Skip to main content
v2026.11,986 entries · CC-BY 4.0

Linguistic Typology: Definition, Word-Order Types and How Typologists Count

Linguistic typology compares languages by structure. See WALS word-order counts turned into shares, with the arithmetic, and R and Python code.

Written and maintained by CASRAI Editorial Board

Published

On this page

Linguistic typology is the branch of linguistics that compares languages by their structural properties, such as word order or the size of a sound inventory, instead of by shared ancestry, in order to describe how languages vary and which patterns recur across them. The Association for Linguistic Typology describes its aim as advancing “the scientific study of typology, that is, of cross-linguistic diversity and the patterns underlying it.” Because typological claims are claims about counts across a sample of languages, the field is quantitative in practice: this guide defines the discipline, then works through real counts from the World Atlas of Language Structures (WALS) to show how typologists turn them into proportions and how to read, report and reproduce them.

What is linguistic typology?

Typology sorts languages into types according to what their grammars, sound systems or vocabularies do. A typologist asks a question such as “where does the verb sit relative to its subject and object?” or “how many consonants does the language distinguish?”, collects the answer for many languages, and then looks at how the answers are distributed. The result is a map of cross-linguistic diversity: which options are common, which are rare and which combinations of features tend to appear together.

The discipline has its own learned society and journal, both named Linguistic Typology (the society is the Association for Linguistic Typology, ALT), and its own reference works, including The Cambridge Handbook of Linguistic Typology, edited by Alexandra Y. Aikhenvald and R. M. W. Dixon (Cambridge University Press, 2017). It sits inside the wider field described in what linguistics is and what linguists study, and it is distinct from the study of language in social context covered in what sociolinguistics is (see also linguistics vs sociolinguistics).

Typological classification versus genealogical classification

A genealogical classification groups languages because they descend from a common ancestor. A typological classification groups them because they share a structural property, whatever their ancestry. Two languages can therefore fall in the same type without being related, and two closely related languages can fall in different types. This is why typological work needs a sample designed to avoid counting one family many times over, a problem treated in the guide to language sampling methods.

Basic word order types

The best-known typological variable is basic word order: the usual relative order of subject (S), object (O) and verb (V) in a simple declarative clause. Logically there are six orders: SOV, SVO, VSO, VOS, OVS and OSV. WALS Map 81A, written by Matthew S. Dryer, adds a seventh value for languages with no dominant order and reports the following counts for a sample of 1,376 languages: SOV 564, SVO 488, VSO 95, VOS 25, OVS 11, OSV 4, and no dominant order 189. The counts add up to the stated total: 564 + 488 + 95 + 25 + 11 + 4 + 189 = 1,376. The data table below the guide gives each count with its share of the sample.

Two features of these counts matter for interpretation. First, two orders (SOV and SVO) account for most of the sample while three orders (VOS, OVS, OSV) are each under 2.1% of it. Second, the “no dominant order” value is a real category, not missing data: 189 languages, or 13.7% of the sample, were coded as lacking a dominant order, so any percentage must state whether those languages are in the denominator.

Worked example: turning WALS counts into proportions

A proportion is the count of languages with a value divided by the number of languages in the denominator you chose, multiplied by 100 to give a percentage. In symbols, p = n / N × 100, where n is the number of languages with the value and N is the number of languages in the sample you are describing. The arithmetic below uses the WALS 81A counts.

  1. Choose the denominator. With all 1,376 sampled languages, N = 1,376. With only the languages that have a dominant order, N = 1,376 − 189 = 1,187.
  2. SOV with N = 1,376: 564 ÷ 1,376 = 0.4099, so 41.0%. With N = 1,187: 564 ÷ 1,187 = 0.4751, so 47.5%.
  3. SVO with N = 1,376: 488 ÷ 1,376 = 0.3547, so 35.5%. With N = 1,187: 488 ÷ 1,187 = 0.4111, so 41.1%.
  4. VSO with N = 1,376: 95 ÷ 1,376 = 0.0690, so 6.9%. With N = 1,187: 95 ÷ 1,187 = 0.0800, so 8.0%.
  5. SOV and SVO together with N = 1,376: (564 + 488) ÷ 1,376 = 1,052 ÷ 1,376 = 0.7645, so 76.5%. With N = 1,187: 1,052 ÷ 1,187 = 0.8863, so 88.6%.

The same count gives two defensible percentages, 41.0% and 47.5% for SOV, depending only on whether languages with no dominant order are kept in the denominator. Neither is wrong; each answers a different question, and a typological statement is only reproducible when it names the denominator.

A second variable: consonant inventory size

Typology also compares sound systems. WALS Map 1A, by Ian Maddieson, classes languages by the size of their consonant inventory. The table below converts the chapter’s value counts to shares of its table total of 563 languages (89 ÷ 563 = 15.8%, 122 ÷ 563 = 21.7%, and so on).

