• No results found

From one perspective, words are nothing special in a usage-based approach to language: If every token of language is stored in a practically unlimited lexicon, and the same language-general processes are used to categorise sounds, words, word chunks, sentences and longer sections of text or speech, there is no reason to see the word as more fundamental than any other larger or smaller portion of language.

Taking a different perspective, several authors have argued that words may be important building blocks: Bybee (2001, p. 30) argues that the word is nevertheless a plausible cognitive entity due to itscognitive autonomy, as it is ‘both phonologically and pragmatically appropriate in isolation’. Similarly, Vihman and Croft (2007, p. 715) argue that a word is ‘the smallest linguistic unit encountered in language use’.

2Vihman and Croft (2007) use the termtemplatesfor these phonological schemas.

2.3. The word 19

2.3.1 Word classes

Bybee’s definition of a word as an autonomous entity is interesting in light of Langacker’s (1987) account of what separates nouns, verbs, adjectives and adverbs: He argues that grammatical cat-egories are semantically definable by theirconceptual autonomy. According to Langacker (1987) a noun designates athing,3 that is, an entity that may be described as a region that is bounded within a domain. As he exemplifies, abeepis a bounded region in time (being relatively short) and pitch (having a clear tone). In the case of concrete or abstract mass nouns, likefurnitureor pain, the region is bounded by effective homogeneity; there may be internal variability (chairs are different from lamps), but we may neverthelessconstruethe mass as homogeneous as long as the variability is not seen as important or relevant. Being a bounded region, a noun is thus conceptually autonomous; it can be conceptualised by itself.

This property sets nouns apart from verbs, adjectives and adverbs: A verb is an entity that designates a [PROCESS], defined as a relation with a positive temporal profile (Langacker, 1987).

An adjective or adverb designates an [ATEMPORAL RELATIONSHIP]; adjectives relate tothings, whereas adverbs relate to other relations (i.e. to a verb, an adjective or another adverb). Processes and atemporal relationships are hence both characterised byconceptual dependence: ‘[O]ne can-not conceptualize interconnections without also conceptualizing the entities that they intercon-nect’ (Langacker, 1987, p. 215). For instance, when we conceptualisethrow, we also concep-tualise the [THING] throwing and the [THING] being thrown, and even our understanding of a colour term such as red will depend on the [THING] it relates to – a red apple has a different colour than red hair.

A process may in turn be profiled as a [PERFECTIVE PROCESS], one that changes through time (e.g.break) or as an [IMPERFECTIVE PROCESS], one that stays constant through time (e.g.

meditate). These two sub-schemas (instances) share a temporal profile, as they both profile a series of component states scanned in a sequential fashion (Langacker, 1987). As a metaphor, we may think of actions in terms of a cartoon of the component states making up the process:

In the case of a perfective process, the picture in the cartoon are profiled as differing from each other, but in the case of an imperfective process, all pictures are profiled as identical.

Langacker (1987, p. 249) suggests a taxonomy of word classes based on these schemas for the semantic pole of nouns, verbs, adjectives and adverbs. Here, processes (verbs) and atemporal relationships, are seen as instances of the schema [RELATION], given their relational profile.

Together with the schema [THING] for nouns, [RELATION] then gives rise to the superschema [ENTITY] (see figure 2.4 on the following page).

3Langacker’snouncategory is wide, covering both nouns, noun phrases, proper names and pronouns, as any of these may be said to profile athing(Taylor, 2002, p. 345).

20 Chapter 2. Theoretical framework

Figure 2.4: Schematic hierarchy of basic classes. Adapted from Langacker (1987, p. 249).

2.3.2 Children’s acquisition of word classes

From the categorisation in figure 2.4, we may then posit that children may acquire nouns easily, as they are conceptually autonomous, whereas verbs and adjectives will be harder to acquire as their semantics are dependent on other words. Gentner (1982) suggests a related, but slightly dif-ferent account, stated asthe Natural Partitions Hypothesis: She posits that nouns prototypically denote concrete objects, whereas verbs prototypically denote abstract and transient events. Due toperceptual-conceptual properties, concrete objects are more likely to be perceived as entities than transient and abstract events are.

As a result of these properties, then, the word–meaning (or word-to-world) mapping is more transparent for nouns than for verbs. Gentner and Boroditsky (2001) add function words to the equation, and suggest a gradient scale of transparency: At the transparent end, we find proper nouns, which show a strong cognitive dominance, as their primary function is to denote unique entities in the real world. At the opaque end, we find function words likethe and and, which show a strong linguistic dominance, as their meanings depend on the linguistic context. Verbs and adjectives are then somewhere in the middle.

Tomasello leans on Langacker’s (1987) word-class schemas as well as on the Natural Par-titions Hypothesis of Gentner (1982) and Gentner and Boroditsky (2001). However, with his focus on the social aspect of language acquisition, he argues that the differences between word classes may not primarily relate to what words children are able to acquire – after all, many of their first words are social terms, not nouns. Rather, he suggests a social-pragmatic modification to Gentner’s hypothesis:

The modified hypothesis is that children learn words most readily in situations in which it is easiest to read the adult’s communicative intentions. Thus, in the right situation they can learn event-type nouns such as breakfast, performatives such as no, and some verbs and other relational words. But concrete nouns, with percept-ible referents, are often used in pragmatically simple situations, in which the adult’s communicative intentions are especially clear – for example, in handling objects or pointing out new objects for shared inspection. (Tomasello, 2003, p. 49)