What is WATP?

The World Atlas of Transitivity Pairs (WATP) is a geo-typological database of morphologically related transitivity pairs such as ak-u ‘to open (intransitive)’: ak-e-ru ‘to open (transitive)’ in Japanese, which participate in the causative alternation (doa-ga aita [The door opened] vs. Taro-ga doa-o aketa [Taro opened the door]). It consists of primary data from about 95 languages contributed by about 80 researchers, which can be downloaded for research purpose.

What does WATP do?

WATP offers visual representation of the geographical distribution of the formal relationship between the members of transitivity pairs from the two perspectives proposed in Haspelmath (1993), viz. (a) preferences of individual verb pairs for different expression types (‘map’ interface) and (b) preferences of individual languages for different expression types (‘chart’ interface). It also permits users to test the validity of the ‘iconic’ explanation pertaining to the direction of derivation between the members of transitivity pairs.

Read more

How to cite WATP

1.	To cite the WATP as a whole: The World Atlas of Transitivity Pairs (2014). Tokyo: National Institute for Japanese Language and Linguistics. (Available online at: http://verbpairmap.ninjal.ac.jp).
2.	To cite the data from individual language: Name of the contributor (2014). Transitivity pairs in LLLL. The World Atlas of Transitivity Pairs (2014). Tokyo: National Institute for Japanese Language and Linguistics (Available online at: http://watp.ninjal.ac.jp). [Example: Pardeshi, Prashant (2014). Transitivity pairs in Marathi. The World Atlas of Transitivity Pairs (2014). Tokyo: National Institute for Japanese Language and Linguistics (Available online at: http://watp.ninjal.ac.jp).]

The making of WATP

WATP is one of the research outcomes of NINJAL’s collaborative project entitled ‘Universals and Crosslinguistic Variations in the Semantic Structure of Predicates’ (PI: Prashant Pardeshi). The data included in this database comes from the following sources: (a) data provided by the members of NINJAL’s collaborative project, (b) data on the 21 languages in Haspelmath (1993) [to be added in the near future], and (c) data from Swedish, Maltese and Tsez provided by Bernard Comrie. Some languages show up in both (a) and (b). The data can be downloaded for research purpose. For data download click here. The database construction and web application development have been done by the Lago NLP (Shiro Akasegawa).

Theoretical background

Many languages of the world, if not all, contain pairs of morphologically related non-causative and causative verbs such as ak-u ‘to open (intransitive)’: ak-e-ru ‘to open (transitive)’ in Japanese or khul-naa ‘to open (intransitive)’: khol-naa ‘to open (transitive)’ in Hindi-Urdu. These transitivity pairs share the lexico-semantic composition, but exhibit crosslinguistic differences in the way they encode the alternation morphologically. After Nedjakov (1969), the issue of morphological relationship and the direction of derivation, if any, between the members of such transitivity pairs has been a topic of intense discussion in the field of linguistic typology (Kholodovich (ed. 1969), Nedjalkov & Silnitsky (1973), Masica (1976), Jacobsen (1985), Croft (1990), Haspelmath (1993), Hook (1996), Kageyama (2000), Nichols et al. (2004), Comrie (2006), Narrog (2007), Haspelmath et al. (2014), among others).

In his pioneering geo-linguistic study, Masica (1976) plots the areal distribution of causatives (transitive verb formally derived from its intransitive counterpart) and anti-causatives (intransitive verb formally derived from its transitive counterpart) alternations (op. cit.: 57, 67). With regard to the motivation behind the direction of derivation, Haspelmath (1993), drawing on insights from Jacobsen (1985) and Croft (1990), advocates an ‘iconic’ explanation: the formally derived (or marked) forms are also semantically derived (or marked) and vice versa. Drawing inspiration from these studies, WATP has been developed at the National Institute for Japanese Language and Linguistics (NINJAL), Tokyo, Japan.

WATP offers visual representation of the geographical distribution of the formal relationship between the members of transitivity pairs from the two perspectives proposed in Haspelmath (1993), viz. (a) preferences of individual verb pairs for different expression types (‘map’ interface) and (b) preferences of individual languages for different expression types (‘chart’ interface). It also permits users to test the validity of the ‘iconic’ explanation pertaining to the direction of derivation between the members of transitivity pairs.

Methodology

WATP consists of data on 31 transitivity verb pairs proposed in Haspelmath (1993) from over 50 languages (including dialects). The morphological relationship between the intransitive and the transitive counterparts of the transitivity pair is classified into the following five types.

