This paper describes Bell Labs' efforts in developing core technologies toward natural language call routing (NLCR) applications. NLCR refers to technology allowing callers of a call center to be automatically routed to their desired destination based on natural spoken responses to an open-ended prompt such as "How may I direct your call?'' Such services are expected to replace interactive voice response (IVR) systems in the future, allowing a better experience for the end user and cost savings for the call center. An NLCR system essentially combines several key technologies, mainly automatic speech recognition (ASR) and topic identification. The role of the ASR system is to convert the input utterance into the corresponding sequence of words. The topic identification module then attempts to reproduce human categorization judgments in order to route the caller to the requested destination, given the hypothesized (possibly partially wrong) word sequence from the ASR system. This paper presents our recent advances in natural language ASR and robust topic identification, focusing particularly on its data-driven aspect and its portability. We also report experimental results from our field trials in the banking domain, illustrating the maturity of the technology and its acceptance by end users, making it an enabler of new revenue-generating services.