was Marion Dexter Learned's assessment of Pennsylvania Dutch. His surname suits his 1889 book which is far more academic than Rauch's. It opens with an ethnographic description of PD speakers followed by sections on phonology, inflection, syntax, and English mixture.

From a standard German viewpoint, PD seems simplified.

I have already mentioned its loss of front rounded vowels.

Noun inflection, already minimal in German, approaches the simplicity of English:

no endings in the singular except for the genitive

-s for the rare genitive singular (cf. English '-s)

-ə(r) or zero in the plural except for the diminutive -li whose plural is -len

Some plurals have umlaut though they have lost the vowel rounding still in standard German:

fuus 'foot' : fiis 'feet' (cf. std Ger Fuß : Füße)

mɒus 'mouse' : meis 'mice' (cf. std Ger Maus : Mäuse)

Note how the English cognates also have umlaut.

Apart from a written den (masc. acc. sg.), both indefinite and definite articles have identical nominative and accusative forms.

The unaccented indefinite article can be ə regardless of gender or case.

PD surprisingly still has a distinction between 'weak' and 'strong' adjective declensions. 'Weak' forms follow definite articles whereas 'strong' forms have no preceding articles:

Weak declension










Strong declension (no genitive endings are listed)














There is also a mixed declension after indefinite articles. Learned's description implies the following endings:






nominative/accusative (same as strong)




(genitive?)/dative (same as weak)

Pronouns still have distinct case forms: e.g.,




















Verbs still have more person and number marking than English:

'to be' singular plural
1st person bin sin
2nd person bischt sin/seid
3rd person is/isch/ischt sin

'to have'



1st person



2nd person



3rd person



The syntax section focuses on the usage of inflected forms and says nothing about word order.

Next: English mixture in PD.


wrote Edward H. Rauch on p. vi of Rauch's Pennsylvania deitsh hond-booch (Rauch's Pennsylvania Dutch Hand-Book (sic), 1879). Orrick is Pennsylvania Dutch for 'many'. Rauch continues:

De Deitsh rule doots net so goot.

Doots corresponds to standard German tut 'does'.

Here's an excerpt from Rauch's English introduction:

The German rule [i.e., a German-based spelling for Pennsylvania Dutch] would not be practical, because from eighteen to twenty percent of all the words commonly used in Pennsylvania Dutch are either English or a compound of English and German, and also because all the youth of our State is taught to read English, and comparatively but few receive any sort of German education. The English rule is therefore decidedly the best for this purpose. Anyone who can read English can also read Pennsylvania Dutch as I have it recorded, and give it, in nineteen out of twenty cases, the correct pronunciation ...

To read it, no study of orthography is at all necessary, because it is simply English. The ch is the only German sound, pronounced as gh in Dougherty or Gallagher.

The only American English pronunciations of those names I've ever heard are

[dowhr̩ti] (gh = [h])

[dakr̩ti] (gh = [k])*

[gæləgr̩] (gh = [g]).

Were they both pronounced with an [x] (i.e., like German ch) or, more likely, with an [h] in 19th century Pennsylvania?

Rauch's English-based spelling for Pennsylvania Dutch reminds me of the English-based orthography for Manx which looks very strange to me since it's unlike the orthographies of its close relatives Irish and Scottish Gaelic.

*The substitution of [k] for Irish [x] in Dougherty < Irish Ó Dochartaigh reminds me of the substitution of Old Japanese k for Middle Chinese *x:

'Chinese': 漢 MC *xan (> Mandarin Han) but Old and Modern Japanese Kan

as in Modern Japanese 漢字 kanji 'Chinese characters'

Old Japanese had no h, so k was the closest available substitute at the time.

Modern Japanese h is from Old Japanese p. DES BUCH ISS MEI

When I visited Pennsylvania Dutch country in 2000*, I was disappointed because I couldn't find anything on the language for sale - not even a pamphlet for tourists.

In 2004, I met Dutchy Digest co-creator Bruce Rosenberger online. Today's his birthday, and I wanted to tell him 'happy birthday' in Pennsylvania Dutch, but a Google search turned up nothing. However, I did find this free book on the language from 1872. The title is taken from p. 36 which contrasts

des buch iss mei

'this book is mine'


's buch iss mei

'the book is mine'

('s is short for des 'the/this')

Being a Germanic language, PD is already relatively close to English and is becoming even more like American English: e.g., the PD vowels ü and ö have been replaced by their closest English equivalents. But not all changes in PD are due to American English influence: e.g, kl- > gl- has no American English parallel, so clue has not become glue.

*By coincidence, I moved to Holland a few weeks later. Needless to say, Dutch proper is not Pennsylvania Dutch which is actually a variety of German. MORE 'NEW' YET OLD KOREAN CLUSTERS IN UNICODE 5.2

Thanks to Andrew West for pointing out that there are even more new hangul characters in the Hangul Jamo (字母 자모 'letter') section on pp. 113-116 of this Unicode 5.2 PDF. Here's the table I posted last night with the new initial characters added in bold:

ns- nc- nh-





















p-h- (not ph-!)









