Language matters: a terminological audit of Australian structural design standards for code calibration

Australian structural design standards are semi-probabilistic: specified values represent random actions and resistances, and partial factors are preferably calibrated so that designs achieve a target reliability. In practice, the factors are largely inherited rather than calibrated, so a standard’s wording is often the only authority on what its numbers mean. We audit the terminology of 25 Australian standards, including draft DR AS 3600:2026, classifying 11,881 occurrences of 28 reliability-bearing terms against a published rubric, using the current Canadian bridge code and 16 Eurocodes as controls. The rubric is a screening instrument: a fault marks wording from which a value’s probabilistic status cannot be recovered. On that basis, 1474 occurrences are counted as faults, chiefly quantile-free ‘characteristic’ values, unquantified modifiers, and unfactored values named ‘design’. Matched by subject, Australian geotechnical, concrete and wind standards carry faults six to ten times as often as European counterparts. Two clause-level case studies show faults no vocabulary audit detects: this lexical audit is a lower bound, and the first step towards the deeper semantic audit the suite requires. We propose a vocabulary consistent with AS 5104:2017, and drafting principles making a value’s probabilistic content recoverable from the clause stating it.