Download The Role and Behavior of 'Do' in English Modal Verbs and more Papers German Philology in PDF only on Docsity!
Some
Do
s and
Don’t
Carson T. Schützes in Child and Adult Grammars
UCLA
[email protected]
Linguistics Colloquium
UMass Amherst
September 20, 2002
0. Abstract This talk focuses on the syntax of dummy ‘do’. An analysis of standard English is proposed that
acquisition facts include children’s early non-adult use of “don’t” with 3‘tun’ in colloquial German and neighboring Germanic dialects, and facts about English acquisition. Thewill be used to also account for historical and dialectal variation, the related phenomenon of periphrastic
person singular subjects,rd^
where I will contrast my approach to INFL underspecification with that of Guasti and Rizzi (2002). The central claim of the analysis is that ‘do’ does not belong under the Tense head, nor does it
- “dummy” (auxiliary, periphrastic) 1.1 Reminder of the basic facts 1. Introduction which free variation of the sort found with ‘tun’ and in child English is expected.“last resort” restriction on ‘do’ is argued to be an historical accident, and a model of INFL is sketched intask is to determine which allomorph of Indicative to spell out in a given environment. The apparentexecuted by the conditions on Vocabulary Insertion (in the sense of Distributed Morphology), whosespecifically an allomorph of Indicative Mood, whose other allomorph is zero. Its “support” function isarise by adding structure in the Spell-Out component. Rather, ‘do’ is an expression of a Mood head,
(^) do 1 in present-day standard English shows up if…^
Subject-Auxiliary Inversion (SAI):• one (or more) of the following occurs:
(^) Does Chris like pizza?
VP-ellipsis:
(^) Chris likes pizza, and Mary does too.
Sentential negation:
(^) Sue does not like pizza.
Prosodic emphasis on the truth of the sentence (aka verum focus; Ac in
LSLT
A particle emphatically marking truth of sentence: Mary DOES like pizza, you dunderhead.
(^) Sue does SO like pizza.
(^) Does NOT.
Does SO! Does NOT!! Does TOO!!!
[Note: hereafter, the latter three are grouped as
(^) Σ (^) heads (aka Polarity heads)]
*Will does Chris like pizza?, *Does Chris will like pizza?)
, and
- the (potentially) finite verb is not
(^) be (^) or auxiliary
(^) have (^) ( *Sue does not be on a diet
, etc.)
Under these conditions
(^) do (^) must (^) be inserted (
*Chris likes pizza, and Mary likes too.
Otherwise (^) do must NOT (^) be inserted (
*Chris do ( es like pizza.
)
Not under discussion: pro-VP forms such as^1
(^) do so, do the same
, etc., which involve main verb
(^) do.
(^) do (^) is called upon to support INFL (Tns+AgrS) when INFL cannot get together
with a verb; perhaps it also supports emphasis and
(^) n’t (as clitics??); it is the last resort for
supporting these elements, when other potential hosts are not around
Questions to be answered: 1.2 Where we are going
Does (^) do correspond to any syntactic position, or is it introduced ex nihilo in Spell-Out?
What ensures that it’s not there when it isn’t needed?2) What ensures that it’s there when it’s needed?
Roadmap of the path to answering them:
- argue on distributional grounds that
(^) do is a syntactic head of a particular category
- the relevant head is always there, it just doesn’t always have
(^) do (^) under it—contextually
acquisition theory, to examine one of children’s non-adult-like uses ofIntermezzo: use the analysis developed thus far, combined with an underspecification-basedgoverned allomorphy, in a Late Insertion model
(^) do
presupposition of question 3 is false: from a broader perspective 3) TRICK QUESTION, which will lead us to data beyond standard English that argue the
(^) do CAN be there when it isn’t
- Why does standard English make it look like—> Replacement questions:needed
(^) do is strictly a last resort?
- How can we capture optionality of
(^) do found e.g. in historical data and another of children’s
non-adult-like uses?
2. How does For much relevant background and like-spirited alternative proposals, see Déchaine 1993
(^) do fit into the tree?
- one standard answer: generated under Tense; another conceivable answer: same as auxiliary 2.1 Heads: the possibilities
(^) be (^) and
have , however they work
- one quick reason to reject the former: the forms of
(^) do (^) contain the regular 3sg and past INFL affixes,
[du + Ø], [dwhich suggests the stem is something separate:
√ (^) + z], [d I (^) + d]
- proposal (not original to me, but more arguments here): that other thing is a Mood (M) head (cf. Roberts 1993, Culicover 1999, i.a.)
- evidence: head and a Tense (^) do (^) is distributionally equivalent to the English modals, which I claim also contain a Mood
(^) morpheme—let’s first establish this last point (^2)
(1) - sequence of tense a. Right now I cannot do it.
b. Yesterday I could not do it. (cf. *Yesterday I cannot do it.)
