When translating a work, it would seem to be useful to pin down exactly what the artist is trying to say, and then express that as clearly and succinctly as possible. But artists aren't generally designing their works for optimal efficient information transfer. As with the "inaccessibility" issues raised in the previous page, an artist often carefully builds ambiguity into the work. What needs to be expressed in the translated work is the content and structure of the work in its intended state of ambiguity.
The ambiguity can be cognitive, where irreconcilable ideas are presented by a work, or it can be perceptual, where the work tempts perception with multiple possible reconciliations of the sensory input from the work. The fight between multiple readings of the work plays out over time, creating a sense of instability. The mind's job is to come to a stable explanation of the sensory inputs it is receiving in order to identify something as either harmless or threat, desirable or undesirable, etc. The work in this case is thwarting the mind's desire for resolution. It is clear that presenting something unambiguously and presenting something ambiguously have entirely different meanings, and so it is essential to maintain this quality through the translation.
In the realm of visual art, examples of work of this sort include most of James Turrell's works in which he creates extremely tangible spatial illusions using light: in some works, the viewer encounters what appears to be a rectangular field of colour applied to the wall. As they get closer and closer to the surface of the field it becomes harder and harder to determine the distance to the field. Almost inevitably, the viewer unconsciously raises a hand to feel for the surface, revealing that there is no surface there at all, but rather a rectangular hole in the wall opening into a second precisely lit room.
Jim Campbell produces works called "Ambiguous Icons" which use very low resolution arrays of LED lights to present moving images that are just at the threshold of our ability to perceive. These works pose an interesting problem. Campbell has tuned them so that a person with normal vision has some initial difficulty making sense of the image. A person with severe myopia would find nothing strange about the image, because as the image is increasingly blurred, the distraction of the detail of the individual lights disappears and image resolves, blurry, but immediately comprehendable.
These particular works create their ambiguity by playing with perception itself and pose extreme challenges with respect to accessibility. One plausible solution in this case might be to find an analogous perceptual ambiguity in the target modality, and create a piece incorporating that.
More generally, there are works which are bound to a particular sensory mode because they present an experience that is specific to that mode. For his 360 degree room for all colours Olafur Elliason created a seamless circular screen backlit by hundreds of red, green, and blue neon tubes under computer control. In many positions of the space, the screen fills your visual field completely and you experience colour in an extraordinary way. My own experience was that I felt my blood pressure rise as the colour shifted to red, then as it settled into green, I felt enormous visceral relief and a distinct calmness. (The description sounds clichˇ. The experience most definitely was not.)
Sound poetry and non-conceptual sound art often rely on sound-centered perceptual experience. Sound Poetry is poetry that plays with sound, turning the mouth of the performer into an extremely versatile instrument. See Paul Dutton's sound poetry for some example recordings. It should be noted that some sound poets make use of graphical notation to write out their sound poems visually.
Other examples would be any of Janet Cardiff (and George Bures Miller)'s soundwalks, where the audience members wear special pairs of headphones which produce the startling impression that she is whispering right in your ear and where sounds in the recording overlay with the actual sounds of the physical space the listener is walking through. But because she uses narrative in her works, there is some part of the work that is not so modality-specific. The sound installations of Maryanne Amacher, in particular "Sound Characters (Making the Third Ear)" provide a more extreme example, where the functioning of the piece is so specific to artifacts of the hearing process that it is highly unlikely that a translation is possible.
Is all music inherently sense-bound in this way? I will be looking much more deeply into this question in the bulk of this inquiry.