Jump to content


Photo

Search Word (Flex) KJV seems bugged


  • Please log in to reply
10 replies to this topic

#1 Enoch

Enoch

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 886 posts

Posted 20 January 2020 - 12:11 AM

With my  12.2.8 Accordance there is what seems a bug to me, namely that if I search a verb by right click Search Word (Flex), the Elizabethan forms are omitted in the results. For example, with ASV or KJV if I put the cursor on trust and chose from the right click pop up menu Search Word (Flex), conjugated forms for trust appear in the results (e.g. trusted), but trusteth is omitted, as in Psalm 115:8.  Do these omissions cause faulty search statistics to be generated? Have these Elizabethan omissions been corrected in the latest Accordance version?

 

#2 Mark Allison

Mark Allison

    Platinum

  • Accordance
  • 1,786 posts
  • Gender:Male
  • Location:Flowery Branch, GA
  • Accordance Version:5.x
  • Platforms:Mac OS X, Windows, iOS, Android

Posted 20 January 2020 - 07:37 AM

A Flex search will often miss Elizabethan forms. That's why (for most cases) I prefer to use a word search with the wildcard character (trust*). 



#3 EricC

EricC

    Gold

  • Active Members
  • PipPipPipPip
  • 330 posts
  • Gender:Male
  • Accordance Version:12.x
  • Platforms:Mac OS X, Windows, iOS

Posted 20 January 2020 - 08:32 AM

Seems to me it'd be possible to program in those early modern English ("Elizabethan") forms, too. It's a finite number of endings. But maybe it's harder than it looks.



#4 Joel Brown

Joel Brown

    Administrator

  • Admin
  • 6,220 posts
  • Gender:Male
  • Location:Houston, TX
  • Accordance Version:13.x
  • Platforms:Mac OS X, Windows, iOS

Posted 20 January 2020 - 11:06 AM

Flex search is designed to easily find related words.  If the goal is accurate, specialized, statistics, then more exact searches should be done, so you know exactly what results you are getting.


Joel Brown

By day: Lead Software Engineer at Accordance
By night: Freelance Trombonist


#5 Enoch

Enoch

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 886 posts

Posted 24 January 2020 - 11:37 PM

Flex search is designed to easily find related words.  If the goal is accurate, specialized, statistics, then more exact searches should be done, so you know exactly what results you are getting.

 

Well the flex search fails to find easily certain related words in an Elizabethan translation. Is it a bug in the system?

 

How would I find all the occurrences of a lexeme (like trust) in an Elizabethan text? 



#6 Solly

Solly

    Gold

  • Active Members
  • PipPipPipPip
  • 459 posts
  • Gender:Male
  • Location:Central Pennsylvania, USA
  • Interests:Far too many for my own good! ;-)
  • Accordance Version:13.x
  • Platforms:Mac OS X, Windows, iOS

Posted 25 January 2020 - 05:46 AM

For ASV or KJV try searching for "trust*". The asterisk means include trust with any ending of any length. Using the KJV, a flex search produced a result of 167 verses found and 170 total words. The search for trust* produced a result of 188 verses found and 191 total words.

 

Words found in the trust* search not found in the flex search were trustedst, trusteth, and trusty.

 

Shalom,

Joseph


Joseph F Sollenberger, Jr


#7 Mark Allison

Mark Allison

    Platinum

  • Accordance
  • 1,786 posts
  • Gender:Male
  • Location:Flowery Branch, GA
  • Accordance Version:5.x
  • Platforms:Mac OS X, Windows, iOS, Android

Posted 25 January 2020 - 06:25 AM

The FLEX search is useful, but it has limitations. It wasn't necessarily designed for precision, but as a "quick and dirty" way to find a broad spectrum of hits.
The WORDS search allows for greater precision, but it's a little more difficult to use, since an understanding of wildcard characters is required.

 

If you learn two simple wildcard characters, you'll probably never use the FLEX search again:

The asterisk ( * ) stands for zero-unlimited number of characters. So a search for trust* (as Joseph has described above) will find both "trust" and "trusteth" and any word that begins with the letters "trust".

 

The question mark ( ? ) stands for exactly one character. So z?? would find all three-letter words in a Text that begin with the letter "z" (Zer, Zia, Zin, Ziv, Ziz in the CSB).

 

So what do you do if you want to find a question mark (or any punctuation)? Put a period ( . ) in front of it. So .? will find all the questions in the text.


Edited by Mark Allison, 25 January 2020 - 06:34 AM.

  • Fabian and Solly like this

#8 Solly

Solly

    Gold

  • Active Members
  • PipPipPipPip
  • 459 posts
  • Gender:Male
  • Location:Central Pennsylvania, USA
  • Interests:Far too many for my own good! ;-)
  • Accordance Version:13.x
  • Platforms:Mac OS X, Windows, iOS

Posted 25 January 2020 - 10:18 AM

Thanks for the additional information, Mark. I had forgotten the role the period played in this (I would have guessed backslash.) The Accordance approach is so much simpler that scripting a regular expression. In my senior years I forget so much of the command line protocol. ;)

 

Happy for GUI,

Joseph


  • Mark Allison likes this

Joseph F Sollenberger, Jr


#9 Enoch

Enoch

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 886 posts

Posted 26 January 2020 - 07:04 PM

Thanks a lot.  Of course "trusty" would mess up your stats if you wanted just verb forms.  Am I wrong in thinking that Accordance discriminates vs Elizabethan English?  Was it intentional to leave out the Elizabethan conjugated forms, or an original overlook?  Am I correct in thinking that Accordance does not want to go to the effort to make searches of Elizabethan texts equivalent to modern text searches?  I mean if there is a method to search lexeme (I don't know if that includes adjectives derived from verbs) for modern English texts, would you not wish to have the same system for Elizabethan texts, so that your stat system would be equivalent?



#10 Enoch

Enoch

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 886 posts

Posted 26 January 2020 - 07:10 PM

The FLEX search is useful, but it has limitations. It wasn't necessarily designed for precision, but as a "quick and dirty" way to find a broad spectrum of hits.
The WORDS search allows for greater precision, but it's a little more difficult to use, since an understanding of wildcard characters is required.

 

If you learn two simple wildcard characters, you'll probably never use the FLEX search again:

The asterisk ( * ) stands for zero-unlimited number of characters. So a search for trust* (as Joseph has described above) will find both "trust" and "trusteth" and any word that begins with the letters "trust".

 

The question mark ( ? ) stands for exactly one character. So z?? would find all three-letter words in a Text that begin with the letter "z" (Zer, Zia, Zin, Ziv, Ziz in the CSB).

 

So what do you do if you want to find a question mark (or any punctuation)? Put a period ( . ) in front of it. So .? will find all the questions in the text.

Thanks for the info on questions.  I am thinking that often the interpretation of a sentence as a question or not, depends mostly on the context; so it may be largely an editorial decision.  But it would be an interesting project to save all the questions of the Bible & classify them (who asked it, to whom asked, rhetorical or not, etc.)  "Yea, hath God said . . . ?" There could be heated questions, as Moses to YHWH.


  • Mark Allison and R. Mansfield like this

#11 Enoch

Enoch

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 886 posts

Posted 11 March 2020 - 01:44 PM

A Flex search will often miss Elizabethan forms. That's why (for most cases) I prefer to use a word search with the wildcard character (trust*).

Thanks so much for the great suggestion.  Of course that would not succeed with so-called strong verbs. Sing* would give you singeth, but not sang or sung. Trust, troost, ist getrusten.






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users