How do you remove the confusion in a conlang when creating and distinguishing nouns, verbs, adjectives, and particles?

Question

I am trying to implement a conlang and am confused how we as humans are able to distinguish nouns, verbs, adjectives, particles, etc. in a sentence. I get that word order is a factor, as well as the nebulous idea of context. But word order seems so flexible, I don't see what I need to do when creating a conlang in terms of how to structure nouns, verbs, adjectives, and particles, to make sure you can understand the sentence. What are the key techniques or procedures to follow in designing this? So you can clearly tell what is a noun, verb, adjective, or particle. How do we even do this in regular natural languages? I don't understand. And I don't know how to search for this in Google to learn more about how we resolve the sentences meaning without having clear and unique patterns that clearly separate nouns, verbs, adjectives, and particles from each other.

Some of examples on why it's confusing.

I see the big black oak tree.

The tree is the noun, and big black oak are modifiers on the noun sort of thing. So you could say they are adjectives or "features" of the tree.

I see the big.

But here big is the noun, like The Big, as in something big, in which you are invoking a lot of context in your listeners mind. Or

I see the black.

Here black is the noun, like all you see is darkness.

In these sentences, there is no change in the structure of the words, but they take on different roles in the sentences. That is, big can be an adjective or noun (or even a verb!) without adding a specific prefix or suffix to that base.

I can't come up with a science on why we are able to properly interpret these sentences. In my conlang, I am thinking that, in order to make it easier to interpret, I would have special endings on nouns, verbs, adjectives, and particles. So you would have the equivalent of:

I see-ami the big-azi black-azi oak-azi tree-ali

In this sort of system, every verb would end in -ami, every adjective with -azi, and every noun with -ali. I keep on thinking I need such a system in order to make the sentences comprehensible. I would use this structure in addition to a strict word order to make it so sentences are comprehensible. So what I keep thinking I need, in summary, is:

1. Specific endings on nouns, verbs, adjectives, and particles, to distinguish them easily.
2. Strict word order in sentences, so you can easily tell what is in the adjective place, in the noun place, in the verb place, etc.

But the problem is, this is not how natural languages work! Even in English which has a somewhat strict word order (even though you can be quite flexible with it if you want), like I demonstrated above, you don't have specific endings on words, and you don't have a super strict 100%-of-the-time word order. (English has some word endings like -ing and -ly and such, but these are not even consistent 100% of the time)

In addition, I am looking across a few languages, and nouns, verbs, and adjectives in each language might be allowed to all start or end with vowels or consonants, and might all have endings which are the same. That is, there isn't even a pattern which clearly separates nouns, verbs, adjectives, and particles into separate categories based on the word's morphology! For example, let's use English. You have these words.

Bully. (Noun, "A mean sort of person", ends in -ly)
Clearly. (Adjective, ends in -ly)
Sully. (Verb, "to make dirty", ends in -ly)

Some rules are more strict, like I can't seem to find any adjectives that end in -ed, which is used for past-tense or state sort of verbs, created, acted, iced, etc. But this -ly ending can go on any type of word. Same with other endings, like -ing.

Bring. (Infinitive verb, ends in -ing)
Clearing. (present participle verb, ends in -ing)
Ring. (Noun, or verb, ends in -ing)
Amusing. (Adjective, ends in -ing)

All kinds of other examples exist (of prefixes and suffixes) where they are used across nouns, verbs, adjectives, etc.. In many other languages too.

Other words, which I'm using in English but might come from other languages, are nouns which have all sorts of endings.

Man. (Noun, ends in consonant n)
Joy. (Noun, ends in consonant y)
Dao. (Noun, "the Way", ends in vowel o)
Glee. (Noun, ends in vowel e)
Rabbi. (Noun, ends in vowel i)

Then same for verbs:

Assign. (Verb, ends in consonant n)
Play. (Verb, ends in consonant y)
Echo. (Verb, ends in vowel o)
Free. (Verb, ends in vowel e)
Ski. (Verb, ends in vowel i)

And same for adjectives:

Green. (Adjective, ends in consonant n)
Happy. (Adjective, ends in consonant y)
Latino. (Adjective, ends in vowel o)
Eerie. (Adjective, ends in vowel e)
Pakistani. (Adjective, ends in vowel i)

So this basically shows, at least for English, there is no relation between the word ending and whether it is a noun, verb, adjective, etc.. So it makes me think in my conlang, I don't need to add the constraint of having nouns, verbs, adjectives, and particles have different endings to make them distinguishable. Somehow we are able to tell the difference in sense of these words based on some other nebulous thing, like context or meaning/semantics during the flow of speech.

The question is, how do I use these facts when constructing a conlang? Can I pick any "base" word and attach any ending on any part of speech (noun, verb, adjective, particle)? And will the language still work? How can I tell if it will work and not be confusing? What do I do when developing the conlang to test if it is confusing or not how I have structured words?

