Question about English Flex searches

[mgvh]

I was trying to find the passage somewhere in one of the Petrine letters about, as I kind of remembered, Jesus preaching to the spirits in prison.

I figured a quick English Flex search for "prison" would find it. I was using the NIV, but the passage didn't come up. I switched to NRSV and was able to find it in 1 Peter 3.19.

It turns out that the NIV has: "he went and made proclamation to the imprisoned spirits."

I.e., the flex search didn't know to look for "imprison" as a word connected with "prison."

Is there some listing of how the words were associated in the English flex search?

[mgvh]

I just noticed too that searching for "save" does not return any instances of "salvation."

I'm losing confidence in helping my students do research who have Accordance but don't know Greek. Any suggestions?

[Daniel L]

I too have been tripped up by assuming flex search is more flexible than it is.

 

From Accordance Help:

• A Flex search finds all verses containing English grammatical forms, plurals, and singulars of all words rather than exact phrases (unless excluded by quotes). For example, a search for "forgive" would find "forgive," "forgiven," and "forgiving."

  When performing a Flex search, anything in quotation marks is treated as an exact search. In this way you can combine a Flex and Word search.

So, it seems to include only number and tense variations. A search for "prudent" will not find "prudence", and vice versa. (I discovered this while using American Heritage.) It doesn't look for related words having prefixes or suffixes. A search for "circumcised" will not find "uncircumcised". A search for "final" will not find "finally".

 

Maybe there should be a "super flex" search mode. Or simply make flex search find everything that most people would expect it to find.

[jlm]

14 hours ago, Daniel L said:

Maybe there should be a "super flex" search mode. Or simply make flex search find everything that most people would expect it to find.

One well-defined alternative is a stemming algorithm, like those implemented in Snowball. This would find “final” in a search for “finally”, but “prudent” and “prudence” wouldn’t match. There are Snowball stemmers for a couple of dozen languages, and an extension for old-fashioned English (-est and -eth) is documented. So more flexible searches would be available for a wider range of texts.

 

 However, it wouldn’t do what mgvh was expecting: that would basically be a root search for English.

[Mark Allison]

The flex search is a good option if you're trying to perform a quick search, where missing a few variations doesn't really matter. It's not intended to be a precision search tool. 

I recommend using wildcards with a Word search (or placing word options in parentheses separated by commas) when trying to be precise.

In the initial example, searching for *prison* in a Word search would find:

 

imprison = 1

imprisoned himself = 10

imprisonment = 4

imprisonments = 1

imprisons = 1

prison = 101

prisoner = 14

prisoners = 24

prisons = 2

Edited May 13, 2021 by Mark Allison

[Solly]

Mark, thanks again for reminding us of the power of the wildcard options in search. I keep forgetting how very useful wildcard searching is. This is a feature that could use some more promotion, though it is always there in the search topic webinars and help info.

 

—Joseph

 

 

[mgvh]

That's a good reminder to use the wildcards.

I just went back to my old BibleWorks to see how they did what they called "fuzzy" searches. They actually used two systems: "Porter Stemming" or "Link Stemming."

Porter stemming does not find the 1 Pet text when searching for "prison" as I found with Accordance.

Link stemming is recommended, and it does indeed find the text.

I don't know what relationship Accordance has with the former BibleWorks folks, but it would be interesting to see if they could obtain their database.

(I checked Logos too, and it works like Accordance, but they have a specific "Fuzzy Bible Search" that is strange to work with but does return the result.)

Edited May 13, 2021 by mgvh

[jlm]

Porter stemming is one of the algorithms implemented in Snowball. It’s not a database, it’s a set of rules for removing characters from the end of words so that the resulting stem of two words will be the same only if they’re likely to be etymologically related.

 

Link stemming, on the other hand, is based on a manually created database of stems for English words found in Bible versions. It will do a better job of the word you’re searching for is in that set, but if you’re searching Bible dictionaries, commentaries, journals, etc., you might be looking for terms that aren’t in any Bible. And if you aren’t searching in English, it obviously won’t help. I find myself doing searches in English Bibles to find verses in other languages that I display in parallel.