The final characters I missed were for

-kn, -kp, -kc, -kkh, -k-h (not -kh!), -nn

I was wondering why there were no characters for new kC- and nC-clusters. They do exist - just not in Extensions A and B where I was expecting them.

I asked,

If none of these characters were normally used to write Korean, what were they intended to represent?

Can you guess why ㅳ -pt might have been created?

I gave three examples of ㅳ -pt which are examples of English in Korean transcriptions from 1935:

ak-sept-e-tɯ [akseptedɯ] 'accepted'

(t is [d] between vowels.)

ak-sept-ing [akseptiŋ] 'accepting'

ak-sept [akseptə] 'accepter'

In modern Korean, accept is written as 악셉트 ak-sep-thɯ without a final cluster ㅳ -pt. English -pt is broken up among two letter clusters:

English -p- > Korean ㅂ -p in 셉 sep

English -t > Korean 트 thɯ

I suspect that clusters like final ㅳ -pt were devised for phonetic notation of

- foreign clusters like English -pt

- native clusters when breaking up words in nonstandard ways: e.g.,

initial -ㄽ -ls- in

-ㄽㅡ록, -lsɯ-rok, normally split up as -ㄹ 수록 -l su-rok 'the more ...' in modern spelling

(I have transliterated ㄹ as l according to modern pronunciation since this example is from p. 48 of 이규영 [李奎榮] Yi Kyu-yəng's 현금 조선문전 Hyən'gɯm Chosən munjən [現今朝鮮文典 Modern Korean Grammar, 1920].)

(Note that initiaipt- did exist in 15th century Korean: e.g, ᄠᆡ ptʌy 'dirt', now 때 ttae.)

Since these clusters were presumably devised ad hoc, it's possible that more clusters will turn up when more old linguistic texts are found. For example, a pre-World War II Korean transcription of Russian здравствуйте zdravstvujte might be

ㅈㄷㄹㅏㅂ ㅅㄷㅂㅟ테


with ctr- and stp-, initial clusters that have no characters in Unicode 5.2.

(The Korean letter ㅿ z would be long extinct at this point, so c would be the closest substitute for foreign z and still is: e.g., 지로 ciro 'zero'. The Korean letter ㅹ v would also be extinct. I have only seen ㅹ v in early hangul spellings of Chinese.) 'NEW' YET OLD KOREAN CLUSTERS IN UNICODE 5.2

Thanks to Andrew West for bringing this document to my attention. I think I vaguely recall hearing about new hangul characters being added to the next version of Unicode, but I never got around to looking for them. You can see the new additions on pages 376 and 431-432 of this PDF draft of the Unicode 5.2 code charts. Among them are many obsolete hangul characters for clusters such as

tm- tp- ts- tc-
rk- rkk- rt- rm- rp- rpp- rβ- rs- rc- rkh-
mk- mt- ms-
psth- pkh- p-h- (not ph-!)
ngr- ngh-

which are absent from my list of 15th century Korean initial clusters. (Note that I use c instead of ts or ch to represent ㅈ.)

(The above table is incomplete. Click here for a complete version.)

There are similarly odd 'new' old characters for final consonant clusters: e.g.,

-nr, -nch

-ttp, -tth

-rkk, -rrkh, -rmh, -rng, -rpph, -rʔh

-mn, -mnn, -mm, -mps, -mc

-pt, -prph, -pm, -pst, -pc, -pch

-sm, -sβ, -ssk, -sst, -sz, -sc, -sch, -sth

-zp, -zβ

-ngm, -ngh

-cp, -cpp, -cc

-phs, -phth

If none of these characters were normally used to write Korean, what were they intended to represent? Here are examples of three words written with syllable-final ㅳ -pt* from p. 283 of 박승빈 (朴勝彬) Pak Sɯng-bin's 조선어학 Chosən əhak (朝鮮語學 Korean Linguistics, 1935):

Syllable 1 Syllable 2 Syllable 3 Syllable 4
Word 1
ak sept e
Word 2
ak sept ing
Word 3
ak sept ə

(Note that ㅇ represents a zero initial consonant in vowel-initial syllables: e.g., ak is spelled 악 Ø-a-k.)

Can you guess why ㅳ -pt might have been created?

*The PDF is nearly illegible, so this is my best guess. I don't have a Unicode 5.2 font yet and my font doesn't allow me to place ㅳ -pt beneath 세 se.

