Some Grammar Notes
Posted: Sat Oct 29, 2016 3:40 pm
Miscellaneous notes on tp grammar in various styles, all tentative but working so far.
Basic sentence:
S => Subject li Predicate
Subject => o/Phrase
Phrase => Word/ Phrase{Phrase}/ Phrase en/anu Phrase / [ni: Display]
Word is ultimately any member of tp vocabulary except la, li, e, o, en, anu, kin, pi. Unofficial words
can also occur, subject to some special rules which need spelling out more precisely.
{} indicates that the enclosed is preceded by 'pi' if it contains more than one word
[] means I don't like this because it screws other things up, but I don't yet know how to deal with it.
Display => Text - / Quote
Text => Sm (i.e. one or more sentences)
Quote => 'Text'
Word => noun / verb / modifier / preposition ...
the tp vocabulary (with the noted exceptions) can represent almost any of these types (see individual entries)
Predicate => Verbal (e Phrase) (,Prepositional)n
Verbal => Phrase / Prepositional / modal ({Phrase})Predicate
Prepositional => preposition ({Phrase}) Phrase
items in parentheses need not occur. Items in parentheses with a following n may occur any number of times.
This grammar is overinclusive, but every simple tp sentence fits in it, with one lowest level obligatory transformation:
[o/mi/sina]Subject li Predicate => o/mi/sina Predicate
That is, 'li' is dropped after these three words standing alone
[]Subject means the enclosed item is the whole of the component Subject
In particular, this grammar does not account for the more or less fixed order of modifiers in Phrase and does not account for the various nature of the added Phrase. These are covered in later transformations (I hope).
Some definitions:
A sentence with [o]Subject is an imperative sentence.
A sentence containing 'seme' is a free choice interrogative sentence and ends with a ? instead of a period. The answer (to be discussed in detail later) is a Phrase replacing 'seme' or the original sentence with such a replacement.
If S is a declarative sentence, S? Is a true/false interrogative sentence, with the answers 'lon', '(lon)ala' or S or the denial of S.
Other simple sentences are declarative sentences and end with a period.
If S is a declarative sentence, 'o S' is an optative sentence, expressing a wish (etc.) that S be true. It ends with a period or !. [Note: this does not claim that the speaker actually has this wish, so an optative is neither true nor false, though it may be honest or not.]
Compound sentences
If S1 is a declarative sentence and S2 a sentence, 'S1 la S2' is a conditional sentence, of the type of S2 and with its punctuation.
If S1 and S2 are sentences, 'S1 anu S2' is a disjunctive sentence, whose type and punctuation is calculated somehow (I think follows S2, but there seem to be other cases)
There are cases where sentences clearly go together, but cannot be compounded because they are either conjunctions and we can't use 'en' between sentences or because they represent English restrictive relative clauses: 'jan ni li kama tawa moku; ona li awen lon tenpo mun' or 'jan li kama tawa moku; jan ni li awen lon tenpo mun' (and other possible formats). Here they are separated by a semicolon rather than a period. This is significant for collapse:
'A li x B; A li y B' => 'A li x li y B'
'A e x B; A e y B' => 'A e x e y B'
otherwise, 'A x B; A y B' => 'A x en y B'
'A x B anu A y B' => A x anu y B'
Hence, 'A x ala anu x B?' => 'A x ala x B?' a yes/no interrogative sentence, answer '(A)(x)ala B' or '(A)x(B)'
'S anu seme?' is an open choice interrogative sentence, answer: any salient sentence, including especially S and its denial.
'Phrase o,' is a vocative and may be added at the very front of any sentence to give a new sentence, of the type of the one to which it was added.
'Phrase o, o Predicate' => 'Phrase o Predicate' i.e., the vocative 'o' and the imperative 'o' fuse. The vocative 'o' and the 'optative 'o' do not.
Some things about 'la'
'A preposition ({Phrase}) Phrase B'=> '(preposition ({Phrase})) Phrase la A B'
It is not perfectly clear what prepositions can be dropped: 'lon' apparently pretty much always and
always before 'tenpo', 'kepeken' pretty much always, others not so clearly.
