470 likes | 480 Views
CxG Lecture 2a. LSA Presession Construction Grammar. http://www.icsi.berkeley.edu/construction-grammar-course/. Oops! We thought the link to this URL was on the LSA page for our course. Apparently it isn’t. It now contains yesterday’s PPT presentation.
E N D
CxG Lecture 2a LSA Presession Construction Grammar
http://www.icsi.berkeley.edu/construction-grammar-course/ • Oops! • We thought the link to this URL was on the LSA page for our course. Apparently it isn’t. • It now contains yesterday’s PPT presentation.
At present, solar power is at least two to three times as expensive as the typical electricity generated in America for retail customers. at present in the past in (the) future [at present] [at least] [two to three] times as expensive as X at least [two to three] times as expensive as X at least [two to three times] as expensive asX [at least two to threetimes] [as expensive] [as X]
Knowing and Figuring Out • There’s a difference in using a language between the things that you have to know outright, i.e., expressions whose properties have to be stipulated, and those things that you can figure out on the basis of other things that you know. • Is that a clear distinction? Can it be made clearer?
green hat (again) • Stipulated: • the meaning, form, and combinatorial properties of of “hat” • the meaning, form, and combinatorial properties of “green” • the construction by which adjectives can modify nouns • Figure-out-able: • the meaning, form, and combinatorial properties of “green hat”
Encoding, Decoding, and Motivation • Decoding mode: Somebody says to me, “I’ll come back next week”. There’s no contextual reference to any particular week, so I figure out that “next week” means the next week coming up. • Encoding mode: I need to know how to refer to the week following the present week. That is, I need to choose the exact phrase “next week” rather than, say, “following week” or “upcoming week”.
Encoding, Decoding, and Motivation • Motivation: As native speakers we can know there’s a reason why the expression is “next week” rather than “Kim’s birthday” • (the sentence stem used for saying this is “It’s no accident that …”), • but we can still recognize that the meaning and use of “next week” has to be stipulated.
Encoding, Decoding, and Motivation • It’s probably easier to see the point in the case of “last week”. “Next” is used to mean “following” in both the deictically and the anaphorically anchored expression: • “next week” vs. “the next week”. • But in the other direction, “last” is not the word we’d use for the preceding week. The phrase “the last week” can’t simply mean ‘the week preceding the week we’ve had in mind’.
wherewithal, the “sufficiency” construction • Somebody says • I don’t have the money to buy an iPhone • We know what it means, we feel there’s a connection with what we know about “the”, and so we see the construction as motivated. • Question: Could we have known that “this is a conventional way to say it” - encoding mode - merely on the basis of our knowledge of the meanings of have, the, money, and the infinitive?
Millions of second-language speakers of English whose mother tongues do not have articles are mystified by the uses of “the”. • Instead of intimidating them further by saying that our native-speaker knowledge of definiteness is subtle enough for the wherewithal facts to simply and naturally “follow from” that subtle knowledge, why not just describe the pattern, explaining that the word “the” is used with the “resource noun” in the way we described. • Okay, that’s not an argument.
Reviewing Yesterday • A sign is a fully specified linguistic object (word, phrase, sentence, …). • Every sign is either a lexical item fresh out of the lexicon, or a “mother” of a construct • A construct is an assembly of a mother sign and one or more daughter signs, and each well-formed construct in the language is licensed by one or more constructions.
the boy’s green hat • the boy’s green hat is a sign licensed by a construction that joins “the boy’s” with “green hat”. Each of these has to be a mother of some construct. • the boy’s is a sign licensed by a construction that attaches the genitive clitic to “the boy” • the boy is a sign licensed by a construction that joins “the” with “boy” • green hat is a sign licensed by a construction that joins “green” with “hat”
the kind of thing you might have done if you’d been ask to draw a phrase-structure tree of the phrase, except that each “mother” is not a grammatical category name but a sign that represents the product of the construction
said more carefully • the boy’s green hat is licensed by a construction that joins the kind of thing “the boy’s” is with the kind of thing “green hat” is • The properties of the full phrase are determined by the construction, by principles related to the construction, and by semantic and phonological principles for integrating the meaning and pronunciation of the specific components.
Locality - “no peeking” • A consequence of this view is that all of the knowledge needed for a construction has to be available in the component signs. • The construction grammar of some decades ago made a point of arguing that some constructions couldn’t be described as “nuclear families” but required “extended families”: granddaughters, nieces, etc.
Prevent NP from VPing VP V NP PP VP from … Ving
Prevent NP from VPing VP V NP PP syn {cat v, infl ing} VP from … Ving
Prevent NP from VPing VP V NP PP VP from … Ving
prevent NP from VPing VP syn {cat p, lexID from2} V NP PP prevent VP from … Ving
prevented me from leaving prevented me from leaving prevented me from leaving The VP structure that licenses this, built on the valence of prevent, takes V + NP + PP/LexID=from2
Other uses of LexID? • How do we recognize the collocations in “take unfair advantage of the situation” (and hundreds of similar things)?
V + NP/LexID=advantage + PP/LexID=of VP V NP PP take A N P NP advantage of
Head feature conventions • The principle that certain features of the head of a phrase get projected to the mother can include the lexical identity of the head daughter. • This mimics an ability that Dependency Grammar has, of showing chains of head-to-head links. • Consider support verbs: take a long bath, make a good impression, wreak terrible havoc, etc. • Coordinate conjunction wreaks havoc with this solution, I think.
