Syntax of Mahal's Loglang

A forum for all topics related to constructed languages
Post Reply
Golahet
cuneiform
cuneiform
Posts: 196
Joined: 12 Aug 2010 16:01

Syntax of Mahal's Loglang

Post by Golahet »

This is some information about the syntax of my main conlang (unnamed as of yet, or named more than once, but these names are discarded). The following is the syntax given in BNF, an ambiguous context-free grammar.


<> = encloses a non-terminal.
| = separates items in a list of exclusive options (either ... or) *
[] = encloses optional item (zero or one time)
{} = encloses optional item that may appear several times (zero or more times)
() = encloses item that appears at least once (one or more times)

Syntax

<grammatical-unit> ::= <sentence> | <noun-phrase> | <vocative-phrase> | conjunction
<sentence> ::= <clause>

<clause> ::= <verb-headed-clause> | <disjunct-headed-clause> | <conjunction-headed-clause>
<verb-headed-clause> ::= verb {<argument> | <adverb-phrase> | <vocative-phrase>} [valency-terminator-adverb]
<disjunct-headed-clause> ::= disjunct {<adverb-phrase>} [{<argument> | <vocative-phrase>} | valency-terminator-adverb]
<conjunction-headed-clause> ::=
  • conjunction (<clause>) [coordination-terminator-adverb] {<adverb-phrase>} [valency-terminator-adverb]


<argument> ::= <clause> | <noun-phrase>

<adverb-phrase> ::= <adverb-headed-phrase> | <conjunction-headed-adverb-phrase>
<adverb-headed-phrase> ::= adverb {<argument> | <admod-phrase>} [valency-terminator-admod]
<conjunction-headed-adverb-phrase> ::=
  • conjunction (<adverb-phrase>) [coordination-terminator-admod] {<admod-phrase>} [valency-terminator-admod]


<noun-phrase> ::= <noun-headed-phrase> | <conjunction-headed-noun-phrase>
<noun-headed-phrase> ::= noun {<argument> | <adjective-phrase>} [valency-terminator-adjective]
<conjunction-headed-noun-phrase> ::=
  • conjunction (<noun-phrase>) [coordination-terminator-adjective] {<adjective-phrase>} [valency-terminator-adjective]


<vocative-phrase> ::= <vocative-headed-phrase> | <conjunction-headed-vocative-phrase>
<vocative-headed-phrase> ::= vocative {<argument> | <adjective-phrase>} [valency-terminator-adjective]
<conjunction-headed-vocative-phrase> ::=
  • conjunction (<vocative-phrase>) [coordination-terminator-adjective] {<adjective-phrase>} [valency-terminator-adjective]


<adjective-phrase> ::= <adjective-headed-phrase> | <conjunction-headed-adjective-phrase>
<adjective-headed-phrase> ::= adjective {<argument> | <admod-phrase>} [valency-terminator-admod]
<conjunction-headed-adjective-phrase> ::=
  • conjunction (<adjective-phrase>) [coordination-terminator-admod] {<admod-phrase>} [valency-terminator-admod]


<admod-phrase> ::= <admod-headed-phrase> | <conjunction-headed-admod-phrase>
<admod-headed-phrase> ::= admod {<argument> | <admod-phrase>} [valency-terminator-admod]
<conjunction-headed-admod-phrase> ::=
  • conjunction (<admod-phrase>) [coordination-terminator-admod] {<admod-phrase>} [valency-terminator-admod]



The syntax is ambiguous in the sense that you could arrive at some strings more than one way. E.g. the string

verb verb adverb

could be produced as (skipping some steps)

<clause>
verb <argument>
verb <clause>
verb verb <adverb-phrase>
verb verb adverb

or as

<clause>
verb <argument> <adverb-phrase>
verb <clause> adverb
verb verb adverb

But the language itself isn't ambiguous, since there is only one way to parse it.

The rule is that parsing never exits the embedding of a constituent unless there is no more syntactically acceptable arguments or modifiers that could continue the constituent. This rule applies without any consideration of the semantics.