'ni: S li x' => 'x la S' (very temporary, until 'ni: Text' gets sorted out)
'x pilin e ni:S' => 'x la S'
'A ni: Display B' => 'A ni B: Display' (ditto)
Building Phrase
Incorporation (not too useful in itself, but handy later on)
'A li Verbal ({Phrase1})n e Phrase2 B' => 'A li Verb ({Phrase1}){Phrase2} B'
'A li Verb ({Phrase1})n preposition ({Phrase3})Phrase2 B' => 'A li Verb ({Phrase1})n {(preposition ({Phrase3})) {Phrase2} B' (exactly when prep, etc. can be dropped needs to be worked out).
A completely incorporated predicate (no free standing objects or prepositional phrases) is indicated by <Predicate>.
Now the basic Phrase building rule is
'A x B; 'x li y' => 'A x {<y>} B'
In this case, the {} indicate not only 'pi' but, if x ends in a pi string, the insertion of a comma before {y}.
The other rule is the relic of the original 'pi' rule, that still sticks in many descriptions of 'pi'. The original 'pi' was a particle meaning “belonging/pertaining to” and could be used with any noun phrase, both as a modifier and as a predicate, so 'ni li pi mi' was legitimate, meaning “this is mine”. Handy though this might be, I doesn't work well in the current language, so we have the replacement rules
'x li jo e y'/'y li ijo {x}; A y B' => A y{x} B'
Incorporation plays another role in Phrase building:
'A ni: Subject li Predicate B' => 'A <Predicate>{Subject} B' if the origins sentence wee S, this new expression is the complete incorporation of S, <S>.
Quotation marks.
By convention (more or less adhered to – mainly less), double quotes are used to surround names not yet fully tokiponized: 'mi jan “John Clifford”'
Otherwise, single quotes are used (they type faster)
Old rant (hopeless, but I have to keep trying): When talking about a word, such as (notably) when saying it is someone's name, we have to use a name of that word (to avoid at least paradoxes and also to keep from talking nonsense). The standard name of a word is formed by enclosing that word in quotes, so for this word: nimi, we use this one: 'nimi' and so on. This quotation name is an unoffical word in tp and so needs a supporting noun, typically 'nimi' and occasionally 'toki'. Just how direct this support has to be is open to some interpretation, however. My totally correct answer to 'nimi sina li seme?' is '(nimi mi li) nimi 'Kipo''. However, pu allows 'nimi mi li 'Kipo'' (ignoring the regular quotation mark screw up), suggesting that noun in the subject carries over to the predicate in this case. By parity of reasoning, the other good answer 'mi li jan Kipo' would allow the variant 'mi Kipo', with the subject noun doing the support role. This is not confirmed – nor denied.
In texts with a lot of dialog, or even soliloquy, the constant repetition of 'x li toki e ni: Quote' is unsightly. So, here are some legitimate variants.
1. A quotation is the name of a bit of text, used to say that someone said it (or heard it or thought it or wrote it or knows it or saw it). Hence, the whole 'e ni: Quote' can be brought into a more standard sentence form as 'e nimi Quote', getting rid of the suspect 'ni: Display' for a messy, but clearly legitimate, Phrase.
2. We can turn this pattern inside out to 'nimi Quote li toki pi x', moving the quotation in front of the speaker, as is often done in natural languages.
3. Building on that, we can split 'x li toki e ni: ' Text1 Text2', to 'nimi 'Text 1' li toki {x}. kin la 'Text2'' This needs a little work, but helps quite a bit with the appearance of texts.
4. With rapid dialog with few intervening descriptive passages, the perennial 'x li toki e ni 'Quote'' can be reduced to 'x la Quote' (cf, 'pilin' in 'la' rules).
5. In really rapid give-and-take with two parties, it is probably enough to just start each new speaker on a new line, without keeping official score of who says what.
Speaking of tidying things ups a bit, there are some dangerous (because introducing many ambiguities) devices to break up the "monotony" of tp sentence forms. Incorporation can get those floating Prepositional, especially things like 'tawa mi' with 'toki' out of the way of the direct object, Topicalization moves a Phrase from and object ('e') position or a Prepositional to the very front:
'A Phrase B' => 'Phrase la A ona B' .
Another sort of connective
S1; <S1> li tan <S2>; S2 => S1 tan ni: S2
S1; <S2> li tan <S1>; S2 => S1. Tan ni la S2.
