Lecture 8: Syntax IVLecture 8: Syntax IV[Note: these are lightly edited versions of overheads and notes used in class. They are provided in order to help you study, but they are not a substitute for coming to class and being prepared. The notes are incomplete, and will often be hard to make sense of if you were not in class.]
In the last class we saw how we could create phrase structure trees for Japanese, a language with very different word order patterns from English, by using exactly the same phrase structure rules that we used for English, but reordering the components of the rules (e.g. S --> VP NP instead of S --> NP VP). By varying the order of constituents in two of our rules, we can generate four different word order patterns.
Notice two things here. First, we are unable to generate the OSV word order characteristic of Yoda's speech (look back to Assignment 2). Given that this is reported to be a rare or non-existent word order, this looks like a positive result. Our phrase structure rules explain why a certain word-order pattern is rarely/never found in human languages.
Second, you will also notice that the 4 orders above do not include some fairly widespread word order patterns, such as VSO (Irish, Tagalog, Hebrew etc.) or verb-second (German, Dutch, Karitiana etc.). We will return to this question at the end of the next section.
Question formation in English: there are two basic types of questions in English. One is used for asking questions where the answer is expected to be 'yes' or 'no'. The other is used in situations in which the questioner is looking for the identity of a person, place, time, object, reason etc.
10. yes/no questions
11. Wh-questions
Generating Wh-questions
First try at a phrase structure rule for generating sentences like (12).
But this won't work, because our VP rule tells us that VP can be expanded to V NP. This should make it possible to generate (14), which is clearly ungrammatical.
We could try to get around this problem by specifying in the rule that the VP must be missing one of its arguments.
[We didn't discuss this in class: but what does this example (16) suggest to us about the structure of wh-questions? Clue: when is coordination possible?
The examples in (18-21) show that it won't do us much good to just say that in a wh-question the VP of the main clause must be missing one of its arguments. This is because we can also generate many kinds of wh-questions in which the missing argument is the subject or object of a more deeply embedded clause. In fact, the missing argument can be arbitrarily deeply embedded in the wh-question. The generalization is always that the questioned argument (or modifier, in the case of 'why', 'where', or 'how' questions) is marked by a wh-phrase at the beginning of the sentence and a 'gap' in the sentence where that argument (or modifier) would otherwise appear. This entails a long-distance dependency between the position of the wh-phrase and the position of the gap. It is not something that we are able to express using the phrase-structure rules we have used so far.
We can, however, easily formulate a procedure for generating wh-questions, if we view it as a two-step process.
Procedure for forming wh-questions in English:
Transformational Grammar
The leading idea behind transformational grammar is that we can give a better account of the complexity of natural language syntax if we view it as the output of relatively simple basic phrase structures plus transformational operations.
[Ideas introduced and developed by Noam Chomsky from 1955-present.]
This view is, of course, by no means uncontroversial among linguists, many of whom have preferred the alternative of enriching phrase structure rules so as to avoid the need for transformational operations.
Auxiliary verb movement
Structure-dependency
Based on examples like (25c) it is tempting to think that yes/no questions require that the first auxiliary of the sentence must be moved to the front. However, the examples in (26-28) show that this cannot be correct, because in (27a-b) the first auxiliary of (26a-b) has been fronted, and yet the questions are wildly ungrammatical.
A more appropriate generalization, which accounts for all of the facts in (24-28), is that yes/no questions are formed by moving the highest auxiliary, i.e. the auxiliary of the main clause. What is wrong with the examples in (27) is that the auxiliary that is moved is taken from inside a relative clause that is a modifier of the subject noun phrase.
yes/no-question formation:
Wh-questions again
Cross-linguistic Word Order again...
Verb-initial: e.g. Irish, Hebrew, Arabic, Welsh, Tagalog
Now we can suggest an answer to the question raised earlier about how to account for the existence of VSO word order, which is found in many languages. One possibility would be to say that languages with VSO word order have movement of the verb to the front of the clause in all sentences: in other words, their basic sentence structures would be very similar to the structure of English yes/no questions. In modern syntactic research, this is one of the leading accounts of VSO language structure.
[We did not discuss verb second word order in class, so this is extra information for students who know German or other verb-second languages, and were wondering how we could derive such a word order. One possibility, some version of which is widely adopted in Germanic linguistics, is that basic sentences of verb second languages are formed in much the same way as English wh-questions, by a combination of verb-movement and movement of a NP (or PP) to the start of the sentence.
Verb-second: e.g. German, Dutch