Speech intelligibility degrades when the listener moves from the source. In a highly reverberant and empty church, scores are measured at different distances from a loudspeaker and compared with those obtained from recordings of the speech at the same location. The phonetic material is a set of phonetically balanced lists of triphonemic French words. Two monophonic (omnidirectional, cardiod) and one stereophonic (ORTF) systems are used. Listening of recorded lists is performed with the same loudspeaker(s) as in the church. Results of the experiment indicate smaller variations of scores with the distance. An important improvement of scores is obtained with the stereophonic system. A second experiment consists of using a sound material composed of two voices performing a singing exercise edited in monophony in order to create two distinct sound planes. Listening tests allow the quantiication of the depth localization variations with distance and sound recording system. Results show a good respect of distance and reveal a sudden change in depth localization for stereophony. Acoustical properties of the church, distance factors of the microphones, properties of the stereophony against monophony, and real listening conditions are considered to discuss the experiments.