This work focuses on AGT Limited* Directory Assistance (DA) Operator Services in Edmonton and is based on recordings of operator-customer dialogues. Twenty five hundred conversations were sampled and transcribed into text. A Directory Assistance State Model (DA-SM) was developed from analyzing the customer-operator dialogues. The model represents the logical stages that exist, and the tasks that are performed by the DA operator while handling a call. Using the model, the major call patterns and subpatterns were identified and their characteristics measured. DA-SM suggested that the operator-customer dialogues follows a logical flow in an organized pattern. The primary operator tasks were then identified for the purpose of finding potential speech recognition applications. From the primary tasks it was identified that the best task for streamlining call flow is early prompting of the location information from the customer. A field study conducted during eight weeks suggested that the operator could prompt the customer to provide the location information during initial dialogue. Keywords: Automatic Speech Recognition; Operator Services; Directory Assistance State Model; Human Factors; Speech dialogues.