Hereafter, unless the distinction is crucial, I use Tense where I should really be saying INFL or T+Agr.^2
a. Nowadays a student will ask the most ridiculous questions if given the chance.
b. In the good ol’ days a student would ask only sensible, important questions.
a. He says he will be late.
b. He said he would be late.
This is not to deny that there are nonpast uses of
(^) could, would
, as well as historically parallel
should, might
; but in this respect the modals are not special: counterfactual conditionals are also
formed with (what look like) past tense forms (
If I knew the answer, …
); cf. Giorgi and Pianesi
readings in (1b)–(3b) are transparently formed.1997. The special (nonpast) meanings of some modals are no reason to deny that the past
- also, the past tense modal forms all end in standard past suffixes
(^) –t (^) or (^) -d
- OK, so (^) do (^) and (^) modals (^) have the structure Mood+Tense
- this makes
(^) do (^) different from finite auxiliary
(^) be / have —whatever they are, they are not Ms; elsewhere I^3
argue they are categorially V
[Note: A fact we will incorporate later is that modals do not take 3sg
(^) -s ; on my account this is not a
deep property of modals, rather a shallow property of
(^) -s .]
2.2.1 Lots of ways in which 2.2 The evidence
(^) do patterns with modals, as against
(^) be/have
- traditionally English modals are said to lack nonfinite forms, or better, can’t appear in nonfinite do (^) tracks modals and these two diverge from auxiliaryclause type of the language, whether it’s in or out; so let’s just look at the data, and see whetherenvironments; but there’s no definition of finiteness around precise enough to tell us, for every
(^) be/have
i) an arguably criterial property of English modals: they are mutually exclusive with
(^) do (4a). This mutual
exclusivity is of course one reason for wanting to place
(^) do (^) in the same class with them; it does
not extend to
(^) be (^) and (^) have (^) (4b).
b. a. *must do (not) go, *do (not) must go, *do must not go must have gone, must be eating, must be in Tulsa
ii) moreover, numerous clause types exclude
(^) do (^) and modals while allowing
(^) be (^) and (^) have .
a. Subjunctives It is vital that John be here on time.
b. It is vital that John be smiling in the photograph.
c. It is vital that Rover have eaten before we arrive.
d. *It is vital that John do not be late.
e. (^) *It is vital that John will not come unprepared.
a. To -infinitives It is important (for everyone) to be on time.
I use “auxiliary^3
(^) be/have ” as sloppy shorthand for ‘all those uses of
(^) be (^) and (^) have (^) that in finite clauses raise across negation’,
thus including copular
(^) be , British English possessive
(^) have , etc.
b. It is important (for a movie star) to be smiling whenever the paparazzi are nearby.
c. It is important (for every applicant) to have finished high school.
e. (^) *It is important (for us) to can be alone.d. (^) *It is important (for us) to do not leave her alone.
a. Small clauses I made him be alone for a while.
b. The director made us be dancing when the curtain opened.
e. (^) *The therapy made her can/could walk again. (cf. The therapy made her be able to walk again.)d. (^) *The conductor made us do not sing the harmony line.c. (^) ?The coach made her not have just eaten when she came to practice.
a. Mad Magazine sentences (see Akmajian 1984) What?? Her be out all night??? That’s inconceivable!
b. What?? Him be drinking at 9 in the morning??? When pigs fly!
e. (^) *What?? Him should/must/could leave the firm?? Never in a million years!d. (^) *What?? Him do/does not pick up the kids on time??? Unthinkable!c. (^) ?What?? John not have finished his homework by 9pm?? Absurd!
a. Why (not) Why (not) be a responsible citizen?
b. Why be working when you could be partying on the beach?
c. (^) ?Why not have made the appointment with her before she has a chance to make one with you? 4
e. (^) *Why should/must stay home? (cf. ?Why be obliged to stay home?)d. (^) *Why do not go to the beach?
a.^ Gerunds b. (^) (*John being drinking was not a surprise.) (^) John being unable to complete 50 pushups is embarrassing.
5
c. John having completed 50 pushups in 45 seconds is impressive.
e. (^) *John canning not complete 50 pushups is embarrassing.d. (^) *John doing not like opera surprises me.
iii) (^) do verb (cf. Rapoport 1987), even though it supports tense morphology:^ has in common with modals that it cannot satisfy the requirement that a finite clause contain a
(^) a. c. (^) *John will (not) tired/out of luck. (cf. I consider John tired/out of luck)b. (^) *John doesn’t out of luck. (^) *John does (not) tired.
(^) a. b. (^) John will be tired/out of luck. (^) John is tired/out of luck.
