Background: Researchers are continuing to investigate the impact of constraint-induced aphasia therapy (CIAT) programs on the spoken language capabilities of people with aphasia at the utterance and discourse level. Currently, there is a lack of consensus on how spoken productions should be measured and there are no investigations that explore the spoken language that people with aphasia use during a CIAT program. Therefore, clinicians and researchers cannot truly know if they are utilizing the most revealing outcome measures to illuminate treatment induced gains in spoken production. Aims: The purpose of this investigation was to examine the within treatment learning of people with aphasia in a CIAT program, by exploring the sensitivity of measures of content, length, syntax, and efficiency at the utterance level and reveal linguistic gains. Methods & Procedures: A repeated measures design was employed to examine 200 utterance samples from eight participants with various types and severity of chronic aphasia between early and late treatment sessions. The measures of correct information units (CIUs), counted words, T-Units, CIUs per utterance, mazes, and mean length of utterance were compared. Outcomes & Results: Nonparametric analyses revealed significant positive gains for the participants in the number of produced CIUs (p = 0.035), counted words (p = 0.012), T-Units, (p = 0.025), and CIUs per utterance (p = 0.012). A significant decrease in the amount of mazes (p = 0.028) was also found. No significant increase was detected in the participants’ mean length of utterance (p = 0.161). Conclusions: The participants displayed advances in the content, syntax, and efficiency of their spoken utterances during the course of a 10-day CIAT program. Specifically, the measures of CIUs, counted words, mazes, and CIUs per utterance emerged as being the most sensitive to capture the within treatment changes made by the participants. The utility of these measures should be further explored to establish treatment baselines, demonstrate within treatment learning, and show posttreatment gains.