Skip to content

02 - Word classes

Words on the page

You might think that words are "things which are separated by a spaces on the page"

But spaces are not necessarily a good indicator of wordhood

E.g. Different dictionaries will list nighttime, night-time or night time. So conventions are fairly arbitrary.

Something which are multiple words on a page, actually behave semantically like a single word

e.g. He has a lot of joie de vivre

Definition

Working definition is "Smallest independent unit of meaning"

Smallest unit

A "minimal" unit = we cannot split it into smaller pieces.

  1. Inactive -> *One sip of coffee and Charles changes from in- to hyper-active.
    It's unsplittable, and therefore a single word.
  2. Underconfident -> ?After eight pints, Angela changes from under- to over-confident.
    It's unsplittable, and therefore a single word.
  3. Walk out -> She walked through the door and out of my life.
    You can split it, and therefore it's two words.
  4. Blackbird (species) -> *She saw a black and a white bird in the tree.
    When we split the word, the sentence is grammatical, but note that black has lost its meaning to denote a species of bird. Therefore it is a single word.
  5. Black magic -> ?The witch practises both black and white magic
    A tricky one. It sounds just about okay, so best categorised as two words. But this is a judgement call.

Independent

This property can be defined in a couple of ways:

(a) Words are interpretable in isolation. They may "stand alone".

  1. A: What was he doing? B: Working
  2. B: How would you describe the experience? B: Unbelievable
  3. Shark!!!!!!!!
  4. Swim!!!!!!!!!

(b) Words can move around the sentence, e.g.

  1. The dog chased the cat
  2. The cat chased the dog
  3. He loves being with his dog

Symbolic

Words have meanings by virtue of an arbitrary relationship between form and function (or sound and meaning)

To test this, we can take a single concept, and look at words to describe this concept across languages. The words will have very different sound properties.

The property of symbolism was listed by the linguist Charles Sanders Pierce (1839 - 194) as a key defining property of human languages

A dirty secret - there is no adequate definition of a word

All of the properties which are used to define a word: minimality, independence, and symbolism, are in fact gradable.

For example black magic is weakly minimal as when the two units are separate it sounds ever so slightly odd, e.g. ?The witch practises both black and white magic

Independence is also highly gradable. Both mood and tizz occur with similar meanings in the phrase, He is in a mood / tizz. But only mood can easily move around a sentence, e.g. His mood was dark. Moroever, if someone were to say tizz in isolation and ask you if it's a word you might find this difficult. You would probably say "yes" on the basis that it occurs in the phrase in a tizz but you would need to think carefully about this.

Some types of word classes lack independence, e.g. determiners such as the.

Finally, symbolism is also gradient. There is lots of interesting work demonstrating that sound symbolism is quite widespread in the world's languages. For example, across languages the word for `tongue' often contains /l/ (Blasi et al., 2016).

So when in clinic you are analysing a sentence, and you're deciding to determine whether a lexical unit is a real word, there may not be a perfect yes/no answer!

Word classes

Words belong to word classes / syntactic categories

The word class determines how we put words together to make sentences

Semantic definitions - why the don't work very well

Typical examples:

A verb is a DOING word An adjective is a DESCRIBING word

But are the following adjectives or verbs?

  1. The raging river tore through the valley
  2. That film really sucks

These examples demonstrate that semantic/functional definitions of words fall short

3 ways to define word classes

(a) Semantics / meaning-based test, e.g. a verb is a 'doing' word.

(b) Morphological test, e.g. we can make a verb progressive by adding -ing

(c) Positional test, e.g. verbs come after subjects and before objects, e.g. she ate the fish, she likes cheese.

So which test?

The positional test is the most reliable, e.g.

  1. The raging river tore through the valley: raging comes before a noun. It is an adjective
  2. That film really sucks: comes after the subject film. It is a verb.

Let's meet the word classes!

Categorization of word classes

  • Content versus function words

Content words = Nouns, Verbs, Adjectives, Adverbs (in Speech and Language Therapy, these are sometimes called 'information-carrying words')

Function words = Determiners, Pronouns, Prepositions (and maybe certain types of verbs, we will come to this later)

  • Open versus closed class.

Open class words permit new members, while closed class words don't. NB all content words are open, and all function words are closed.

Open = Nouns, Verbs, Adjectives, Adverbs

Closed = Determiners, Pronouns, Prepositions.

  • What the words do in the sentence?

Referring words = Nouns, Pronouns

Modifying words = Adjectives and Adverbs

  • Where do they appear in the sentence?

The verb complex tends to contain auxiliary verbs and adverbs in addition to the verb itself

The Noun Phrase tends to contain determiners and adjectives in addition to the noun itself

Prepositions are used before the Noun Phrase.

  • Super- and sub-types

Types of modifying words: adjectives and adverbs

Types of referring words (nominals): Nouns and Pronouns

Word classes and language impairment

Children with language impairments

  • Tend to miss out function words
  • Have difficulties understanding prepositions (when used to express location), e.g. on, in
  • Have reduced verb diversity
  • Sometimes omit arguments of the verb, e.g. 'She put the books'

5 minutes exercise

What's unusual about the following sentence: "That film was absobloodylutely brilliant!"?

A student says 'I am going to a lection'. What does 'lection' mean, and how has this error arisen?

Bibliography

Blasi, D. E., Wichmann, S., Hammarstrom, H., Stadler, P. F., \& Christiansen, M. H. (2016). Sound-meaning association biases evidenced across thousands of languages. Proceedings of the National Academy of Sciences, 113(39), 10818-10823. https://doi.org/10.1073/pnas.1605782113