For a language consciously designed by people, see Constructed language.
A synthetic language uses inflection or agglutination to express syntactic relationships within a sentence. Inflection is the addition of morphemes to a root word that assigns grammatical property to that word, while agglutination is the combination of two or more morphemes into one word. The information added by morphemes can include indications of a word's grammatical category, such as whether a word is the subject or object in the sentence. Morphology can be either relational or derivational.
While derivational morpheme changes the lexical categories of words, inflectional morpheme does not. In the first example below, faster remained an adjective when followed by the suffix; however, teacher becomes a noun after the suffix is added. Therefore, the first case is an example of inflection and the latter derivation.
In synthetic languages, there is a higher morpheme-to-word ratio than in analytic languages. Analytic languages have a lower morpheme-to-word ratio and higher use of helping verbs and word order. The four subtypes of synthetic languages are agglutinating languages, fusional languages, polysynthetic languages, and oligosynthetic languages.
Language exhibits synthesis in two ways: derivational and relational morphology. These methods of synthesis refer to the ways in which morphemes, the smallest grammatical unit in a language, are bound together. Derivational and relational morphology represent opposite ends of a spectrum; that is, a single word in a given language may exhibit varying degrees of both of them simultaneously. Similarly, some words may have derivational morphology while others have relational morphology. Some linguists, however, consider relational morphology to be a type of derivational morphology, which may complicate the classification.
In derivational synthesis, morphemes of different types (nouns, verbs, affixes, etc.) are joined to create new words. That is, in general, the morphemes being combined are more concrete units of meaning. The morphemes being synthesized in the following examples either belong to a particular grammatical class - such as adjectives, nouns, or prepositions - or are affixes that usually have a single form and meaning:
In relational synthesis, root words are joined to bound morphemes to show grammatical function. In other words, it involves the combination of more abstract units of meaning than derivational synthesis. In the following examples note that many of the morphemes are related to voice (e.g. passive voice), whether a word is in the subject or object of the sentence, possession, plurality, or other abstract distinctions in a language:
Agglutinating languages have a high rate of agglutination in their words and sentences, meaning that the morphological construction of words consists of distinct morphemes that usually carry a single unique meaning. These morphemes always look the same no matter what word they are in, so it is easy to separate a word into its individual morphemes. Note that morphemes may be bound (that is, they must be attached to a word to have meaning, like affixes) or free (they can stand alone and still have meaning).
Fusional languages are similar to agglutinating languages in that they involve the combination of many distinct morphemes. However, morphemes in fusional languages are often assigned several different lexical meanings, and they tend to be fused together so that it is difficult to separate individual morphemes from one another.
Oligosynthetic languages are a theoretical notion created by Benjamin Whorf with no known examples existing in natural languages. Such languages would be functionally synthetic, but make use of a very limited array of morphemes (perhaps just a few hundred). Whorf proposed that Nahuatl was oligosynthetic, but this has since been discounted by most linguists.
Synthetic languages combine (synthesize) multiple concepts into each word. Analytic languages break up (analyze) concepts into separate words. These classifications comprise two ends of a spectrum along which different languages can be classified. The present-day-English is seen as analytic, but it used to be fusional. Certain synthetic qualities (as in the inflection of verbs to show tense) were retained.
The distinction is, therefore, a matter of degree. The most analytic languages consistently have one morpheme per word, while at the other extreme, in polysynthetic languages such as some Native American languages  a single inflected verb may contain as much information as an entire English sentence.
In order to demonstrate the nature of the analytic–synthetic–polysynthetic classification as a "continuum", some examples are shown below:
|Literal translation||tomorrow day||I||of||friend friend||will||for||I||do||birth day||egg cake|
|"Tomorrow my friend(s) will make a birthday cake for me."|
However, with rare exceptions, each syllable in Mandarin (corresponding to a single written character) represents a morpheme with an identifiable meaning, even if many of such morphemes are bound. This gives rise to the common misconception that Chinese consists exclusively of "words of one syllable". As the sentence above illustrates, however, even simple Chinese words such as míngtiān 'tomorrow' (míng "bright" + tīan "day") and péngyou 'friend' (a compound of péng and yǒu, both of which mean 'friend') are synthetic compound words. This is of interest for reconstruction of hypothesized nostratic and proto-world languages.
The Chinese language of the Classic works, and of Confucius for example, is more strictly monosyllabic (and southern dialects to a certain extent): each character represents one word. The evolution of modern Mandarin Chinese was accompanied by a reduction in the total number of phonemes. Words which previously were phonetically distinct became homophones. Many disyllabic words in modern Mandarin are the result of joining two related words (such as péngyou, literally "friend-friend") in order to resolve the phonetic ambiguity. A similar process is observed in some English dialects. For instance, in the Southern dialect of American English, it is not unusual for the short vowel sounds ĕ and i to be indistinguishable before nasal consonants: thus the words "pen" and "pin" are homophones (see pin-pen merger). In this dialect, the ambiguity is often resolved by using the compounds "ink-pen" and "stick-pin", in order to clarify which "p*n" is being discussed.
It has been argued that analytic grammatical structures are easier for adults learning a foreign language. Consequently, a larger proportion of non-native speakers learning a language over the course of its historical development may lead to a simpler morphology, as the preferences of adult learners get passed on to second generation native speakers. This is especially noticeable in the grammar of Creole languages. A 2010 paper in PLOS ONE suggests that evidence for this hypothesis can be seen in correlations between morphological complexity and factors such as the number of speakers of a language, geographic spread, and the degree of inter-linguistic contact.