When designing a dictionary for multiple languages, what should be a single "entry" (focusing on languages like English)?

Question

I am looking at this dictionary guide and just looked up homophones to see how you might handle unrelated words with same spelling and pronunciation and part of speech. Google has basically separate main chunks, with the word as the header, followed by pronunciation, followed by parts of speech nested inside, and for each part of speech, a set of definitions.

I'm wondering though how it works in other languages. Can you create a separate entry for each part of speech instead?

In English, you might have a word like "what" which has 3 parts of speech: pronoun, determiner, and adverb. They are pretty much the same meaning, just different angles because of the part of speech difference. So it kind of makes sense to have 1 entry with 3 parts of speech instead of 3 entries. But I am contemplating making them separate entries each.

I am imagining URLs like:

/word/what/pronoun
/word/what/determiner
/word/what/adverb

Or for the word "rose" (flower, or past tense of rise):

/word/rose/1/noun (flower)
/word/rose/1/verb (make rosy)
/word/rose/2/verb (past tense of rise)

They would theoretically be on different pages.

Can you have a word with several different definitions under one part of speech, in any language? I can't think of any in English really yet. Something like:

/word/foo/1/verb/1 (means to brush hair)
/word/foo/1/verb/2 (means to cook dinner)
/word/foo/1/verb/3 (means to stand still)
/word/foo/1/noun/1 (means tree)
/word/foo/1/noun/2 (means freedom)

Basically, there would be unrelated meanings for same sounding and same written word, for same part of speech? Does that ever happen?

I don't really like how Wiktionary has a page like https://en.wiktionary.org/wiki/he which links to definitions in all kinds of languages, where in English there could be several unrelated meanings like with tear. I think at some level it would make sense to have separate pages for each meaning sector.

Even if the meaning is:

/word/foo/1/verb (to burn)
/word/foo/1/noun (a burn)

I think they should maybe be separate and then link to each other as related.

The main question is, for a web app sort of functionality (where you have the luxury of links and URLs, not limited to pages in a book), should you create separate entries for parts of speech when they are similar in meaning (like "what")? Or "created" (adjective, or past tense verb of "create").

/word/created/adjective
/word/created/verb

Likewise, can you have multiple different definitions under the same part of speech?

I guess my "brush hair" and related examples like /word/foo/1/verb/1 doesn't make sense. Each of those would be a separate entry basically, like /word/foo/5. So then the question that remains is, should you break apart parts of speech into separate entries too?

Answer 1

In this realm, the primary desideratum of dictionary design is having related words being easily connected. In a print dictionary, this happens by structuring entries so that the commonality is represented via alphabetical order. For example in Monier William's Sanskrit dictionary, words are sorted according to root, and prefixes are not the basis for ordering. An html-ish dictionary has much wider possibilities so that you can link an entry to other entries and not rely on physical order of presentation.

There are at least three entries that a comprehensive English dictionary would have for "burn" (a verb and two nouns). The most important distinction is between noun and verb, so at least two entries would be recommended. A single entry would approach the unwieldy. This is even clearer with the word "mean", because there isn't even semantic unity behind that set of words. On the other hand, a full deconstruction of "mean" would lead to many pages of entries, which would suggest that "mean" is a meaningless word.

The desired product simultaneously recognizes unity yet difference. At a minimum, the "semantic" entries would be distinguished from "cruel" and "average". The OED is a reasonable model of structured subdivisions for English. I don't think the concept of "entry" is coherent in dictionary construction, but "level of organization" is. You should present words in a sensible (accessible) speech-based order (spelling or pronunciation, depending on whether speakers can know how a word is spelled), but in a language with a lot of morphology, you will need to make organizational accommodations for ignoring prefixes (e.g. Sanskrit), stem-variations (e.g. Arabic) and zero derivation that changes grammatical class (English). I think that the expanding-menu approach is best so that you initially have two entries "burn" then expand the noun entry into the "product" noun and the "activity" noun. For physical dictionary or a low-tech electronic one, this is / may be impossible to implement.

In other words, subdividing is both good and bad, and whether unifying vs. subdividing is preferable depends on the language, the intended audience / use, the presentation medium, and your skills.