Linguistic ethnocentricity is pervasive. When scholars and lay people identify a ``General American English'' (GAE) they generally mean, ``the way I talk'', whether they are from Buffalo or Dallas or Philadelphia, just as speakers of modern Arabic dialects, when asked where the purest Arabic is spoken, frequently give their own place of origin (Ferguson, p.c.). This dialect, which everyone seems to believe in, is in fact no real dialect at all. Does General American have a distinction between the sound classes in cot and caught?3.2 Between morning and mourning? Between those in Mary, and marry? Or Mary and merry? Answers to these questions are not to be found, because General American, as popularly conceived, is not a well-defined object, either linguistically, socially, or geographically. Speakers of this supposed dialect may come from Texas or New York. But the dialects of Texas and New York are not phonetically identical.
The closest relation to reality of ``standard'' or ``General American'' English, is often to be found in the self-monitored, self-perceived speech of the individual speaker who is using the term. Explicating the term ``standard'' in this way is controversial, since most people who use the term think they are referring to some actual dialect. Phonologists, phoneticians, linguists and lay people all refer to this imaginary dialect. While it is certainly conceptually possible, its existence, much less linguistic or phonetic uniformity, has not been established.
An actual dialect is the speech form used by a speech community for communication in its day-to-day linguistic activities. It is not to be defined as the self-monitored imaginations of academics. The term, ``General'', seems to imply that GAE is a uniform dialect spoken throughout the U.S. This is contrary to the facts. The work of dialect geographers and sociolinguists for the past 70 years has established, if nothing else, that standard, or General American English does not exist. Hans Kurath, after conducting a large-scale study of American dialect geography, stated: ``The widely accepted assumption that there is a `General American' type of English proves to be equally unfounded in fact; no Southerner or New Englander would ever have made such a generalization.''(1949:vi) He established Northern, Midland, and Southern as three major, distinct dialects of (Eastern) American English: ``There is an extensive Midland speech area that lies between the traditionally recognized `Northern' and `Southern' areas.''(1949:v)(emphasis in original.) Depending on the differences that are considered to count as crucial, more or fewer dialects may be distinguished.
For some, ``General American'' could mean the conglomeration of all American dialects. But this step makes analysis of surface phonological structure impossible, since the different patterns contained in this conglomeration lead to very different structural analyses. Compare, for example, the structure of Alabama English given in that chapter with the structure of Chicago White English.
Research has shown that phonetic and phonological diversification of dialects is increasing, despite the widespread belief in the homogenizing effects of television, radio, and other modern forms of mass communication. Even relatively local subdivisions of these larger dialects, down to the level (at least) of large cities, may differ in important ways when phonetic/phonological behavior is studied. A merger may be in progress in one town, while twenty miles down the road, the phonological distinction is completely robust (Herold 1990).
Paying attention to geographic details is not yet enough; it still ignores the major social subdivisions within urban communities which are also phonetically highly differentiated. Within many U.S. cities, for example, the social division that is correlated with the greatest linguistic and phonetic differences is between (most) blacks and (most) whites. Where phonetic change is in progress, the speaker's age may be systematically related to variations in phonetic details. Gender is another important variable, which has had more attention paid to it in laboratory studies than the variables of age and ethnicity. Finally, speech style is an important variable which has a significant influence on phonetic behavior. For example, speakers behave differently in important ways, often distorting their unmonitored (vernacular) sound pattern, when paying increasing amounts of attention to their speech (Labov 1966). Phonetic differences across these social dimensions have been found in vowel formant-frequencies (Labov, Yeager, and Steiner 1972), in consonant cluster articulation (Guy 1980), pitch range (Tarone 1973), etc. It is therefore impossible to analyse ``General American'' as a single linguistic object.
How can we best approach the study of American English dialects, then? One approach is to analyse each dialect on its own, from scratch. But no English-speaking linguist comes to the study of phonology without intimate knowledge at some level of a particular English sound system. Another approach is to describe a single reference dialect in detail, and then show, for each other dialect examined, the differences between it and the reference dialect. Sound changes and sound correspondences may be insightfully explored by showing how the patterns of a particular dialect differ from those of a relatively well-understood reference pattern. This is true both at the level of phonological structure and of low-level phonetic realizations.
