Keeping track of Pennsylvania German: A discussion of evidences found in A linguistic atlas of Pennsylvania German

David Bowie
Linguistics 460
Final paper

Table of contents

0. Introduction. In 1954, Carroll E. Reed and Lester W. Seifert released A linguistic atlas of Pennsylvania German (hereafter LAPG), a slim volume that, judging from the introduction, was intended as a preliminary report on research on variation in the Pennsylvania German (hereafter PG) language. Unfortunately, it seems that a more comprehensive report on Reed and Seifert's dialectological research was never published, and so LAPG is all that exists for someone trying to get a handle on variation in PG in the early to mid-twentieth century; even so, it is possible to use information gleaned from LAPG to draw various conclusions about the settlement by the Pennsylvania Germans of the area surveyed, as well as bringing up some interesting questions about the language itself. It becomes a particularly interesting item to look at in light of the fact that reports from the field seem to be unanimous in reporting the death of PG among the non-plain PG groups, and its perseverance among the plain groups1 (see, for example, Van Ness 1980; Huffines 1989; Meister Ferré 1991).

1. The phonemic system of PG. The particulars of the phonemic system of PG has occasionally been the subject of some debate; however, most sources seem to agree with the one laid out by Buffington and Barba (1965), which is laid out in Appendix 1. For the purposes of the current discussion, the most important disagreement with this system is the one raised by Beam (1982) and Stine (1990),2 who claim that Buffington and Barba's long [æ] is actually pronounced either as [ei] or [e:] (which one is not entirely clear).3 Bowie (forthcoming) has noted that in the Pennsylvanian dialects of PG (that is, the dialects that LAPG covers) there has been a change in the vowel system (which is either in progress or was frozen at some point before it reached completion) such that pre-rhotic short front vowels have been lowered from the positions shown in diagram (1); this is not exactly the phenomenon that this paper deals with (see section 2), but it is a closely related one. Bowie did not deal with the variation within the system, however, and the contents of LAPG give one a chance to look at the variation as it is distributed geographically to see whether geography had had any impact on the changes occurring in PG by the time that the interviews for LAPG were made.


2. The focus of the current study. The focus of this paper is the variation between æ and other vowels before r. As already noted, Bowie (forthcoming) has already dealt with some of the history of part of this alternation, finding that there has been a lowering of short front vowels to æ before r in eastern Pennsylvanian dialects of PG; this paper extends that by looking at other environments reported in LAPG, specifically the variation between long a: and æ: in barig/baerig4 'mountain,' the variation between short a and æ in farschde/faerschde 'heel,' and the variation between long e: and æ: in kere/kaere 'to sweep.' (Although these are by far not all of the words in LAPG showing the alternation between a vowel and æ before r, these were chosen as the best to give this report a solid grounding.) In addition, this paper deals with a pair of lexical variants, bodem/bode versus flor 'floor' and weschkessel versus weschbailer/weschboiler 'laundry boiler,' in an attempt to find out if there are parallels between lexical borrowing and the movement of pre-rhotic vowels to æ.

3. The atlas used in this study. The atlas used for this study is, as already noted, A linguistic atlas of Pennsylvania German (Reed and Seifert 1954). LAPG covers part of southeastern Pennsylvania, with a total of 104 informants in the following counties (the number in parentheses after each county is the number of informants from there): Berks (27), Lehigh (21), Lancaster (15, all from the northern half), Schuylkill (7), Snyder (6), Dauphin (5), Northumberland (5), Lebanon (4), Bucks (3), Montgomery (3), Carbon (2), Juniata (2), Monroe(1), Northampton (1), Union (1), and York (1). There are seventy-nine maps of this area showing variants of individual lexical items by informant, and an additional five showing isoglosses in this region, of which only one is not derived from other maps in the volume, along with a base map of the region (showing the location of all but two of the informants, who were located off the map) and two colonial-era historical maps showing the county borders in the region and a very rough sketch of the roads in the area, both as of 1776, as well as a map of Pennsylvania showing all of the counties of the state with the PG speech areas shaded in; from this last map it is apparent that the atlas looks only at non-plain speech communities, as Washington and Chester counties are both completely unshaded, and both had significant numbers of PG-speaking plain communities at the time. Finally, the atlas contains ten maps of southwestern Germany showing parallels to some of the maps of southeastern Pennsylvania, as well as one base map of southwestern Germany and one map of the Rhine-Palatinate in the eighteenth century. It should be noted that LAPG was apparently released as a preliminary report on the authors' research into variation in the area it covers, and perhaps as a result unfortunately contains no sociolinguistically useful information about the informants such as age, occupation, educational level, &c. If there were other atlases that could be used to add to the information available in LAPG it would be welcomed, but both my own search as well as searches by Richard C. Beam and Jennifer L. Griffith (personal contact) attempting to find one have to date been fruitless.5

