What fits the criteria: ah, okay, good, better, best, green, child, children, student, to study, slow

What does not fit the criteria: greener, greenest, students, studying, studied, slow down


  • greener is a regular comparative form that can be derived from green + er. The same goes for greenest.
  • better is irregular; in a sense that you cannot make it via good + er. The same goes for best.
  • child and children fit, because the first is a dictionary form, the second is an irregular plural
  • student fits, students - not, because it is just +s
  • to study fits, because it is a dictionary form of the verb (whatever one we choose, in my example it is infinitive, in Bulgarian it will be present tense 1st person, for example)
  • slow fits (adjective), but slower won't
  • slow down does not fit, because it is a combination of words, both of them fit (slow - adjective, down - preposition or adverb), but as a combination, I would want to treat it differently, so it does not match.

What comes to my mind is "regular dictionary single word form", but is there any definition terser and more precise than that?

The idea is that I try writing software that can decline, conjugate and manipulate words. In this paradigm, I intend to oppose concepts of an atomic word and a combination of words, where all of the latter are atoms linked by some rules and defined beforehand parts of speech:

  • good student = good (adjective, non-comparative) + student (noun, singular)
  • better students = better (adjective, irregular comparative of good) + students (noun, plural form of student)

I try to work in terms of domain-driven design, and name entities in my code professionally, so I would appreciate your suggestions.

3 Answers 3


"Regular citation form" is probably good enough. The problem is that you're asking about something that isn't really a concept of linguistic theory (that is, it isn't special enough that we have a term for it). "Citation form" or "lemma" does not specifically identify "regular" as contrasted with "not as regular", but from your examples you seem to specifically want to treat "irregular" forms differently.

The problem is that there are degrees of regularity, where "good; better" is on the "most irregular" end of the scale, and "bushes" (plural of bush) is on the "least irregular" end. What non-linguists typically see as "irregular", linguists see as "a less obvious rule". In terms of spelling and pronunciation, plural "cars, cats, dogs" are very simple – they show the effect of one rule of morphology ("add /z/, spelled s") which can be subject to a rule of pronunciation ("pronounce /z/ as [s] after voiceless consonants"), a rule that is not reflected in the spelling. Words like "bush" involve another complication, where the plural is spelled and pronounced with an additional vowel. This is light-weight irregularity, that is, there is an additional rule. Possessive 's is basically the same underlying form and set of rules as plural s, except that the rule of spelling is different (you don't add the letter e).

Patterns like "ring, rang, rung" with the meaning "ring a bell" are not completely random – there is a rule. It's just that some words like think don't follow that rule (at least in prestige dialects). The fact that "ring" meaning "encircle" is not inflected like the other verb "ring" is itself a regularity (nouns turned into verbs do not follow the ablaut pattern). So the question is, what kind or (ir)regularity are you interested in?

  • first of all, accept my sincere thanks for taking the time to look into my question! You raise good points, I'll try to address some of them, at least.
    – noomorph
    Oct 15, 2019 at 17:06
  • Regarding the continuum of "most/least irregular", that's a nice catch. I intend to introduce an "irregular" form every time I cannot consistently enough replace it with a programmatic set of rules. Of course, I'll need to think about how to link those regular and irregular forms in the database, but that's a separate challenge.
    – noomorph
    Oct 15, 2019 at 17:08
  • As for "ring -> rang" and "ring -> ringed", I think I already have this concept in my domain model. Each regular citation form can have definitions, and those definitions do not have to necessarily share morphological and other various traits. I am aware of the concept, where a word has a homonym which behaves similarly in most, but not all situations. So, "word definition"-level granularity theoretically should cover those cases.
    – noomorph
    Oct 15, 2019 at 17:13
  • Last but not least, the language in the subject is a zonal constructed language, so it should not have way too many irregularities.
    – noomorph
    Oct 15, 2019 at 17:17

In morphology, this is called a lemma (plural lemmata if you want to be pretentious, lemmas if you don't). In lexicography, it's also sometimes called a citation form; in psycholinguistics, it's sometimes called a lexeme (though that more often refers to the whole mental object containing all the different forms, not just the single form you list in a dictionary: the single form is just a shorthand for that mental object).

Notably, a lemma is generally an actual word in the language, while a "stem" or "root" doesn't have to be. For example, the lemma for "river" in Latin is rīvus, while the stem is rīv-. In English, stems are generally valid words in and of themselves, but this isn't always the case in other languages.


Yes, a lemma. The construction form, not a philosophy, is the point of its form, for the construction phase. We do not intricately define anything else here out of tandem and out of the constructs of the design, out of the places. The direction must be precise and transcription is a more formal approach to actual paper work. It does mean a creating part is necessary and that may make more law available, but the negative of that is a power to create a law that has an ever reaching state to something else and that's how we define a reach for anything that doesn't have heavy works of carrying capacity like English does; we can carry things like this in tandem to create a use of some formative intuition we may have from language. And the assumption of it creating a series of any kind, the descriptive event may offer us the possibility, in tandem, with other descriptive properties, that with rule in law, the creatjng purpose is more for purpose of ammendment, not the qualities of law creating particular, useful, and alloted rules of process; This is all for formal thought before we eequire a neutral linguistics, and the application of it should be in its native language...to assume a course of nature is intertwined with the vessel in its design. If the language changes in increasing a single and or median value, the construction of values ceases in parts what we have in it for "cognition". We would seek to propose things instead and we may lose a quality of unestranged realities as they mix in to add in parts what they with minimum use and etiquette, they were able to ascertain was a manner of inscription and not just one of the powers of listening. To estrange you in the guidelines of thought as they encounter the ideal within meanings. The words are and would be a strict requirement to the strictness you would think worked in the same way, but that would not be the same requirement we may actually have in common. Offering someone, officially, one more than you may actually have, you may be estranging that whole quality to you for them to fail in, and you may be needing to require yourself the greater constraints without failing a mean to describing median as it maintains a formula for all rules and the guidelines may push on boundaries you may need to remain in for further processing to be even, on appendix, a vertabrate experience, so you don't dangle the word gold in some hue or shade.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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