At last year’s ASA Fall Meeting the authors proposed a new approach for the correction of errors introduced by an automatic speech recognizer (ASR) during the decoding of difficult, large vocabulary speech tasks. This approach involves overwriting the incorrectly decoded words (or character strings) using a pressure-sensitive pen, on a transparent paperlike interface (PLI) capable of recognizing handwriting on-line. This solution combines the best features of both keyboard entry and mouse selecting: It is appropriate even for out-of-vocabulary words, yet the user’s attention remains concentrated on the screen at all times. In fact, further analysis shows that four intrinsic forms of complementarity between speech and handwriting add to the viability of this solution. Here evidence of complementarity in information, performance, functionality, and robustness is presented. Such complementarity contributes to the relative speed and ease of use of PLI to correct ASR errors. This in turn suggests an extension of this idea toward a jointuse of both ASR and PLI inputs. In this situation, ASR and PLI are used simultaneously, and the results of handwriting recognition are fed to the speech recognition module as apriori information to help reject unlikely possibilities and thereby guide the selection of promising candidates. Experimental results indicate that this approach may lead to substantially higher message recognition accuracy. As it requires users to speak and write consistent messages, psychological experiments that demonstrate this willingness are reported provided the increase in recognition accuracy effectively eliminates the need for error correction.