Extensions

So far, we have not discussed any of the language features you'd likely encounter in a grammar textbook for a typical human language: plurals, verb tenses, pronouns, etc. Those features are not part of OGTRTA proper; they are left up to the individual language designer.

One of the nice things about OGTRTA is that it provides a consistent backdrop against which to compare different implementations of various language features. Using OGTRTA, we can clearly see how different implementations make tradeoffs between conciseness, flexibility, and ambiguity.

Still, it can be somewhat tricky to figure out how to work some of these features into an OGTRTA language without disrupting the syntactic and lexical architecture that OGTRTA provides. This page aims to solve that problem with how-to guides and examples.

General strategies

Affixation

Compounding

Subordination

Null morphemes

Post-serialization processing

OGTRTA proper does not dictate any particular morphology. However, there are certain morphological patterns that I have found to work well with OGTRTA. Those are outlined here.

Verbal morphology

The dictionary form of verbs is the active participle. This allows adjective-like and preposition-like verbs to modify other words with no special inflection.

A verb used as the predicate (main verb) of a sentence is marked, usually with a morpheme glossed FIN (finite), or with a tense marker.

I tend to mark only finite verbs for tense. However, you might want to consider whether participles should also have past, present, and future tenses, as in Esperanto.

The infinitive form of a verb (INF) converts it to a noun, keeping its valence the same. The gerund (GER) is similar, but has a valence of zero: all complement slots are removed. GER is thus equivalent to the composition of MID (see below) and INF.

I usually have a set of valence changing affixes that mark the passive, middle, and causative voices. The active voice is unmarked.

A bit more detail on how this works:

A noun or verb can be thought of as having a number of syntactic slots which can be filled by complements. Verbs typically also have a syntactic slot for a subject NP. Each syntactic slot is associated with a semantic role which defines how the complement in that slot relates to its complend. The semantic roles are lexically determined by the complend — that is, they are part of the dictionary meaning of the complend.

For example, a verb like give/2 might have the following syntactic slots:

give/2   __________ __________ _______
complend complement complement subject
         gift       recipient  giver

That is, the complement noun phrase in the first slot following give/2 plays the role of gift, the next complement plays the role of recipient, and the subject plays the role of giver.

As discussed below in the section on recommended verbal morphology, derivational affixes on a verb can rearrange the semantic roles of its slots. For example, by applying a middle-passive inflection to give/2, we might obtain the following structure:

give-MIDPASS/1 __________ _______
complend       complement subject
               recipient  gift

The MIDPASS inflection causes the subject role to disappear. The first complement's role is moved to the subject slot.

Or, we might use a middle-voice inflection to remove all complements:

give-MID/0 _______
complend   subject
           giver

Modifiers

Because OGTRTA languages tend to be either consistently head-initial or head-final, the modifiers of multiple heads frequently get stacked on top of each other, making it hard to see which modifiers go with which head. I like to have some way to disambiguate which head a modifier attaches to. There are a few strategies for doing this:

You could also use some form of noun class agreement, but I find that that creates more problems than it solves.

Interfixed modifiers can help solve some ambiguity problems.

Copula

In standard OGTRTA, adjectives and prepositions are verbs and do not require a copula ("to be") to be used as the predicate of a sentence. However, a verb glossed as be/1 is still needed to form sentences that equate nouns ("a toad is an amphibian").

See adjectives and prepositions for hints on how to make predicative adjectives and prepositions use a copula.

Noun morphology

I find that requiring nouns to be marked for number leads to awkward situations where you don't know if something should be singular or plural. (In written English we have constructions like "car(s)" and "child(ren)" to deal with these situations, but there's no convenient equivalent in the spoken language.)

The solution is to let the dictionary form of a noun be transnumeral. Number marking is required on definite nouns only. I like to do this by having two definite articles, one for singular and one for plural. Number can optionally be indicated for indefinite nouns with a numeral ("one") or a modifier ("many", "several").

Numbers

I find it convenient to make numerals nouns; this makes elisive phrases like "I ate four" and arithmetic ("three plus two is five") work grammatically with no additional complications.

To express ideas like "three apples" you'd use a preposition-like verb: "three of apples". The same preposition can also naturally be used for multiplication: "five of three of apples makes fifteen apples".