Without having the conlang completed and working, (while it is being developed), as I try to generate example sentences, it seems like it would lead to pure confusion not having more rigid patterns in the words to distinguish nouns, verbs, adjectives, and particles.

How should I be thinking about this to make some progress? I am stuck on the step of how to create words in different parts of speech, and place them in an order in sentences. I have created a pattern for how word "bases" are defined (i.e. they have to start and end with a consonant), but that is even limiting, I would like it if some can start or end with a vowel too...

How should I conceptualize the differences between nouns, verbs, adjectives, and particles, so I can generate words that flow together into a sentence but aren't confusing? It seems that adding specific word endings would remove the confusion, but what else will remove the confusion so I don't need to do that, and I can be like English in that words of various parts of speech can have any vowel/consonant start or end the word?

Answer 1

Fundamentally, parts of speech are lexically determined. What this means is that in general, there is no way to find the part of speech of a word knowing only the word itself. You have to memorise it separately for each word.

A good way of seeing this is to observe that two languages may assign different parts of speech to exactly the same word when it is borrowed from one language to another. For instance, boil is a verb in English. But when it was loaned into the Papuan language Komnzo, it was borrowed as a member of the Komnzo class of ‘property nouns’ (Döhler 2018). There is nothing inherently ‘nouny’ or ‘verby’ about the form or function of this word; you simply have to memorise that boil is a verb in English and a property noun in Komnzo. And this can even happen within a language: the English word release, for instance, is both a noun and a verb.

So, if word classes cannot be reliably distinguished based on their form or their semantics¹, then what is the difference between different parts of speech? The answer is simple: parts of speech specify the environments in which words can be used. For instance, consider this sentence:

I saw the _____.

This sentence has a gap in it. You can make a grammatical sentence by filling this gap with nouns: I saw the dog, I saw the cat, I saw the chair, I saw the release, I saw the idea. (Note that, even though the last example doesn’t really make sense, it still counts as it is a grammatical sentence.) On the other hand, if you try this with a non-noun, the result is ungrammatical: *I saw the sit, *I saw the eat, *I saw the is. So we can use this example to give a definition of a noun: a word which makes a grammatical sentence when preceeded by I saw the.

This definition can be made a lot more general though: a noun is simply a word which can follow a determiner. What is a determiner? Well, you could define it as something which can precede a noun, but that would be circular. Instead, simply note that there are a fixed set of 20 or so words which can occur in such a position: a, the, one, some, much, many and the like. A determiner is then just one of these words. Determiners are thus a ‘closed class’: a part of speech which has a fixed number of words, and into which no more words can be added. Note that each language has a slightly different set of closed classes, though there are similarities: articles, numbers, adpositions and adjectives, for instance, are often closed. (Komnzo unusually has verbs as a closed class, explaining why my example of boil above was loaned as a noun rather than a verb.)

But you can go further. English nouns can be used after a determiner — but there are several other environments in which nouns can occur. They can take the plural marker -s: dogs, cats, chairs, releases, ideas. They can also modify another noun: dog leash, cat collar, chair wood, release method, idea food. (The last few examples are a bit contrived, but illustrate the point well enough). So, you can use this to define nouns in English: English nouns are that word class which can be used after a determiner, take a plural marker, and modify another noun. Or, to put it another way, if you find a word which can occur in all these places — why then, that word must be a noun!

So, to answer your question ‘How should I conceptualize the differences between nouns, verbs, adjectives, and particles?’: conceptualize it in terms of morphosyntactic environments (the fancy name for what I’ve been talking about). A word class is distinguished by the environments it can occur in: if you know what word class a word occurs in, then you know in which environments you can use in, and vice versa.

I will use one final example to illustrate the point. Consider the well-known example sentence:

The gostak distims the doshes.

From this sentence, I can tell that gostak and dosh are nouns, while distim is a verb — even though I may never have seen these words before. Why? Just look at the environments:

• Gostak occurs in the environment ‘the ____’. Thus we know it is a noun.
• Dosh occurs in the environment ‘the ____’, and also takes the suffix -s. Thus it must be a noun.
• For distim, we can look at the position of the word in the sentence. It occurs between two noun phrases; this is obviously sufficient to make up a complete sentence. Any word with this property in English must be a verb.

(Further reading: François’s analysis of Hiw word classes is a particularly illuminating example of these ideas. I highly recommend reading through this paper for a detailed discussion of how word classes can be defined with reference to syntactic environments.)

---

¹ Well, this is a slight lie. There are clearly semantic distinctions between word classes such as nouns, adjectives and verbs. Also, I vaguely remember that there are some languages in which some parts of speech do have phonological correlates, though I can’t remember their names. My point is that neither of these are particularly reliable ways to determine parts of speech, which should ideally be defined using only morphosyntactic criteria.