Why there is only one entry for lexical item 'train' in noun category?

I have run the sentence ‘Express trains don’t stop at Princeton Junction’ and I am got the following output,
SENT: Express trains don’t stop at Princeton Junction.
[ LTOP: h0
INDEX: e2 [ e SF: prop TENSE: pres MOOD: indicative PROG: - PERF: - ]
RELS: < [ udef_q<0:14> LBL: h4 ARG0: x3 [ x PERS: 3 NUM: pl IND: + ] RSTR: h5 BODY: h6 ]
[ compound<0:14> LBL: h7 ARG0: e8 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: x3 ARG2: x9 ]
[ udef_q<0:7> LBL: h10 ARG0: x9 RSTR: h11 BODY: h12 ]
[ _express_n_1<0:7> LBL: h13 ARG0: x9 ]
[ _train_n_of<8:14> LBL: h7 ARG0: x3 ARG1: i14 ]
[ neg<15:20> LBL: h1 ARG0: e2 ARG1: h15 ]
[ _stop_v_1<21:25> LBL: h16 ARG0: e17 [ e SF: prop TENSE: pres MOOD: indicative PROG: - PERF: - ] ARG1: x3 ARG2: i18 ]
[ _at_p_loc<26:28> LBL: h16 ARG0: e19 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: e17 ARG2: x20 [ x PERS: 3 NUM: sg IND: + ] ]
[ proper_q<29:48> LBL: h21 ARG0: x20 RSTR: h22 BODY: h23 ]
[ compound<29:48> LBL: h24 ARG0: e25 [ e SF: prop TENSE: untensed MOOD: indicative PROG: - PERF: - ] ARG1: x20 ARG2: x26 [ x PERS: 3 NUM: sg IND: + ] ]
[ proper_q<29:38> LBL: h27 ARG0: x26 RSTR: h28 BODY: h29 ]
[ named<29:38> LBL: h30 CARG: “Princeton” ARG0: x26 ]
[ named<39:47> LBL: h24 CARG: “Junction” ARG0: x20 ] >
HCONS: < h0 qeq h1 h5 qeq h7 h11 qeq h13 h15 qeq h16 h22 qeq h24 h28 qeq h30 >
ICONS: < > ]

Upon going through the output I thought that the relation ‘_train_n_of’ is incorrect. I have tried few more sentences with lexical item ‘train’ and I noticed same. And to know what’s going on, I have checked the ‘lexicon.tdl’ file and found only one entry in noun category i.e., ‘_train_n_of’, which is irrelevant in many instances, especially ‘train’ in the sense of ‘a vehicle used for public transport’. I think you can include on more entry ‘_train_n_1’ excluding the preposition ‘of’.

Kindly, explain if there is any other reason,

Thank you,

The of sense indicates an optional prepositional complement, so it includes cases where there is no complement.

The complement is possible in the sense of a vehicle used for public transport, e.g. “a train of empty carriages”.

Thank you Sir, thanks a lot for your response. But I have few more doubts,

  1. I have seen few entries (about 47) in lexicon.tdl with both ‘_n_of’ and ‘_n_1’ relations. And I have tested few sentences with those entries (e.g., ‘He is sitting on the chair’) and found the output to be neat and clear without any confusion. My question is why there are few entries both with and with out ‘of’ prepositional complement.
  2. I have also seen a few entries with ‘_n_i’ (e.g., ‘_hold_n_i’) relation. I have searched for documentation, but I couldn’t trace any. My question is, are these entries representing some kind of idiomatic sense, please explain.
    Kindly clarify my doubts,

Thank you

For some words, prepositional complements correlate in some way with other syntactic behaviour. In the case of “chair”, the sense of being a chairperson (of something) can be used without an article, e.g. “she is chair” is grammatical only in the sense of being a chairperson, not a piece of furniture, even without an explicit complement. So “she is chair” is parsed with _chair_n_of rather than _chair_n_1. In general, the ERG will only make distinctions between senses where there is also a syntactic difference.

That’s correct. And I think you’re right that idiom documentation is currently missing from the ERG semantic documentation. I believe these predicates can trigger special processing rules (see: http://svn.emmtee.net/erg/tags/2020/idioms.mtr) but perhaps @Dan could say more about that.

Hi @guyemerson and @Rajini , thank you for this providential discussion; very related to my work on the glosstag corpus, the annotation and semantic parsing of the wordnet glosses (definitions and examples).

As you may guess, correctly handling multi-word expressions is a “pain in the neck”! In the current stage, I am trying to devise a proper approach to combine the sense annotation layer with the ERG semantic representations.

Idioms are among the kinds of MWE I need to deal with, and I haven’t paid attention to the _i quantifiers and predications before this discussion. The expression out of view is one that I found:

[1] OpenWordnet-PT Browser "out of view"

Wordnet documents only the adverbial sense of it at OpenWordnet-PT Browser 00510105-r but none of the three occurrences of the expression in the list above seems to be adverbial phrases. ERG analyzed the first example from [1] above as an idiomatic expression in all the 16 readings it produced. Wiktionary only says “out of view” is a prepositional phrase.


Maybe @Dan can add more explanations, or I may need to reread the “pain in the neck” paper. In particular, I would like to understand the introduction of the quantifiers and WHAT triggers the idiom reading in this case. Wordnet does not explicitly mark idiomatic senses. Could this particular sense of view be said to be idiomatic? On what grounds?

Once the quantifier is introduced, the variable it introduces may be the best place to plug the sense associated with the expression. I found cases where we have idiom_q_i linked to predicates without the _i suffix and cases where idiomatic predicate senses were used. I didn’t find occurrences of _i predicates without properly introducing the variable with idiom_q_i.