(^) do hosts Tns but doesn’t save the sentence;
(^) be (^) saves the sentence even when it
doesn’t host Tns
This sentence is already blocked as a violation of the Doubl-ing restriction.^5 Example from Wachtel (1979), contra Akmajian, Steele and Wasow (1979).^4
- M[indic] has two allomorphs:
(^) do (the stem), and Ø [conceptually close to accounts based on
(^) do
deletion, e.g. Emonds (1970)]
- certain (^) Σ (^) heads prevent T & V from uniting, stranding T affix
(22)- vocabulary items look like this: [epist, necess] <—>^ Paradigm for M
(^) must
[(indicative) …
] <—>^6 (^) do (^) /___ affix/clitic (something that needs morphological support)
[(indicative)]
(^) <—> Ø (elsewhere)
(^) Pat runs
: (^) [ MP (^) Pat Ø [ TP (^) –s [ ΣP (^) Ø [ VP (^) run ]]]]
(^) Pat does not run
: [ MP
(^) Pat do [ TP (^) -s [ ΣP (^) not [ VP (^) run ]]]]
(^) Pat DOES run
: [ MP
(^) Pat do [ TP (^) -s [ ΣP (^) -Ø’ [ VP (^) run ]]]]
4. Application to children’s nonagreeing
(^) don’t
“Non-agreeing 4.1 The phenomena
(^) don’t ”
(^) a. (^) so Paul doesn’t wake up
(^) (adam28)
b. Robin don’t play with pens
(^) (adam28)
(27)“Medial neg”
CHI: you not have one.URS: alright I think I have one.
(^) (adam19)
- thus, (^) do can be omitted, or it can apparently be left unagreeing, alongside the adult option
[An old theory, that
(^) don’t (^) is just a variant of
(^) not for the child, can be ruled out on distributional grounds]
- on my analysis of adult INFL, minimal divergence from target would imply that nonagreeing
(^) do has M
pronounced as
(^) do , therefore triggered by some affix/clitic, but something about Tns/Agr does not
yield 3sg (^) -s (^) spellout
- medial neg should mean M spelt out as zero, there was no trigger for
(^) do ; something about Tns/Agr
prevented it from being an affix
- to narrow down the possibilities: look at another
(^) do -support environment, viz. SAI: do we find
nonagreeing
(^) do (^) there? No
Parentheses indicate this feature specification is not actually needed in the final model. It is here to remind us what sorts of^6 cases will fall under the relevant vocabulary entries.
Table 1 : Distribution of agreement on
(^) do (^) with 3sg subjects
in negative declaratives versus nonnegative questions with inversion
pooled from five children
(Table 3 in Guasti and Rizzi 2002)
Environment
do
does
Before (^) n’t
Question with inversion
a.^ Guasti and Rizzi’s finding: (^) so Paul doesn’t wake up
(^) (adam28)
b. Robin don’t play with pens
(^) (adam28)
c. Does dis write?
(^) (adam28)
d. # Do Robin like that?
(^) (unattested)
- thus, whatever is nonadult about child INFL shows up in declaratives with
(^) n’t but not in (inverted
positive) questions
- two immediate hypotheses: has to do with nature of
(^) n’t or with raising of I to C; I pursue the former,
Guasti & Rizzi (G&R) the latter; for a different approach see Roeper (2002)
- proposal: 4.2.1 Vocabulary entries 4.2 My proposal (^) n’t can demand support independent of INFL (because it’s a clitic, or maybe an affix, cf.
- we never see this in adults because for themZwicky & Pullum 1983)
(^) n’t is restricted to (fully) inflected clauses
- what the paradigm gap, i.e. no nonagreeingsupport- but medial neg tells us kids can do something nonadult with INFL such that it no longer needs
(^) do , tells us is that kids do NOT have a non-adult
INFL that requires support; thus their nonadult INFL is [-affix]
(^) do as support for
(^) n’t when Tns/Agr is funny, otherwise the zero M
allomorph comes out
- why would child’s non-adult Tns/Agr not require
(^) do -support?
- here we need to fill out the adult paradigm for Tense+Agr (fused), considering more environments than Halle & Marantz do
- first, to keep -s off the modals & subjunctives, we specify it as occurring to the right of indicative Mood only
- [–past] is null, but it is an affix, i.e. something that needs support; that’s because we get
(^) do -support
with all subjects in present tense, not just 3sg:
(^) We run —> We do not run
- to motivate fact that both [+past] -ed and [-past] -Øaff must have that tense feature specified; if they did not, they would show up in subjunctive, for example
- note subjunctive Agr (and Tns, if it’s present at all), must be non-affixal, as must M
- to motivate this and what’s to come, consider subjunctives: - elsewhere allomorph is Ø, but not an affix support, which might include small clauses, Mad Magazine sentences, the- otherwise we predict ANY clause type with Tense or Agr would have an affix in need ofSeems reasonable since there is no tense contrast here, at least in Englishgiven our existing vocabulary entries, Tense is either absent or has some value other than ±past.- if Tns+Agr is here, it does not need support; given NOM subject, at least Agr must be here;It is vital that he (*do(es)) not be late.
