English Auxiliaries and the Mapping to Phonological Words

 

We’ve been discussing ways in which a hierarchical structure of morphemes created by recursive Merge could be ordered, first to last.  A major observation is that morphemes that surface inside words seem to generally obey Brody’s “Mirror”; from stem to the right, one goes up the structural hierarchy (the mirror of the order of phrases, which go down the hierarchy as one goes from left to right = LCA).  The general approach of head movement and adjunction, as well as Brody and NanoSyntax, involves consideration of a “head projection” starting at the root of a word in the Merged tree and moving head to head up the tree to create complex heads.  However, this general approach doesn’t really deal well with the situation in English (and many other languages) in which a sequence of auxiliaries (or other functional words) have complex internal morphological structure but are ordered according to the functional sequence that governs the ordering of affixes within words, in languages in which the same functional heads that emerge as auxiliaries in English occur within a single phonological word.  English “been” in “John has been handing Mary a book” has the stem-suffix organization we would expect from a hierarchical structure that puts the participle head above the verbal stem and orders them à la head-movement or its replacement.  However, the Perfect morpheme within “be-en” is ordered with respect to the Tense morpheme on “has” and the Progressive morpheme on “hand-ing” in line with what we would expect from LCA ordering:  Tense Perf Prog.  The intuition we have is that the verbal parts of auxiliaries are there to support the functional morphemes and are not ordered with respect to these morphemes or with respect to one another according to an f-sequence or extended projection ordering.  But there does seem to be a type of cyclic word and phrase formation process going on, where first we create the auxiliaries (via “Mirror”) and then we order the auxiliaries.  In German (etc.) the auxiliaries themselves appear in Mirror order, suggesting two cycles of ordering along a head projection that might employ Mirror.

 

I’ll examine here aspects of the creating of an auxiliary sequence in English to highlight the theoretical issues we have to confront.  One involves how to chop up a head sequence within a phase into phonological word sized units – the creation of the English auxiliary verbs themselves.  The second is how to order the auxiliary verbs.  Do we do this as we move up the head projection creating the auxiliaries, or, once finished creating the auxiliaries, do we start over from the bottom of a head projection and order the auxiliaries along with other constituents?  Crucially here is the question of ordering the verb and the auxiliaries with respect to the complements of the verb.  On most assumptions, objects and indirect objects at least may appear in the hierarchical structure between pieces of what will become the verb.  [I’m bracketing here the question of whether adverbs and negation are part of the head sequence.]  If we want to order the verb with respect to its complements via a head projection what do we do about this apparent interleaving of verb morphemes and complements?  In addition, in V last languages, we might want to order the complements with respect to the verb via Mirror:  the complements start out lower but end up higher than the verb (pieces).  But we don’t necessarily want to force a Mirror order of complements in German with respect to English to get the complements before the verb in German.

I will use my example “John has been handing Mary a book” to highlight the issues here.  I’m assuming that Merge will create the following head-projection for the sentence, where left to right order corresponds to hierarchical order.  Note that I’m using “Tr” as the head that replaces both voice and v.

Tense Tr Tr[+DP] Perf Tr Tr[-DP] Prog (John) Tr Tr (Mary) Tr √HAND ([a book])

          has                    been                                     handing

Above the two objects in this structure, which need to Merge as phrases with the Phrase taking Tr heads, there’s a question of how the heads other than the complement-taking verbs Merge into the trees, i.e., via a head-complement relation or via a head-head relation.  I’ve claimed in my work that Tr is the only head that Merges with phrases as complements and that Tr is what makes up verbs (there is nothing syntactic to verbs other than the Tr heads).  To follow these assumptions consistently, I would have the Tr’s that make up the auxiliaries Merge with what they combine with as heads to complements.  That is, the auxiliary “be” would be two Tr’s, each taking a phrasal complement.  Moreover, if we claim (as I do) that the perfect participle (and the gerund) are nominal while the passive and progressive participles are adjectival, we explain the choice of auxiliaries – a nominal participle phrase requires a [+DP] (transitive) Tr, which it gets in “have” (which is the spell-out of a rootless combination of +DP Tr phrases) while an adjectival participle phrase is content with “be” (which is the spell-out of a rootless combination of voice (Tr) and [-DP] Tr).  However, the “Aspect” heads Perf and Prog (which are actually just nominal and adjectival heads syntactically) may Merge with TrP’s treating the TrP as a (complex) head; not being Tr heads, the Perf and Prog could not Merge with phrases qua phrases.

Suppose we start with the root √HAND and project a head projection that goes as far as Tense.  From the existence of root suppletion for the past tense of GO, we have every reason to suppose that all the heads in this head projection are within a single phase – a single phonological spell-out domain (that includes all the heads from Tense down to the verbal root).  If we move up the head projection from the root, doing Mirror order at each head, will construct the complex head underlying “handing.”  We want stop after the Prog morpheme and start building the next word.  Two questions:

1.         Why do we stop?

2.        Do we keep the verb we created in play, ready to be ordered with respect to the next word that we’ll create as we create the words?  Or do we just move on from each created word?

Let’s suppose that there is something about the participle head that tells the system to stop.  If this property is a property of vocabulary items, we would need to be doing Vocabulary Insertion as we form the syntactic complex heads.  If instead it’s a property of morphemes in English, we need to know what kind of feature this is that ends syntactic word formation in English at the perfect morphology but doesn’t end it at a similar morpheme in languages that don’t use auxiliaries.  My own treatment of English treats the syntactic label of these Aspect morphemes as “Participle,” suggesting that it’s something other than aspect that identifies these morphemes in the syntax and would stop word formation along the relevant head projection (but I don’t know what “Participle” is).

Let’s also suppose that when the end of a word is reached as we move up the head projection, we just leave the word alone, waiting for another process that orders the words with respect to one another.

So, perhaps, after creating the complex heads “has” “been” “handing,” we move to a next step of linearization. Do we start now at the lower object (“a book” in our example) and project it as a head into a head projection for linearization (as a D) along with the other object, the subject, and the complex heads we created in the first go-round?

Crucially, the pieces of the verb “hand” are interleaved with the two complements in the hierarchical Merged structure.  In any head projection we create from the hierarchical structure after we have created our complex heads, we need to decide where the complex heads lie with respect to the complements.  The most promising alternative is to place the complex head for linearization at the point of the highest head they contain.  This would on the face of it recreate the prediction, “bigger (more morphemes) is higher.”  On standard assumptions about where items like negation and adverbs fall within the extended projection of the verb, the “bigger is higher” generalization is known to be false.  However, we can re-think those standard assumptions.  For example, I don’t believe that we need a Harizanov and Gribanova or Svenonius and Adger type story about complex heads appearing lower than the highest position of all their morphemes to account for the relative position of tensed verbs and negation in English and French (see my SubStack posts).

We still need to come up with clear mechanisms for dealing with the order of complements with respect to each other and with respect to the verb and the remainder of the extended projection of the verb.  We’d like, for example, the goal over theme order that seems to be the Merged order in double object English sentences and DAT ACC German sentences to survive as is, even if they are, as in German, subject to Mirror ordering with respect to the verb.  That is, the DAT ACC structure seems to Mirror over the verb as a unit.  However, it’s possible that the ACC DAT order in French Theme Goal double complement constructions (as well as the DP PP order for English “give the willies to everyone that entered”), where the DAT seems to c-command the ACC, could result from linear Mirror ordering of the ACC over the DAT during the linearization of a head projection along the extended projection of the verb.  Food for thought.