Basic sentence:
S => Subject li Predicate
Subject => o/Phrase
Phrase => Word/ Phrase{Phrase}/ Phrase en/anu Phrase / [ni: Display]
Word is ultimately any member of tp vocabulary except la, li, e, o, en, anu, kin, pi. Unofficial words
can also occur, subject to some special rules which need spelling out more precisely.
{} indicates that the enclosed is preceded by 'pi' if it contains more than one word
[] means I don't like this because it screws other things up, but I don't yet know how to deal with it.
Display => Text - / Quote
Text => Sm (i.e. one or more sentences)
Quote => 'Text'
Word => noun / verb / modifier / preposition ...
the tp vocabulary (with the noted exceptions) can represent almost any of these types (see individual entries)
Predicate => Verbal (e Phrase) (,Prepositional)n
Verbal => Phrase / Prepositional / modal ({Phrase})Predicate
Prepositional => preposition ({Phrase}) Phrase
items in parentheses need not occur. Items in parentheses with a following n may occur any number of times.
This grammar is overinclusive, but every simple tp sentence fits in it, with one lowest level obligatory transformation:
[o/mi/sina]Subject li Predicate => o/mi/sina Predicate
That is, 'li' is dropped after these three words standing alone
[]Subject means the enclosed item is the whole of the component Subject
In particular, this grammar does not account for the more or less fixed order of modifiers in Phrase and does not account for the various nature of the added Phrase. These are covered in later transformations (I hope).
Some definitions:
A sentence with [o]Subject is an imperative sentence.
A sentence containing 'seme' is a free choice interrogative sentence and ends with a ? instead of a period. The answer (to be discussed in detail later) is a Phrase replacing 'seme' or the original sentence with such a replacement.
If S is a declarative sentence, S? Is a true/false interrogative sentence, with the answers 'lon', '(lon)ala' or S or the denial of S.
Other simple sentences are declarative sentences and end with a period.
If S is a declarative sentence, 'o S' is an optative sentence, expressing a wish (etc.) that S be true. It ends with a period or !. [Note: this does not claim that the speaker actually has this wish, so an optative is neither true nor false, though it may be honest or not.]
Compound sentences
If S1 is a declarative sentence and S2 a sentence, 'S1 la S2' is a conditional sentence, of the type of S2 and with its punctuation.
If S1 and S2 are sentences, 'S1 anu S2' is a disjunctive sentence, whose type and punctuation is calculated somehow (I think follows S2, but there seem to be other cases)
There are cases where sentences clearly go together, but cannot be compounded because they are either conjunctions and we can't use 'en' between sentences or because they represent English restrictive relative clauses: 'jan ni li kama tawa moku; ona li awen lon tenpo mun' or 'jan li kama tawa moku; jan ni li awen lon tenpo mun' (and other possible formats). Here they are separated by a semicolon rather than a period. This is significant for collapse:
'A li x B; A li y B' => 'A li x li y B'
'A e x B; A e y B' => 'A e x e y B'
otherwise, 'A x B; A y B' => 'A x en y B'
'A x B anu A y B' => A x anu y B'
Hence, 'A x ala anu x B?' => 'A x ala x B?' a yes/no interrogative sentence, answer '(A)(x)ala B' or '(A)x(B)'
'S anu seme?' is an open choice interrogative sentence, answer: any salient sentence, including especially S and its denial.
'Phrase o,' is a vocative and may be added at the very front of any sentence to give a new sentence, of the type of the one to which it was added.
'Phrase o, o Predicate' => 'Phrase o Predicate' i.e., the vocative 'o' and the imperative 'o' fuse. The vocative 'o' and the 'optative 'o' do not.
Some things about 'la'
'A preposition ({Phrase}) Phrase B'=> '(preposition ({Phrase})) Phrase la A B'
It is not perfectly clear what prepositions can be dropped: 'lon' apparently pretty much always and
always before 'tenpo', 'kepeken' pretty much always, others not so clearly.
