Selecting text across forty pages by hand is miserable, and PDF selection rarely behaves. Taking the whole text out at once is faster.
Extract Text reads embedded characters from a PDF; it does not recognise letters in page pictures. If the output is empty, first check whether text can be selected in a reader.
For scans, use Image to text instead. Its current OCR language is English, and it can read image or PDF input locally. Review names, numbers and punctuation because recognition can make mistakes.
Extracted text does not preserve table structure or visual layout. Multi-column documents can produce an unexpected reading order. Compare important quotations and figures with the original page.
The text arrives in the order the PDF stores it, with a line break wherever the PDF has one. That means sentences may be broken across lines and some words may be split with a hyphen. Paste it into an editor and use find and replace to tidy it before you use it.
Tables and columns are not laid out again, so a table becomes a run of values and a two column page can interleave. For a table, paste into a spreadsheet and separate the columns there. If the result is empty, the PDF is a scan and the words need Image to text instead.
Yes, in Image to text. Extract Text is for existing PDF text layers.
These text tools export or copy plain text; they do not add a searchable layer to the original PDF.
A PDF stores each line of text where it was placed on the page, not as paragraphs. The line breaks come with the text and need tidying in an editor.