All your meaning are belong to us

Sometimes it seems to me that software tester in general have a certain nag for linguistics and semantics in particular. Just have a look at all those testing Vs checking and DevOps articles (I’ll come back to the latter discussion later). Which is not overly surprising as semantics is about meaning, which on the other hand is the core aspect of every testing oracle. So let’s have a short look at meaning from a linguistic point of view. Dating back to de Saussure and early semiotics meaning consists of two parts, a signifier and something signified. Or a title and a concept if you want.
And the relationship between those two sides is totally arbitrary. So, how come we understand each other? Well, that’s about a group of people agreeing on a certain relationship and using it frequently (this is pretty much why I say “Stuhl”, while you say “chair” or some other word and that thing still looks more or less the same). A group in this case can range from all speakers of a language to a team of colleagues. The latter will probably share quite some vocabulary with the former, but will certainly establish a set of signifier signified relationships of their own. The problem now is that agreeing on a relationship is easier for concepts that have a real word reference like tree or apple. For abstract concepts like peace, love or software testing this gets harder and needs to be made explicit (i.e. talk to each other). This often correlates with the frequency a word is used within the bigger group. Those infamous irregular verbs of English stay because they are frequently used. If the frequency goes down, the more likely it is that the irregular form will be given up in favor of a regular one (I was taught that the past tense of to learn is learnt, modern dictionaries say it is learned. Or both.).
But there is of course more to words in terms of their usage and communication. You can find my take on that here. Two points I would like to stress here are word formation (moving away from core semantics to the morphology/semantics cross-section) via compounding and cultural context of speakers. And these nicely come together by the DevOps term. A few weeks ago Amy Phillips blogged on why she hates DevOps. Not necessarily the concept, but the term. It was because she thinks that it excludes everybody not development or operations. And that is something that I totally disagree with. Yes, there is no reference to testing in the title, but to me, that doesn’t mean that testing (or requirement engineering as well) are not part of a DevOps process. Let me push a little deeper on that: DevOps is a portmanteau noun, meaning that two words and their meaning blending into one word and one meaning. Eurasia would be another example: the landmass consisting of Europe and Asia. This blending is a subset of the word formation process of compounding, which, on a simplified level, means putting words together to form a new word. Testmanager is an example of this (and heck, German is famous for this: ever heard of Donaudampfschifffahrtskapitänsmütze? Well, it refers to the hat of a captain working for the steamboat company on the river Danube, but never mind, no one uses it outside of dreadful linguistic examples, but it is a well formed word within the German language). The meaning can easily be derived: a testmanager is a manager who deals with test(ing). Semantically there is a head B that carries the meaning (manager), but is specialized in A (test). So A+B is a special kind of B. This is called an endocentric compound. The meaning is within the constituents.
So what about DevOps now? It is not neither an operation that develops nor a development that operates. So A+B as a special kind of B doesn’t work here. You could argue that A and B are equal and both carry the meaning. Bittersweet would be an example of such a compound. This a special case of endocentrism called copulative.
But I don’t think that this is the right category either. Let’s have a look at the word baldhead. Semantically it is not a bald head, so it’s not of the A+B is a special kind B type and it certainly not of the A and B are equal type. The real meaning (person without much scalp hair, usually male and a bit derogative) lies outside of the constituents. This is called exocentric.
I would argue that DevOps belongs into this category as it is a full-fledged software development process including all major aspects including testing that puts a certain emphasis on the collaboration between all participants including the operations side, but I do not feel excluded because it is called DevOps and not ReqDevTestOps or something like that. From my point of view the concept represented by the word includes all these other aspects. So much for the compounding part, which brings me to the cultural contexts of speakers, or speech act participants if you want. Linguistically you could also argue that DevOps is an acronym where initial letters or syllables (as in this case) of a phrase are combined to form a new word. NATO, laser, AIDS, Benelux come to mind as examples. As I did some reading on this and talked to some former fellow students of mine, it struck me that English and German linguistics (and probably language understanding of native speakers, too) seem to differ in terms of dealing with acronyms and endo- vs. exocentrism. While there was no hint in English sources about the possibility of acronyms possessing an exocentric meaning and thus making Amy’s point of view fully understandable in terms of linguistic categorization, German sources referred to acronyms being able to sport exocentric meaning (supporting my point of view). I haven’t conducted any detailed research on this, so it rather seems to be a hypothesis at the moment, but it reminded me nevertheless that whenever you are crossing cultural borders, you need to keep in mind that even though words may seem to be understood the same way, there is always a chance that some part of the perceived meaning differs due to different cultural backgrounds and understanding of linguistic concepts.
To sum it up:
  • Meaning is arbitrary. There is no direct connection between a word and the concept it represents.
  • Meaning of words or phrases consisting of other words does not necessarily have to be within the limits of the single word meaning.
  • Meaning and its perception may vary when crossing cultural borders.
Sources:
Booij, G.: The Grammar of Words. OUP, 2005.
Haspelmath, M.: Understanding Morphology, Arnold, 2002.