Given linguistic units U_u and U_m which are equal except that U_m has a feature which U_u lacks; then U_m is marked against unmarked U_u. This holds both for distinctive and for significative units.

Markedness in distinctive units

As an example of phonological markedness, consider the pair of [a] and [ã], as in .

They share all their phonetic features except that [ã] is nasal, which [a] is not. Thus, [ã] is marked against [a].

Likewise, the pair [s] and [z], as in , shares the entire set of articulatory features and differs only by the additional vibration of the vocal chords characterizing [z]. Thus, [z] is marked against [s]. Markedness implies additional internal complexity, with the consequence of a more limited distribution and a lower overall frequency.

Markedness in significative units

In significative units, the marked member of a pair bears an additional grammatical or semantic feature providing a certain bit of information, while the unmarked member by itself provides no information in the relevant respect, providing the opposite information only in contrastive contexts. The category of number is a case in point. The nominal expression wife in a is in the singular, but this says nothing about the number of persons meant.

Again, in b it is understood that there is one wife per person, and this works specifically by contrast with the plural expression wives in #c. The plural is marked against the singular by its additional feature of plurality.

Another example is found in tense. The past tense used in b limits the assertion to some time in the past. The present tense used in #a does not limit the assertion to present time; it is timeless.

The past tense is marked against the present tense by specifying past time as the period for which the predication holds. The unmarked present tense does nothing of the kind and refers to present time only if a contrast to other points of time arises.

The feature distinguishing the marked member may also be a semantic one in a lexical opposition. In pairs of antonymous adjectives, commonly one member designates a pole of the dimension in question, while the other member covers the entire dimension and designates the opposite pole only by contrast. For instance, the adjective old covers the dimension of age, while the antonymous adjective young designates the low pole of the dimension. Therefore old is used in such constructions as in , while young is not generally so used; and if it is (), it implies reference to the low pole of the age continuum.

Consequently, in this antonymous pair, old is unmarked, young is marked. Other pairs of antonymous adjectives illustrate the same principle. Thus short is marked against long by the additional feature of designating the pole of the continuum of length, while long is used to refer to the continuum as such.

Therefore long is normal in , while short in the same contexts, as in , sounds weird and requires special contextual conditions to be acceptable.

For significative units, the same markedness principles hold as for distinctive units: marked units have a more limited distribution and an overall lower frequency. Singular noun forms are several times more frequent in texts than plural forms; and present verb forms are much more frequent than past-tense forms. Likewise, the unmarked member of an antonymous pair appears in contexts where the opposition is neutralized, as in and , and is textually much more frequent than the marked member.

In the clearest and simplest cases, the marked member of an oppositive significative pair bears an additional element on the expression side which codes the additional functional feature. This is the -s suffix for plural, which has no counterpart in the singular, and the -ed suffix for past tense, which has no counterpart in present tense. Putting it simply, additional information is imparted by additional expression elements. However, this is not necessarily so, as shown by the examples of lexical markedness. The additional feature characterizing the marked member may be not separately represented in the expression and manifest itself only by the fact that the distribution of this member is properly included in the distribution of the unmarked member.

Unilateral foundation

If complexity is built up gradually, the marked member of a pair presupposes presence of the unmarked member in the system. Thus, to take up the examples given, a phonological system can include nasal vowels like /ã/ only if it already possesses oral vowels like /a/; and it can include voiced obstruents like /z/ only if it already possesses voiceless ones like /s/. Likewise, plural marking in a language presupposes the presence of a numberless form, and past-tense marking presupposes a form neutral to tense. There are entire languages which lack nasal vowels, voiced obstruents, number or tense, while all languages have oral vowels and voiceless obstruents, numberless and tenseless forms. Again, there are languages which lack the categories of number or tense. Finally, to conceive a property that varies continually between two poles, one adjective suffices. While English has the antonymous pair of marked shallow contrasting with unmarked deep, Spanish only has profundo ‘deep’ opposed to poco profundo ‘not too deep’.

This build-up of complexity is regulated by the principle of unilateral foundation: presence of the marked member of an opposition in a system presupposes presence of the unmarked member (Jakobson 1941). Thus, a language first acquires the unmarked member, then the marked member of an opposition. This principle is operative in language evolution, language acquisition, language loss and language change.