How can I find which words don't exist in a language?

Question

This answer got me thinking about words that exist in many languages but don't exist in a few. It states that there was no word for and in PIE. I had previously read that Some languages do not answer yes–no questions with single words meaning 'yes' or 'no'. Instead they typically employ echo answers. E.g. Is it hot? might be answered with It is [hot] or It isn't [hot] instead of yes or no.

How can I find out which words are missing unrepresented in a language, where the same concept is commonly represented in other languages as a single word?

Answer 1

It's actually next to impossible to get that information, unless you're doing in-depth personal investigation of the language. You might expect there to be a word for "tree" in every language, unless the language is spoken only in a place that has no trees. Even when there are trees, the boundaries between categories are not universal, and there may not be a distinction between "tree" and "shrub", in which case you would conclude that the language doesn't have a word for tree or for shrub (but we don't have a word for treeshrub). Most human languages do not have a word for "reptile" or "machine", though in principle any language could borrow a word from another language, if they wanted to have such a word. The problem is that it is extremely difficult to get detailed semantic information on the precise referents of words in languages, and instead you'll get approximations via western language translations.

For abstract and functional words, the notion of "equivalent word" is even more problematic. There are, as far as I know, no Bantu languages that have a word that is used exactly the way "and" is used in English. There is a common preposition na- which is used both for phrasal conjunctions and for instruments and comitatives (whereas in English you can't say *"I danced and Lucy"), and the English consecutive usage ("He bought a car and drive it into the wall") is usually rendered with a distinct verb tense. So in these language, there is no word "and", that is no word that is used in exactly the same way as "and" is in English.

The idea that words can be "missing" presupposes that there is a set of words that languages normally have. English is missing the word for "all-white reindeer", which is gabba in North Saami. The reason for that gap should be obvious. The notion of a word being "missing" ought to be paired with the notion (equally false) that there are languages with "extra" words, e.g. the rich reindeer vocabulary of N. Saami, owing to the cultural importance of reindeer among the N. Saami. If you combine these concepts, you could derive an interesting empirical question, namely, what are examples of cross-linguistic word pairs, where the referents of the words overlap substantially but not completely -- or, the referents of a word in one language properly includes the referents of a word in another language. Making that determination requires substantial field work, and can't be based just on how speakers translate words of their language into English.

[EDIT]

I note that the original question has been modified to shift from "missing" to "unrepresented". I don't see any substantive difference between "missing" and "unrepresented", although there are connotations of "failure" associated with "missing" that are blurred with "unrepresented". If something is unrepresented in X, that means that it does not exist in X, which is what it means for a thing to be missing in X. Regardless of the word you use to describe that relationship, the idea of being "unrepresented" is meaningful only if there is an expectation that some specific word-referent pairing is normally found, using something as the standard -- what, then, is that standard?

Answer 2

The only way you can really determine the lack of an equivalent lexeme in a language is through the process of translation (by a competent, experienced translator). The answer by @user6726 is absolutely correct. Let me just offer an additional perspective.

First, you can only ever really compare two (or possibly a small group) languages to get a picture of alignment. Many people start with the (implicit) assumption of a cultural or linguistic default from which languages diverge. In today's context, the default is usually described by English (an assumption shared even by many non-native speakers). So you get the typical newspaper articles using the 'no word for X' trope which the Language Log rails against.

It is not clear what equivalence looks like. There are countless features of the real (and social) world that are unproblematically not represented lexically in a language - e.g. names for exclusively local fauna or local customs. But there are many more, where the social and physical worlds are the same but the lexical inventory is not. I can see at least three types of misalignment:

1. Complete lack of lexeme / construction or corresponding reality (names of physical features, flora/fauna, local customs, etc.). In larger languages with recent academic literature, there are often specialist vocabularies that do contain such lexemes - but they are only available to a very small group of speakers.

2. Lack of lexeme / construction for a shared cultural feature exposed or introduced through contact.

3. Misaligned usage or category scope of shared lexeme / construction and social reality, arising due to lack of contact.

I consider a broader definition of word - including the traditional 'lexemes' as well as the broader concept of 'construction' from construction grammar (which removes the endless debate over 'what is a word').

Also, this classification is just a heuristic - any individual case will be a better or worse fit into one or two of these categories.

With the two caveats above, if you look closely at any two languages, you will find that words that do not fit at least to some degree into one of the above are the exception in the core of each of the languages' construction inventory (I suspect that many of the more peripheral constructions are probably more likely to align better in closely related languages - but research would be needed).

All of these misalignments may cause smaller or greater communication or translation difficulties.

