Sentences

Chapter 4 built words from morphemes. This chapter builds sentences from words. The study of how words combine into phrases and sentences is syntax. Its starting observation is that a sentence is not just a string of words in a row: Dog bites man and Man bites dog contain the same words and mean opposite things, and *Bites man dog is not English at all. Speakers carry in their heads a system that decides which combinations work and what they mean, and syntax tries to describe that system.

Word Classes

Words fall into classes (traditionally called "parts of speech") according to how they behave: which other words they combine with and which endings they take. The classes are defined by behaviour, not by meaning. A noun is not "a word for a person, place or thing"; it is a word that can follow the, take a plural -s and serve as the subject of a sentence.

The classic proof that syntax is independent of meaning comes from the linguist Noam Chomsky: Colourless green ideas sleep furiously is nonsense, but it has an English structure (Adj Adj N V Adv). *Furiously sleep ideas green colourless uses the same words in no structure at all. A speaker recognises the first as grammatical and the second as not, even though neither makes sense.

Phrases and Constituents

Words group into phrases, and the groups, called constituents, are the real building blocks of a sentence. A phrase is named after the class of its most important word, its head: a noun phrase (NP) is built around a noun, a verb phrase (VP) around a verb, a prepositional phrase (PP) around a preposition.

How Do We Know a Group Is a Constituent?

Three simple tests, all of which a group of words passes only if it is a unit:

1. Substitution: the group can be replaced by a single word, usually a pronoun. The old dog chased the cat → It chased it. So the old dog and the cat are NPs.
2. Movement: the group can be moved as a block. The cat, the old dog chased works; *Old dog chased the, the cat does not.
3. Coordination: the group can be joined to a similar group with and. The old dog and the young cat works.

Tree Diagrams

Linguists draw sentence structure as a tree, with the whole sentence (S) at the top, which splits into constituents, which split further, down to single words. Here is the sentence The old dog chased the cat as a tree and as a labelled bracketing, which says the same thing in one line:

The tree makes two things visible. First, the sentence has two main parts: a subject NP (the old dog) and a VP (chased the cat). Second, the VP contains another NP (the cat), which is the object. Notice that recursion (Chapter 1) is built in: an NP can contain a PP, which itself contains an NP, as in the dog in the garden of the house, and so on without limit.

Structural Ambiguity

Some sentences are ambiguous not because a word has two meanings but because the words can be grouped in two ways. Structure decides meaning.

The two readings use exactly the same words in the same order. The difference is whether with the telescope attaches to the verb phrase (how I saw) or to the noun phrase (which man). Only the tree tells them apart. Another case is old men and women, which can mean old men plus women, or old men plus old women.

Word Order Across Languages

The basic order of subject (S), verb (V) and object (O) differs across languages. English says Taro ate the apple (SVO). Other languages put the verb elsewhere. Almost all of the world's languages fall into one of the three common orders: SOV and SVO between them cover the large majority, VSO is a distant third, and the other orders are rare.

Word order is not an independent choice. A language that puts its verb after its object (SOV) tends to put its helper words after the noun too: Japanese has postpositions (Tōkyō ni, literally "Tokyo to") where English has prepositions (to Tokyo). Languages tend to be consistent about whether the head of a phrase comes first (English: drink beer, in Tokyo) or last (Japanese: biiru o nomu "beer drink", Tōkyō ni). This regularity was first noticed by the linguist Joseph Greenberg.

German is a mixed case: in a main clause the finite verb comes second (Er trinkt Bier), but in a subordinate clause the verb goes to the end (dass er Bier trinkt).

Case Endings and Free Word Order

English needs strict order because it has no endings to mark who did what to whom: only position tells you the subject from the object. A language with case endings marks the role on the word itself, so position is freer. Esperanto marks the object with -n; Hungarian marks it with -t.

In Esperanto and Hungarian the roles stay fixed whichever way you order the words, and order is used for emphasis instead. This is the accusative -n you met in Esperanto Basic Conversation 1: Mi vidas la hundon and La hundon mi vidas say the same thing. Free order does not mean "anything goes". In both languages the endings must still be right, and some orders sound more natural than others.

Hands-On Exercises