Misc. corrections

[funderburkjim]

Corrections and/or issues noticed recently.

mw72:aShezas,7238:aSezas:t:
ap:uddac,8980:udfc:t: dda/df
ap90:vizvadyac,26668:vizvadryac:p: The ending ligature appears to be drya
pw:lakac,95266:lakaca:t: no virAma on 'c'
mw:upaDOkita,34614:upaQOkita:t: D/Q
mw:upaBrita,35263:upaBfta:t: ri/f
mw:upaBrita,35264:upaBfta:t: ri/f
mw:upayazwri,:upayazwf:p: ri/f. Text missing dot under 'r'. cf mw72, ap, etc.
mw:arcisArcizmat,15734:arcizmat:t: 
mw:arcisArcizmat,15735:arcizmat:t: 
mw:arcisArcizmatI,15736:arcizmatI:t: 
mw:ambuDakAminI,14508.1:ambu--DakAminI,ambu-Da-kAminI:t: refactor key2
mw:amBaHsyAmAka,14574:amBaHSyAmAka:t: s/S
mw:bEjavApayana,146456:bEjavApAyana:t: pay/pAy

[funderburkjim]

; Notes on some dualplural forms in MW:
; aDyayanatapasI:  n.du. of aDyayanatapas

[funderburkjim]

See below for resolution

; notes on likely HxA that need to be HxB in MW
;1A,alpena,16828 has different key1 than previous:alpa
;2A,utkzepO,30980 has different key1 than previous:utkzepa
;1A,upapakzO,35046 has different key1 than previous:upapakza
;1A,kAkutsTO,47217 has different key1 than previous:kAkutsTa
;1A,kuSIlavO,53471 has different key1 than previous:kuSIlava
;3A,kfcCrAtikfcCrO,54760 has different key1 than previous:kfcCrAtikfcCra
;2A,kruYcO,58224 has different key1 than previous:kruYca
;2A,bEjavApayana,146456 has different key1 than previous:bEjavApAyana
;1A,samyak,237309 has different key1 than previous:samyaYc
;1A,samyak,237310 has different key1 than previous:samyaYc
;1A,samyak,237311 has different key1 than previous:samyaYc
;1A,samyak,237312 has different key1 than previous:samyaYc

[funderburkjim]

mw:parAgagin,116779:parAgin:t:  key2 should be parA<sr/>gin
mw:parAgagata,116781:parAgata:t: key2  parA<sr/>gata
mw:parAgagantf,116782:parAgantf:t: key2 parA<sr/>gantf
mw:parAgagama,116783:parAgama:t: key2 parA<sr/>gama
; these are probably the tip of an iceberg.  The last three in particular are cases where
; (a) there is a prefixed verb, classed as an H1 or an H2
; (b) there is a block of H3 verbal derivatives WHICH MW CODES IN A VERY SIMILAR WAY TO
;       a block of compounds
; (c) the digitization incorrectly infers the full form of the compound
;      e.g. from parAgam and gantf the wrong inference is parAga-gantf

[gasyoun]

the tip of an iceberg

Any chance to batch weed them out?

[funderburkjim]

I'm attempting that now.

[drdhaval2785]

Very interesting and long awaited corrections of compounds. I am sure, refinent in this topic would need some grammatical algorithm. Will chip in at that juncture for auto suggestion algorithms.

Best wishes

[funderburkjim]

An identification of errors like those under prefixed root parigam has now been done. Fortunately, the number of similar errors is much smaller than anticipated. Here they are:

;1  117828  parinirvA   parinirvA   VERB:K ***
mw:parivARa,117829:parinirvARa:t: key1 error
mw:parivApayitavya,117833:parinirvApayitavya:t: key1 error
mw:parivAyin,117834:parinirvAyin:t: key1 error
;1  118977  parisaMcakz parisaMcakz VERB:K 
mw:paricakzya,118978:parisaMcakzya:t: key1 error
;1  118983  parisaMtap  parisaMtap  VERB:K
mw:paritapta,118984:parisaMtapta:t: key1 error.   ALSO remove HOm; also remove hom on 117633
mw:pratyupaBupaBoga,134403:pratyupaBoga:t: key1 error

[funderburkjim]

The working documents for this preverb analysis is here.

• the preverb1.py program reads the file all.txt from the MWderivations repository. This is a listing of almost all the headwords of MW. There is classification information of the headwords which in this case allows the identification of prefixed verbs (VERB:K). There is also H-code, so that the parent-child relations implied by the H-code can be examined. From this information, preverb1 program identifies preverbs that have children, and prints these to preverb1.txt.
• Then, preverb1.txt was examined manually to look for cases like under parigam. The results were collected in preverb1_prob.txt file.
• When this was done, the preverb1_change.py program helped in generating standard form corrections, as shown above.

