In which encoding 'dep_en_tb.txt' is saved ? If it is e.g. utf-8, I think you could try the following:
$textsectb = file_get_contents('dep_en_tb.txt');
header('Content-type: text/plain; charset=utf-8');
This code would not encode anything at all. Still if the document is transferred with the proper encoding, I think you should not get squares or question marks instead of the entities...
Of course I supposed that you have been working with responseText (not responseXML) property of XMLHttpRequest object.
You said you would not like to encode the HTML. Please notice, that in some tasks not encoding HTML could create certain security problems.
Edit: if you really would like to actually encode the entities, you could do it like (please remove spaces from strings like '& #233;' - the system did not allow me to post otherwise converting entities to characters):
$textsectb = str_replace(array('é','è','ì'),array('& #233;','& #232;','& #236;'),$textsectb); // please remove spaces from strings like '& #233;'!
Please see this table
for reference how entities could be encoded ...