The roots of verbs and most nouns in the Semitic languages are characterized as a sequence of consonants or "radicals" (hence the term consonantal root). Such abstract consonantal roots are used in the formation of actual words by adding the vowels and non-root consonants (or "transfixes") which go with a particular morphological category around the root consonants, in an appropriate way, generally following specific patterns. It is a peculiarity of Semitic linguistics that a large majority of these consonantal roots are triliterals (although there are a number of quadriliterals, and in some languages also biliterals).
The following are some of the forms which can be derived from the triconsonantal root k-t-b (general overall meaning "to write") in Hebrew and Arabic:
Note: The Hebrew fricatives transcribed as "ḵ" and "ḇ" can also be transcribed in a number of other ways, such as "ch" and "v" , which are pronounced [χ] and [v], respectively. They are transliterated "ḵ" and "ḇ" on this page to retain the connection with the pure consonantal root k-t-b. Also notice that in Modern Hebrew, there is no gemination.
In Hebrew grammatical terminology, the word binyan (Hebrew: בנין, plural בנינים binyanim) is used to refer to a verb derived stem or overall verb derivation pattern, while the word mishqal (or mishkal) is used to refer to a noun derivation pattern, and these words have gained some use in English-language linguistic terminology. The Arabic terms, called وزن wazn (plural أوزان, awzān) for the pattern and جذر jaḏr (plural جذور, juḏūr) for the root have not gained the same currency in cross-linguistic Semitic scholarship as the Hebrew equivalents, and Western grammarians continue to use "stem"/"form"/"pattern" for the former and "root" for the latter—though "form" and "pattern" are accurate translations of the Arabic grammatical term wazn (originally meaning 'weight, measure'), and "root" is a literal translation of jaḏr.
Although most roots in Hebrew seem to be tri-radical, many of them were originally bi-radical, cf. the relation between:
The Hebrew root ש־ק־ף - √sh-q-p "look out/through" deriving from ק־ף - √q-p "bend, arch, lean towards" and similar verbs fit into the shaCCéC verb-pattern.
According to a study of the Proto-Semitic lexicon, biconsonantal roots are more abundant for words denoting Stone Age materials, whereas materials discovered during the Neolithic are uniquely triconsonantal. This implies a change in Proto-Semitic language structure concomitant with the transition to agriculture. In particular monosyllabic biconsonantal names are associated with a pre-Natufian cultural background, more than 16,500 years ago.
A quadriliteral is a consonantal root containing a sequence of four consonants (instead of three consonants, as is more often the case). A quadriliteral form is a word derived from such a four-consonant root. For example, the abstract quadriliteral root t-r-g-m / t-r-j-m gives rise to the verb forms תרגם tirgem in Hebrew, ترجم tarjama in Arabic, ተረጐመ "täräggwämä" in Amharic, all meaning "he translated". In some cases, a quadriliteral root is actually a reduplication of a two-consonant sequence. So in Hebrew דגדג digdeg means "he tickled", and in Arabic زلزال zilzāl means "earthquake".
Generally, only a subset of the verb derivations formed from triliteral roots are allowed with quadriliteral roots. For example, in Hebrew, the Piʿel, Puʿal, and Hiṯpaʿel, and in Arabic, forms similar to the stem II and stem V forms of triliteral roots.
Another set of quadrilateral roots in modern Hebrew is the set of secondary roots. A secondary root is a root derived from word that was derived from another root. For example, the root מ-ס-פ-ר m-s-p-r is secondary to the root ס-פ-ר s-p-r. סָפַר saphar, from the root s-p-r, means "counted"; מִסְפָּר mispar, from the same root, means "number"; and מִסְפֶּר misper, from the secondary root מ-ס-פ-ר, means "numbered".
Traditionally in the Semitic languages, forms with more than four basic consonants (i.e. consonants not introduced by morphological inflection or derivation) were occasionally found in nouns — mainly loanwords from other languages — but never in verbs. However, in modern Israeli Hebrew, syllables are allowed to begin with a sequence of two consonants (a relaxation of the situation in early Semitic, where only one consonant was allowed), and this has opened the door for a very small set of loan words to manifest apparent five root-consonant forms, such as טלגרף tilgref "he telegraphed". But, -lgr- always appears as an indivisible cluster in the derivation of this verb, so that these five root-consonant forms do not display any fundamentally different morphological patterns from four root-consonant forms (and the term "quinqueliteral" or "quinquiliteral" would be misleading if it implied otherwise).
In Amharic, there is a very small set of verbs which are conjugated as quinquiliteral roots. One example is wäšänäffärä 'rain fell with a strong wind' The conjugation of this small class of verb roots is explained by Leslau. Unlike the Hebrew examples, these roots conjugate in a manner more like regular verbs, producing no indivisible clusters.