Locality problems • There are some constructions that we offered as making a locality commitment difficult: WXDY, conditional sentences, and correlative conditionals.
be X doing what Y “What’s X doing Y?” Constraints: “be” is finite dependent is “doing” object is “what” Y is “secondary predicate” Meaning: X is Y that is anomalous, needs explanation
be X doing what Y • BE+DOING does not have ‘progressive’ meaning. • What’s that scratch doing on my violin? • WHAT has to be initial in the clause • What were you doing in my kitchen? (‘why were you…”) • You were doing what in my kitchen? • X+BE relation is available for linearization through other constructions. • What’s she doing with my shoes on? BE+X • I wonder what she’s doing with my shoes on. X+BE • Modifications and elaborations are possible • What in heaven’s name are you still doing out there?
be X doing what Y • Y can be locative • What are you doing in my kitchen? • Y can be adjectival • What are you doing naked? • Y can be adjectival PP • What’s that flag doing at half mast? • Y can be a with/without absolute • What are you doing without any shoes on? • Y can be participial • What are you doing standing out in the rain?
Conditional sentences - briefly • There’s a huge literature on conditional sentences. One careful study in the cognitive linguistics framework is that of Eve Sweetser and Barbara Dancygier, Mental Spaces in Grammar: Conditional Constructions. • One feature that interests me is the semantic consequences of paired verbal forms between the two clauses.
Set aside pragmatic conditionals: • if you’re hungry, there’s a gallon of ice-cream in the fridge • if I don’t see you again, have a nice trip • and epistemic conditionals: • if the patient dies, I gave him the wrong medicine • and stick with “straight” conditionals, conditionals that imply some kind of causal connection between two events or situations.
Some types: • If you eat it, you’ll die. • If you ate it, you’d die. • If you had eaten it, you would have died. • If she’s here, she’ll help us. • If she were here, she would help us. • If she was here, she helped us. • If she had been here, she would have helped us. (when?)
Some issues • How to represent the TAM features in the two clauses (Tense-Aspect-Modality). • How to associate such pairings with the right interpretations.
Some issues • AND THEN, how to do that when the clauses are not sisters: • If you had eaten it, I think you would have died. • If you ate it, my guess is that you’d die. • And what about “then” • If you are sane, then I think you should stop buying Diesel goods • Me, I’m just pointing out the problems.
The Correlative Conditional • a.k.a. the comparative conditional, the correlative comparative, etc. the more I drink, the more charming I become
Name dropping • Fillmore 1986 • Fillmore Kay O’Connor 1988 • McCawley 1988 • Michaelis 1994 • Culicover & Jackendoff 1999 • Borsley 2004 (details will be added on the course website)
Several versions • the more the merrier • the older the man, the longer the kilt • the more people come, the better • the bigger they come, the harder they fall • we’ll look at the full clause versions
Each part begins with the followed by a comparative expression. • The comparative expression is a left-isolated (extracted) argument or adjunct of its clause. • The faster we drive, the sooner we’ll get there.
The comparative can be of a variety of categories: • complete NP: the more we eat … • quantifier in NP: the more beans we eat … • adjective: the older they are … • adjective in NP: the bigger a box they request … • adverb: the faster we drive • adverb: the more carefully I explain it …
LDD • The compared element can have a long-distance relation to its predicate: • the more books [they try [to get me [to read __]]]
NegPol • The first clause is a negative polarity context, the second clause isn’t: • So the sooner anyone who believes this stops breathing, the better off the rest of us will be! • (Of course the second clause could have its own polarity trigger, the comparative less.) • …, the lessanyone would be interested.
SAI • The second clause, but not the first, permits an inverted auxiliary. Hard to find. • I have been taught to reverence and revere my departed ancestors. The sooner you have the good sense to join them, the sooner will I revere thee. • The less we seek to fathom what He has been pleased to conceal from us, the sooner will He vouchsafe its revelation to us through His divine spirit.
THAT • The first clause allows the that-complementizer but the second (subject to the Google curse) doesn’t. • The sooner that privatization happens, the sooner water customers will have reliable water on a consistent basis. • The sooner that we can detect it, the sooner we can begin to treat it.
THEN • The second clause allows an initial “then”: • The sooner we can accept the fact that this is a war, then the sooner we can get about the task of winning it. (National Review Online) • The sooner you register, then the sooner you’ll start to get some rewards.
COP-DEL • If the comparative phrase is the complement of the copula, the copula can be omitted in both parts. • The older the man, the longer the kilt. • The more intelligent the students, the better the marks. (Borsley example)
{TAM-protasis, TAM-apodosis} • The TAM pairings in the two clauses match SOME of those found in conditional sentences, but NOT those related to counterfactuals. • the more he drinks, the more charming he becomes • the more he drinks, the more charming he’ll become • the more he drank, the more charming he became • *the more he had drunk, the more charming he would have become
Semantics • The semantics determined by the construction is a monotonic correlation between an independent variable, expressed by the first clause, and a dependent variable expressed by the second clause. • the more I listen to this, the more bored I get
Summarizing:Semantics similar to conditionals.TAM pairings similar to non-CF conditionals.
I have no idea what a locality-respecting solution could look like.