You will need to say more about your computer system and the websites that present this problem. Are you using the Zotero import metadata button? Are you importing from a file?

I have had no problem importing those characters from publisher sites nor from a database such as PubMed. There have been a few problems with importing (or copy/paste) from WorldCat.

Could we get some example URLs? I'm not seeing this e.g. here: https://en.wikipedia.org/w/index.php?title=Ł

OK, so I _am_ seeing this on google scholar, but not on any of the other ones. I think I know what the google scholar problem is and can take a look, but I'm very puzzled about the other sites.

How about the Wikipedia article? Same issue?

I use Zotero in English and system language is Czech. The problem is only with the Google Scholar.

If your problem is with letter characters when you download an item directly from Google Scholar, try linking to the publisher from the GS listing.

Because of the way that GS compiles the item list (mostly by capturing online versions of the publications themselves but also by grabbing citations with OCR from article reverence lists) the characters can be rendered in a peculiar way.

GS and Crossref have different methods of populating their databases. CrossRef and OCLC Worldcat also occasionally have character encoding problems. This will extend to many other characters beyond the velarized L that you mentioned.

I may be wrong about this but Zotero seems to do a really good job of recognizing and correcting character encoding even when it is different from what is declared in the page header. "Funny" characters displayed on the screen and in the webpage headers can miraculously be correctly captured by Zotero translators.

The problem with Google Scholar is that the records do not necessarily have the same character encoding scheme even within the same record.

The way that GS populates its database, by uncurated capturing of its contents from across the web, will cause errors. I could go on a long-winded further explanation of the problem with character encoding and how, from experience with my own online bibliographic database, even some publishers' websites get it wrong online and in the metadata they provide to databases but this isn't the place for that.

A speed-priority workflow especially that based on uncritical collection of raw metadata from GS will have not only errors due to character encoding -- it will also have volume, issue, and pagination errors. It will have omitted author names and authors in the wrong order. Then, if you openly share your bibliography online, you will reinforce the error when GS scans your site -- further compounding the problem.

Looks like there's a bug either in the BibTeX returned by GS, or in the BibTeX translator that's being called underneath. Or, and BTW, in my case it turns it to "\l", not "/l" (this is on Linux).

Specifically, here's what the GS translator gets as the BibTeX:

@book{derbis1998poczucie,
title={Poczucie jako{\'s}ci {\.z}ycia a swoboda dzia{\l}ania i odpowiedzialno{\'s}{\'c}},
author={Derbis, Romuald and Ba{\'n}ka, Augustyn},
year={1998},
publisher={Stowarzyszenie Psychologia i Architektura}
}

And here's what the BibTeX translator conferts that to (which is essentially what is being used as the title at the end of the day):

@book{derbis_poczucie_1998,
title = {Poczucie jakości życia a swoboda dzia{\textbackslash}lania i odpowiedzialność},
publisher = {Stowarzyszenie Psychologia i Architektura},
author = {Derbis, Romuald and Bańka, Augustyn},
year = {1998},
}

OTMH I don't know whether that {\l} is correct or not, if yes, then it must be the BibTeX translator, which does a lot of low-level parsing inside.