From the pages of Xin Tang, Mark Swofford has resurrected a classic piece by John DeFrancis entitled "Homographobia." Here's Mark's post. The entire essay may be found here. A pdf of the whole issue of Xin Tang 6, in which John's essay appears, is available here.
This is the opening paragraph of John's essay:
Homographobia is a disorder characterized by an irrational fear of ambiguity when individual lexical items which are now distinguished graphically lose their distinctive features and become identical if written phonemically. The seriousness of the disorder appears to be in direct proportion to the increase in number of items with identical spelling that phonemic rendering might bring about. The aberration may not exist at all among people favored by writing systems that are already closely phonemic, such as Spanish and German. It exists to a mild degree among readers of a poorly phonemic (actually morphophonemic) writing system such as English, some of whom suffer anxiety reactions at the thought of the confusion that might arise if, for example, rain, rein, and reign were all written as rane. It exists in its most virulent form among those exposed to Chinese characters, which, among all the writing systems ever created, are unique in their ability to convey meaning under extreme conditions of isolation.
Since the so-called homograph problem is invariably brought up by opponents of Chinese romanization, John's demolition of this bugaboo in XT 6 is crucial for allaying the irrational fear of romanization that plagues many otherwise reasonable individuals. In his inimitable style, John uses a combination of powerfully marshaled evidence and clever wit to allay fears of homography in romanized Chinese. His citation of homographs and near-homographs in Hawaiian and in Vietnamese is particularly entertaining.
My rule of thumb is always this: if homography were a problem in (more or less) phonetic scripts based on real, spoken languages, then homophony would be a problem in the speech of such languages.