In PDFs, IPA and mathematical symbols (including superscript and subscript) are not parsed correctly. They seem to be OK if the source is html, but see for example this paper: https://www.cambridge.org/core/journals/journal-of-linguistics/article/abs/why-underlying-representations/9E09BCAB142EA5B0C08B49C72CA59435

...in which it would have been good to parse arrow representations as pictures instead of making them disappear. and this one: https://www.kuojennifer.com/files/publications/2024_wccfl_proceedings_preprint.pdf

where some IPA symbols are not read correctly.