4. Materials used for mapping. The particular maps chosen for mapping in this study have already been mentioned in this report; they are barig/baerig (LAPG map 8), farschde/faerschde (LAPG 10), kere/kaere (LAPG 12), bodem/bode/flor (LAPG 37), and weschbailer/weschboiler/weschkessel (LAPG 39). The individual points on these maps were entered into an Excel spreadsheet (a copy of which is appended to this paper in Appendix 2) with seven columns; the first column is the nearest town along with a rough approximation of where the point lies in relation to that town (if necessary), the second column is the ZIP code of that town, and the final five columns contain one-letter codes indicating what was spoken by each of those informants.6 The values from this spreadsheet were them transferred into MapInfo 4.0 in order to produce the maps attached to this report.

5.0. Maps showing regional variation. Seven maps were produced with MapInfo 4.0 and the ADC WorldMap North America CD-ROM, and they are contained in Appendix 3 (and, in the HTML version, with their descriptions below).7 Each of these maps will be discussed separately below, with discussion of the interaction of the information contained on the maps being reserved until later in this report.

base map--click for full size image

5.1. Base map. This map simply shows the area shown in the other maps, along with a black square where each informant was located. (In area where the same town held two or three informants, they are placed in a vertical line near each other.) It should be noted that state borders are shown on this map (Pennsylvania-Maryland to the south, Pennsylvania-Delaware to the southeast, and Pennsylvania-New Jersey to the east) to give a better idea of where the informants were located; in addition, one can see the foothills of the Allegheny Mountains stretching from the west side of the map toward the northeast corner, the Delaware River in the southeast, the Schuylkill River running into the Delaware River in Philadelphia flowing southwest, and the Susquehanna River flowing from the Susquehanna Valley in the foothills of the Allegheny Mountains down into Maryland in about the center of the south edge of the map.

barig/baerig map--click for full size image

5.2. Map 1: barig/baerig. This map deals with the distribution of long a: and æ: in the PG word for 'mountain' (note that Stine (1990) gives barig as the spelling and pronunciation of the word). By looking at the map one can see that the use of æ in this word is not very widespread, even to the point that æ does not seem to predominate anywhere (except possibly far into the Susquehanna Valley), and isogloss 1-a shows the line south of which one hears barig exclusively.

farschde/faerschde map--click for full size image

5.3. Map 2: farschde/faerschde. This map deals with the distribution of short a and æ in the PG word for 'heel' (for which Stine (1990) gives farschde as the spelling and pronunciation). This map shows that the use of æ in this word is much more common than in barig/baerig; isoglosses 2-a1 and 2-a2 show the lines separating the central a-pronouncing region from the northern and southern æ-pronouncing areas (note that these isoglosses might actually ought to be a single line, but there is not enough information from the Schuylkill River Valley to show where in that region the exact border of the production of a or æ in these words actually is). In addition, isoglosses 2-b1, s-b2, and 2-b3 are included to show areas in which there are exceptions to the area's general pattern; it should be noted that 2-b2 and 2-b3 encircle places in which there is variation between the informants who were very close together.

kere/kaere map--click for full size image

5.4. Map 3: kere/kaere. This map shows the production of long e: and æ: in the PG word for 'to sweep.' In this case æ is by far the rarer form, much like in Map 1, but unlike Map 1 æ appears only (with one exception in the Susquehanna Valley) in Lancaster County in the south, Isogloss 3-a marks the limit of consistent æ-production, and isogloss 3-b marks the limit of the communities in which one hears variation. Isogloss 3-c encircles the one community outside of Lancaster County in which one hears the æ form.

