At a time of high demand for engineering graduates, the mean graduation completion rate of engineering undergraduates in Australia has been identified as approximately 54% (with considerable variation across institutions and sectors). Such a proportion of non-completions has been viewed as an excessive loss to the qualified workforce of Australia. Broad brush, government-collected statistics do not, however, provide the level of detail required to understand who leaves, when and why they leave and where they go. This paper reports on a pilot study undertaken to precede and inform final decisions on research design and methodology for a multi institutional project seeking to understand and reduce student attrition from engineering degrees across Australia. The aim of the project is to produce guidelines on curriculum formulation and delivery strategies to reduce attrition from engineering programmes while meeting course outcomes.The pilot study was conducted at an institution which has a relatively diverse range of students (a high proportion of whom study part time) and engineering degree structures incorporating traditional and internship-based degrees. Results from a cohort analysis which tracked pathways to completion or non-completion of the degree for the cohorts from two specific entry years are presented. From this analysis, groups of students who "persisted over long periods", "switched to another degree" or "withdrew from the university" were identified and interviewed. Their experiences and stories formed an essential pathway to a better understanding of the dynamics of retention/attrition and factors which required further investigation before the multi institutional study began.

Industry engagement, commonly implemented as a 12 week industry placement during a vacation towards the end of the degree, has traditionally been a provider-mandated component of externally accredited professional engineering degrees in Australia. Such placements are intended to bridge knowledge and capability gaps between academic study and engineering employment and contextualise the final phase of academic study. Changes in the composition of Australia’s engineering industries have made it progressively harder to source such placements. In-curriculum exposure to engineering practice has also been expected, but has been delivered with considerable variability. In 2014 the authors completed a national project, led by the Australian Council of Engineering Deans (ACED), with peak industry bodies and several partner universities, funded from the Commonwealth Department of Industry Workplace Innovation Program, to explore how improving industry engagement could contribute further to engineering graduates’ learning outcomes and employability. The data collected from the engineering students and employers, reported in this paper, can now be regarded as baseline data on industry engagement, against which subsequent developments can be referenced. For the first time, students’ ratings of the value of different methods for industry engagement are shown to be related to their ‘authenticity’. Several industry-inspired in-curriculum interventions were also trialled at partner universities. Guidelines for good practice were developed from melding the experiential findings with theoretical perspectives. In the years since completing the project, the accreditation body, Engineers Australia, has updated and intensified its focus on engagement with practice (including changing its language from ‘exposure’ to ‘engagement’), and many engineering faculties have significantly enhanced their models and requirements for work integrated learning and industry engagement. This paper outlines these changes and examples of new implementations, including virtual and electronically-mediated methods that also reflect ongoing changes in engineering industry practice.

English speech based on accent dependent parallel phoneme recognition (PPR) has been developed. The classifier is designed to process continuous speech and to discriminate between native Australian English (AuE) speakers and two migrant speaker groups with foreign accents, whose first languages are Lebanese Arabic (LA) and South Vietnamese (SV). The training of the system can be automated and is novel in that it does not require manually labelled accented data. The test utterances are processed in parallel by three (AuE, SV and LA) accent-specific recognizers incorporating the accent-specific HMMs and phoneme bigram language models to produce accent discrimination likelihood scores. The best average accent classification rates were 85.3% and 76.6% for accent pair and three accent class discrimination tasks, respectively. Analyses of the contributions to accent discrimination by the phoneme level processing, and by the language model, are described.

No abstract