THE FOLLY OF ISOGLOSSES IN
DIALECTOLOGY AND THE 
SPURIOUS MIDLANDS 
DIALECT IN THE USA

© 1998, 2002 by Orchid Land Publications

[20020510]

     There are two subtypes of a discredited approach to dialect analysis; one is to draw lines (isoglosses) around areas or regions; the other, to draw isoglosses around social types.  While this is not the place to give all of the details, the reader can learn why this approach is discredited in Bailey, Essays on time-based linguistic analysis (1996, Oxford University Press), Ch. 4.  Perhaps most important of the reasons against this approach is that in its averaging over many exceptions (within the area or social characteristic lined off), it does not tell us what is in an individual's internalized language ability--at least in most situationa of linguistic complexity. 

     The approach that works where others fail is:

First find what goes with what.

If we find scattered around the regional or social landscape the following data (where a, b, c, d represent diverse phenomena)

ba   dcba  abc  a adcb ba cabI and suchlike

we determine the INTERNAL LANGUAGE PATTERN that the data reduce to without fudging--first

a ab abc abcd

Then generalize the formula that fits the situation.

In the foregoing example, the appropriate formula is:                         

                   d implies c implies b implies a

i.e. wherever d is found, c will be found, but not necessarily conversely; wherever c is found, b will be found, but not necessarily conversely, and wherever b is found, a will be found, but not necessarily conversely--where "wherever" refers to a given style of a given speaker.

     This approach has solved (as the chapter referred to illustrates) extremely complex problems found in the North of England and in ancient Greece--problems totally unamenable to current issoglossic methodologies.  (Numerous patches have been suggested to improve the obvious defects of these methods, but none has ever succeeded in making the needed improvements.)  Indeed, researchers have looked at the Greek problem and come up with almost every possible answer except the one maintained by the ancient Greeks themselves; but this approach agrees with those Greeks.  See the reference already given.
     This approach tells us more, predicting patterns not actually uncovered in the data to hand (it tells us what to go looking for, in this sense making a theoretical prediction); it can actually uncover differen waves, i.e. waves spreading (at different times) from different centers.  The data themselves reveal the fallacy of believing that neat lines on a geographical or social map actually hold valid in the sense that the speakers of, say, c all fall within the geographical or social line, there being no area having ac without b--except in paltry instances that can be accounted for.  Even where outsiders speaking a different form of the same language move into a another area, their "dialect" will not (more than trivially) violate the generalizations that d implies c,   c implies b, and b implies a.  
    This methodology not only captures the real scientific generalizations that areal and social isoglosses have either never been able to capture or, with good luck, capture only irrelevantly.

CLICK HERE FOR ILLUSTRATION 

     One of the worst flights of fancy in linguistics* has been the spurious Midlands dialect in the USA (the illegitimate offspring of the Midland dialect in England).  The researchers devoted to this mythology found what they claim are three dialect boundaries in non-coastal America.  (Strangely enough the northern boundaries of the ones to the south at each point are more northerly than the southern boundaries of the contiguous more-northerly region!  That hasn't deterred them.)  The four regions separated by these three "lines" could have been more rationally (insofar as there is much rationality in the whole business) termed upper and lower Northern States and upper and lower Southern; but no, they chose to say that there is between the North and the South a Midland type, subdivided into upper and lower or more northern and more southern subdivisions.  And lo, this has gotten into otherwise respectable dictionaries!  No one disputes that there are recessive areas along the Gulf and Atlantic coasts of the US--one centered on New Orleans, the Tidewater area (with some differences between Richmond and Charleston), New York City (with northern New Jersey), and New England.  And there can be little  doubt that places near to the Great Lakes have traits in common that differentiate them from other Northern types.  For that matter, Western States (clear out to San Diego) have Southern traits mixed with (more) Northern ones.  As "r-lessness" dies out in many regions among whites (it seems to be having a rebirth in New York City), many African American speakers still preserve something that is not all that basically different from the old Southern and Tidewater speech.  Of course, African American English varies a little from place to place in outside of the South, as in the South, but it is generally recognizable.  One need only mention the fortis (even ingressive) occlusives and overrounded "o" common to many African Americans and many older Deep Southern and Appalachian Whites.  These are more than just items in a listing; they say something about the fundamental architectonics of the kinds of English in question--though little about its current social or regional repartitions.
     Anyhow, as it needs to be said, every lectal analysis should be, if valid, something that can be fitted into the overall analysis of English.  This consideration is of course rejected by the minilectalists (the overwhelming majority of linguists)--those who refuse to analyse anything over and beyond one style used by one speaker at one time.  Such scholars reject the idea that there is "a language," e.g. English, in any sense other than a sociological one--and an abstract sociological one at that (cf. Chomsky's notion of studying a non-existent, idealized homogenous community of speakers).   This mythology makes one wonder if such linguists are not really dreaming when they get around to discussing these foundations of linguistic analysis.  But they are the majority--in a statistical sense, how could they be so wrong.  You only have to go to the history of science to see that such is indeed possible.
     But the method of just running around and listing a few items (the way dialectologists and sociolinguists do) can never rise to a systematic, let alone scientific, analysis of the internal architectonics of language--what alone would have the potential of rising to the level of scientific analysis.  Even if isoglosses were a rational way to go about investigating dialects and other kinds of language variation, it could hardly have been managed less rationally than in the invention of the spurious Midlands dialect in the USA.  It is passing strange that, while scientists took a quarter of a century to admit the existence of earth plates after they had been first proposed (other parallel examples from science are available), non-scientists like linguists will right off adopt the silliest idea that people  trained in obsolete ideologies come up with.  Why non-science is so different from science in this respect--one might have antecedently expected that scientists, at least, would recognize truth more quickly--must be left to our philosophers.

---

    *It has been second, perhaps, only to the notion [engendered in a climate of antitheoretical, positivistic thinking long gone except among dialectologists and sociolinguists] that the only kind of language amenable to rational analysis is one style of one speaker at one moment--the idiolect or minilect; or perhaps the accompanying absurdity that  binary oppositions among those mythological phonemes could create a "system."  No refinement, no amount of fudging, not even watering down binarity, has ever made  minilectal phonemic analysis doable without a lot of hedging and compromising with the idea.

CLICK HERE FOR MORE ABSURDITIES ACCEPTED BY LINGUISTS