Request for sample input for Jacy and Zhong

Perhaps someone will feel generous and help me out here.

I’d like to see some analyses of questions, polar and/or constituent, from Jacy and Zhong.

In particular, I am interested in:

A. How the question particles were implemented and whether there was a need in a feature which basically means that a WH-word is present in the clause and so the particle is not possible there (Mandarin?)

B. I am also interested in whether there is any contrast between ka and no that was implemented in Japanese with respect to main and embedded clauses (some literature says ka is only possible in embedded questions, in informal speech).

C. How do no and ka in Japanese (and perhaps there is something similar in Mandarin?) interact with the subordinator to? Can both occur in the same clause?

Now, looking at Jacy and Zhong would be easiest tif I had some input sentences ready and would then start from exploring the parses (as opposed to exploring the grammar files generally). Alas, I do not speak Japanese or Mandarin, and I do not even know the alphabets.

I have some Japanese sentences which I found in the literature or constructed, using Latin alphabet.

(1) Would someone be so kind to provide me with analogous input sentences which are supposed to be parsed by Jacy?

(2) Would someone be so kind to supply analogous or similar sentences in Mandarin which are supposed to be parsed by Zhong?

If you simply comment on the grammaticality judgments below, that would also be helpful on its own. If you have time to only provide me with one sentence, please do!

Thanks very much in advance.

(1) Kim wa  nanio    yonda no
    Kim NOM what.ACC read  Q
  What did Kim read?

(2) Kim wa  nanio    yonda
    Kim NOM what.ACC read 
  `What did Kim read?

(3) Sandy wa  Kim ga  nanio    yonda no sitteiru
    Sandy NOM Kim NOM what.ACC read Q   know
   What did you know that Kim read? (Or does this mean "Do you know what Kim read"?)

(4) dare ga  kuru
    who  NOM came
    Who arrived?

(5) dare ga  kuru no
    who  NOM came Q
    Who arrived?

(6) *dare ga kuru ka
     who NOM came Q
    Who arrived?  -- I am told this is not possible in informal speech?

(7) Sandy wa  Kim ga  nanio    yonda ka sitteiru
    Sandy NOM Kim NOM what.ACC read  Q  know
   What did you know that Kim read? -- This, however, is possible, in an embedded clause?

(8) Sandy wa  Kim ga  nanio    yonda sitteiru
    Sandy NOM Kim NOM what.ACC read  know
   What did you know that Kim read? (Or does this mean: Do you know what Kim read?)

Finally, is something like this possible and what would it mean?

(9) Kim wa Sandy ga nanio yonda to sitteiru

(10) Kim wa Sandy ga nanio yonda to sitteiru no

(11) Kim wa Sandy ga nanio yonda ka to sitteiru 

Hi Olga, I’m not sure how helpful this is and I don’t know Japanese or Chinese to test this myself, but some of these analyses might be available in the demo grammars here: http://delph-in.github.io/delphin-viz/demo/

I used a transliterator for ‘dare ga kuru’ (so I’m not sure if this is right) and was able to parse だれ が くる in the demo Jacy grammar.

1 Like

That is very helpful Kristen; which transliterator?

I just googled and found this: https://www.translitteration.com/transliteration/en/japanese/hepburn/
So I can’t vouch for it, but it seems to work

1 Like

That’s awesome; thanks. It does seem to work. I will just need some other nouns, not Kim and Sandy but something that actually is in Jacy. That should not be too hard to find.

I see that Jacy licenses (6) which does not surprise me much; I was really not sure about the claim that it is not possible. Would still love to hear a native speaker’s and a Jacy developer’s input of course.

Hm, doesn’t seem like I can just grab something from lexicon.tld (from Jacy) and plug it instead of my Kim and Sandy successfully. Must be missing something. So, still hoping for some help :).

Here’s the sentences in Japanese, with random first names pulled from the Jacy lexicon:

(1) 秋高 は 何 を 読ん だ の
Kim wa nanio yonda no
Kim NOM what.ACC read Q
What did Kim read?

(2) 秋高 は 何 を 読ん だ
Kim wa nanio yonda
Kim NOM what.ACC read
`What did Kim read?

(3) 知恵 は 秋高 が 何 を 読ん だ の 知っ て い る
Sandy wa Kim ga nanio yonda no sitteiru
Sandy NOM Kim NOM what.ACC read Q know
What did you know that Kim read? (Or does this mean “Do you know what Kim read”?)

(4) 誰 が 来る
dare ga kuru
who NOM came
Who arrived?

(5) 誰 が 来る の
dare ga kuru no
who NOM came Q
Who arrived?

(6) 誰 が 来る か
*dare ga kuru ka
who NOM came Q
Who arrived? – I am told this is not possible in informal speech?

(7) 知恵 は 秋高 が 何 を 読ん だ か 知っ て い る
Sandy wa Kim ga nanio yonda ka sitteiru
Sandy NOM Kim NOM what.ACC read Q know
What did you know that Kim read? – This, however, is possible, in an embedded clause?

(8) 知恵 は 秋高 が 何 を 読ん だ 知っ て い る
Sandy wa Kim ga nanio yonda sitteiru
Sandy NOM Kim NOM what.ACC read know
What did you know that Kim read? (Or does this mean: Do you know what Kim read?)

Finally, is something like this possible and what would it mean?

