[COUNT 5 =k] may be broken for original language texts

[dandennison]

I can [COUNT 5 =k] for any translated text, but any original language text refuses.

 

What exactly does the default COUNT mode count in an original language text? Does it differ in Letters mode?

[Mark Allison]

You don't need to add the =k when you're using the COUNT command in an original language text. That =k is specifying a Key Number search. So it's simply [COUNT 5].

Edited January 17, 2022 by Mark Allison

[dandennison]

What exactly does the default COUNT mode count in an original language text? Does it differ in Letters mode?

 

[Mark Allison]

From the help documentation:

 

Using [COUNT ?] in Tagged Texts

The [COUNT ?] command searches only for words that occur the specified number of times in the search range. This command always uses a list of exact words (sensitive to case, punctuation, and accent).

The default COUNT search in a tagged text uses the list of lemmas that occur the specified number of times. Additional modifiers can be used inside the brackets to choose a different list for the search. These modifiers (Keys, Inflected, and Tags) are entered with an equal sign directly after the numeral. Only the first character of each modifier needs to be entered.

[COUNT 1] uses the list of lemmas that occur only once.

[COUNT 1=i] uses the list of inflected forms that occur only once.

[COUNT 1=t] uses the list of entire tags that occur only once.

[COUNT 1=k] uses the list of Key numbers that occur only once.

Example 1: To find all hapax legomena (words that occur only once) search for [COUNT 1].

Example 2: To find any complete tags that appear less than 10 times in the range, search for [COUNT 1-9=t].

Note The COUNT command normally searches for the words that occur the specified number of times in the search range. However, when COUNT is combined with the RANGE command as in [RANGE Gen 1-3] <AND> [COUNT 1], and the Search range pop-up menu is set to All Text, the search result is words in Gen 1-3 that occur only once in the entire text

[Mark Allison]

Also:

 

[COUNT ?]

The COUNT command searches only for the words that occur the specified number of times in the search range, as specified in the Range pop-up menu. Replace the selected question mark with the number or range of numbers of your choice. Use a plus (+) or minus (-) sign to search for all words occurring more or less than the number entered.

Examples:

• [COUNT 3] finds all the words that occur exactly three times in the current range.
• [COUNT 1-3] finds all the words that occur up to three times in the current range.
• [COUNT +3] or [COUNT 3+] finds all words that occur three or more times in the current range.
• [COUNT -3] or [COUNT 3-] finds all words that occur three or less times in the current range.

COUNT is similar to the HITS command in that it limits the search to a list of words. It can be combined with the @ symbol, connecting commands, and the internal modifiers =k, =i, and =t after the numerals.

[COUNT 1 =k]	uses the list of Key numbers that occur only once
[COUNT 1 =i]	uses the list of inflected forms that occur only once
[COUNT 1 =t]	uses the list of entire tags that occur only once

Combined Examples: Entering lov*@[COUNT 1 =k] searches for words beginning with lov that are tagged with a Key number which occurs only once throughout the range.

However, when COUNT is combined with the RANGE command as in [RANGE Gen 1-3] <AND> [COUNT 1 =k] (when the Range pop-up is set to All Text) the search result is Key numbers in Gen 1-3 that occur only once in the entire text.

This command is also an item in the Construct tab. When used in a Construct tab, the Set word count dialog box opens for entry of the number and any modifier that applies.

[dandennison]

ok, so does  [COUNT 1] *+ count only the roots that appear once?

[Mark Allison]

1 minute ago, dandennison said:

ok, so does  [COUNT 1] *+ count only the roots that appear once?

 

Try [COUNT 1]@*+ 

Without the @ sign, Accordance is treating this as two different words: 1) a hapax, followed by 2) any root. 
With the @ sign, Accordance is looking for a single word with two different criteria: 1) it has to be a root form, and 2) the this root form can only appear once in the text.

 

[dandennison]

Both [COUNT 1]@* and [COUNT 1]@*+ yield 3035 hits in HMT-W4.

 

I imagine if the + operator is doing anything at all, that one of these searches will be smaller than the other.

[dandennison]

The help clarifies this somewhat:

Quote

The plus sign (+) before a word specifies that it is a root so Accordance finds all occurrences of words that come from that root.

 

Therefore, it seems a search of +* would treat every word as a root.

 

I think I'm back to my original idea of searching on the key numbers as a not-great approximation to the roots.

 

Is there a way to get trilateral roots out of a wildcard search?

[Mark Allison]

So it turns out that I was wrong, and simply using the [COUNT 1]@*+ command didn't work (as you discovered). However, after talking this over with @David Lang, it looks like  [COUNT 1] @ [verb]@??? should give you what you're looking for.

If there's something specific you're trying to find, perhaps you could give us a little more detail?