(^) Why (not) (^) construction
[3sg, -past] <—>^ Paradigm for T+Agr:
(^) -s / M (^) ___
[indic]
[+past] <—>
(^) -ed
[-past] <—> -Ø
affix^
[] <—> Ø (by default, not an affix)
- now update M: presumably there is a subjunctive M head that blocks out modals,
(^) do ; then can still
treat [indic] as default value if we want
[epist, necess] <—>[subjunct] <—> Ø^ Paradigm for M:
(^) must
[] <—> …
(^) do (^) / ___ [+affix]
[] <—> Ø (by default, not an affix)
(^) Paradigm for
[negative] <—>
(^) not , (^) n’t cl
[positive] <—>
(^) so, too (^) / ___ [emphatic]
[emphatic] <—> -Ø
cl^ @
[]
<—> Ø
- what is nonadultlike about INFL—T or Agr (or both)? From my other work, both. ATOM (Agr/Tense4.2.2 Underspecification Omission Model): Agr and Tense can be independently underspecified. Table 2: (^) Distribution of 3sg pronoun subject case as a function of auxiliary form
pooled from five children
(Table 10 in Guasti and Rizzi 2002)
Subject case
don’t
doesn’t
Nom
nonNom
- consider, for each vocabulary entry, what combinations of feature presence/absence could result in insertion of that vocabulary item.
10^ Table 3
(+ = has a specified value; – = value unspecified)
Tns+Agr
Children’s Infl
(^) Case of subject
[indic]^ M
[3sg, -past] <—>
(^) -s / M (^) ___
[indic]
Irrelevant
+Tns, +Agr
Nom
do
[+past] <—>
(^) -ed
Irrelevant
+Tns, -Agr+Tns, +Agr
nonNOMNom
do
[-past] <—> -Ø
affix
Irrelevant
+Tns, -Agr+Tns, +Agr
nonNOMNOM
do
[] <—> Ø
-n’t not, Ø
-Tns, -Agr -Tns, +Agr
nonNOMNOM
do Ø
Effects of readjustment rules: /du/ + /z/ —> /d
√z/
/du/ + /d/ —> /d
Id/
/du/ + /nt/ —> /dont/
- we see that the INFL that does NOT need support arises when Tense is missing, whether or not Agr is missing; this predicts both NOM and nonNOM subjects with nonagreeing
(^) don’t (^) and with medial
neg; former is true; latter probably (see Schütze 1997)
UG says: 4.3 Guasti & Rizzi’s proposal not is a property of the language-specific system of morphological rules.(ii) If a feature is left unchecked in the overt syntax, whether it is morphologically expressed or(i) If a feature is checked in the overt syntax, then it is expressed in the morphology.
(^) a. this, always do-sg it the childrenb.^ Questo, lo fa sempre i bambinithis, the children always do-pl it^ Questo, i bambini lo fanno sempre
(Anconetano)
(^) a. it is come-sg the your sistersGl’è venuto le tu’ sorelle
b. the your sisters they are come-plLe tu’ sorelle le son venute
(Fiorentino)
Against this backdrop, Guasti and Rizzi can account for children’s nonagreeing
(^) don’t (^) in two steps:
Table 6 : Sarah
EnvironmentFiles 50–137 (3;2.23–5;0.25)
do does
SAI
Before (^) n’t
(^40) 55
Bare noninverted
Table 7 : Adam
EnvironmentFiles 11–33 (2;8.0–3;5.29)
do does
SAI
Before (^) n’t
(^12) 8
Bare noninverted
Table 8 : Ross
EnvironmentFiles 24–50 (2;6.18–4;3.15)
do does
SAI
Before (^) n’t
(^20) 72
Bare noninverted
- there are numerous systems that look very much like standard Eng 5.1 Data 5. Last resort vs. Optionality
(^) do -support but do not have the last
resort character; rather they have free variation between
(^) do and inflected main verb:
he does come
(^) = (^) he comes
English in 15
& 16th^ centuryth^
(^) do (^) is attested for English throughout the 16
century and persisted into the 18th^
(Visser,th^
1969; Warner, 1993).
Ellegård (1953): “
do (^) + the infinitive was functionally synonymous with the finite full verb” (p. 151)
- considering how frequently periphrastic
(^) do was used with certain verbs in the middle of the 16
th
Whendo-form at this time conveyed a special shade of meaning, differentiating it from the simple form.Century, Ellegård concludes (p. 167) that we can “dispose effectively of any hypothesis that the (^) do (^) is practically always used [with a given verb], it cannot fill any such function and is
absolutely nothing but a mark of tense.”
Palsgrave (1530): “
I do (^) is a verbe moche comenly used in our tonge to be put before other verbes, as it is
all one to say ‘I do speake…’ and suche lyke, and ‘I speake…’.”
- contemporary translators would put the periphrastic construction in correspondence with a Latin sentence that gave no indication of emphasis.
The paradigm in (1) persists to this day in South-Western dialects of British English (Klemola, 1998).Modern dialects Klemola says: in Southwestern English English,
(^) do (^) is NOT a habitual aspect marker, although in
with inflected main verbs, whatever their meaning. The one restriction Klemola notes onthe past tense this may represent the majority of its uses; rather, it seems to be in free variation
(^) do (^) in
Southwestern English English is that it cannot occur with
(^) be (^) (p. 41), but other stative verbs are
fine.
a.^ South-Western England (Wright 19xx) c. (^) I do go.b. (^) I do eat. (^) Thee do look.
Many researchers have documented uses of Child English
(^) do that, at least according to the child’s prosody, were not
pattern in whichBohnacker, 1999; and work cited there). Crucially, these errors are not part of a more generalinvoked by any of the standard triggers (Hollebrandse and Roeper, 1996; Zukowski, 1996;
(^) do is widely overused; that is, (38) is attested, (39) is not. On the hypothesis
that child grammars must conform to UG, this is further evidence that spurious
(^) do (^) is a
possibility in human languages that are otherwise fairly similar to (adult) English.
(^) a. f. (^) You did make my bed a little fan.e. (^) I did paint this one and I did paint this one…and I did paint this one.d. (^) Who did take this off?c. (^) I do have juice in my cup.b. (^) I did wear Bea’s helmet (^) A witch did look like it has slippers.
(Tim 2;11–3;0, Roeper corpus)
(^) a. b. (^) #He did runs. (^) #He does ran.
(Roeper 1991)
d. (^) #John doesn’t can play alto-sax.c. (^) #It does is.
(Hollebrandse and Roeper 1996)
(40)Perhaps even a corner of our English (^) a. plaintiff’s home…without regard for public safety drive a motorcycle through the front yard of thed.^ Your Honour, we intend to prove that the defendant, John Doe, did willfully andc.^ We, the employees of Unity Airlines, do hereby announce our intention to …b.^ I, John Hancock, do solemnly swear to uphold the duties of the office of President…^ I, the undersigned, being of sound mind, do this day hereby bequeath …
tun
- parallels dummy& counterparts throughout Germanic
(^) do (^) in several respects: we can find it in questions with SAI,
tun and in imperatives, [can’t check VP ellipsis], but NOT in nonfinite environments [interestingly, cannot support emphatic positive polarity, a topic for future work]
The same phenomenon that is documented for 16
century English emerged in German around the sameth^
(Erb, 1995, 2001, in press; Schönenberger and Penner, 1995).prescriptive pressure), as well as in Swiss-German, far-flung German dialects, and Dutch dialectstime and is robustly attested to this day in the spoken language all over Germany (4) (despite
(^) a. (^) Sie liest (^) ein Buch.
she reads a
(^) book
b. Sie tut (^) ein Buch lesen.
she does a book
(^) read
(^) a. (^) Ds Ching tuet sech scho säuber aalege.
[Bärndütsch]
‘The child already gets dressed by himself.’the child does self already independently get-dressed
b. ‘The mother thinks about what she wants to buy.’the-mother does self think what she wants to-buyD’Muetter tuet sech überlege, was si wott choufe.
(Schönenberger & Penner 1995: 318)
(43)Question: (^) Tuesch iez bald mëlchen?t^
‘Are you going to milk (the cows) soon?’do you now soon milk
[Swiss-German]
(44)Imperative: ‘Do not make yourselves dirty!’do you-pl not soil^ tut euch nicht bekleckern!
[Upper Saxonian]
(45)Negative: I do not knit^ Ich du net strige.
[Pennsylvania German]
- spurious (^) tun (^) is not limited to V2 clauses: Erb documents it in V-final embedded clauses, so it is not
crucially tied to the C position.
(^) a. b. (^) Da hab ich sie gefragt, wie ihr denn Kärnten so gefällt.‘Then I asked her how she liked Kärnten’then have I her asked how she PART Carinthia PART to-like does (^) Da hab ich sie gefragt, wie ihr denn Kärnten so gefallen tut.
- Both (^) do (^) and (^) tun have homophonous main verb counterparts that differ from the auxiliary use in not
allowing stative interpretations. Spurious
(^) tun is like dummy
(^) do in being restricted to finite
clauses, despite the fact that it has an infinitival form
“the (^) tun grammaticality ofthe tense and mood paradigm. It was also excluded that the type of eventuality plays a role for the-periphrasis as a whole shows virtually no restrictions with respect to the sentence type and to
(^) tun
.. .non-agentive verbs co-occur with
(^) tun ” (Erb 2001, pp (190–191)
- this is not to claim that a speaker’s choice on a given occasion is random; Erb (2001) summarizes itself distinguishes them.conjugated form of the main verb would be. The point is that nothing in the syntax or semanticsseveral plausible factors that play into this choice for German, including how obscure the
- Erb says about the only thing that’s impossible for most people is
(^) tun with aspectual auxiliaries:
(^) a. (^) ??/*Ich tue es gesehen haben. I do it seen have
b. ??/*Du tust dort gewesen sein. you do there been be
- The one difference between the languages is of course that since German has verb raising, there is always an alternative to the use of
(^) tun —namely raising the main verb to Infl rather than leaving it
in VP and inserting
(^) tun (^) in situ—hence
(^) tun is never obligatory (except in some fronting/clefting
constructions).
There are some obvious potential alternatives that should be considered, but that Erb (2001) shows do not go through for
(^) tun. For instance, one might attempt to argue the alternants come from
apparently anyone who speaks German cannot help but have the spuriousanything, with a sociolect lacking geographical boundaries, but this would still miss the fact thatcontact with Germany for centuries, e.g. the Pennsylvania Dutch. We would have to be dealing, ifcommunities where it has been seriously looked for, including those that have lacked closeargues against this on the grounds that the construction has been attested in all German speechdifferent dialects and that all the speakers in question are bi-dialectal (e.g., Watanabe, 1994). Erb
(^) tun option in their
parsimony leads us instead to pursue a single grammar approachgrammars, as witnessed also by the fact that it continues to be actively proscribed in schools;
German auxiliaries that have the finiteness restriction (Erb, to appear): a subset of those in English, viz. future (^) werden (^) ‘will’ + only epistemic readings of modals +
(^) tun ‘do’ (she adopts the view that
werden (^) is actually an epistemic modal)
(^) a. (^) Er (^) tut ’s (^) nicht glauben.
(Colloquial & dialectal German)
he does-it not
(^) believe
‘He doesn’t believe it.’
b. (^) *Er braucht das Buch nicht lesen (zu)
(^) tun .
he needs (^) the book not
(^) read (to) (^) do- INFIN
(‘He doesn’t have to read the book.’)
c. (^) *Sie hat das Buch nicht lesen
(^) getan .
she has the book not
(^) read (^) done- PART
(‘She hasn’t read the book.’)
(^) a. (^) Sie (^) wird (^) Flamenco tanzen.
she will F
dance
‘She will dance Flamenco.’
b. (^) *Sie versprach, pünktlich kommen zu
(^) werden .
she promised
(^) on-time (^) come (^) to (^) will- INFIN
(‘She promised to come on time.’
c. (^) *Sie ist pünktlich kommen
(^) (ge)worden
.
she is on-time
(^) come (^) will-( IPP/) PART
(‘She was going to come on time.’)
(^) a. (^) Das (^) muss (^) der Postbote sein.
‘It is necessarily the case that this is the postman.’that must the postman be
b. (^) *Morgen
(^) wird das der Postbote sein
(^) müssen .
tomorrow will
(^) that the postman be
(^) must- INFIN
(‘Tomorrow, it will be necessarily the case that this is the postman.’)
- thus, whatever version of uniting we adopt, I need it to have the property that it is forced to apply or interleaved with syntactic operations]lag between syntax construction and pronunciation; that is, I need spellout to be tightlyproperty, though it could be stipulated on top; more generally it could be stipulated as a maximalnot at exactly this point, no later [most extant methods of uniting don’t inherently have this
- that decision point could be within SpellOut, if the latter is strictly cyclic, except that spellout can generally look at the whole tree
- if we allowed the main verb to look into the Numeration to check for a modal (which would be tricky tree after syntax finishes the whole tree there would be no problembusiness), there would no longer be a problem; also if we allowed Spellout to work on the whole
- VS. comparing ‘don’t raise
(^) be , unite T&M’ with ‘raise
(^) be , unite (^) be (^) and T’, the choice is simply to raise
or not; what’s to compare is in syntax: we could have head raised but didn’t
- normally the movement option blocks no movement, but in imperatives it appears to be the opposite, (14) vs. (15); presumably really the movement option is completely unavailable
To explain imperatives I essentially follow Emonds (1994). The idea is that
(^) *Be not scared
(^) reflects a sort
of paradigm gap:
(^) be (^) has no raised imperative form (15), perhaps because imperative morphology
consumes the ability of
(^) be (^) to be moved above
(^) Σ. Thus, unlike in finite clauses,
(^) be (^) cannot act as
way to realize the structure using just a form ofsupport if it is blocked from combining with the main verb. Only in that case, when there is noan INFL supporter in imperatives, but imperatives still contain some INFL material that requires
(^) be , can (^) do (^) and (^) be (^) co-occur.
8
- languages/dialects differ on enforcing Emonds’s “minimize # of words” edict—learner who starts by assuming it is obeyed easily resets based on positive evidence, i.e. hearing the “spurious” use
1) (^) does 6. Concluding remarks (^) do (^) correspond to any syntactic position, or is it introduced ex nihilo in Spell-Out?
Answer: to where finite (^) do (^) is in the syntactic position where modals go, which crucially is not Tense and is not identical
(^) be (^) & auxiliary
(^) have (^) go
Answer: The need is one of morphological support, the environment for vocabulary insertion of 2) (^) what ensures that it’s there when it’s needed?
(^) do
- child data on nonagreeing makes reference to that
(^) don’t (^) support this: we get
(^) do before clitic negation, but in other
environments we find zero and we do not find nonagreeing
(^) do , when (by hypothesis) INFL
doesn’t need support
(i) The class “imperatives” over which this generalization holds is fairly broadly defined (e.g. Denison 1993): (^8) b. (?)If you don’t be more careful, something’s going to get broken.a. (^) Why don’t you be more careful?!
Answer: Nothing! Fact that children don’t enforce this last resorthood suggests it is not the only 3) (^) what ensures that it’s not there when it isn’t needed? possible, perhaps not the unmarked, property of a system of this type in UG
4) (^) Why does standard English make it look like
(^) do (^) is strictly a last resort?
Answer: A violable, Gricean principle that effectively compares representations cross-derivationally (in Binding; also, acc. to Schütze 1997, in case/agreement).this, in that it is frequently suggested that they have trouble with such comparisons (e.g. ina very constrained way, perhaps built from the same numeration). Child data are consistent with
5) (^) How can we capture optionality of
(^) do found in various dialects etc.?
Answer: A particular kind of look-ahead problem prevents one derivation from blocking the other, even out is the problemthough they start from the same numeration; perhaps lookahead from the syntax into the spell-
(^) do on this story—the affix property of INFL is missing, because some of its
features are underspecified
- what this means for kids using spurious
(^) do : either they don’t know/compute the Gricean principle, or
something makes inflecting the main verb (which in my story means raising it) unavailablethey misset the quasi-parameter, or on some occasions there’s no choice for them because
- elaborating, children’s apparent (mis)use of spurious
(^) do (^) could be seen in two ways. First, it could be an
particular, one in which the alternative to spuriouscontexts in which it can be used. Second, it could reflect an intermittently nonadult grammar, ininterface error, i.e. a failure to map correctly between a grammatically valid structure and the
(^) do , namely a structure with verb raising, is not
Root/Optional Infinitive utterances are clearly missing INFL features, whereas in a spuriousdifference, however, is that in the languages cited (though arguably not always in child English)suggestion that failure of V raising is indeed at the heart of the phenomenon.) An importantfail to raise verbs in main clauses, thereby making them nonfinite. (See Phillips, 1995, for thechild languages such as French, Dutch, German etc. (Rizzi 1994; Wexler 1994), whereby childrenalways available. This would closely analogize it to the Root/Optional Infinitive phenomenon in
(^) do
utterance the features are syntactically present, merely spelt out on a different word.
- by claiming, unlike Hollebrandse & Roeper, that spurious
(^) do (^) is not the same phenomenon as tense
that’s been testeddoubling, I would predict possibility of dissociations in time and across children; don’t know if
Some dialects allow “fewest words” to be overridden by other factors that make the use of spurious
(^) do
impact on the jury. advantageous. For example, in the legalistic context, delaying the main verb may heighten its
Appendix: (^) do + be
(^) be (^) is a main verb that does not raise, or when
(^) do (^) is meaningful. There are some
isolated dialectal exceptions, but in the cases I am aware of it is clear that either
(^) do (^) or (^) be (^) has a
different status from that of the standard language.
For example, where
(^) do (^) carries habitual meaning, as in Southern Hiberno-English, (52) are possible. Here
do (^) is behaving like
(^) will (^) or any other modal
(^) a. b. (^) They do be angry. (^) He does be here every Friday.
And in Belfast and other Northern Hiberno-English dialects
(^) be (^) is treated as a regularly inflected main
verb that carries habitual meaning, so it is not surprising to find it co-occurring with
(^) do ,
(^) a. (^) He bes here every Friday.
[Hiberno-English, Alison Henry and Siobhan Cottell p.c.]
b. Does he be here every Friday?
c. (^) *Bes he here every Friday?
(^) a. b. (^) *He bes not here every Friday. (^) He doesn’t be here every Friday.
He DOES be here every Friday.
Something similar is true in Black English, though main verb inflection is generally Ø and so cannot provide independent evidence for the status of
(^) be (^) (Green, 1993; Déchaine, 1993).
(^) a. (^) Bob doesn’t be angry.
[AAE]
‘Bob isn’t usually angry.’
b. ‘Sue is SO usually reading books during class.’Sue DO be reading books during class!
Thus, the generalization to which I know of no exceptions is that when
(^) do (^) is semantically a true dummy
and be (^) carries only its pure grammatical meaning, they cannot co-occur.
- Roeper (1991) reports kids saying “This didn’t be colored = ‘This wasn’t colored’”, “Does the fire be on every day?”, “Didn’t be mad”. His conjecture is that this is because children still think
(^) be
same ones at the same ages, say things like “He bes here” inflects like a main verb, and hence does not raise; some children, we don’t know if they’re the
Denison (1993) reports his kid at age 5 was still saying “Did you be quiet?, I didn’t be naughty” I would exclusively in those forms “Be quiet” “Don’t be naughty”speculate that “be quiet” and “be naughty” were learned as main verbs, having been heard almost
- it’s true also in modern English that
(^) be/have (^) are compatible with
(^) do only when they are (forced to be)
nonfinite, namely in imperatives, as seen above
[recall from English imperative that
(^) do (^) is compatible with (copula at least)
(^) be :
Do (not) be alarmed if you hear someone talk about “bringing it down”.]