9.22.0:46: Why were the 'new' old Korean initial characters (Hangul Jamo Extended-A) placed on p. 376 between the Southeast Asian Rejang and Javanese characters in the Unicode 5.2 draft? Why not place them before the 'new' old Korean vowel and concharacters (Hangul Jamo Extended-B) on pp. 431-432 after the Hangul Syllables characters? K-OLLAPSE IN KOREAN

In my last post, I mentioned how Korean aspirates orignated from clusters which in turn originated from consonant-vowel-consonant sequences: e.g.,

hVk > hk > kh

The evidence for that shift is straightforward. But how do we know that k-clusters as well as h-clusters became later Korean aspirates?

15th century Korean (15K), the first stage of the language written in the hangul alphabet, had the following initial clusters and aspirates:

Aspirates and geminate hh ph- th- tsh- kh- hh-
s-clusters and geminate ss sp- st- ss- sk-
p-clusters pt-, pth- ps- pts-
ps-clusters pst- psk-

Notice that only two of the clusters (in bold) contain k.

One might initially conclude that those two clusters came from earlier Korean sequences



which later collapsed into the clusters sk- and psk- and that earlier Korean had no sequences like




But why would pre-15K forbid *k if it were not preceded by *s? Such a constraint has no parallel in known languages and would be arbitrary.

It is more likely that pre-15K did have instances of *k that were not preceded by *s. But why doesn't 15K have more k-clusters like pk- or kt- from earlier *pVk- or *kVt-? Perhaps *k became aspiration in 15K unless preceded by *s:

Stage 1 Stage 2 Stage 3 Stage 4 Stage 5: 15K
*pVk- *pk- *px- *px- ph-
*kVp- *kp- *xp-
*tVk- *tk- *tx- *tx- th-
*kVt- *kt- *xt-
*pVtVk- *ptk- *ptx- *ptx- pth-
*pVkVt- *pkt- *pxt-
*(pV)sVk- *(p)sk- (p)sk-
*(pV)kVs- *(p)ks-
*t(V)sVk- *tsk- *tsx- *tsx- tsh-
*kVt(V)s- *kts- *xts-
*kVk- *kk- *xk- or *kx- *kx- kh-

Using the above table, we might expect 15K phi 'blood' to be from either *pVki or *kVpi.

If Korean phi has any cognates among the 'Altaic' languages which are Korean's most likely relatives, those cognates should either resemble *pVki or *kVpi. A proposed cognate that looks like phi is 'too good to be true' since no other Altaic language has undergone Korean-like consonant shifts as far as I know. Here are the words for 'blood' in other Altaic languages:

Turkish kan

Classical Mongolian chisun

Manchu senggi

Japanese chi < Old Japanese ti < *tVi

Obviously, none of these words are related to 15K phi < *pVki or *kVpi, and none are related to each other.

(The resemblance of Japanese chi to Classical Mongolian chisun is coincidental, as J chi is ultimately from *tVi which doesn't look like Mongolian chisun at all. And even if one were to claim that M chisun were from *tVisun, one would have to explain why M has -sun and Japanese doesn't.)

This comparison is not definitive counterevidence against relating Korean to any of these languages, but it means that if their common ancestor shared a word for 'blood', most of them no longer preserve it with the same meaning.

Next: What other 15K clusters would you expect, and why are they missing?

9.21.0:49: Here's an example of how reconstructing the origin of Korean aspirates might point toward a relationship with a foreign word.

Korean for 'burn' is 타- tha-, which vaguely looks like Japanese 炊く tak-u 'burn'. Dropping the ending -u from the Japanese form results in a stem tak- which could be from an earlier *taka-. If I didn't know that Korean th- could be from tVk-, I might regard

K tha- : J *tak(a)-

as a coincidence since there's no reason for Korean to lack a -k. Lots of Korean words have -k: e.g., ttak 'exactly'. However, one of the two possible early Korean sources of tha- does resemble the early Japanese root:

K *tVka- : J *tak(a)-

One could even guess that the first vowel of the early Korean root was *a on the basis of Japanese:

K ?*taka- : J *tak(a)-

Does this one comparison mean that Korean and Japanese are related? Not necessarily.

First, it is possible that Korean tha- is really from early Korean *kVta- and has nothing to do with Japanese *tak(a)-. Having the same consonants in different orders and one possible shared vowel is not enough to link the two words.

Second, it is possible that the first vowel of early Korean *tVka- was not *a and doesn't match the first *a of Japanese *tak(a)-.

Third, it is possible that Japanese borrowed *tak(a)- from some early Korean peninsular language like 百濟 Paekche a.k.a. Baekje or Kudara (now extinct). I think this is the most likely scenario.

Whenever two words with similar or identical meanings in different languages look alike, consider the possibilities:

1. They may look alike now, but they may have been totally different in the past.

2. They may always have looked alike, but their resemblance is sheer coincidence.

3. One language may have borrowed the word from the other.

4. The two words really were inherited from a common ancestor.

Before jumping to conclusion 4, I prefer to consider the first three possibilities.

Tangut fonts by Mojikyo.org
Tangut radical font by Andrew West
All other content copyright © 2002-2009 Amritavision