commerce students of the University of Malta. A corpus of erroneous and error-free structures is appended and preferred target forms given for erroneous NPs. The study takes into account those features that characterise the NP’s as the output of Maltese learners of English and to a lesser extent also those that characterise them as NP’s within the academic register. To this end the data is also classified as specialised or non-specialised.

Erroneous NP’s are grouped by structure and analysed according to their developmental or cross-linguistic nature. An attempt is made to identify fossilised errors. The number and type of errors suggest that very few subjects have achieved near native-like competence, although the general picture constructed from the analysis suggests that a good number of subjects exhibit features in their production indicative of advanced learner status, such as a generally high proportion of developmental errors in relation to transfer errors. The analysis also provides an insight into the language usage and language preferences of this restricted bilingual community through the use of a questionnaire.