bodem/bode/flor map--click for full size image

5.5. Map 4: bodem/bode/flor. This map shows the production of lexical variants for the PG word for 'floor,' of which two, bodem and bode, are clearly descended from European German Boden 'floor,' and the other is a borrowing from English floor. It is worth noting that the flor form is generally mixed with the other forms, so that isogloss 4-a simply shows the southernmost limit of the borrowing. Isogloss 4-b splits the German-descent forms between those that have a final nasal and those without one.

weschbailer/weschboiler/weschkessel map--click for full size image

5.6. Map 5: weschbailer/weschboiler/weschkessel. This map shows the production of lexical variants of the PG word for 'laundry basin,' of which two, weschbailer and weschboiler, are borrowed in part from the English laundry boiler, and one is more closely related to the European German Waschkessel 'laundry basin.' Unlike the case in Map 4, in this case the borrowed forms often appear completely separate from the German-descent form, so that isoglosses 5-b1 and 5-b2 show the rough limits of where one hears the borrowed forms exclusively. Isogloss 5-a splits the borrowed forms, showing the limit between where one hears the form farther from the English (weschbailer) exclusively from where one hears both it and the form closer to the English (weschboiler).

isoglosses from maps 1-5--click for full size image

5.7. Isoglosses from Maps 1-5. This map simply shows all of the isoglosses from Maps 1-5 along with their identifying numbers; it is worth noting that isoglosses 2-b1, 2-b2, 2-b3, and 3-c are different sorts of isoglosses than the rest, as they mark points where there are exceptions to the patterns the other isoglosses on their maps delimit. The remainder of this report will deal with drawing conclusions about the variation in PG from these isoglosses and other information found in LAPG.

6. Co-occurrence and non-co-occurrence of PG variants. A look at the map with all of the isoglosses reveals some interesting co-occurrences. Most strikingly, there is a bundle of isoglosses 2-a2, 3-a, 5-b2, and perhaps 4-b at the northern border of Lancaster County, and 2-a2 and 5-b2 continue together as they cut across the Susquehanna Valley. Other points where isoglosses bundle together can be seen in the northeastern section of the map, where isoglosses 1-a and 5-a bundle together for a short distance, and in the north central region of the map, where isoglosses 1-a and 2-a1 bundle together as they cross the foothills of the Alleghenies; Also important to note, although perhaps of less interest, is the fact that the variation surrounded by isogloss 2-b3 covers much of the same region as the variation found between isoglosses 3-a and 3-b. On the other hand, some items that might be expected to co-occur do not; except for the short stretches noted earlier, the isoglosses from Maps 1-3 do not match, and the isoglosses showing borrowing versus non-borrowing of particular words (that is, isoglosses 4-a, 5-b1, and 5-b2) do not match at all.

7.1. The effects of geography: Barriers and apparent barriers. There are two major geographical features in the area under study that would seem likely to have an effect of blocking the spread of innovations among speakers of PG--the Susquehanna River and the beginning of the Allegheny Mountains. However, the only bundling of isoglosses that has anything to do with these features is the merging for a short stretch of isoglosses 1-a and 2-a1, and this is a rather unsatisfactory result, as they cross the mountains together rather than are blocked by the mountains together. However, one of the historical maps in LAPG explains this problem--LAPG's Map 4 gives a very rough sketch of the roads in the region as of 1776,8 and one of the roads crosses the mountains at approximately that point (stopping short of the Susquehanna River). Then, once one thinks to look at the roads, other isoglosses are explained--beyond the example already mentioned, isoglosses 4-a and 5-a follows roads for their entire length. Interestingly, these last two are cases of lexical diffusion, and even more interestingly, isoglosses 4-a and 5-a are those that mark the farthest spread of the most English-like form. In those two cases it appears that the roads actually marked the borders for speech regions, such that for some reason heavy borrowing from English was blocked by particular roads. In the case of isoglosses 1-a and 2-a1 crossing the Allegheny foothills together at a road, it may be that the road in question did not function as a barrier, but actually served as one conduit (out of multiple ones, most likely) for innovations to be spread, but that the innovations (as well as the older forms) were carried by those passing through the towns in the Allegheny foothills into the Susquehanna Valley beyond. It should be noted that the road these two isoglosses follow ended with the descent into the Susquehanna Valley (and it was the only eighteenth-century road to cross any part of the Alleghenies within the area being studied), so it is unfortunately not possible at this time to determine what effect migration patterns had in the Susquehanna Valley.

