An infix is an affix inserted inside a word stem (an existing word or the core of a family of words). It contrasts with adfix, a rare term for an affix attached to the outside of a stem such as a prefix or suffix.[note 1]
When marking text for interlinear glossing, most affixes are separated with a hyphen but infixes are separated with ⟨angle brackets⟩.
The present tense of some Proto-Indo-European verbs, in the case of a certain number of roots, adds a nasal infix (m, n) to the basic root. The stems of the other tenses have the root without the infix, and thus these verbs are called nasal-presents. This phenomenon is inherited, and preserved to varying degrees, by some of the early daughter languages such as Sanskrit, Ancient Greek, Latin language, etc.
In Nicaraguan, Costa Rican, and Honduran Spanish, the Spanish diminutive affix becomes an infix ⟨it⟩ in names: Óscar [ˈoskar] → Osquítar [osˈkitar] (cf. standard Oscarito); Edgar → Edguítar; Victor → Victítor.
Arabic uses a common infix, ⟨t⟩ ت for Form VIII verbs, usually a reflexive of Form I. It is placed after the first consonant of the root; an epenthetic i- prefix is also added since words cannot begin with a consonant cluster. An example is اجتهد ijtahada "he worked hard", from جهد jahada "he strove". (The words ijtihad and jihad are nouns derived from these two verbs.)
Infixes are common in Austronesian and Austroasiatic languages. For example, in Tagalog, a grammatical form similar to the active voice is formed by adding the infix ⟨um⟩ near the beginning of a verb. The most common infix is ⟨in⟩ used to make an intentional verb, as in 'giniba', meaning ‘ruined’ (from ‘giba’, an adjective meaning ‘worn-out’); 'binato’, meaning ‘stoned’ (from ‘bato’, ‘stone’); and 'ginamit’, meaning ‘used’. Tagalog has borrowed the English word graduate as a verb; to say "I graduated" a speaker uses the derived form grumaduate.
Khmer, an Austroasiatic language, has seven different infixes. They include the nominalizing infix ⟨b⟩, which derives lbɨən 'speed' from lɨən 'fast' and lbɑɑng ' trial' from lɔɔng 'to test, to haunt', or the agentive ⟨m⟩ deriving cmam 'watchman' from cam 'to watch'. These elements are no longer productive, and occur crystallized in words inherited from Old Khmer.
In Seri, some verbs form the plural stem with infixation of ⟨tóo⟩ after the first vowel of the root; compare the singular stem ic 'plant (verb)' with the plural stem itóoc. Examples: itíc 'did s/he plant it?' and ititóoc 'did they sow it?'.
Tmesis, the use of a lexical word rather than an affix, is sometimes considered a type of infixation. These are the so-called "expletive infixes", as in abso-bloody-lutely. Since these are not affixes, they are commonly disqualified from being considered infixes.
Sequences of adfixes (prefixes or suffixes) do not result in infixes: An infix must be internal to a word stem. Thus the word originally, formed by adding the suffix -ly to original, does not turn the suffix -al into an infix. There is simply a sequence of two suffixes, origin-al-ly. In order for -al- to be considered an infix, it would have to have been inserted in the non-existent word *originly. The "infixes" in the tradition of Bantu linguistics are often sequences of prefixes of this type, though there may be debate over specific cases.
The Semitic languages have a form of ablaut (changing the vowels within words, as in English sing, sang, sung, song) that is sometimes called infixation, as the vowels are placed between the consonants of the root. However, this interdigitation of a discontinuous root with a discontinuous affix is more often called transfixation.
When glossing, it is conventional to set off infixes with ⟨angle brackets⟩, rather than the hyphens used to set off prefixes and suffixes:
which contains the suffix -ly added to the word original, which is itself formed by adding the suffix -al to the root origin.