Who’s on First? (Or "Verb Second without counting")
At the end of the last posting I mentioned the following conundrum: if language/grammar cannot count (a generalization that is a corrolary of the Structure Dependence Principle), how come it can make references to the second position? Such references are needed for rules that define the operation of the so-called second-position clitics (found in many of the world’s languages, e.g., the interrogative clitic li in Russian), as well as the verb-second phenomena in Germanic languages, described in earlier postings.
Let us briefly review of the verb-second phenomena. It is found in such Germanic languages as Dutch, German, Swedish, Norwegian, Icelandic etc. As shown in the examples below, the finite (that is, tensed) verb — ‘read’ in the examples below — must appear in the second position in the main clause. This is true whether the element in the first position is a subject, a fronted direct object or an adverbial of place, time (as in the examples below) etc.
a. Dutch: Gisteren las ik dit boek.
yesterday read I the book
b. German: Gestern las ich dieses Buch.
yesterday read I the book
c. Swedish: Igår läste jag denna bok.
yesterday read I the book
all: ‘Yesterday I read the book.’
Note that in the English translation the verb does not appear in the second position: instead, it follows both the adverbial yesterday and the subject I. Thus, English is the only Germanic languages that does not use verb-second under normal circumstances. (On the residual verb-second in English, read here.)
Let us attempt a solution to this conundrum: how can we refer to verb-second without counting. To do so, we must redefine the second position (and also the first) in structural terms rather than in terms of linear word order.
Note that when we talk about the verb in the second position we mean just one word: the finite (tensed) verb. If there is more than one verb in a clause — a verb cluster — only the first auxiliary is subject to the verb-second rule. Other verbs in the cluster appear later in the clause, after the subject. The placement and order of these other verbs is subject to language/dialectal variation in Germanic. In contrast, when we talk about the first position, we mean something that can be more than just one word, that is, a phrase. Thus, the subject can be a multi-word phrase (e.g., ‘that man over there’); the direct object can be a multi-word phrase (e.g., ‘the book that you recommended’); and the adverbial can be a multi-word phrase (e.g., ‘sometime last year’).
This brings us to the all-important concept of phrase structure. In recent decades linguists have shown that words in sentences do not just follow each other in a linear order, like beads on a string. Instead, they form units — phrases — which become parts of larger units/phrases, and so on up to the level of the sentence. And probably the most important discovery is that such phrases all have the same structure schematized below. Each phrase X (X Phrase, or XP for short) consists of a specifier and another sub-unit, which in turn consists of the head and the complement. Importantly, the head is the only obligatory component of a phrase and the only single-word component. Both specifier and complement (differing only in their degree of closeness to the head) are optional and both are potentially multi-word phrases.
While arguing for this blueprint of phrase structure would take me too far afield, this schema is exactly what allows us to recast the verb-second (and other second-position phenomena) in structural terms. If we assume that a sentence — like smaller phrases that constitute it — has this template, we can define “the first position” as the specifier of the XP which is the sentence, and “the second position” — as the head of that XP.
Thus, it is not mysterious at all that language can refer to “the first” and “the second” position without counting, even to 2. But without counting (and given the template above), no grammatical rule can ever refer to “the third”, “the sixth” or any other position!