6.4: Semantic Features

6.4.1 From 10.2 Intensions in the Mind, in Anderson's Essentials of Linguistics

Video Script

In the last unit, we saw that one important piece of a word’s meaning is the intension: the attributes or properties in your mind that you use to decide whether a thing in the world can be labelled with that word. In this unit, we’ll think about how those intensions might be organized in the mind.

One theory suggests that intensions might be organized in our minds as sets of binary features. So the intension for the word bird might be made up of features like [+living], [-mammal], [+wings], [+eggs], [+flying]. The intension for the word fish would have some features that are the same as the intension for bird, like [+living], [-mammal], [+eggs]. But the intension for fish would have [-wings] and [-flying]; instead, it would have [+swimming]. Some of these features could be shared across intensions for words that refer to quite different things in the world, so the intension for the word airplane, for example, probably includes [+wings] and [+flying], but [-alive].

The nice thing about using feature composition (also known as componential analysis) to represent intensions is that it can capture some of these similarities and differences across categories of things in the world using the simple, efficient mechanism of binary features. It may well be that our intensions for words describing the natural world are made up of some binary features. But can you think of any problems with this way of organizing meanings? Think about a penguin. A penguin is a member of the category of things that can be labelled with the word bird, and it shares some of the features of the intension for the word bird: it’s a living thing, it has wings, it lays eggs. But a penguin can’t fly. In fact, a penguin has the feature that’s associated with our intension for fish: it can swim. So it’s definitely a bird, but it definitely doesn’t have all the features associated with the intension for bird. If our intensions were organized in our minds just as binary features, then we wouldn’t be able to represent the meaning of the word penguin in our mind, but clearly, we do have an intension for the word penguin. So how might penguins be represented in our minds?

Another theory of intensions suggests that we have fuzzy categories in our minds. These categories contain exemplars, which are basically our memories of every time we’ve encountered an extension of the word. Some members of the category are prototypical exemplars: they have all the typical attributes of members of that category, so they’re near the center of the category. For most North Americans, a robin is about as prototypical as it gets as an exemplar of the category bird. Some exemplars are more peripheral: they have fewer of the defining attributes and they might have some attributes that aren’t typical. So a penguin, for example, is more peripheral because it doesn’t fly, and an ostrich is peripheral because it’s so darn big. Because the category has fuzzy boundaries, we might even have some exemplars in our mind that aren’t really members of the category at all, but share some attributes with category members, like bats: they’re small and they fly, but they’re not actually birds. In the next unit, we’ll talk about some of the evidence we have that our intensions might be organized in fuzzy categories with prototypes.

Check Yourself

Exercise \(\PageIndex{1}\)

Thinking about the category of fruit for most speakers of Mainstream American English, pomelo is probably:

• more peripheral than apple
• more prototypical than apple

Answer

"more peripheral than apple"

Hint: Is pomelo similar to an apple? Possibly not; while they are both fruit, they are different types.

Exercise \(\PageIndex{2}\)

Thinking about the category pets for most speakers of Mainstream American English, tarantula is probably:

• peripheral
• prototypical

Answer

"Peripheral"

Hint: Is a tarantula a typical pet in North America? Not so much.

Exercise \(\PageIndex{3}\)

Which of the following attributes are likely to be part of the average English-speaker's intension for the word pencil?

• ink
• writes
• lead
• erasable

Answer

"Writes" and "Lead" and "Erasable"

Hint: Think about the attributes of a pencil. The only one that doesn't link is ink.

6.4.2 Semantic Features, from Sarah Harmon

Video Script

Catherine did a really great job of explaining semantic features. This is an area that I’ve got quite a bit of experience. Therefore, I'm going to give you a little bit more richness, as it were, with respect to semantic features. Specifically, why it's so important when we are doing any kind of linguistic analysis on a language.

To start off, Catherine talked about common/semantic features. She talked about how we can combine them to describe all sorts of possible setups. One of the things that we use semantic features for is to analyze how a given language groups nouns together. In some languages that's based off of gender; biological gender is the origin and then grammatical gender follows. If you've ever taken a class in an Indo-European language that was not English, and you wondered why a table was feminine or a pencil was masculine or a scooter was neuter, that has to do with semantic features. It's usually based off of whatever patterns you see for biological gender, and then everything else follows suit.