This means that in the above example the adverb will ALWAYS be a modifier of the second verb, never the first. To get the other possibility you could reorder the words as

verb adverb verb

or

verb verb valency-terminator-adverb adverb

"But", you may ask, "shouldn't the second verb of the first of those options be an argument of the adverb, instead of what we wanted: an argument of the first verb?"

The presence, number, and semantic roles of the arguments of a head is explicitly marked on the head. This means that terminators are never needed to disambiguate to which head an argument belongs. The parse rule above and the need for terminators applies to modifiers, not to arguments. (Well, it kind of applies to arguments as well, but it doesn't matter for well-formed strings.)

Argument structure is an inflectional category, i.e. the roots themselves doesn't specify the argument structure, and the morphology that handles it has absolute semantic precision. That means that the words "take", "get", and "give" all are different forms of the same lexical item.

Thus, if we have "verb verb adverb" and simply reorder the words to "verb adverb verb", then the second verb can't be an argument of the adverb, since the first sentence established that the adverb has no arguments.


Parts of speech, or Terminals

The language has only one part of speech - using John Quijada's terminology: formatives. (There might be rhetorical benefits from calling the part of speech "verbs".) But it still has words that functions as verbs, nouns, etc. What I mean is that part of speech is an inflectional category. The morphology has absolute semantic precision, which means that you never have to guess the meaning of a word, you just have to look at the morphemes that build it up. Nouns always refer to the subject of corresponding verb, so if you have the verb "look at" and form a noun from it, then the noun means "a looker". This is the only nominalization process that exists in the language. To get other nouns from the same verb, other morphological things need to happen. This applies to all nouns, without exception, and to all other words in the entirely language. But this isn't about the morphology, so I won't go into detail.

But this means that any root could combine with any morphology, so roots for e.g. "and" and "xor" could be nouns or whatever, and you could combine the root for "goat" with the morphology for a conjunction (ofcourse, some combinations may be semantically anomalous). I will pretend that this isn't true.


conjunctions
As can be seen above, a single conjunction can be an independent grammatical unit. And as stated, the valency is explicitly marked on the conjunction. This also applies to adverbs, adjectives, and admods. They are not modifiers of the conjunction, they are arguments. That is, the modifiers that precede the coordination-terminator. The adverbs, adjectives, and admods that follow the coordination-terminator are modifiers of the whole expression.

Note that "conjunction admod admod" will have the second admod to modify the first, and either the conjunction only takes a single argument, or the string is incomplete or ungrammatical. (In either case it is incomplete "to the left", i.e. an independent grammatical unit, if headed by a conjunction, can only be a single conjunction and nothing more.)

Conjunctions are coordinating conjunctions like "and" and "xor".

verbs
Verbs are like verbs in English (except for what I already have said).

disjuncts
Disjuncts are a category that include tense, aspect, and mood. Some disjunct may, when a sentence is translated to English, be labeled as adverbs, conjunctions, or prepositions, or some other hard-to-define things, like "not". There is a disjunct that could be labeled "a polar question particle". The difference between a disjunct and a verb is the difference between "not" and "is false", the former a disjunct and the latter a verb, of the same stem. As you can see above verbs and disjuncts doesn't have the same syntax.

"yes" and "no" are examples of disjuncts that doesn't take arguments.

adverbs
Adverbs above include what corresponds in English to adverbs as well as adverbial prepositions. Any adverb that takes an argument is a preposition. Adverbs that take clauses as arguments may be labeled "conjunctions" or "complementizers" when found in English.

nouns
Nouns behave roughly in the same way as in English. As stated, the nouns could be conceptualized as derived from verbs (or simply as non-finite verbs). If that corresponding verb is bivalent or polyvalent, then the noun when in the "construct state" will take the objects of that verb. As already stated, the noun itself refers to the subject. So, from the verb "look at", we get the noun "a looker", and when in "construct state" it takes the thing looked at, e.g. a bird: "looker bird" = "a looker of a bird", "someone (or something) that looks at a bird". Nouns as "father" or "foot" are also bivalent verbs in their cores, "is a father of" and "is a foot of".

