Languages differ based on how they package meaning into a single morpheme, and how they package morphemes into a word. Traditionally, languages have been classified into four morphological types based on the structure of the word: isolating, agglutinative, fusional, and polysynthetic.

Traditional morphological types

Isolating

Agglutinative

In agglutinative languages, there is a much higher morpheme-to-word ratio. In other words, there are many morphemes in a single word. Usually, there is little interaction between each morpheme and each morpheme contains a single piece of grammatical information; in other words, there is more-or-less one-to-one matching between morphemes and meaning. However, there is still usually only one root morpheme in most words (with the exception of compound words).

Turkish, as shown in Table 1, is an example of an agglutinative language.

In Turkish, nouns inflect for number (singular or plural) and case (nominative, definite-accusative, etc.). Case marks for the role that the noun plays in the sentence (e.g., subject, object, object of preposition, etc.). Each morpheme in Table 1 has exactly one meaning. For example, -ler means plural, -i means definite-accusative, and so on. To form a word in Turkish, multiple morphemes are strung together.

Fusional

Fusional languages, on the other hand, have morphemes that express multiple pieces of grammatical information simultaneously. There is almost no one-to-one match between pieces of meaning and number of morphemes. There also tend to be multiple morphemes in a single word.

Latin is an example of a fusional language, as shown in Table 2.

Latin nouns, like Turkish, also inflect for case and number. However, unlike Turkish, we cannot separate the parts of the word that indicate number and the parts that indicate case. The morpheme -am, for example, simultaneously encodes both accusative and singular, while the morpheme -ārum simultaneously encodes both genitive and plural. We cannot separate either -am nor -ārum into two parts; they are each one morpheme with two meanings. There is no morpheme that repeats down either the singular or plural columns, nor any morphemes that repeat across each case row.

Polysynthetic

Finally, polysynthetic languages often have a single word representing meanings that are associated with whole sentences in other languages. These words contain long strings of roots and affixes. Often the object, and sometimes the subject, are inflected onto the verb.

Inuktitut is an example of a polysynthetic language, as shown in (2).

(Inuktitut; Compton and Pittman 2010: 2168)

Both of the Inuktitut words in (2) contain multiple morphemes. The example in (2a) contains root words that are associated with more than one part of speech, including the verbs ayag- ‘go’ and ni- ‘claim’ and the adverbial yugnarqe- ‘probably’. The example in (2b) contains inflectional morphemes that are associated with nouns, like the nominative case marker vi- as well as inflectional morphemes that are associated with verbs, like the declarative first-person singular marker -tunga. Each of these words are translated into a whole English sentence. Example (2a) even has two clauses in the translation!

In speech, we don’t usually pause in between words, so why do we think that the examples in (2) are each single words?  There are a few reasons. First, speakers have intuitions that they are single words. Second, the morphemes are mostly bound morphemes that cannot stand on their own. Third, many phonological processes cross morpheme boundaries but do not cross word boundaries, giving us a big clue about where word boundaries are.

Polysynthetic languages are sometimes confused with agglutinative languages. Unlike agglutinative languages, a polysynthetic language will often have more than one root. Compound words, you may recall, also have more than one root. However, unlike compound words, words in polysynthetic languages contain inflectional morphemes from more than one part of speech in a single word. Another potential difference between polysynthesis and compounding is that polysynthesis is productive, while compounding varies in productivity across languages.

Synthesis, fusion, and exponence

The four morphological types above really conflate three properties: synthesis, fusion, and exponence. Not all languages fit perfectly into one of the four traditional categories, so it is sometimes useful to consider each of these properties independently. Some languages may even differ depending on the kind of morpheme. For example, Chapter 21 of the World Atlas of Language Structures (WALS) surveys how different languages package both noun and verb inflection. If you cross-reference the two maps associated with Chapter 21, you’ll find that some languages, including Chukchi, Finnish, German, West Greelandic, Nenets, Russian, and Yaqui, combine case with number in a single morpheme, similar to the Latin examples in Table 2, but do not combine their tense markers with another category. Conversely, you’ll find that other languages, including Aymara, Brahui, Georgian, Hindi, Spanish, and Tibetan combine their tense markers with agreement in a single morpheme but do not combine case with another category (Bickel and Nickels 2013b). It may be enough to classify both of these sets of languages as fusional in some situations, but in other situations it may be useful to make more nuanced distinctions.

The four traditional categories are discrete categories, meaning a language either fits into the category or does not. The three categories of synthesis, fusion, and exponence, on the other hand, are scales, which means that a language might fall on the high end of the scale, on the low end of the scale, or somewhere in the middle.

