Integrating Roots and Grammatical Morphemes into syntactic trees
Joseph Emonds. Palacký University. Czech Republic November 13, 2017, Masaryk University Roots Workshop
QI:    Are there Roots? (How are they defined?).....................YES (?)
Q2:    What is the role of features of Roots in Syntax..............NONE
A good model of vocabulary storage and insertion into syntactic derivations needs to make a distinction between distinct kinds of morphemes.
• There is an attested dichotomy between "lexical" and "grammatical/ functional" categories
• This dichotomy is sharp, not a continuum.
(1)     one'won: back (P) < back (N), may/ May. would, wood, not/knot, etc.
I        CLOSED vs. OPEN classes of root vs. functional/ grammatical morphemes
Exhaustive lists of classes of grammatical morphemes.
(2) a.       Jackendoffs (1977) Determiners in Pet:
the. this, these, that, those, every, each, all, both, a(n) some, any, no. which, what, we. us, you
b.       Pronouns in Pet with no N: he. she, it. they, I, who Total: 24
(3) a.       English modals in I possible with no agreement (includes a null 'present subjunctive'I):
will, won'/. would, can, could, may. might, shall, should, must, need, dare, ain 7, 0
b.      Items in I with (mostly) irregular agreement:
is, are. am. was, were, do, does, did, don't, have, has, had Total: 26
(4) Morphemes used to make plural cardinal numerals NUM (work by J. Hurford)
a. Free:   zero, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, hundred, thousand, million, billion, trillion, zillion, dozen
b. Bound: thir-, twen-, -teen, -ty, -score Total: 24
(5) Degree/ grading specifiers of adjectives:
very, so, too, enough, rather, quite, this, thai, how, real, pretty, damn, darn, awful, mighty, hella, more, most, less, least, -er, -est Total: 22
Apart from the functional heads in (2)-(5) above, the "lexical" categories also have at most 20-30 grammatical members. This means, "items with no features other than those that function in syntax."
(6) English grammatical verbs V/v, which are not roots (except in idiomatic combinations);
a. Free: be, have, get, do, go, come, need, dare, make, let, see, hear, watch, bring, lake, say. help, Iwant
b. Bound derivational: -en, -ize, -ify, -ate, 0 ("conversion") Total 22+
(7) English grammatical nouns N/n, not roots (except in idiomatic combinations):
a. Free: one(s), stuff, olher(s),fact, place, time, way, people, man
b. Bound derivational: -self, -thing, -ment, -tton, -age, -th, -ness, -ity, -t, -ist, -ism, -y, -ee. -ette, -er, -ess
Total 25+
(8) Functional/grammatical categories are not Roots.
Roots can only be associated with these four "lexical head" categories: N/n, V/v, A/a, P/p.
Lexical categories can select different categories of phrasal sisters (N. Fukui and M. Speas).
• Thousands of) Nouns. Verbs. Adjectives. Hundreds of Prepositions (H. van Riemsdijk's work on Dutch)
• There are 3. at most 4. open class categories. (9) gives examples of P.
(9)     a.       Transitive P:
in case of in spite of. due to. out of. off of. up to. alongside, by means of. by way of given, instead of concerning, by virtue of. by dint of. according to, except for, ...
b. Intransitive P.- outside, aside, overhead, ahead, aboard, upstairs, abroad, away. back.
c. Plus ordinary prepositions, to a total well over 50.
The distinction between roots and grammatical/ functional morphemes will be empirically adequate only if it predicts or is consistent with the (mostly well known) empirical distinctions.
11       Properties distinguishing roots from grammatical morphemes
At least eleven distributional, semantic, and phonological properties divide the two types. (Notice that free vs. bound distinction is not a relevant distinction.)
(10) Distinctions   (i) semantic
(ii) formal
(11) Semantic distinction between roots and grammatical morphemes:
a. purely semantic features/
b. conceptual (i.e. interpretable), grammaticalized (syntactic) features F
(12) 3 examples of "purely semantic" root features/
a. Damage vs. destruction: damage, harm, injure, hurt, mess up vs. destroy, ruin, wreck, total, disable
b. liquid nouns vs. solid nouns: drink the aspirin, break the butter
c. Color: dark, light turquoise/ rust vs. * dark, light iron/ fire
(13) 3 examples of "conceptual/grammatical" features F for lexical heads.
a. ±ANIMATE
b. GRADABLE. for Adjectives
c. STATIVE for Verbs
(14) Roots can be associated only with lexical heads and contain purely semantic features/ Grammatical morphemes have no purely semantic" features/, they have only F. Many interjections have no syntactic F, they have only/
(15) A model of vocabulary storage reflects the presence/ absence of features// F
• Dictionary (roots): repository of items with /
• Syntacticon (grammatical morphemes): repository of items with only F
HI     A Model of vocabulary insertion: Deep vs. Late insertion
Bottom up derivational model of recursive PDs (= phasal domains)
Think of PDi as some embedded clause or maximal nominal phrase, and of PD2 as a higher or matrix clause. (16)   ... [pro X-PD,-Y] ...
When a derivational phase on PDi terminates, PDi is ready for the LF interface ,but not vet for PF.
(17)
Steps in the phase of PD2 (after PDi is processed and sent to the LF interface):
Deep or Phase-Initial Insertion. Insert a lexical head of PD2 into X-Y—often a root, but not necessarily. Such insertion might also include items besides heads.
Transformational derivation of PD2, including possible raising/ control of items in PD|,
Late Insertion. Grammatical morphemes (not roots!), can now be inserted bottom up in (16). o  PF Insertion into any unlicensed empty nodes remaining in PDl These are invisible in LF. o  Syntax Insertion into empty nodes in PDi. These are visible at LF, and PD2 is now ready for the LF interface.
JLLSL
SYNTACTICON: syntactic/conceptual features: (T) lexical addressing by syntactic category and context featires
C(F)
PF: INTERFACE WITH LANGUAGE USE/SPEECH
LF: INTERFACE WITH UNDERSTANDING/SEMANTICS
(19)    What is the role of Roots?
it What is the level of insertion of Roots?
ROOTS: DICTIONARY    semantic features: (Q ,j      addressing by phonological content
INTERFACE WITH MEMORY. AND OTHER COGNITIVE SYSTEMS
Discussion of the handout Time levels of Insertion.
• PD2 in all the trees is the highest NP.
• For the result nominals. there is no embedded PD|.
• For the complex event nominals and the gerunds, the shaded VP and V are the embedded PD| (20) Distinctions between Roots and grammatical morphemes (Emonds (2000: Ch. 3 and 2015)).
	Syntactic properties:	Roots	Grammatical morphemes
i.	Grammatical categories which can spell them out	N, V, A, P	ALL
ii.	Items with PF features dissociated from their LF positions	NO	possible
iii.	"Late insertion" possible during syntax and at PF; see handoul	.: NO	possible
	Phonological properties:		
iv.	Items conform phonologically to "primary vocabulary" Example: Germanic uniform stress on initial syllable	NO	YES
v-	Bound right hand heads in compounds have inherent stress (careful vs. care-free: postman vs. posl-man)	YES	NO
VI.	Phonetically zero morphemes possible. E.g. in lists (3) and (61	NO	YES
	Storage and processing properties:		
vii.	Interface with non-linguistic memory, psychology and culture	YES	NO
viii.	Open classes; hundreds / thousands of members (vs. closed classes with 2-3 dozen). Adults can coin neologisms.	YES ., .	NO
ix.	Can have purely semantic featuresfnot used in derivations.	YES	NO
X.	Processing look-up in terms of initial consonant clusters ("cohort theory" in psycholinguistic processing)	YES	NO
xi.	Full suppletion inside paradigms (go/went; bad/ worse)	NO (several languages)	possible
How does the model (17)-(19) relate to the empirical generalizations in Table (20) ???			
- accounts for (i)	Roots (= open class items, the dictionary) only accessible for phase-initial insertion.
- explain (ii), (iii)	Dissociations (=Alternative Realizations) result from derivations, not from storage.
- accounts for (iv)	Since there are so many roots, phonology must diversify to increase their number.
- explains (v), (vi)	Obligatory grammatical heads are expected, so need no phonetic prominence.
- explains (vii),(ix)	The purpose of syntax is to relate stored psychology and culture, not to create them.
- accounts for (x), (xi).	The indices/ stored roots are underspecified phonological feature matrices. They are needed to make roots in trees pronounceable.
IV     The 'extra baggage' of phonological and purely semantic features/on roots in derivations
• The indices of roots in pronounceable trees are minimal specifications of how to find the pronunciations.
• Integers are not minimal specifications. The set of integers contains a successor function, two closed operations of addition and multiplication, etc. These are not part of the language faculty.
Another name for a stored item's index is 'classificatory matrix':
(21) "The more direct the relationship between the classificatory and phonetic matrices [ the latter are directions for pronunciation, JEJ, the less complex—the more highly valued will be the resulting grammar." (Chomsky and Halle 1968: 381).
• Since the feature/ category specification (including insertion contexts) of each grammatical formative is unique, pronunciations are just paired with them in the Grammatical Lexicon/ Syntacticon.
• Since the syntactic features of Grammatical Morphemes have meaning (in their canonical LF positions), there can easily be sentences that are fully interpretable, even though they contain no roots.
(22) You can do that when 1 come back.
Those people aren 'I saying what ire should bring with us today.
Some references
Anderson, Stephen (1982). Where's Morphology? Linguistic Inquiry 13. 571-612.
Aronoff. Mark (1976). Word Formation in Generative Grammar. Cambridge: MIT Press.
Chomsky. Noam (1970). Remarks on Nominalization. In Readings in English 'Transformational Grammar.
184-221. Eds. R. Jacobs and P. Rosenbaum. Waltham: Ginn.
Chomsky. Noam and Morris Halle (1968). The Sound Pattern of English. New York: Harper and Row Emonds. Joseph (2000). Lexicon A Grammar: the English Syntacticon. Berlin: Mouton de Gruytcr.
— (2004). How do we construct Convergent Numerations? Theoretical and Applied Linguistics at Kobe Shorn 7. 1-20.
— (2015). The Computational Lexicon. English Linguistics
Halle. Morris and Alec Marantzf 1993). Distributed Morphology and the Pieces of Inflection. In The View from Budding 20. Ed. K. Hale and S. J. Keyset. 111-176. Cambridge: MIT Press.