"The man who came to dinner stayed" (I think the modern rule -- i.e., since I was in High School -- is "the man that came to dinner") shows an example of a restrictive relative clause, "who came to dinner", not set off by commas. It is a relative clause because of the "who" which serves as subject of the clause and yet connects it to the earlier "the man". It is restrictive because it is a part of the specification of which man it is we are talking about, rather than adding information about an already identified man. It contrasts with a non-restrictive relative clause, as in "The man, who came to dinner, stayed", where it is assumed that we already know who the man is and are just adding more information. It is, then, factually -- if not rhetorically -- interchangeable with "The man, who stayed, came to dinner", while a similar shift would not work in the first case, since "the man who stayed" might very well be someone other than the man who came to dinner in terms of identification.
In English, the difference between restrictive and non-restrictive is the commas -- and a different voice pattern, we hope. In Logic, the difference is more marked: the relative clause in the restrictive case is part of the description, "the x (x Man & x came to dinner) stayed". while the non-restrictive cases is merely a parallel sentence or predicate: "The x(x Man ) stayed & the x(x Man ) came to dinner" (or, in collapsed form, "The x (x Man )(stayed & came to dinner)"). In tp there need be no distinction at all: 'jan li kama tawa moku li awen' or, in an effort to make the distinction, 'jan li kama tawa moku; jan ni [or 'ona'] li awen'. The 'jan ni' form is closer to restrictive, the 'ona' to the non-restrictive, but neither is clearly one or the other. It does seem, however, that this distinction is made, somehow, in every language and is important (for us Montagovian, if no one else). Another version is also possible: 'jan ni li awen: jan/ona li kama tawa moku', which feels closer to the restrictive ideal, even if the restriction is mentioned later. Indeed, at one point, when 'pu' was still an hypothetical word said to be some sort of punctuation, it was suggested that it step in here for the connection in this case and other 'ni' -':' connections (the only obviously missing punctuation in tp). The positioning of 'pu' elsewhere stpped this line of thought, which still, in this case, was rather unsatisfactory, since the restriction is remote and perhaps misplaced.
And, besides, the restriction is a property of the thing in question, so ought to go into the modifier chain somewhere (close, in the speciation position, not just in the property slots, ideally). So, that gives 'jan pi kama moku [or kama pi tawa moku' to avoid the potential food reading] li awen' . But what if the property involved does not take our target as subject "The man that I saw yesterday is injured". 'mi lukin e jan ni lon tenpo suno pini; ona li pakala', etc. but (tp has no passives) nothing in the modifier string directly.
Stephan Schneider, pursuing his program of learning tp by trying to do everything a totally different way, hit upon the notion that, since PPs, which are also sometimes VPs, could be inserted into modifier strings with just a 'pi', why couldn't other VPs, so 'jan pi kama tawa moku li awen' That didn't present a problem, but something like "The man who ate the cow died" 'jan pi miku e soweli li kama moli.", went against a rule we had been enforcing against others who had tried this more or less accidentally: "'e' can't occur in a noun phrase" or something like that.
sonja stepped in at this point and suggested the possibility that a predicate could occur in NP and so the 'e' would be justified as in the predicate. but that meant that the line would have to be 'jan pi li moku e soweli li kama moli.' This does trick the eye used to finding the 'li' and dividing the sentence at that point, but it does not otherwise screw up the grammar (as the 'pi PP' cases show). Unfortunately, it does not go much further than this, since getting the subject of the inserted predicate into the modifier string does break the patterns too much: 'jan pi mi li lukin e ona li kama', burns out the parser circuits, however modified in ways so far conceived. As long as the target object is subject of the predicate we could (this is not yet a permission and certainly not a recommendation) insert the predicate into the modifier string (what happens to the other modifiers we leave to later investigation). But if the expression is an object of any sort, DO or OP, this approach cannot be used and we are left -- until someone cleverer comes along -- with the two sentence solution and a feeling of dis-ease.
Last edited by janKipo
on Fri Dec 18, 2015 2:09 pm, edited 1 time in total.