Dictionary Pronunciations: Mine or Theirs or Yours?

The conventional dictionary as a source of at least verified, if not correct (right, standard, best, proper), information about the English language has been with us for well over two centuries. Much ink has been used to discuss the status of the information to be found there. Most philosophical perspectives fall in the dichotomy of PRESCRIPTIVE versus DESCRIPTIVE views of the function of the dictionary. Prescriptivism views the dictionary as sponsoring or perhaps even creating the best knowledge so as to authorize this knowledge; thus, the dictionary sets a standard. Descriptivisim views the function of the dictionary as collector and reflection of what users of the language actually do.

Closer thought, however, suggests that prescription must be based on something: Reasoning from principle? Inspiration? Observation of written texts, including the observations by prior observers (tradition)? Likewise, description can never be complete (that is, a reflection of every instance), which means that selection of what language use to observe, or what to represent of what has been observed, becomes necessary. Thus, all dictionary philosophies must yield to the practical reality of discrimination, the choice of one or some from among many. The inherent complexity of language and its usage underlies this necessity. Literacy, especially print literacy, functionally moved to limit the effects of individual and group variation in various parts of the language. Thus, the creative, often orally based, spelling of the Middle English manuscript tradition (where blood could be spelled five ways in one paragraph) rapidly gave way to selecting one or a few variants as the ones to be used in print. These selections eventually became established as the correct, right, standard way.

In spite of the oddities of contemporary standard spelling (as in cough, tough, and through), one place where users generally trust the authority of the dictionary is in matters of spelling. Such issues as British versus American conventions (gaol, colour), all the variable treatment of compounds (two words, hyphenated, one word), or the common knowledge of unofficial variants found in many public venues (Kit Kat Klub) still do not undermine the common judgment (or is that judgement?) that a dictionary is the source of the correct spelling. Limited kinds of variations are not strong enough to shake the visual conventions of standard spelling, as testified to by failed attempts at spelling reform. Complications of both fact and theory are set aside for the literate functional value of maintaining a high degree of visual consistency for words.

But what of pronunciation? Unlike spelling and unlike meanings and unlike syntactic usage, there are no visual sources (printed texts) from which to derive evidence (save whatever evidence spelling itself suggests). The collection of evidence, sifting and winnowing, must turn to observation of speakers speaking in real time or electronically stored time. And if correctness, from either a descriptive or a prescriptive point of view, is to be sought, what manner of selection is to be followed here? Unlike our long tradition of print literacy and the voluminous array of scholarship of observation, selection, and commentary, the underpinning for dictionary representations of pronunciation is much thinner, in part because repeated observation is possible only within about the last century and in part because the status of individual speakers is not easily comparable to that of printed sources. And so while dictionaries since the eighteenth century have moved from being the observations and efforts of one or a few people to the products of research staffs working from planned observation of print sources, pronunciation characterization is still often the result of a pronunciation editor working from long-term personal observation, others' scholarship, and prior editorial efforts. But with approximately one billion users of the language worldwide and with varying styles often used by any single speaker, what observation, what sampling is possible and desirable?

Arguing from a prescriptivist point of view, the "best pronunciations of the best speakers" would yield a desirable model for users of the dictionary. Defining a...