In the first case - classification of components of language (a) - attention is directed toward a particular construction that arises in language - for example, reflexive verbs, oral stops, or discourse particles. Then, using cross-linguistic data, all the types of these specific phenomena are determined. The goal is to better comprehend how this facet of language operates by identifying the degrees of similarity and the degrees of variance that one finds among languages. There is also keen interest in determining whether there are correlations between the various patterns that one finds in a language.

For instance, we might do a typological investigation on oral plosive sounds.These are sounds, also called “stops,” that are produced when the airstream is completely impeded in the vocal tract, as in English [p] and [g]. If we were to examine the distribution of oral stops in the world's languages, we would immediately be struck by the fact that all languages have at least one plosive sound. Thus, we would have discovered a universal about sound systems in human language. It is important to realize that this fact is not a logical requisite for language because we can easily conceive of a language that does not have any oral plosives. Therefore, our empirical discovery that all languages have at least one stop leads to an ontological question: Why should language be structured in this way? I return to the problem of explanation presently, but first let us determine what other sorts of facts about stops we would learn from our typological investigation.

There are over 50 distinct oral stops that occur in language, but individual languages utilize only a small proportion of this universal set, with languages such as Punjabi (Indo-Iranian: India, Pakistan), which has 24 plosives, being highly exceptional in how many plosives it contains (Gill and Gleason 1963). As we continued our investigation, we would further discover that plosive sounds are not equally distributed in the languages of the world. Some are extremely common, such as [p], [t], and [k]. In fact, nearly all languages have at least one of them. In contrast, some plosives are relatively rare, such as the voiced uvular stop [G], which is found, for example, in Somali (Cushitic: Somalia). We now might notice certain intriguing facts about our emerging typology of plosives such as “unexpected” gaps. For instance, plosives created by bringing the lower teeth into contact with the upper lip are nonexistent, even though they are physically possible to articulate. Finally, we might identify certain stops as being “dominant”—for example, the voiceless alveolar stop [t] appears to be an especially dominant plosive sound because, if a language has only two voiceless stops, one of them is bound to be [t].

From this simple typological study we have learned a host of important facts about sound systems. Not all these facts are of the same sort. For example, some were absolute universals (e.g., all languages have at least one stop); some were universal tendencies (e.g., almost all languages have [p], [t], or [k]); and some were implicational universals (e.g., if a language has two voiceless stops, then one is a [t]). Implicational universals have played a particularly prominent role in typology because they commonly suggest connections between two or more aspects of language.