Value Consonants Languages Share of 563
Small 6-14 89 15.8%
Moderately small 15-18 122 21.7%
Average 19-25 201 35.7%
Moderately large 26-33 94 16.7%
Large 34 or more 57 10.1%

The chapter reports a range from 6 consonants (Rotokas) to 122 (!Xóõ), with a mean of 22.7, a mode of 22 and a median of 21. The “average” class is defined as 19 to 25 consonants, the largest single class at 35.7%, while the two lowest classes together hold 37.5% (211 languages) and the two highest 26.8% (151 languages).

Interpretation: how to read a typological count

A count of languages describes the sample, and only indirectly describes the world’s languages. The table below gives reading guidance for the figures above; each row is an editorial reading rule, not a statement taken from WALS.

What you see What it supports What it does not support
SOV 41.0% and SVO 35.5% of 1,376 languages These two orders are the most frequent values in this sample. That 41.0% of all languages are SOV, or of all speakers.
VOS, OVS, OSV each under 2.1% These orders are rare in this sample. That they are impossible; the sample contains at least one language for each.
189 languages with no dominant order A sizeable group does not fit a single-order label. That the data for them is missing.
Consonant inventory: average class largest (35.7%) Mid-sized inventories are most common in the 563-language table. A causal explanation for why.

Assumptions and common mistakes

Typological percentages rest on assumptions that are easy to forget once the numbers are on a page.

  • Treating a sample as a census. WALS records the languages that its contributors coded for a given feature. Different chapters have different sample sizes (1,376 languages for word order, a table total of 563 for consonant inventories), so shares from different chapters are not directly comparable.
  • Ignoring relatedness. If many sampled languages descend from one ancestor, a frequent value may reflect inheritance more than any general tendency. The guide to language sampling explains how samples are built to limit this.
  • Dropping the “no dominant order” group silently. As the worked example shows, removing 189 languages moves SOV from 41.0% to 47.5%.
  • Mixing denominators in one table. Report every percentage in a table against the same N, and say which N.
  • Over-reading a coded value. A WALS value is a classification decision by the chapter author for a language, and a single label compresses real variation within a language.

How to report a typological proportion

Name the feature, the database and chapter, the number of languages, the denominator and the percentage, and cite the chapter author. A sentence in APA-style prose might read: “Of the 1,376 languages coded in WALS chapter 81 (Dryer), 564 (41.0%) have SOV as the dominant order; excluding the 189 languages with no dominant order, the share is 47.5% of 1,187.” Place the formal reference entry for the chapter in your reference list in the format your style guide prescribes for a chapter in an online database; the guide to using and citing WALS covers the database and its citation conventions. State the date you retrieved the figures, because online databases can be revised.

How to calculate it in Excel, R or Python

Each snippet computes the share of each word order among all 1,376 languages and among the 1,187 with a dominant order.

Excel

Put the seven counts in B2:B8 (SOV, SVO, VSO, VOS, OVS, OSV, no dominant order). In C2 enter =B2/SUM($B$2:$B$8) for the share of all languages, and in D2 enter =B2/SUM($B$2:$B$7) for the share among languages with a dominant order (rows 2 to 7 only). Fill both down; format as percentages.

R

counts = c(SOV=564, SVO=488, VSO=95, VOS=25, OVS=11, OSV=4, none=189)
round(100 * counts / sum(counts), 1)
dominant = counts[names(counts) != "none"]
round(100 * dominant / sum(dominant), 1)

Python

counts = {"SOV": 564, "SVO": 488, "VSO": 95, "VOS": 25, "OVS": 11, "OSV": 4, "none": 189}
total = sum(counts.values())                       # 1376
dom = {k: v for k, v in counts.items() if k != "none"}
dom_total = sum(dom.values())                      # 1187
print({k: round(100 * v / total, 1) for k, v in counts.items()})
print({k: round(100 * v / dom_total, 1) for k, v in dom.items()})

Sources

Data: WALS Map 81A basic word order: language counts and shares

Counts of languages per basic word order from WALS chapter 81 (1,376 languages), with two derived shares computed by the author: share of all sampled languages and share of the 1,187 languages that have a dominant order.

WALS Map 81A basic word order: language counts and shares
Word orderLanguages (count)Share of all 1,376 languages (percent)Share of 1,187 languages with a dominant order (percent)
SOV5644147.5
SVO48835.541.1
VSO956.98
VOS251.82.1
OVS110.80.9
OSV40.30.3
No dominant order18913.7not applicable

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Linguistic Typology: Definition, Word-Order Types and How Typologists Count

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →