Broken Swedish words (split by accidental spaces) breaking translations

Hi

There is an issue in the text where single Swedish words have random spaces in the middle of them. Because the words are split, clicking them only selects half the word, which breaks the translation tool.

Here are examples found in the text:

  • “använ der” should be “använder”
  • “om fattar” should be “omfattar”
  • “svensk ans” should be “svenskans”
  • “me ningarna” should be “meningarna”
  • “del vis” should be “delvis”
  • “exem pel” should be “exempel”
  • “ur sprungslandet” should be “ursprungslandet”
  • “behärs kar” should be “behärskar”
  • “kod växling” should be “kodväxling”

Please fix the text formatting so words are joined correctly.

I generated ai generated reading text. There I saw such issues. I also see such issue in public text where text seems to be taken from pdf or other sources. Because of this issue, reading becomes very tough as even known words seems unknown suddenly as its broken down

Thanks for reporting this, and sorry it made the reading hard.

The spaces are in the text itself rather than something Readlang is adding when it displays it. All nine of your examples come from one text in the public library, “I språkens värld”, which was pasted in from a PDF. In that PDF the words were hyphenated across line ends, and when the text was copied the hyphens were lost and the line breaks became spaces, so “använ-der” ended up as “använ der”. Readlang stores and shows what it was given, so the split words stay split and each half is treated as a separate word.

For a text you uploaded yourself, you can fix this without leaving Readlang if you have Premium. Open the text, click the three vertical dots icon at the top, choose “Edit text…”, then click “AI Refine”. Untick “Fix spelling and grammar” so only “Clean up formatting” is selected, click “Polish Text”, then “Save Changes”. I tried this on a copy of the affected paragraphs and it rejoined all nine words correctly. It works on texts up to 7,000 words. More about it here: NEW AI Polish feature

For a public text that someone else uploaded, you can’t edit it directly. You can flag it with “Leave feedback…” in the same three dots menu. If you want to read it now, copying the text into ChatGPT with something like “remove the spurious spaces inside words without changing any wording” and uploading the result as your own document works too.

On the AI generated text: I looked through the Swedish texts generated by Readlang’s AI and couldn’t find this happening in any of them. If you have one where you saw it, could you post a link? That would help a lot.

1 Like

Hi. I don’t think this is a problem specific to Swedish. Just to add to what Steve said about Readlang’s own AI-generated texts, I have over 60 such texts in Swedish, mainly short stories, and don’t remember having had any problems of this nature. Nor have I had any problems with the some 200 Swedish texts I’ve uploaded.

As for public texts uploaded from PDF, this is unfortunately not so uncommon, and I’ve seen it, for example, in a French upload to the public library.