While human tutors typically interact with students using spoken dialogue, most computer dialogue tutors are text-based. We have conducted two experiments comparing typed and spoken tutoring dialogues, one in a human-human scenario, and another in a human-computer scenario. In both experiments, we compared spoken versus typed tutoring for learning gains and time on task, and also measured the correlations of learning gains with dialogue features. Our main results are that changing the modality from text to speech caused changes in the learning gains, time and superficial dialogue characteristics of human tutoring, but for computer tutoring it made less difference.

This project augments an existing intelligent tutoring system (AutoTutor) that helps learners construct explanations by interacting with them in natural language and helping them use simulation environments. The research aims to develop an agile learning environment that is sensitive to a learner’s affective state, presuming that this will promote learning. We integrate state-of-the-art, nonintrusive, affect-sensing technology with AutoTutor in an endeavor to classify emotions on the bases of facial expressions, gross body movements, and conversational cues. This paper sketches our broad theoretical approach, our methods for data collection and evaluation, and our emotion classification techniques.

Abstract. This study shows that affect-adaptive computer tutoring can significantly improve performance on learning efficiency and user satisfaction. We compare two different student uncertainty adaptations which were designed, implemented and evaluated in a controlled experiment using four versions of a wizarded spoken dialogue tutoring system: two adaptive systems used in two experimental conditions (basic and empirical), and two non-adaptive systems used in two control conditions (normal and random). In prior work we compared learning gains across the four systems; here we compare two other important performance metrics: learning efficiency and user satisfaction. We show that the basic adaptive system outperforms the normal (non-adaptive) and empirical (adaptive) systems in terms of learning efficiency. We also show that the empirical (adaptive) and random (non-adaptive) systems outperform the basic adaptive system in terms of user perception of tutor response quality. However, only the basic adaptive system shows a positive correlation between learning and user perception of decreased uncertainty.

This paper investigates the reliability of detecting a learner’s affective states in an attempt to augment an Intelligent Tutoring System (AutoTutor) with the ability to incorporate such states into its pedagogical strategies to improve learning. We describe two studies that used observational and emote-aloud protocols in order to identify the affective states that learners experience while interacting with AutoTutor. In a third study, training and validation data were collected from three sensors in a learning session with AutoTutor, after which the affective states of the learner were identified by the learner, a peer, and two trained judges. The third study assessed the reliability of automatic detection of boredom, confusion, delight, flow, and frustration (versus the neutral baseline) from sensors that monitored the manner in which learners communicate affect through conversational cues, gross body language, and facial expressions. Although the primary focus of this article is on the classification of learner affect, we also explore how an affect-sensitive AutoTutor can adapt its instructional strategies to promote learning. 1.

This paper evaluates dialogue-based student performance in a controlled experiment using versions of a tutoring system with and without automatic adaptation to the student affective state of uncertainty. Our performance metrics include correctness, uncertainty, and learning impasse severities, which are measured in a “test ” dialogue after the tutoring treatment. Although these metrics did not significantly differ across conditions when considering all student answers in our test dialogue, we found significant differences in specific types of student answers, and these differences suggest that our uncertainty adaptation does have a positive benefit on student performance.

Our research goal is to investigate whether previous findings and methods in the area of tutorial dialogue can be generalized across dialogue corpora that differ in domain (mechanics versus electricity in physics), modality (spoken versus typed), and tutor type (computer versus human). We first present methods for unifying our prior coding and analysis methods. We then show that many of our prior findings regarding student dialogue behaviors and learning not only generalize across corpora, but that our methodology yields additional new findings. Finally, we show that natural language processing can be used to automate some of these analyses.

Abstract not found

The relationship between emotions and learning was investigated by tracking the affective states that college students experienced while interacting with AutoTutor, an intelligent tutoring system with conversational dialogue. An emotionally responsive tutor would presumably facilitate learning, but this would only occur if learner emotions can be accurately identified. After a learning session with AutoTutor, the affective states of the learner were classified by the learner, a peer, and judges trained on Ekman’s Facial Action Coding system. The classification of the trained judges was more reliable and matched the learners much better than the low scores of untrained peers. This result suggests that peer tutors may be limited in detecting the affective states of peer learners. Classification accuracy was poor at constant intervals of polling (every 20 seconds) but much higher when individuals declared that an affect state had been experienced.

Abstract. Human observations and classifications have shown to provide substantial leverage for developing models of students ’ motivation, attitudes, and strategic choices, as a student interacts with an intelligent tutoring system. However, human observation and classification is highly timeconsuming, which has limited its use. We present a technique for conducting human classification on “low-fidelity ” text-based replays of student behavior derived from logs of tutor usage. We show that low-fidelity classification is much faster than live classification, and that low-fidelity classification is approximately as accurate as live classification for detecting a behavior known as gaming the system, using a machine-learning detector of this behavior as the gold standard.

We investigate using the PARADISE framework to develop predictive models of system performance in our spoken dialogue tutoring system. We represent performance with two metrics: user satisfaction and student learning. We train and test predictive models of these metrics in our tutoring system corpora. We predict user satisfaction with 2 parameter types: 1) system-generic, and 2) tutoringspecific. To predict student learning, we also use a third type: 3) user affect. Alhough generic parameters are useful predictors of user satisfaction in other PAR-ADISE applications, overall our parameters produce less useful user satisfaction models in our system. However, generic and tutoring-specific parameters do produce useful models of student learning in our system. User affect parameters can increase the usefulness of these models. 1