Here are some examples from my experience of translating between Czech and English:

1. Complete lack of shared lexemes and cultural/physical reality (often only present in specialist vocabulary in one language)

When I translated Lakoff's Women, Fire and Dangerous Things, I struggled with the names for Australian animals - it was easy to find the taxonomic names but none that would explain what they meant to the non-specialist reader without footnotes.

Another word that may fit into this category is 'gerrymandering' which is a fairly unique word - and to some extent practice (as a particular kind of electoral fraud).

2. Lack of lexemes with shared cultural/physical reality (often leading to loans or calques)

I've written about the Czech word 'prozvonit' - to call just to let the phone ring with the purpose of alerting someone to action - usually prearranged. It arose with mobile phone usage to save money and is still used to arrange meetings. I have experienced the practice in the UK and even more prominently Albania (and I'm sure it's common elsewhere) but neither developed a corresponding lexeme.

Other examples mentioned (such as lack of 'and' in Bantu or 'yes/no' in Mandarin) may fit here.

There is also a large group of words which seemingly have the corresponding lexeme but the social realities they refer to are sufficiently different that speakers of one or the other language feel the need to adopt a new construction (either a calque or loan).

My favourite example of this is 'training'. The English corporate use of training (both world and reality) was so different from 'školení' that many people in Czech companies with English-speaking connections, started simply using the English word. Czech bankers also felt that the word 'cash' was more descriptive because of collocations like 'cash flow' that they preferenced it over 'hotovost'. In the voluntary sector, a similar problem arose with the word 'community'. I have seen one person in Kazakhstan interpret 'community service' as have a party. Another good one is 'fun'. I have heard a Russian speaker say 'Poydomte, budet fun.'

Some of these may be good candidates for category 3 but I chose to include them here because the lexemes feel sufficiently different to lead to loans rather than just misunderstandings. Some more are listed in the 'Czechs have a word for it' list.

An example from specialist language that straddles categories 2 and 3 I discovered when translating Lakoff. Czech does not really have a word for 'calculus'. The mathematical discipline obviously exists as does a term for 'differential calculus'. But nothing to describe the subject English students can take. I've looked, asked mathematicians, researched encyclopedias and textbooks. This one is complicated because it can lead either to borrowing or misunderstanding depending on who is experiencing the contact (specialist will borrow or calque, non-specialist will misunderstand).

3. Usage or category misalignment for mostly corresponding lexemes (leading to misunderstandings)

The vast majority of examples will fit into this category. For instance, the word 'bread' / 'chleba' translates straightforwardly between Czech and English when applied to loaves, however, English also uses it as a category label for which Czech has a different word. There are countless examples of this in either direction - for instance in the expression of manner of motion. Czech 'pivo' means 'lager' by default where as the (Br) English 'beer' means 'ale'. Czech does not even have a good word for 'ale' (which would go in category 2).

George Bush was mocked for saying something like 'The French don't have the word for entrepreneur.' but it is entirely possible that that is the case. In Czech, 'podnikatel' is a good translation but certainly does not have all the corresponding connotations or collocations.

A very common manner of misalignment is in collocation and what has been called 'semantic prosody' or connotation. The word 'volunteer' is a good example. In my work with the Peace Corps (in 15 countries), this was a word that caused the most issues - since in many of these contexts (former Soviet Union) 'volunteer' had the semantic prosody with non-optional public service - ie a negative connotation. (Community is the other one but it seems to better fit in category 2 because of the frequent loans).

Sometimes, it is the cultural classification of a shared cultural reality that is misaligned. For instance, the Czech 'hluboký talíř' which is used for soups only and can either be described as 'plate' or 'bowl' in English without any of the correct cultural connotations.

A similar case is 'závora' which is a type of gate that goes up and down (often not fully covering the entrance - e.g. level crossings) which simply does not belong in the category of 'gate' in Czech. Both Czech and English 'have a word for it' but its use is more ambigous in English and will leave Czech speakers confused. (Many examples in the opposite direction exist.)

To give an example of a misaligned construction, I would list lack of gender reference in anaphora. The English epicine 'they' makes translation very difficult when the gender of the referent is not known. The same for genderless profession names (secretary - Czech has a different word for things like 'Secretary General, doctor).

Sometimes, etymology gets in the way. E.g. the English 'nurse' has mostly lost its association with gender in the medical context, so a 'male nurse' is fine. The Czech 'sestra' (sister) did not, so no 'male nurse' expression is possible. Going the other way, articles in English specify a lot of things that Czech leaves ambigous (even though it expresses definiteness in other ways - it often just does not make it clear).

All of the above (and much more) make the claim for linguistic universalism problematic - and lies behind the constructionist reluctance to look for UG-like or even Greenbergian universals. While the reality behind these differences can be described in both languages (in most language pairs), can those languages be meaningfully described as lexically or constructionally equivalent?