Synthesis

Synthesis is the measure of how many morphemes may combine into a single word. A language with a high degree of synthesis will have many morphemes contained into a single word, while a language with a low degree of synthesis will have few morphemes combined into a single word. Languages with a low degree of synthesis are often called analytic languages.

Fusion

Fusion is the measure of how phonologically separable morphemes are from their hosts. In a language with a low degree of fusion, it is easy to identify the boundaries between different morphemes. Isolating languages can be classified as low fusion languages because most of the morphemes are separated by a word boundary, which is a very clear phonological boundary. Agglutinative languages also have a low degree of fusion. On the other hand, a language with a high degree of fusion will have morpheme boundaries that are less clear. English exhibits a higher degree of fusion in some irregular verb inflections, such as catch ~ caught, fall ~ fell, and bend ~ bent. In these forms, there is no part of the word that we can clearly identify as either a present tense or past tense morpheme. Semitic languages such as Hebrew and Arabic are well-known for using vowel changes as one of their main morphological processes (instead of just in irregular verbs like English), and so would be classified as having an even higher degree of fusion than English.

Note that this meaning of fusion is very different from the category fusional discussed above, which is more closely related to the concept of exponence.

Exponence

Exponence is the measure of how many meanings can be encoded into a single morpheme. Wari’, a Chapacura-Wanham language spoken in Brazil, has high exponence on the word na, which encodes the person and number of the subject, as well as several categories of verb inflection (realis, non-future, and active), for a total of five grammatical categories expressed in a single morpheme.  When multiple meanings are encoded in a single morpheme in this way, it is known as cumulative exponence.

(Wari’; Everett and Kern 1997: 339, as cited in Bickel and Nickels 2013b)

Low exponence, on the other hand, will only have one meaning encoded in each morpheme.

Redefining the traditional categories

Let us now revisit the four traditional morphological categories and redefine them in terms of their synthesis, fusion, and exponence.

Isolating languages have one or close to one morpheme per word, which means they have low synthesis. The morphemes, belonging as they do to independent words, are phonologically separable, and therefore isolating languages have low fusion as well. Finally, most morphemes in isolating languages encode only one meaning, and so they also have low exponence.

Agglutinative languages have many morphemes per word, and therefore have high synthesis. However, the morphemes are still phonologically separable and encode one meaning per morpheme; in this way, agglutinative languages have low fusion and low exponence.

Fusional languages are characterized by having many meanings encoded into a single morpheme, which means they have high exponence. The definition of fusional languages doesn’t specify their degree of synthesis or fusion, but fusional languages will tend to have fewer morphemes per word than agglutinative languages since each morpheme is carrying more information.

Finally, polysynthetic languages are highly synthetic. The definition of polysynthesis does not specify the degree of fusion or exponence, but polysynthetic languages tend to have a small degree of fusion, with phonological processes crossing morpheme boundaries, and medium amounts of exponence.

Check yourself!

References and further resources

Comedy and satirical linguistics

🎉 Phlogiston, Phineas Q. 2007. Cartoon theories of linguistics part 3: Morphological typology. Speculative Grammarian CLII (3). https://specgram.com/CLII.3/09.phlogiston.cartoon.3.html.

Sources for examples

Compton, Richard, and Christine Pittman. 2010. Word-formation by phase in Inuit. Lingua 120: 2167–2192.

Lieber, Rochelle. 2016. Introducing Morphology. 2nd edition. Cambridge: Cambridge University Press.

Academic sources

🔍 Bickel, Balthasar, and Johanna Nichols. 2013a. Fusion of Selected Inflectional Formatives. In WALS Online (v2020.3) [Data set], ed. Matthew S. Dryer and Martin Haspelmath. Zenodo. https://wals.info/chapter/20.

🔍 Bickel, Balthasar, and Johanna Nichols. 2013b. Exponence of Selected Inflectional Formatives. In WALS Online (v2020.3) [Data set], ed. Matthew S. Dryer and Martin Haspelmath. Zenodo. https://wals.info/chapter/21.

🔍 Bickel, Balthasar, and Johanna Nichols. 2013c. Inflectional Synthesis of the Verb. In WALS Online (v2020.3) [Data set], ed. Matthew S. Dryer and Martin Haspelmath. Zenodo. https://wals.info/chapter/22.

Mattissen, Johanna. 2004. A structural typology of polysynthesis. Word 55 (2).