7.2. The effects of geography: Conduits. West of the Susquehanna River and southeast of the Allegheny foothills, it does not appear that there are any natural features in the PG-speaking region to block the expansion of innovations, and looking at the eighteenth-century roadway system, it seems that certain roads worked as conduits for the expansion of phonological change in southeastern Pennsylvania. The most striking of these is isogloss 5-b1, which runs perpendicular to a cluster of roads emanating from Bethlehem (in Northampton County) and Hosensack (in Lehigh County). This is so striking because one of the roads it runs perpendicular to--and therefore one of the roads it appears to have used as a path for spreading--is the same road that isogloss 5-a found such a barrier. Similarly, isogloss 4-b cuts directly across several roads, suggesting that its spread was aided by lines of communication marked by roads, and in this case again the roads cut across were roads that were a barrier to the spread of lexical forms borrowed from English.

8. The special status of Lancaster County. To this point mention has been avoided of the most striking bundling of isoglosses, which occurs at about the northern border of Lancaster County. Upon seeing such a thing, one might expect that there is some sort of geographical feature that marks that line, but there are no natural features to cause such isoglosses, nor is there any close association with a particular road or roads, either as a barrier or a conduit for the spread of change. One explanation for this, however, may exist in the population of PG speakers in Lancaster County--although Lancaster County (particularly northern Lancaster County) has consistently been the home to the greatest concentration of plain PG speakers in Pennsylvania, it does not seem to be mentioned very often as a center for large numbers of non-plain PG speakers, which means that the difference in Lancaster County may be that the non-plain PG speakers (who, as was mentioned earlier in this paper, are the population LAPG focuses on) would be more likely to have more contact with plain speakers of PG than the non-plain PG speakers in other counties. The only problem with this sort of idea is that it makes it seem likely that the non-plain PG speakers in Lancaster County would be more likely to speak like their plain counterparts, when this is not the case--in general, reports from the field have found that plain PG speakers are more likely to use English borrowings than the non-plain (see Huffines 1988 and Fuller 1996 for discussion and further references), and so if the extensive contact between plain and non-plain PG speakers in Lancaster County is the cause of the differentiation of Lancaster County speech from that of the rest of the region then it would have to be the non-plain speakers marking themselves linguistically as non-plain. In the end it seems likely that this contact is at least part of the cause for the differentiation of Lancaster County speech from the rest of the region, but further research needs to be undertaken in this area.9

9. Summary. In the end, there are not many firm conclusions that can be drawn from this preliminary excursion into LAPG, but there are a few hypotheses that can be drawn to explain the variation in PG that LAPG captures. These hypotheses are listed in (2), in some cases along with a couple of additional thoughts on the matter.

  • Natural geographical features such as major rivers and mountains have had less influence on the spread of variation in PG in southeastern Pennsylvania than features such as roads
  • Particular lines of communication helped spread phonological changes while at the same time blocking lexical borrowings from spreading
  • Lancaster County acts as a separate linguistic area, perhaps as a result of there being more extensive contact between plain and non-plain PG speakers there than elsewhere
10. Future research. There are several ways in which future research would be most helpful in shedding light on the questions raised in this report (as well as the many more questions that exist). Some of these have been listed in footnotes throughout this paper, and the most significant of them10 along with various other items future research should take are listed in (3).

  • Find more information about the informants in LAPG; particularly important is determining whether all the informants were, in fact, non-plain (see footnote 4)
  • Get more accurate ZIP codes for both those cities currently listed as 13000 and those whose current ZIP code number places them a few miles from where they actually were (see footnote 5; this may result in isoglosses slightly different from those presented in this report)
  • Plot more maps with the same variables as Maps 1-3 and compare whether the same segment consistently behaves similarly in different words
  • Develop/find an overlay for the maps which allows county boundaries to be placed on the maps, allowing for easier references to counties
  • Develop an overlay for the maps showing eighteenth- and nineteenth-century roads for the region; this is an extremely important thing to do (see footnote 7)
  • Find information on the settlement patterns of the area, not just by Germans, but also by those who preceded them
  • Carefully compare pairs of maps in LAPG such as numbers 43 and 43a, which show the distribution of the same item in both southeastern Pennsylvania and southwestern Germany
  • Look carefully into the language contact situation between plain and non-plain PG speakers in Lancaster County and elsewhere, particularly between the Industrial Revolution and about 1950
These items are probably too much for one follow-up to this report, but that many are included to give an idea of the wide range of openings left for future research in this field.


Beam, Richard C.
1982. Pennsylvania German dictionary.
Bowie, David.
Forthcoming. "Voah mei daett sei deitsh: Developments in the vowel system of Pennsylvania German." Penn working papers in linguistics, Volume 4 (Proceedings of the 21st annual Penn Linguistics Colloquium) .
Buffington, Albert F. and Preston A. Barba.
1965. A Pennsylvania German grammar. Revised ed. Published as Yearbook of the Pennsylvania German Folklore Society. V27.
Dorian, Nancy, ed.
1989. Investigating obsolescence: Studies in language contraction and death.
Fuller, Janet.
1996. "When cultural maintenance means linguistic convergence: Pennsylvania German evidence for the Matrix Language Turnover hypothesis." Language in society. 25:493-514.
Huffines, Marion Lois.
1988. "Pennsylvania German among the plain groups: Convergence as a strategy of language maintenance." Pennsylvania Mennonite heritage. 11(3):12-16.
1989. "Case usage among the Pennsylvania German sectarians and nonsectarians." Dorian 1989:211-226.
Meister Ferré, Barbara.
1991. Stability and change in the Pennsylvania German dialect of an Old Order Amish community in Lancaster County. University of Georgia PhD dissertation.
Reed, Carroll E. and Lester W. Seifert.
1954. A linguistic atlas of Pennsylvania German.
Stine, Eugene S.
1990. Pennsylvania German to English dictionary.
Van Ness, Silke.
1980. Changes in an obsolescing language: Pennsylvania German in West Virginia.
Note that in addition to the items listed here and in the introduction to LAPG, it would likely be worthwhile to sift through issues of American speech from the 1920s through the mid-1950s. There was a great deal of interest among dialectologists in PG at that time, and although some of what was reported then has been debunked, a careful reading of those materials should provide a solid background in the state of dialectological study into PG preceding the publication of LAPG.

Appendix 1

The sound system and orthography of Pennsylvania German
(following Buffington and Barba (1965))

Accented short vowels:
aaglatt 'smooth,' ball 'soon'
æaeaernscht 'earnest,' Maetsch 'match'
eegewwe 'to give,' Wedder 'weather'
(i)idick 'thick,' Kind 'child'
(wedge)oDochder 'daughter,' oft 'often'
(u)uBuch 'book,' Budder 'butter'
Accented long vowels:
(oh):aa/aah/oNaas 'nose,' Raahm 'cream'
æ:aeAerd 'earth,' gaern 'gladly'
e:e/ee/eeh/ehlese 'to read,' See 'sea'
i:ie/iehgrie 'green,' frieh 'early'
o:o/oo/ohrot 'red,' Boot 'boat'
u:u/uu/uhHut 'hat,' Schuh 'shoe'
Unstressed vowels:
(schwa)ehawwe 'to have,' brode 'to fry'
iischeeni 'pretty,' gleeni 'little'
aerBruder 'brother,' vergesse 'to forget'
auauGaul 'horse,' Haus 'house'
aieiheit 'today,' mei 'my'
(oh)ioiHoi 'hay,' Moi 'May'
bbbeisse 'to bite,' Lumbe 'rag'
j/xchBicher 'books,' Dach 'roof'
dddrei 'three,' Bruder 'brother'
ffFenschder 'window,' uff 'up/on'
g/jggewwe 'to give,' Regge 'rain'
hhhawwe 'to have,' Hund 'dg'
kc/ck/ghwacker 'awake,' gheere 'heard'
llLoch 'hole,' alt 'old'
mmMesser 'knife,' Raahm 'cream'
nnnee 'no,' Mann 'man'
(engma)ngRing 'ring,' englisch 'English'
pp/bhPeif 'pipe,' bhalde 'to keep'
rrrufe 'to call,' Biere 'pears'
ss/ssSeel 'soul,' besser 'better'
(esh)schscheisse 'to shoot,' wesche 'to wash'
t(esh)tschTschab 'job,' rutsche 'to slide'
ttTax 'tax,' Tee 'tea'
fvvier 'four,' viel 'much'
vw/wwwas 'what,' gewwe 'to give'
ksxHex 'witch,' nix 'nothing'
jyyung 'young,' Yaahr 'year'
tsz/tzzwee 'two,' Zeit 'time'

