Collocational investigation of Nepali postpositions

Home | About | Projects | Publications | Links | Contact | Nelralec | EMILLE

This project is a corpus-driven analysis of the postpositions which mark the semantic and/or grammatical role of Nepali NPs. The underlying theoretical asssumptions made by this investigation are, firstly, that grammar is continuous with, or emergent from, the lexicon; and secondly, that grammatical categories emerge from or are defined by the distributional behaviour of words or groups of words in discourse.

These principles are found in a number of widely different approaches to grammar, including Construction Grammar (see e.g. Croft 2001) and Lexical Priming theory (Hoey 2005). In this investigation, following the general methodology presented by Hoey (2005), the distributional behaviour of Nepali postpositions is analysed by means of examining those postpositions' collocations.

A word's collocations are the links that exist between that word and the company that it keeps. Statistically, a collocation can be said to exist if a collocate word occurs significantly more frequently in the vicinity of the node word than it does elsewhere in the corpus. Using a sub-section of the Nepali National Corpus, we investigated the collocations of Nepali postpositions.

The issue of enquiry here is whether the Nepali postpositions are a single category of grammatical words, or whether they fall into two categories, as much literature on Nepali grammar has suggested. If two categories of postposition do exist, we should expect them to be evident in the distribution of the two types of postposition, for which the patterns of collocation in the corpus are evidence.

For example, consider the ten most significant collocates of the ergative-instrumental postposition le, as shown in the (romanised text) table below:

Table showing the collocations of the Nepali ergative-instrumental marker le

(Click here for a PDF document describing the transliteration system employed for the Devanagari writing system)

It is immediately striking that eight of the top ten collocates of this word are pronouns (marked with a † ). We can say, then, that this association with pronouns is a key distributional feature of le. By examining other postpositions to see if they share this feature, we can determine whether this distributional feature is one that determines a morphosyntactic category or subcategory within the broader set of Nepali postpositions.

The first results of this investigation will be published in the near future (Hardie, forthcoming).


Croft, W (2001) Radical Construction Grammar. Oxford: Oxford University Press.

Hardie, A (forthcoming) A collocation-based approach to Nepali postpositions.

Hoey, M (2005) Lexical Priming. London: Routledge.

Home | About | Projects | Publications | Links | Contact | Nelralec | EMILLE

Last updated 17th May 2006.