That's not the only way that we can classify or group nouns together. For example, if you learn a Sino-Tibetan language or any of the languages of Southeast Asia, you will run into the setup where you have classifier. I’m going to use Mandarin as an example—I'm not going to pronounce it because I’m going to be terrible at it. For both of these examples, you have a count noun and you have a mass noun, and then in each case you have a different type of classifier that goes with it. If you want to talk about only having one book, you do not in Mandarin say ‘one book’; you say ‘one-the classifier that means that this is a countable thing-book’. However, if you're talking about something that needs more specification—meaning, you need to specify the quantity of some grouping or mass of that item—then you need to use a specific quantify. A classifier that describes what kind of situation you're talking about, so the word for ‘one’, followed by the classifier for box, and then ‘lightbulb’. Note that because there is no inflection in Mandarin for plural, you don’t say ‘one box of light bulbs’, instead you use different phrasing to say ‘this is a specific grouping of lightbulbs’. Instead of tacking on additional information by phrases in Mandarin and other East and Southeast Asian languages, you use a classifier, in this case, the one that means box.

This is also done in a number of Niger-Congo languages, where you have what is frequently called multiple genders, but it really ties into semantic features. You have a morpheme that says that something is animate or not; you have another morpheme that says that it's a plant versus an animal versus some kind of mineral or rock. You might have another thing that says, whether it's female or male, and along it goes.

Semantic features are really helpful when we try to describe a situation that happens with respect to different types of nouns and how they combined with other aspects. You have here family terms at the bottom part of the slide, in the case you see Latin. In Latin, you actually specified what side of the family that the given aunt or uncle was from. That's still really common in a number of languages, although not the Romance languages, as it died out early in their formation. I'm showing you at the bottom of the screen the word for ‘cousin’; there are three different setups. We have an English, which has no differentiation for gender or family side or anything.

In the case of Spanish, you do differentiate for gender, so you do say that this is a female or a male cousin; notice that the terms are exactly the same, just the different vowel the end which signifies gender. In German, you actually have different terms, depending on whether the cousin is female or male.

What is really cool is when this gets showcased in different ways, even within a given language. Many of you are Spanish speakers, either natively or not, and you're looking at these data and they are hard to parse out for you. This is Asturian Spanish. Asturias is the province in the north central part of Spain. They showcase something called a mass neuter, which means that a mass noun that we're talking about it has a different gender associated with it, and you see it in the direct object pronoun. Yes, Spanish speakers, in fact all Romance speakers, you're looking at this a little oddly because you are used to these pronouns going in front of the conjugated verb. That is not the case in Asturian Spanish; it actually goes afterwards, and we can talk more about this later. But if it's a ‘package’, which is masculine and count (you can count it individually, e.g., ‘one package’, ‘five packages’, ‘10 packages’), then the pronoun that gets associated with it is [lu]. I know, Spanish speakers, that's different; just bear with me. If it's a feminine, count noun, like a ‘mare’ (female horse), then the pronoun is [la]—Spanish speakers, you would expect this. However, if it's a mass noun, like ‘grass’—think about it, you can't just say ‘one grass’, you have to say ‘one field of grass’, ‘one blade of grass’—you have to quantify it somehow. The pronoun that goes with it is [lo]. For those who don't know Spanish, [lo] is typically the masculine direct object pronoun. It is not usually used with feminine anything'. In the history of Spanish, and also in a couple of dialects of South-Central Italian and in a couple of dialects of Rhaeto-Romance, which is a Romance language is spoken in the Alps, this phenomenon happens in these remote locations. It's not very common and we don't know where exactly it comes from, because Latin didn't do this and very few dialects of any Romance language do this. There is a history of it in the history in Spanish, going back to the 13th century, and in fact in South central Italian we also see it going back to the 12th and 13th century. We don't really see it much before that, and so we don't really know where it comes from.

What makes this fascinating is to be able to look at the semantic features and how they play in a given language, and in some cases, how one dialect will differ from another.