That is an interesting question. I can only add for the moment, that from K. Pohl's book about Req. Eng., he considers that the NFR really doesn't exist and these are quality requirements poorly specified.

Well, but then they exist, don't they ? :D

If you ask me, yes they do exist. I like the Sommerville's NFR taxonomy very much. It appears in Chp. 6 or 8 of the Sommerville's Software Engineering book.

I know the term from John Mylopoulos' papers. It later influenced research on goal modeling. If course, NFR exist. Whether or not they come with quality metrics is a question at which level of detail a system is specified and/or which stage of the analysis you are considering.

As I mentioned above, Mylopoulos cited an AI journal for the term. Unfortunately I do not have access to that one yet. I also wrote an e-mail to Pamela Zave, but she can't remember where the term came from.

I had a look at the Paper by Mylopoulos, Chung and Nixon and noticed that I misinterpreted footnote 1 for reference 1. That renders the AI journal useless for my search, I'll remove the reference in my question above.

My current state of knowledge is the following: at least in terms of published research, Yeh and/or Zave have first used the term in a paper. I am currently waiting for some 'ancient' conference proceedings (COMPSAC 1982) that will maybe shed light on this.

Yeh uses the term in his 1982 COMPSAC paper, but he does not cite anybody for the term.

If found an occurrence of the term here, but I have no access to the report:

http://www.dtic.mil/docs/citations/ADA047473

Does somebody have this report and can look into it?

Check out this. http://books.google.de/books?id=9fYFiZ7QeEcC&pg=PA149&lpg=PA149&dq=%22Requirements+elicitation+methods+are+now+including+elicitation+of+nonfunctional+requirements%22&source=bl&ots=6csTczimm2&sig=1EHE6boIY_zBMkd2UQOJneOmlIQ&hl=zh-CN&sa=X&ei=WGHyUvGsBoLmswbsnYDoCg&ved=0CCkQ6AEwAA#v=onepage&q=1975&f=false A book published in 1975 mentioned the terms "nonfunctional requirements"

The book you link to is of 1994. See what I copied from Google Books:

Encyclopedia of Computer Science and Technology: Volume 30 - Supplement 15: Algebraic Methodology and Software Technology to System Level Modelling

Allen Kent, James G. Williams

CRC Press, 08.02.1994

Did you find something in it? I did a full text search and found nonfunctional, but not the book from 1975 you were mentioning.

sorry, i posted the wrong link. this book has several version. "non-functional requirements" was also mentioned in the 1975 version. http://books.google.de/books?ei=EGryUqHdBcPbswbLzYG4Ag&hl=zh-CN&id=ZhBLAQAAIAAJ&focus=searchwithinvolume&q=nonfunctional

But i also googled and found out that functional and non-functional requirements were also mentioned in mechanical engineering. Probably it's not a so specialized terms and has already been there (in the field of engineering, probably) for a long time before the emergence of software engineering.

I'm afraid the snippets that Google provides are from a newer edition. One snippet where "non-functional" occurs cites "Tardieu (55)" and that reference refers to a work of Tardieu from 1991.

Regarding your second remark, yes, I also think that is very likely. But I did not find a good source from e.g. mechanical engineering yet.

@Ina: My hope was that it would help to understand the original intended meaning of the term.

I found the term awkward in the first place, let me draw an analogy. It occurred to me like calling all vehicles apart from cars, "non-cars". So bicycles would be classified as "non-cars". I was wondering where the dichotomy functional/non-functional came from and why it seems nobody criticized it. Boehm in 1976 already mentions the term "quality requirement" which would be a more appropriate term, but then the argument would start NFR != Quality Requirement and so on.

Maybe the answer is, simply put, that it does not make so much of a difference if the commonly accepted taxonomy of requirements is well-designed.

I currently assumed that Yeh took the term from somewhere and used it for his publications. Looking at the below cited publication, my assumption is that he wanted to point out that there are many more classes of requirement than functional ones. For that purpose, I guess using "non-functional" is exactly right.

R. T. Yeh, “Requirements Analysis - A Management Perspective,” presented at the IEEE Computer Sixth International Computer Software and Applications Conference (COMPSAC), 1982, pp. 410–416.

And a final question to Ina, again: Can you elaborate why it would make a difference to discuss NFR in different scopes? Would the definition change then? My understanding was that the definition is domain-independent. E.g. from van Lamsweerdes textbook : "Non-functional requirements define constraints on the way the software-to-be should satisfy its functional requirements or on the way it should be developed." Exchange the term "software" with "product" and you obtain a general definition.

The publications mentioned so far are chiefly in software engineering sub field. It is important to note that the term 'non-functional requirements'  was used in earlier publications in other (older) engineering sub fields such as mechanical engineering design.

Eliab, can you point me to a publication from mechanical engineering? Thanks a lot!

Hello Florian, I have interest in non-functional requirements too.

Did you found something about the origins of the term?

@Victor: The earliest software engineering-related publications that I found are all mentioned in this thread.

Do you have the paper  R. T. Yeh, “Requirements Analysis - A Management Perspective”?

Could you send me?

http://agilemodeling.com/artifacts/technicalRequirement.htm

Article Characterizing Architecturally Significant Requirements

https://dl.acm.org/doi/book/10.5555/371246

https://www.academia.edu/8234038/Applied_Software_Project_Management_by_Jennifer_Greene_and_Andrew_Stellman_November_2005

These may help you.