Direction of derivation	Expression type	Formal relationship between the members of the transitivity pair
Directed	A (Anticausative)	Transitive verb is basic from which the intransitive counterpart is formally derived (Detransitivization).
	C (Causative)	Intransitive verb is basic from which the transitive counterpart is formally derived (Transitivization)
Non-directed	E (Equipollent)	Both transitive and intransitive verbs are marked
	L (Labile)	Transitive and intransitive verb are formally identical
	S (Suppletive)	The transitive and intransitive verb are formally distinct and underived

The data which cannot be classified as any of the above-mentioned types is classified as ‘O (other)’.

The data included in this database comes from the following sources:

(a)	data provided by the members of NINJAL’s collaborative project
(b)	data on the 21 languages in Haspelmath (1993) [to be added in the near future]
(c)	data from Swedish, Maltese and Tsez provided by Bernard Comrie

Some languages show up in both (a) and (b). The duplication of data from the same language is an advised choice since the possibility of different classification of the same data by different scholars cannot be ruled out. The data can be downloaded for research purpose. For data download click here.

Contributors and languages

Visualizations

WATP offers visual representation of the geographical distribution of the formal relationship between the members of transitivity pairs from the two perspectives proposed in Haspelmath (1993), viz. (a) preferences of individual verb pairs for different expression types (‘map’ interface) and (b) preferences of individual languages for different expression types (‘chart’ interface). It also permits users to test the validity of the ‘iconic’ explanation pertaining to the direction of derivation between the members of transitivity pairs.

The ‘map’ interface of the database offers a visualization of the preferences of individual verb pairs for different expression types. In this function, the 31 verb pairs ranked on the basis of A type / C type ratio in Haspelmath (1993: 104, Table 4) are arranged as a horizontal scale on the top of the world map. By clicking on a particular verb pair, the expression type for that verb pair for all the languages included in the database is displayed on the map at the representative location where the language in question is spoken. Further, for each verb pair, the percentage of languages in which it is expressed in each of the five expression types is displayed in the form of a pie chart in the top left corner. In the case of verb pairs on the left end of the scale the likelihood of the basic event to occur spontaneously is higher, hence the chances for the pair to be of the C type alternation are greater. Conversely, in the case of verb pairs on the right end of the scale, the likelihood of the basic event to occur with the instigation of human causer is higher, hence the chances for the pair to be of the A type alternation are greater. The percentage of number of languages of a particular derivation type for a particular pair shown in the pie chart serves to test the validity of the iconic explanation for the direction of derivation.

The ‘chart’ interface of the database offers a visualization of the preferences of individual languages for different expression types, which Haspelmath refers to as the ‘typological characterization’ of a language. A practical problem one encounters in collecting data for the 31 transitivity pairs is that most of these verbs can be translated in the target language in multiple ways. This is for two reasons: (i) many of them are polysemous and (ii) they have been elicited in isolation without any context. Further, sometimes the translational equivalents of a verb pair (say rise/raise) may belong to a directed alternation with the opposite direction of derivation, viz. A and C. To represent such variations, two versions of ‘chart’ viz. the tile chart and the bubble chart are provided. While the tile chart can display only one translational equivalent of the verb pair in question, the bubble chart can distinctly show as many as four translational equivalents. In the tile chart and bubble chart representations it is possible to rank the languages in ascending or descending order of a particular expression type, which makes it possible to check if a language has a dominant preference for a particular expression type concerning these 31 verb pairs.

Haspelmath (1993: 106, Table 4) ranks the 31 verb pairs in the ascending order of the ratio of A (anticausative) divided by C (Causative) expression type in the data pool of 21 languages and claims that the verbs on the top end of this scale are more likely to participate in the causative alternation, while those on the lower end are more likely to participate in the anticausative alternation. The ‘slopegraph’ interface is a tool which enables comparing the ranking of an individual verb pair on the basis of the A/C ratio in Hasplemath’s data pool of 21 languages on the one hand and that in the data pool of 61 languages in the WATP on the other hand and thereby tests the effectiveness of the ranking of an individual verb on the scale.

With the ‘comparison’ interface, any two languages out of the data pool of 82 languages can be chosen and the data from these two languages can be compared by placing them side-by-side.

User Manual

Click the following link to open the user manual in a new window.

WATP User Manual (PDF File)

References

Operating environment

WATP can be best viewed in the following browsers: Firefox, Chrome, Explorer (Version 9 onwards). We recommend using Firefox or Chrome.

Update history

June 6, 2014	ver. 1.00（48 languages）
April 29, 2015	ver. 1.10（12 languages added）
June 23, 2016	ver. 1.30（1 language added, 21 languages from Haspelmath 1993 Appendix added, Slopegraph and comparison interfaces added）
November 3, 2020	ver. 1.40（28 languages/dialects added, Interfaces updated）

Contact

Comments and suggestions for further improvement are welcome. Please contact us at: