Lexical search and wild cards

[sang]

Hello
I am trying to find certain symbols in a Maxqda document while excluding others that contain them from my search. For example, I have instances of / and of // and /~ and only want to find (and autocode) the first. Unfortunately, I haven't found a way to exclude the other two from my search. It doesn't work with the extended search (/ but not //) or with wild card characters in the lexical search <(/)> (as a matter of fact, the wild card search with < > doesn't seem to work at all). It also doesn't help to tick "whole words only" as it is not really a word. Is there a way of doing this?
Thanks a lot and best regards
Silja Ang

[Stefan Rädiker]

Hi Silja,

I think we found a solution for getting the results you are looking for. You need to create multiple search items in one search and integrate all characters that may follow "/".
Please search for

/a
/b
/c
/d
...
/z
/[space] => please type in a space instead of [space]

The list of search strings will be long but you can search for them at once. Please note that you can save such a search request and re-load it later if you need it again.

By the way:
<() => search for word begin and
()> => search for word end
cannot be combined. This may have led to the impression that the search with wildcards didn't work at all.

Best regards on behalf of the MAXQDA team,
Stefan

[sang]

Dear Stefan

Thank you very much for your answer. I tried it out, but unfortunately, it does not work for me, because if the / is at the end of someone's turn, it will also include the next turn in the autocoding, and I need to count the individual symbols in each speaker's turn.

I started autocoding with the AND NOT option (// AND NOT //~), which works for single-line turns at least. In longer turns it does not code // if there is a //~ nearby, of course. If there was the option to choose "within one line" rather than "within one sentence" in the extended search, it would work for me, as I have one symbol at the end of each line. I assume there is no way of doing this?

Another option I tried was to autocode //~ and autocode //, then do a coding query and look for intersections, so that I could delete all the //~ that overlap with // from the list. However, this did not work as I wasn't able to delete all the instances in the table at once (and doing it one at a time takes too long for several thousand instances).

I also assume there is no wildcard that represents the end of a line?

I am adding an extract from my transcription below, maybe this will explain my issue a bit better. Each of the four turns makes up one paragraph.

Best wishes
Silja

[i:19pkbg0x]EXAMPLE:

C: <H::>
<F<HI ^yeah but HI>F> \~
<W I I <HI still HI> think people are quite ^ignorant W> \\
<P<CRK<LO when it ^comes to LO>CRK>P> \
… ((37:24))
<P<HSK<LO your^selves LO>HSK>P> /~
… ((37:26))
<P<HSK the whole .. <LO ^p:roblem LO>HSK>P> /
<PP<LO<HSK about ^politics [and all] that HSK>LO>PP> \
L: [<P<HSK ^yeah HSK>P>] \~
C: <P<HSK<W and .. I <HI ^think HI> that’s W>HSK>P> \
<H:>
… ((37:31))
L: and it [a^nnoys you] _

[/i:19pkbg0x]

[sang]

I found a (slightly complicated) way of doing this, in case anyone has a similar problem:

1. I first looked for all signs of one kind with the extended search (AND NOT), e.g. // AND NOT //~ AND NOT //\\~. Now I only had signs left that co-occurred in the same sentence as the AND NOT signs. The more complex signs like //\\~ (that contain others but are not contained in other signs) I could auto-code completely like this.
2. I automatically coded all signs of one kind in a separate code (e.g. // , coded as //+), starting with the longest/most complex ones.
3. I then activated all codes that contain // (e.g. \\//~, //\ etc.)
4. I did a code query, picking ONLY THIS CODE and chose //+ as the single code.
5. Like this, I found all instances that are coded with // and nothing else; I coded these auto-matically as //.
6. Finally I deleted the new category //~ again and checked if there were any overlaps left (ac-tivating everything containing //  code query  intersection (set) 2 overlaps).
7. I then did this for \\, then for / and \ etc.

[Stefan Rädiker]

Hi sang,

we are glad you have found a way to get the information you need. We were still working on a solution, but your post was faster

Good look for finishing the analysis!
The MAXQDA team
Stefan