Akmajian, Adrian (1984). ‘Sentence types and the form-function fit’, Selected References
(^) Natural Language and Linguistic
Theory (^) (^2) , 1–23.
Linguistic Inquiry Akmajian, Adrian, Susan M. Steele & Thomas Wasow (1979). The category AUX in universal grammar.
N. Beckman (ed.),Arnold, Mark D. (1995). Notions of economy in language change: The spread of periphrastic ‘do’. In: Jill
(^) Proceedings of the North East Linguistic Society 25
, Volume two:
(^) Papers from the
workshops on language acquisition and language change
Bohnacker, Ute (1999). Children, tense and auxiliary
(^) do. Newcastle and Durham Working Papers in
Culicover, Peter W (1999.Linguistics 5, 41–74.
(^) Syntactic nuts: Hard cases, syntactic theory, and language acquisition
, Oxford
Déchaine,University Press, Oxford. (^) Rose-Marie
(^) (1993). (^) Predicates
(^) across (^) categories.
(^) Ph.D. (^) dissertation,
(^) University
(^) of
Denison, David (1993).Massachusetts at Amherst.
(^) English historical syntax: Verbal constructions
. London: Longman.
Ellegård, Alvar (1953).
(^) The auxiliary ‘do’: The establishment and regulation of its use in English.
and R. Zanuttini (eds.),Emonds, Joseph E. (1994. Two principles of economy. In: G. Cinque, J. Koster, J.-Y. Pollock, L. RizziMIT.Emonds, Joseph (1970). Root and structure-preserving transformations. Unpublished Ph.D. dissertation, Stockholm: Almqvist & Wiksell.
(^) Paths towards universal grammar
, 155–172. Washington, D.C.: Georgetown
Erb, Marie Christine (1995). Eine Theorie expletiver Verben: DieUniversity Press.
(^) tun -Periphrase im Deutschen. MA
Erb, Marie Christine (in press). Auxiliaries and finiteness in German.Brabant.Erb, Marie Christine, 2001). Finite auxiliaries in German. Ph.D. dissertation, Katholieke Universiteitthesis, Universität Frankfurt am Main.
(^) Lingua .
Giorgi, Alessandra, and Fabio Pianesi (1997).
(^) Tense and aspect: From semantics to morphosyntax
Green, Oxford University Press, New York. Lisa J. (^) (1993). (^) Topics in African American English:
(^) The (^) verbal (^) system (^) analysis. (^) Ph.D.
Guasti, Maria Teresa and Luigi Rizzi (2002). Agreement and tense as distinct syntactic positions: Evidencedissertation, University of Massachusetts, Amherst. from acquisition. To appear in Guglielmo Cinque (ed.),
(^) Functional structure in DP and IP: The
cartography of syntactic structures
, vol. 1, Oxford University Press.
Hale and Samuel Jay Keyser (eds.),Halle, Morris and Alec Marantz (1993). Distributed morphology and the pieces of inflection. In: Kenneth
(^) The view from Building 20: Essays in linguistics in honor of
Sylvain Bromberger
, 111–167. Cambridge, MA: MIT Press.
Heidi Harley and Tony Bures (eds.),Halle, Morris and Alec Marantz (1994). Some key features of distributed morphology. In: Andrew Carnie,
(^) Papers on phonology and morphology, MIT Working Papers in
Linguistics (^) 21, 275–288.
acquisition. In: Charlotte Koster and Frank Wijnen (eds.),Hollebrandse, Bart and Thomas Roeper (1996). The concept of DO-insertion and the theory of INFL in
(^) Proceedings of the Groningen Assembly on
Language (^) Acquisition
(^) held (^) at the University (^) of Groningen, (^) 7– (^) September
(^1995) , (^) 269–271.
Klemola, Juhani (1998). Semantics ofGroningen, The Netherlands: Centre for Language and Cognition.
DO
(^) in South-Western dialects of English. In: Ingrid Tieken-Boon
van Ostade, Marijke van der Wal and Arjan van Leuvensteijn (eds.),
(^) Do (^) in English, Dutch and
German: History and present-day variation
, 25–51. Münster: Nodus Publikationen.
Ganger and Kevin Broihier (eds.),Phillips, Colin (1995). Syntax at age two: Cross-linguistic differences. In: Carson T. Schütze, Jennifer
(^) Papers on language processing and acquisition, MIT Working
Papers in Linguistics
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik (1985).
A
comprehensive
grammar of the English language
. London: Longman.
dissertation, MIT. [Distributed by MIT Working Papers in Linguistics.]Rapoport, Tova R. (1987). Copular, nominal, and small clauses: A study of Israeli Hebrew. Ph.D.