[funderburkjim]

While looking for cases like parigam, I also noticed a couple of other kinds of cases that seemed like markup errors . These were of only two kinds:

• H errors. The general idea of the H-codes is that it provides a hierarchy. There are H1,2,3,4 codes. So the principle is that entries marked as H2 are somehow related to a parent H1 entry, and similarly H3 under H2 and H4 under H3. In the cases changed according to this, it seemed that an H3 for instance was not a child, but rather a sibling of the alleged parent H2, and so the change was to make that H3 an H2. For example:

  2    25549   AmuKIkf AmuKIkf VERB:K
  3    25550   AmuKIBU AmuKIBU VERB:K  ** H2

  Once you think of this notion, many additional potential correction candidates will be noticed.
  So, the current batch is by no means exhaustive. But I think they are 'safe' changes.

• The other kind of error has to do with key2 markup. Look at the scan image above to note how MW shows 'gantf' under 'parigam'; he does it with a little circle preceding 'gantf'. In the notation of MW.xml, this little circle is coded as <sr/> or <sr1/> (I'm not sure why there are both tags; I don't think there is a semantic difference between the two.) I the all.txt file, these have been rendered with a tilde '~'.

  What was noticed is that often the current markup differs from that of the print. For instance

  1    138438  prasraMs    prasraMs    VERB:K
  3    138439  prasraMsa   pra-sraM~sa m   ** pra-sraMsa

The corrections of this type were noticed in a random manner, and are definitely incomplete -- there are numerous other similar corrections that should be made. It just requires an effort to examine the cases.

However, this distinction (between a '-' and a '~' ) DOES have signifcance in MW, as it indicates a different relation between the parent and child. My current working principle is that:

• the '-' indicates a compound (samAsa) relation. For instance,

  guru/-karman [p= 359] : n. any affair of a spiritual teacher

• the '~' indicates
  • nouns which are primary derivatives of verbs. The relation between parigantf and its parent parigam is of this type.
  • a tadDita affix applied to the parent. MW gives the example mOda°kika under mOdaka:

[funderburkjim]

The main purpose of the MWderivation repository mentioned above is to use the information of the H-hierarchy and the key2 markup to 'explain' all the headwords of MW. In its current incarnation, all but about 10,000 of the 200,000+ headwords are explained. The application of more Sanskrit knowledge would likely reduce this number, as well as improve the derivations of some words whose current explanations might be suspect.

The analysis2.txt file shows the current state of this MWderivation work.

[funderburkjim]

vcp:anukrISa,1804:anukroSa:t: I/o

[gasyoun]

10,000 of the 200,000+ headwords are explained. The application of more Sanskrit knowledge

Hint, hint - Dhaval is missing here.

[funderburkjim]

In the course of #296 investigation, noticed a possible mis-spelling in other dictionaries:

; ziNga:ziNga:BUR,PW,SCH,SHS;ziNgaH:SKD
 PW, SCH, SHS, SKD should prob. all be ziqga (libertine)

Since N and q in Devanagari are quite similar (differing only by an extra 'dot' in N), it MAY be that this word is mistyped in PW, SCH,SHS,SKD, as it was in BUR. Will examine later.

[funderburkjim]

June 6, 2016. These corrected.

in MW under piNga, there are errors in gender for the feminine piNgA, for L=123567 - 123573: All of these should be piNgA: f..

(H2B) pi/NgA : f. a bow-string RV.  viii, 58, 9 (Sāy.  ; cf. piNgala-jya) [L=123566]
(H2B) piNga : m. a kind of yellow pigment (cf. go-rocanA) [L=123567]
m. the stalk of Ferula Asa Foetida L.  [L=123568]
m. turmeric, Indian saffron L.  [L=123569]
m. bamboo manna W.  [L=123570]
m. N. of a woman MBh.  [L=123571]
m. of durgA W.  [L=123572]
m. a tubular vessel of the human body which according to the yoga system is the channel of respiration and circulation for one side ib. [L=123573]

[funderburkjim]

Here is a question about German (old and new), coming from a user-submitted correction for Grassman. Under headword paYcadaSa, the user said that funfzehn needs to be changed to fünfzehn. And Google translate also says the spelling is u-umlaut. However, Grassman print shows several instances without the u-umlaut under these words paYcadaSa, paYcadaSan, and paYcASat.

Incidentally, in the totality of our digitization, these are the only places were 'funf' appears. By contrast, fünf appears 32 times, though never in form fünfzehn.

