I have in mind examples such as the Scheingallizismus (lit. appearance of Gallicism) in German which are words/phrases constructed from French origins but are themselves unknown in French speaking countries. For instance, the German word blamage meaning "disgrace" was coined from French blamieren (to embarrass) and the nominal suffix -age, and is even pronounced like French. For non-alphabetic writing systems, I have in mind the Japanese kokuji/wasei kanji (国字/和製漢字) which are characters coined mainly based on logographical rules of Chinese characters (interestingly, while many of these kokuji are only seen in Japanese, a few of them did wind up adopted back into simplified Chinese); for instance, the kanji 凪 meaning "calm" was coined from the radical (i.e. component form) ⺇ and the character 止 meaning "to stop".
In such cases, although the constructed words/phrases/characters are mostly unknown in their adopted source languages, native speakers of the adopted languages are often able to recognize some parts of these coined words/characters, but only to a limited extent.
My question: is there a linguistic term specifically accounting for the phenomenon described above?
I've thought about calques, loanwords, semantic loans, false cognates, neologisms, phono-semantic matchings, and so on; but none of them seems to accurately describe it. I do admit that this is a field that I am relatively ignorant of, and please correct me if I have made any mistakes.
Update: After some further research, it seems to me that the concept of pseudo-loan would be an appropriate umbrella term for all above-mentioned instances. There is another rather rare term, allogenism, which essentially describes the same phenomenon. However, I am not certain as to whether these two are accepted as linguistic terms, or rather just convenient designations.