This might be a strangely framed question for a liguist, since I'm a physicist and know very little about this field.

My question relates to how changing a symbol (for example a letter) in a word changes its meaning. For example, in English the word "CAT" means something completely different if we "mutate" one letter. Even if RAT, CUT and CAR are one letter appart from CAT the meaning is far away from the original. I picture this mathematically like a function (which converts symbols into meaning) that in the case of English should be a discontinuous or jagged function.

I was asking myself if there is a human language that continuously transforms symbols into meaning. That is, a language that when a small random change is performed over a word the meaning suffers a continuous small change.

Maybe it sounds weird but a non-human language like DNA has this property. If you change one nucleobase the change in the phenotype is not very large. If it were very large, then natural selection wouldn't work, since any small mutation would give drastic changes and gross deformities that wouldn't make for any stable correlation between the environment and the fitness to it. So my question is, are there any human languages with that property? And does this feature has a proper name in the field of linguistics?

Perhaps one could say that there is a small subset of "continuous morphology" (let's call it like that, for now) in many languages, since there are things like prefixes and sufixes that mantain some of the abstract meaning between different words, while mantaining morphological similarities (since the prefix or sufix is the same configuration of letters)?

Are there entire languages like this or large subsets of a language that behave in this way?

  • 1
    I am doubtful. The issue is the completely arbitrary mapping between sign and signified. There is nothing “tree”-like about a tree; this is why we can call it a “tree” and the Spanish can call it an “arbor”. Language is faced with the impossible task of assigning sounds to everything in the universe — as well as everything we’ve conceived of which is not in the universe. And, of course, faced with this task, we do the only sensible thing: we make it up as we go along. Language is no more than a series of accidents if history, constrained only by human hardware (physiology).
    – Dan Bron
    Jun 29 '21 at 11:52
  • 1
    That said, there are constructed languages (conlangs) which attempt this, precisely because it’s not seen “in the wild”. See, for example, Ithkuil. Of course, if it ever became a “natural” spoken language, such regularity and self-consistency would be the first casualty.
    – Dan Bron
    Jun 29 '21 at 11:56
  • @DanBron I think this is a good explanation. If you could make an answer out of it I will probably vote for it (after waiting some days in case other might want to contribute).
    – Swike
    Jun 29 '21 at 13:06
  • @DanBron, what about some sign language? Jun 29 '21 at 16:44
  • @StefanOctavian What about sign language? How does that overcome or change the considerations I laid out for verbal languages?
    – Dan Bron
    Jun 29 '21 at 16:45

No. Human language is by nature discrete or even discretised. There are some parameters that can in principle vary continuously, like the length of vowels or the quality of vowels in the formant diagram, but all natural languages pick a discrete selection of values (vowels can be long or short, giving a binary distinction, in many languages [ternary schemes with three levels of vowel length are also known, and there are a lot of languages without phonemic vowel length at all]; a fixed set of vowel phonemes is used in any language). When you present human subjects continuously varying stimuli, you can determine a language (and maybe subject) specific kipppunkt where the perception suddenly changes between one or the other representation. When you have heard of sigmoid activation functions—this is the manner how it happens mathematically.

P.S. I don't think that your DNA analogy holds water. Point mutations can be quite disruptive and their effect is mitigated only by the fact that we carry two copies of each gene, one paternal and one maternal.

  • You are right in the sense that continouos is not a very good wording of mine, since it will always be discrete in nature, but at least monotonical in specific regions. I wanted to convey some sort of general smoothness in my question rather than strict continuity.
    – Swike
    Jun 29 '21 at 14:36
  • @JanusBahsJacquet Added an explication to the answer. Jun 29 '21 at 14:36
  • @swike My point is: Even when there is the potential for continuity, human language is by choice discrete. Jun 29 '21 at 14:38
  • Regarding natural selection I agree that there can be large mutations produced by swaping just one nucleotide, but in general the genome is way more robust to such small perturbations. Evolutionary gradualism is the way of thinking here, and if the mapping is not as smooth then heritability would not be there to give natural selection a way to operate. In fact sexual reproduction (your example) is a way to increase variability not a way to damp huge changes in phenotype. And even if that was the case you still have asexual reproduction as an example (which dominates the biosphere).
    – Swike
    Jun 29 '21 at 14:41
  • 1
    Differentiability, on the other hand, has no possible analog in natural language, because there are no known functions (i.e, there are no relations with unique values for given arguments).
    – jlawler
    Jun 30 '21 at 15:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.