'ni: S li x' => 'x la S' (very temporary, until 'ni: Text' gets sorted out)
'x pilin e ni:S' => 'x la S'
'A ni: Display B' => 'A ni B: Display' (ditto)
Building Phrase
Incorporation (not too useful in itself, but handy later on)
'A li Verbal ({Phrase1})n e Phrase2 B' => 'A li Verb ({Phrase1}){Phrase2} B'
'A li Verb ({Phrase1})n preposition ({Phrase3})Phrase2 B' => 'A li Verb ({Phrase1})n {(preposition ({Phrase3})) {Phrase2} B' (exactly when prep, etc. can be dropped needs to be worked out).
A completely incorporated predicate (no free standing objects or prepositional phrases) is indicated by <Predicate>.
Now the basic Phrase building rule is
'A x B; 'x li y' => 'A x {<y>} B'
In this case, the {} indicate not only 'pi' but, if x ends in a pi string, the insertion of a comma before {y}.
The other rule is the relic of the original 'pi' rule, that still sticks in many descriptions of 'pi'. The original 'pi' was a particle meaning “belonging/pertaining to” and could be used with any noun phrase, both as a modifier and as a predicate, so 'ni li pi mi' was legitimate, meaning “this is mine”. Handy though this might be, I doesn't work well in the current language, so we have the replacement rules
'x li jo e y'/'y li ijo {x}; A y B' => A y{x} B'
Incorporation plays another role in Phrase building:
'A ni: Subject li Predicate B' => 'A <Predicate>{Subject} B' if the origins sentence wee S, this new expression is the complete incorporation of S, <S>.
Quotation marks.
By convention (more or less adhered to – mainly less), double quotes are used to surround names not yet fully tokiponized: 'mi jan “John Clifford”'
Otherwise, single quotes are used (they type faster)
Old rant (hopeless, but I have to keep trying): When talking about a word, such as (notably) when saying it is someone's name, we have to use a name of that word (to avoid at least paradoxes and also to keep from talking nonsense). The standard name of a word is formed by enclosing that word in quotes, so for this word: nimi, we use this one: 'nimi' and so on. This quotation name is an unoffical word in tp and so needs a supporting noun, typically 'nimi' and occasionally 'toki'. Just how direct this support has to be is open to some interpretation, however. My totally correct answer to 'nimi sina li seme?' is '(nimi mi li) nimi 'Kipo''. However, pu allows 'nimi mi li 'Kipo'' (ignoring the regular quotation mark screw up), suggesting that noun in the subject carries over to the predicate in this case. By parity of reasoning, the other good answer 'mi li jan Kipo' would allow the variant 'mi Kipo', with the subject noun doing the support role. This is not confirmed – nor denied.
In texts with a lot of dialog, or even soliloquy, the constant repetition of 'x li toki e ni: Quote' is unsightly. So, here are some legitimate variants.
1. A quotation is the name of a bit of text, used to say that someone said it (or heard it or thought it or wrote it or knows it or saw it). Hence, the whole 'e ni: Quote' can be brought into a more standard sentence form as 'e nimi Quote', getting rid of the suspect 'ni: Display' for a messy, but clearly legitimate, Phrase.
2. We can turn this pattern inside out to 'nimi Quote li toki pi x', moving the quotation in front of the speaker, as is often done in natural languages.
3. Building on that, we can split 'x li toki e ni: ' Text1 Text2', to 'nimi 'Text 1' li toki {x}. kin la 'Text2'' This needs a little work, but helps quite a bit with the appearance of texts.
4. With rapid dialog with few intervening descriptive passages, the perennial 'x li toki e ni 'Quote'' can be reduced to 'x la Quote' (cf, 'pilin' in 'la' rules).
5. In really rapid give-and-take with two parties, it is probably enough to just start each new speaker on a new line, without keeping official score of who says what.
Speaking of tidying things ups a bit, there are some dangerous (because introducing many ambiguities) devices to break up the "monotony" of tp sentence forms. Incorporation can get those floating Prepositional, especially things like 'tawa mi' with 'toki' out of the way of the direct object, Topicalization moves a Phrase from and object ('e') position or a Prepositional to the very front:
'A Phrase B' => 'Phrase la A ona B' .
Another sort of connective
S1; <S1> li tan <S2>; S2 => S1 tan ni: S2
S1; <S2> li tan <S1>; S2 => S1. Tan ni la S2.