Based on this, I suspect the 'funf' form should be considered a print error, and be changed to fünf .

The question is - is this to be treated as a printing error , or is it some variant old German spelling that should remain unchanged in our digitization?

[gasyoun]

Based on this, I suspect the 'funf' form should be considered a print error, and be changed to fünf .

Agree.

The question is - is this to be treated as a printing error , or is it some variant old German spelling that should remain unchanged in our digitization?

Not aware of such (there are too many, actually, to note them), but this is not one of them for sure. Add u-umlaut.

[drdhaval2785]

Hint, hint - Dhaval is missing here.

I need to understand what is the left over and what is the methodology to help @funderburkjim . Readme needs to be explained in short, brief points.

[funderburkjim]

@drdhaval2785 I'll get to that brief readme in the not too distant future.

[funderburkjim]

@gasyoun Thanks for input re German - will add the umlauts.

[funderburkjim]

@drdhaval2785 A user submitted a correction to PWG, under headword anyatra, section 7) c); col 2, top of this page

His point was that c) should say 'abl.' rather than 'instr.'

This seems right to me, as the example shows 'manuzAt', which is abl.

However, it is confusing that under (d) another ablative example appears.

Do you agree that c) should be 'abl', and is thus a print error ?

[funderburkjim]

Found a situation under 'jaya' similar to that under piNga above - the entries after the first feminine form jayA were incorrectly noted as jaya, m. Corrected.

[drdhaval2785]

It doesnt seem to be an error to me. The word to look for is 'anyatra' and not 'mAnuza'. anyatra doesnt take case suffices. It has to be conjured from context. Context seems suitable to me.

[gasyoun]

anyatra doesnt take case suffices. It has to be conjured from context.

Why without?

[funderburkjim]

@drdhaval2785 Re anyatra - Agree that anyatra is indeclinable.

I was thinking that the 'als instr.' meant 'with (another word) in instrumental', and that the 'als abl.' meant 'with (another word) in ablative', and so forth.

However, I think you are saying that 'als. instr.' means 'when this word (anyatra) is used in an instrumental sense' , etc.

The 'als. instr.' example has:

• Sanskrit: tasya nAnyatra mAnuzAt
• German: durch Niemand ausser durch einen Menschen
• English (via Google translate): by Nobody except through a man

At least based on this indirect translation, the instrumental case phrase is 'by Nobody' which I take to refer to 'tasya na anyatra' in the Sanskrit.

So, this confirms @drdhaval2785 's interpretation.

Another confirmation is in the description of AP, where, among other uses, the authors say that '[anyatra is often used] with the force of the nom. case'.

This is the only indeclineable word I recall which has this property of being used as if it is one of various cases. I wonder if there are other such words (is there an anyatrAdi list?) or if it is somewhat unique.

[funderburkjim]

Examination of ziNga as mentioned above. Only Sch shows ziNga. Others are typos

pw:ziNga,116428:ziqga:t:  Author states that KiNga. Kiqga are 'vgl.' (variant reading?)
sch:ziNga,25483:ziNga:n:  Author mentions that pw has ziqga
shs:ziNga,41465:ziqga:t:  confirm by alph. order
skd:ziNgaH,37026:ziqgaH:t:  confirm by alph. order

[funderburkjim]

Resolution of the cases above alpena thru saMyak:

mw:alpena,16828:H1A>:H1C>:t: markup change
mw:utkzepO,30980:H2A>:H2B>:t: markup change (dual)
mw:upapakzO,35046:H1A>:H1B>:t: markup change (dual)
mw:kAkutsTO,47217:H1A>:H1B>:t: markup change (dual)
mw:kuSIlavO,53471:H1A>:H1B>:t: markup change (dual)
mw:kfcCrAtikfcCrO,54760:H3A>:H3B>:t: markup change (dual)
mw:kruYcO,58224:H2A>:H2B>:t: markup change (dual)
mw:bEjavApayana,146456:bEjavApAyanat: pay/pAy
mw:samyak,237309:H1A>:H1C>t: markup change
mw:samyak,237310:H1A>:H1C>t: markup change
mw:samyak,237311:H1A>:H1C>t: markup change
mw:samyak,237312:H1A>:H1C>t: markup change

[funderburkjim]

These corrections have now been installed:

       #t  #p  #n
ap90    1   0   0
ap      1   0   0
mw     30   1   0
mw72    1   0   0
pw      2   0   0
sch     0   0   1
shs     1   0   0
skd     1   0   0

[gasyoun]

Ger. als = Eng. as (= in sense).