Subject motion (SVO/OVS word order)

To create an SVO language, start from the standard VOS word order described on the syntax page, and add the following syntax rule:

S -> NP VP

That is, a sentence can consist of a subject noun phrase followed by a verb phrase.

To make an OVS language, reverse all the syntax rules as described in the section on reversibility.

Subject-as-modifier (VSO/OSV word order)

To create a VSO language, start from the standard VOS word order described on the syntax page, and replace the expansion rules for S nodes with the following:

S -> NP
   | VP
   | S CONJ S

The subject of a verb can then be expressed by a modifier phrase attached to the verb. The head of this modifier phrase can be a null morpheme.

To create an OSV language, reverse all the syntax rules as described in the section on reversibility.

Sentence-level modifiers

Sentence-level modifiers can provide attitudinal or evidential information: "Supposedly, Hamlet saw the ghost last night"; "Luckily, no one noticed us".

To add sentence-level modifiers to OGTRTA, add the following syntax rule:

S -> VP S

Optionally, you can allow sentence-level modifiers after the main clause, though this will create ambiguity unless you have some kind of modifier disambiguation.

S -> S VP

To disambiguate modifier attachment, you can mark sentence-level modifiers that follow their modifiee. One way to do this is to separate the clauses with a sentence particle (glossed SP). You might also mark sentence-level modifiers with a morpheme ADS (adsentence) or SM (sentence modifier) (note: these glosses are nonstandard).

Determiners

The core syntax of OGTRTA does not define a part of speech or syntax rules for determiners, but you could add them.

Replace the syntax rules for NP with:

NP -> DET? N/n VP* NP{n}
    | NOM S
    | NP CONJ NP;

Interfixed modifiers

If your language has determiners, you might allow "short modifiers" (modifier phrases headed by zero-valence verbs) between a determiner and its noun.

Syntax rules:

NP -> DET? SM* N/n VP* NP{n}
    | NOM S
    | NP CONJ NP;
SM -> V/0 VP*
    | SM CONJ SM;

Adjectives and prepositions

OGTRTA languages can emulate adjectives and prepositions by dividing verbs into two inflection paradigms, e.g.:

Form Class 1 Class 2
Infinitive maiten "to eat", sen "to be" sen-hujo "to be red"
Present Tense maita "eats", asa "is" asa-hujo "is red"
Past Tense omais "ate", oas "was" oas-hujo "was red"
Active Participle maitio "eating", asho "being" hujo "red"

Class 1 verbs are thus similar to what we would call verbs in English, and Class 2 verbs are similar to adjectives and prepositions.

Possessives

Possessive pronouns can either be determiners as in English, or verbs.

Example of possessives-as-determiners:

Example of possessives-as-modifiers:

Alternatively, you may not have separate possessive pronouns, and instead use a preposition-like verb to mark the possessive/genitive for both nouns and pronouns:

Valence-changing

Verb nominalization

Imperatives as NPs

Questions as NPs

Trace pronouns

Relative clauses

Relative clauses can be implemented in terms of nominalized sentences:

Here, the pronoun li "he," which refers back to studior "student," is left in place in the subclause. In this hypothetical language, you'd say, in effect, "I know a student such that he speaks Latin."

This is nice and regular syntactically, but it's a bit wordier than many natural languages.

To get a system closer to how English works, you could do the following:

That all sounds rather abstract, so let's look at an example:

You'd form the question "Who speaks Latin?" like this:

And you'd turn that into a relative clause like this:

Two null morphemes in one noun phrase! Is there nothing nothing can't do?

Demonstratives

Demonstratives like "this," "that," "these," "those" can be represented in a few ways:

Numbers

Cardinal numbers can be expressed in a few ways:

Example of numbers-as-determiners:

Example of numbers-as-verbs:

Example of numbers-as-nouns:

In choosing among these, consider how elisive phrases like "I bought three" will work. All the options but numbers-as-nouns require a dummy word (glossed unit below) to make the syntax work:

The dummy word can of course be a null morpheme.

Ordinal numbers can work similarly.

Ordinals-as-determiners:

Ordinals-as-verbs:

I don't think ordinals-as-nouns makes sense, but maybe I'm wrong about that.

Inchoative and cessative

Prepositions "to" and "from", and variants like "onto" and "from out of" can be replaced by inchoative and cessastive inflections of other prepositions. E.g. "to" can be at1#INCH. "for" (dative) can be of1#INCH.

Glossing abbreviations for morphemes mentioned on this page