Appendix 2

Coding spreadsheet

CommunityZIP Codebarig/ baerigfarschde/ faerschdekere/ kaerebodem/ bode/ florweschbailer/ weschboiler/ weschkessel
Lower Heidelberg(NW)13000aeama
Lower Heidelberg(S)13000aeama
Lower Heidelberg(W)1300000a00
Mauch Chunk(at)13000aaama
Mauch Chunk(S)1300000000
Mount Aetna(E)19544aaamo
Mount Aetna(N)19544aaaea
Mt Pleasant Mills(SE)17853aaafa
Mt Pleasant Mills(SW)17853aaaea
New Berlin(S)17855eaamo
New Tripoli(E)18066a0afo
New Tripoli(S)18066a0afo
New Tripoli(SE)18066a0afk
North Heidelberg(at)13000a0aea
North Heidelberg(SE)13000aaafo
Port Ann1300000amo
Schuylkill Haven17972aeafa
Stines Corner13000aeamo

Appendix 3

Maps of the PG-speaking region

  1. Base Map
    base map--click for full size image

  2. Map 1: barig/baerig
    barig/baerig map--click for full size image

  3. Map 2: farschde/faerschde
    farschde/faerschde map--click for full size image

  4. Map 3: kere/kaere
    kere/kaere map--click for full size image

  5. Map 4: bodem/bode/flor
    bodem/bode/flor map--click for full size image

  6. Map 5: weschbailer/weschboiler/weschkessel
    weschbailer/weschboiler/weschkessel map--click for full size image

  7. Isoglosses from Maps 1-5
    isoglosses from maps 1-5--click for full size image

Appendix 4

List and location of files

This section will be completed as soon as I have reorganized the files used for this project into a more sensible arrangement; they will also be placed on a diskette to be added to the paper copy of this report.


1 The plain groups are the "old order," or highly conservative, Amish and Mennonite groups, as well as many of the Hutterites and Moravians.

2 Stine's (1990) mention of this is basically simply a reaffirmation of Beam's (1982) claim.

3 This claim regarding the pronunciation of the digraph ae could possibly be looked at more closely in future inquiries into variation in PG vowels.

4 Please note that here and elsewhere in this paper I follow the Buffington and Barba (1965) orthographic system (see Appendix 1) by writing as ae what LAPG gives as ä. Either way the sound is the same (æ); the orthography of LAPG appears to simply be a reflection of the fact that it was published in Germany, where the umlauted vowel symbols are generally preferred to the digraph vowel+e.

5 Of course, further information may be hiding in particular university archives somewhere. For such searches, the first places to look would likely be the Universität Marburg (where LAPG was published), the University of Washington (where Reed was when it was published), the University of Wisconsin (where Seifert was at the time), and Johann-Wolfgang-Goethe-Universität (where Reed wrote the introduction to LAPG).

6 Some points were given a ZIP code of 13000 in this table; this is because they have no listing in the Postal Service's listings. Most of these towns do exist, though, but finding the ZIP code for them in order to add them to the maps remains a project for the future.

7 Appendix 3 of the paper version of this report contains fourteen maps, while in the HTML version there are only seven maps. This is because both color and black-and-white maps are attached to the paper copy of the report, while the HTML copy includes only color maps. The black-and-white maps were included in the paper edition because they can be more easily photocopied.

8 It is such a rough sketch of the roads, in fact, that I have avoided including a copy of it for this report, feeling that the whole question of when what roads were where would best be served by finding other sources with a clearer picture of what the situation actually was (as well as finding out what the roads were like in the mid-nineteenth century).

9 It would be most interesting to find out whether the supposition that all speakers interviewed for LAPG were non-plain PG speakers is, in fact, correct. If it is not, and some of the (usually Lancaster County) speakers who preferred German-descent forms were plain PG speakers, that would shed some historical light on when plain PG speakers began converging toward English as a means of maintaining their PG (Huffines 1988).

10 In my opinion, at least.

1-1 The "sound" is gleaned from Buffington and Barba's (1965) description; the orthography and examples are those given by them. Their orthography is given because LAPG uses a closely related one.