This comparison was implicitly done at the phonetic level, for example, in the foundational ``Quantitative Study of Sound Change in Progress'' (Labov, Yaeger and Steiner 1972). There, the differences between fine-grained phonetic realizations of sound-classes, found in the vernacular speech of particular speakers, was sometimes compared to the phonetic forms of a reference dialect. For example, if the nucleus of the /uw/ vowel in GOOSE overlaps in F1-F2 space with the nucleus of the /I/ vowel in KIT, it is said to be ``fronted''. But ``fronted'' only has meaning relative to some reference point where the vowel is not fronted. In cases where older, more conservative speakers of the dialect were not studied, and unavailable for comparison, differences between the discovered patterns and those of an implicit reference dialect were taken as evidence of sound changes. Thus the reference dialect is implicitly considered to be the historical antecedent of the dialects studied. Such a claim is difficult to test directly, since we do not have much in the way of phonetic records for antecedents of these communities. Comparative reconstruction, to the extent that it has consequences for phonetics, may be able to shed light on the reality of such reference dialects, but until further work is done, we can only claim at best that this reference dialect is an inferred reference point that assists us in understanding how modern dialects got to be the way they are now. Perhaps further research will show that the inferred earlier dialects have more than inferred reality.
It is clearly very useful to be able to describe dialects in relation to such a reference dialect. However, ``Standard English'' or ``General American'' are names that unfortunately suggest a real, existing dialect -- to repeat, a dialect that has been shown not to exist. Or if the many actual dialects that these terms correspond to are analysed as one, the result is an incoherent analysis. If some reference dialect is to be characterized, it should be a well-defined object, at least linguistically. This chapter phonologically defines a reference dialect of this kind, which I will call ``Reference American'', or ``RA''. This dialect is explicitly a useful fiction. Like all phonologists that work on on their native language, my starting point is my own dialect. My Western U.S. dialect has a number of mergers between vowel classes that many English dialects keep separate. In order to characterize a system that is applicable to more dialects than my own, I include a number of distinctions in Reference American phonological structure which are absent in my own dialect.3.3 Some distinctions made in certain American dialects (for example, Southern morning vs. mourning) and in various non-American dialects are not included. The goal here is to describe a single, coherent, synchronic, phonological system, not to include, willy-nilly, all the differences in all known dialects, which may co-occur no single dialect.
Thus Reference American is that dialect from which much of American English can be derived, containing all the distinctions I know of that are likely to have occurred in a single relatively conservative American English dialect. It may be considered as a conservative form of (Northern) American English, or on the other hand as merely a convenient fiction. It is not the non-existent uniform dialect which all cultivated Americans speak. Many of the differences among English dialects, which in this chapter are partly and temporarily ignored, are given their due in later chapters.
The point of this discussion is that we must be clear about the correspondence of what we study to actual, living, spoken dialects. Commonly they are two different things. Clearly, the goal of linguistic research is to elucidate the grammars of well-defined languages, rather than confused conglomerations of different dialects or unreal dialects which are claimed to be real. I do not deny the usefulness of convenient analytical fictions, since Reference American is such a fiction. But claiming that this dialect is one which is generally spoken by all (or educated, or upper-class, or Mid-Western, etc., etc.) Americans would be misleading.
We will find that the phonological structures of Reference American English are applicable to quite different dialects; the fact that the phonology of one dialect should be similar, in some fundamental ways to the phonology of another is quite intriguing. The differences between dialects are statable in terms of a small number of mergers and splits (often in particular environments), plus a number of differences in the rules which fill in the phonetic details of a rather underspecified phonological structure. In some cases, a considerable re-organization of the phonological structure is indicated (as in Alabama and Jamaican Creole, for example). This suggests the very strong claim that, despite the considerable differences in phoneme inventory and in phonetic realization rules, the principles governing the structure of the surface phonology of many English dialects are the same. This is supported by the observation that English speakers are often able to understand each other without a great deal of trouble throughout the English speaking world, despite their considerable differences.
Next I will enumerate a number of sound classes (and discuss the theoretical status of certain kinds of enumerations), and then describe Reference American surface-phonological structure. In later chapters I will apply the basic structures and dimensions argued for in this chapter to several quite different dialects.