(9) 知恵 は 秋高 が 何 を 読ん だ と 知っ て い る
Kim wa Sandy ga nanio yonda to sitteiru

(10)知恵 は 秋高 が 何 を 読ん だ と 知っ て い る の
Kim wa Sandy ga nanio yonda to sitteiru no

(11)知恵 は 秋高 が 何 を 読ん だ か と 知っ て い る
Kim wa Sandy ga nanio yonda ka to sitteiru

1 Like

I should say that involves some guessing as to where the white space is … let me know if you have problems.

(3), (7), (8) – there’s no ‘you’ here, because both clauses have over 3rd person subjects.

Regarding (3) — there is no marker of that as a Matrix polar question, it would have to be intonation. I’m not sure about no there on the embedded clause, but I’m not a native speaker!

Regarding (7) — if you want that to be a matrix polar question, the ka is completely out of place on the embedded clause. Or rather, it would have to mean something like “What does Sandy know whether Kim read”, and I don’t think it can.

Thank you, Emily!

From your sentences, I can parse (1) and (2) which is an improvement as I did not have parsable versions of those before. But I cannot parse any of the others. (But I still have (4)-(5)-(6) if I simply transliterate using Kristen’s link).

At least the ones with kuru had to do with the segmentation (fixed above now).

Hi, Olga…

The testsuite of Zhong (Mandarin Chinese) is here.

These do not include the WH-words, but there are several related items including the A-not-A and topic-comment constructions.

Probably, I think you can refer to Zhenzhen’s thesis.

The wh-questions are dealt with in Section 5.8.

Sanghoun

1 Like

Thank you @sanghoun!

Zhenzhen’s thesis section 5.8 is very nice and clear.

So in Zhong, QUE is used to indicate that a question is a constituent question, and a unary rule is used to ultimately license the question. That is also what I am doing in the Matrix right now.

Then, in Zhong:

To prevent attaching MA to content questions, the
NON-LOCAL.QUE of the complement must be empty, indicating it’s not already a question.

What I don’t understand however is, what is the QUE value on the mother of the unary question-phrase rule? Is it not empty? If it is empty (and I think it is, from looking in Zhong), then why doesn’t MA attach to it? (And if it were not empty, then what prevents the question-phrase rule from applying to itself?)

I am going to look in Zhong now, I think I have parsable examples from Zhenzhen’s thesis.

But I wonder which grammar to load? In the Zhong repo that I got from github there is (1) cmn -> zhs; (2) cmn -> zht; (3) yue

Does anyone know?

Regarding MA and QUE … perhaps there’s something else that prevents MA from attaching to the mother of the in situ rule, that is, the in situ rule has to be the highest thing?

As for which grammar: zhs is probably Mandarin with simplified characters, zht is Mandarin with traditional characters. If your word for ‘who’ looks like 誰 you want zht. If it’s 谁 you want zhs. Likewise, the question particle ma 嗎 in zht and in 吗 zhs. I think yue is the Cantonese grammar.

1 Like

OK, here’s what I found about Zhong.

Here’s what seems to prevent MA from attaching to a constituent question:

So it does not look like it actually is QUE.

As for ma/not-ma, there is a hierarchy in Zhong:

spart := avm.
not-crs := spart.
not-ma := spart.
crs := not-ma.
ma := not-crs.
no-spart := spart.
not-crs-not-ma := not-crs & not-ma.

And then here’s the question particle:

; Subtype for question particles. Constrains SF to ques directly. MA.
qpart-lex-item := spart-lex-item & nonque-item &
  [ STYLE robust,
    SYNSEM [ SPART ma,
             PUNCTUATED -,
	     LOCAL [ CAT.VAL.COMPS < [ SPART not-ma,
                                       LOCAL.CAT.VAL.SPR olist,
                                       NON-LOCAL.QUE 0-dlist ] >,
		     CONT.HOOK.INDEX.SF ques ] ] ].

And finally question phrase:

;SPART ma to stop MA being combined with questions
;using QUE 1-dlist now since 1-plus-list doesn't give consistent results yet.

question-phrase := unary-phrase-super & headed-phrase &
"""
<nex>他 买  苹果
<ex>谁 买 了 苹果
<ex>他 买 了 什么
<ex>谁 买 了 什么
<ex>他 哭 不 哭
<ex>他 吃 不 吃 苹果
<nex>他 吃 不 吃 苹果 吗
"""
  [ SYNSEM [ LOCAL [ CAT [ MC +,
			   HEAD +vjc &
			        [ FORM finite ],
                           VAL #val & [ SUBJ < >, 
                                      COMPS < > ] ],
                     CONT.HOOK [ LTOP #ltop,
			         INDEX #index & [ SF ques ],
			         XARG #xarg ] ],
             NON-LOCAL [ SLASH  #slash,
                         QUE 0-dlist & [ LIST < > ],
                         REL #rel ],
             SPART ma ],
    C-CONT [ HCONS <! !>,
	     ICONS <! !>,
             RELS <! !> ],
    ARGS < #hdtr >,
    HEAD-DTR #hdtr & 
             [ SYNSEM [ LOCAL [ CAT.VAL #val,
                                CONT.HOOK [ LTOP #ltop,
				            INDEX #index & [SF prop-or-ques],
				            XARG #xarg ] ],
                        NON-LOCAL [ SLASH #slash,
                                    QUE 1-dlist,
                                    REL #rel ] ] ] ].