Some spatial or temporal nouns when they take arguments may be labeled "prepositions" when found in English.

vocatives
Vocatives are nouns, except that they behave syntactically different. And that they are vocatives, ofcourse. They are sort of outside of the rest of the sentence. The truth value of a clause doesn't change depending on the presence of a vocative. While the parse rule does apply to vocatives as with arguments, when the vocative is unambiguously the last part of a clause it doesn't apply, and it will be assumed to be part of the outer clause.

adjectives
Adjectives are like the adverbs: when they take arguments they are adjectival prepositions. Among the adjectival prepositions we find an important one that serves both as a genitive adposition and a relativizer, lets call it R:

car-DEF R king-DEF = "the king's car"
car-DEF R PAST buy I-DEF c-ANAPH-DEF = "the car that I bought"

Adjectives only appear as modifiers of nouns (incl. the vocative kind).

admods
"Admod" is short for "admodifier". When found in English these would be called "adverbs" when not having any arguments, and "prepositions" when they have arguments, possibly "complementizers" when taking clausal arguments. Note that they are their own modifiers. While adjectives aren't modified by other adjectives and adverbs aren't modified by other adverbs, admods are modified by other admods.

An important representative is "very" as in "a very big person".


Some more word order

The subject precedes the objects (i.e. it's VSO, not VOS)

The language is nominative-accusative-dechticaetiative, but has no morphological case, unless you take the "part of speech" inflection as cases, which isn't that unreasonable. Well, you could consider the vocative as a case as well, which happens to operate in the same morphological slot as the part of speech inflection.

It has a topic-comment structure. It works together with above syntax by having a topic particle, which really is a verb, that takes the topic as the subject and the rest of the clause as the object, the embedded clause.

TOPIC goat-DEF PAST buy I-DEF g-ANAPH-DEF ≈ "As for the goat, I bought it"

= verb noun disjunct verb noun noun

But this structure is only used when for a heavy topic, when you want to emphasize it. Ordinary VSO is the most common.

The same syntax is also used for a scope-rising particle.

not PAST see-1SG the.dog each = "I have never seen every dog"
SCOPE the.dog each not PAST see-1SG it = "I have never seen any dog"

The language is wh-in-situ, which might be obvious.



This will do for now.

(And yes, "see-1SG" and "see I-DEF" is two ways to say the same thing.)



edit

* the original was "xor (either or)"
Last edited by Golahet on 03 Sep 2010 17:36, edited 1 time in total.
User avatar
MrKrov
banned
Posts: 1929
Joined: 12 Aug 2010 02:47
Location: /ai/ > /a:/
Contact:

Re: Syntax of Mahal's Loglang

Post by MrKrov »

That's more syntax than we normally see.
User avatar
Testyal
roman
roman
Posts: 936
Joined: 12 Aug 2010 19:47
Location: England

Re: Syntax of Mahal's Loglang

Post by Testyal »

xor actually means exclusive or
:deu: :fra: :zho: :epo:
Golahet
cuneiform
cuneiform
Posts: 196
Joined: 12 Aug 2010 16:01

Re: Syntax of Mahal's Loglang

Post by Golahet »

Yes.

Maybe "either" in "either or" above could be misinterpreted as in "either of them will do". Exclusive or is indeed meant, and the "either A or B (but not both)" construction the one referred to.


Actually, "a | b | c" means that only one of those options will be, unlike "a xor b xor c" which also allows all three to be at the same time, but not two of them. Some suggestions how to make it more clear formally?
peterofthecorn
cuneiform
cuneiform
Posts: 91
Joined: 14 Aug 2010 02:32

Re: Syntax of Mahal's Loglang

Post by peterofthecorn »

Mahal wrote:Actually, "a | b | c" means that only one of those options will be, unlike "a xor b xor c" which also allows all three to be at the same time, but not two of them. Some suggestions how to make it more clear formally?
I thought I was pretty sure that a xor b xor c would not allow all three to be true at the same time, but that it would only allow one, and exactly one, to be true.
Post Reply