1. Defining the values

The present chapter provides an overview of the relations that hold between comitatives and instrumentals in the languages of the world. Comitative  and instrumental  are labels for grammaticalized semantic relations between participants of an event. For the purposes of coding such relations, a wide variety of relator morphemes is used, such as case affixes and adpositions. For convenience, all of the relevant relators are glossed by with in the examples below. For the present purpose, we have looked only at the most grammaticalized overt morphological expressions. Stylistic and other variation has been discounted for obvious reasons. Moreover, we restrict this presentation to prototypical instances of comitatives and instrumentals.

Prototypically, a comitative relator morpheme is employed to encode accompaniment. An example is the Inga (Quechuan; Colombia) suffix -hua(n)  in (1a), where the third-person plural subject is the accompanee and alcalde  ‘mayor’ is the companion. An instrumental relator morpheme normally marks a noun phrase as the instrument used by an agent in a given situation to carry out the action designated by the lexical verb. An example is (1b), where the same Inga case suffix connects the third-person plural agent with the instrument caspi  ‘stick’ used to carry out the action of measuring depths.

(1) Inga (Levinsohn 1976: 124-125) 

a.	Accompaniment:
	alcalde-huan-ta-si	pueblo-ma	samuncuna
	mayor-with-even-quo	town-all	come.3pl
	‘They came to town together with the mayor.’
b.	Instrument:
	caspi-hua-si	tanteancuna	ñayapa	junda
	stick-with-quo	measure.3pl	how	full
	‘It was so full, they were measuring it with a tree-trunk.’

Inga is in accordance with the situation found in English insofar as both languages make use of only one relator – the case suffix -hua(n)  and the preposition with, respectively – to encode the two relations under examination. Accompaniment and instrument receive the same coding and, thus, comitative and instrumental are morphologically indistinguishable. This identity of coding is rather common in the languages of Europe and also attested elsewhere on the globe (cf. §2). However, comitative-instrumental identity is not the only way for languages to organize the relation between the two categories. In many languages, we encounter the opposite of identity, viz. comitative-instrumental differentiation, as well as a mixture of features of identity and differentiation in one and the same language. The three patterns we have been able to identify in earlier publications (see Stolz 1996) are as follows:

Values of Map 52A. Comitatives and Instrumentals

Go to map
	Value	Representation
	Identity	76
	Differentiation	213
	Mixed	33
Total:		322

1.1. Identity 

This is the pattern valid for English and Inga (cf. (1) above). The relators for comitative and instrumental are identical. The two categories therefore are considered to be part of a syncretistic chain. A non-Indo-European language of Europe characterized by identity is Estonian, cf. (2). 

(2) Estonian (Lavotha 1973: 96) 

a.	Instrument: "Comitative -ga"
	ma	kirjuta-n	sule-ga
	I	write-1sg	pen-with
	‘I am writing with a pen.’
b.	Accompaniment: "Comitative -ga"
	Villem	jaluta-b	isa-ga
	Villem	go.for.a.walk-3sg	father-with
	‘Villem is going for a walk with his father.’

The nominal case suffix –ga  ‘with’ traditionally bears the label “comitative” (Stolz and Stroh 2001). However, it is used to encode both of the relations in focus: it occurs in accompaniment situations on the companion noun (e.g. isaga  ‘with (the) father’ in (2b)), and also as the relator marking the instrument (cf. sulega  ‘with (a/the) pen’ in (2a)). There are no alternative ways of expressing the two relations with an equally high degree of grammaticalization.

1.2. Differentiation 

The most frequent solution for the encoding of comitatives and instrumentals obeys a pattern maximally different from that represented by identity. Differentiation requires (at least) two different relators for comitative and instrumental, neither of which can replace the other. This type clearly dominates outside of Europe, although there are also some instances in the Old World, e.g. (3). 

(3) Finnish (Karlsson 1978: 125, 133) 

a.	Instrument: "Adessive -llä"
	Hän	kirjoittaa	kynä-llä.
	s/he	write.3sg	pen-with
	‘S/he is writing with a pen.’
b.	Accompaniment: "Comitative -ine"
	Läsnä	oli	V. V.	vaimo-ine-en.
	near	be.pst.3sg	V. V.	wife-with-poss.3
	‘V. V. was present with his wife.’

In Finnish, there are two inflectional cases which divide up the domain of the single relator of languages of the identity type. The so-called adessive serves inter alia the function of marking instruments (e.g. kynällä  ‘with (a/the) pen’ in (3a)), whereas the inflectional comitative -ine-  indicates accompaniment (e.g. vaimoineen  ‘with his wife’ in (3b)). (In recent years, the inflectional comitative has been giving way to an alternative construction with the postposition kanssa  ‘with’ governing the genitive case. Irrespective of this ongoing change, the pattern of differentiation has been preserved, because the new construction is also used exclusively to encode comitative function.) Note that Finnish and the closely related and immediately neighbouring Estonian opt for two different solutions: where Finnish employs the pattern of differentiation, Estonian is characterized by identity.

1.3. Mixed 

There is a third group of languages that combines features of the two foregoing types, which we call simply mixed. In contradistinction to the two above-mentioned patterns, languages of this type have two or more different relators, at least one of which is syncretistic in the sense that it encodes comitative as well as instrumental, and one of which is specialized so that it encodes only either comitative or instrumental. This mixture of features is relatively rare among the world’s languages. However, there are instances of "mixed" in Europe, e.g. (4).

(4) Hungarian (Bánhidi et al. 1975: 235, 345) 

a.	Instrument: "Comitative-Instrumental"
	Tol-lal	ír-ok.
	pen-with	write-1sg
	‘I am writing with a pen.’
b.	Accompaniment: "Comitative-Instrumental"
	Jan	is	megjelenik	barát-já-val.
	Jan	also	appear.3sg	friend-poss.3sg-with
	‘Jan too shows up with his friend.’
c.	Accompaniment: "Associative"
	Csónak-ház-ak	sport-és	játszóter-ek	vár-ják
	boat-house-pl	sport-adj	playground-pl	wait-3pl
	család-ostul	gyerek-estül	az	ember-ek-et.
	family-with	child-with	def	man-pl-acc
	‘Boathouses and sports grounds are waiting for the people with family and children.’

In Hungarian, two inflectional case suffixes in the noun paradigm partially compete with each other when it comes to encoding comitatives. First, there is the so-called comitative-instrumental suffix -val/-vál  ‘with’. This suffix is polysemous because it may either have a comitative reading marking the companion (e.g. barátjával  ‘with his friend’ in (4b)), or it may mark an instrument (e.g. tollal  (< toll + -val ) ‘with a pen’). Thus, -val/-vál  reflects the familiar pattern of identity. In addition, there is the so-called associative  case marked by -stul/-stül  ‘with’. The functional domain of this suffix is restricted to the comitative proper, i.e. it marks companions (e.g. családostul  ‘with family’, gyerekestül  ‘with child’ in (4c)). Put differently, -stul/-stül  has the typical traits of a morpheme showing differentiation. It is important to note that the associative can never be used with an instrumental reading. However, -stul/-stül  and –val/-vál  can be substituted one for the other if a comitative reading is intended. Note that Hungarian is a member of the same genealogical group as Finnish and Estonian, though more distantly related to them. Notwithstanding these genealogical ties, Hungarian prefers a third solution which is different from both of those found in Finnish and Estonian.

2. Geographical distribution

Our sample is made up of 322 languages. These languages are asymmetrically distributed over the three types identified above. With a total of 213 languages (= 66%), differentiation clearly dominates on a world-wide scale, leaving only 76 languages, i.e. slightly less than a quarter of the sample, as representatives of identity, and only 33 languages (= 11%) displaying mixed structures. However, the three types show markedly different areal preferences. Notwithstanding the fact that identity is a minor solution as compared to differentiation, it is very strong in Europe, where 24 languages (exactly half of the sample languages spoken in Europe) lump together comitative and instrumental. Outside of Europe, the situation changes dramatically. The share of the identity type drops to 30% in Africa, 19% in the Americas, 17% in Asia and just 8% in the Pacific, whereas the vast majority of non-European languages keep comitative and instrumental formally apart: differentiation characterizes 60% of the languages in Africa, 70% of the Asian languages, 70% of the languages in the Americas and as much as 86% of the languages of the Pacific. Differentiation dominates everywhere except Europe. On the other hand, the third type – mixed – is always a minority solution, and never reaches beyond the 20% mark. The highest score for mixed is 18% in Europe; this relatively high percentage probably reflects the tendency (see below) of erstwhile members of the differentiation class to acquire the identity pattern via language contact in Europe. Notwithstanding the marked disproportions in the geographical distribution, none of the three types is totally absent from any of the five areas, cf. Table 1. With the exception of Europe, every continent favours one of the patterns – more precisely, differentiation – with far more than 50%.

It is interesting to note that even within Europe, where identity has its one real stronghold, the pattern itself seems to be a relatively recent innovation spreading from the Standard Average European languages to their neighbours (Stolz 2001c). The three Finno-Ugric languages from which the above examples are drawn probably differ in their organization of the relation between comitative and instrumental precisely because two of them – Estonian and Hungarian – have been subject to contact influence by Indo-European languages (most prominently German in both cases), where identity is the established pattern. 

3. Theoretical implications

Our findings bear upon current linguistic theory for the following reason. Formerly, the view was widely held that identity is the most frequent pattern world-wide, if not the only possible structural solution for comitatives and instrumentals (Lakoff and Johnson 1980; for other pertinent titles, cf. Schwarz et al. 2001). However, this putative universal of human language structure turns out to be nothing more than an areal preference of languages of Europe – especially those which are counted among the so-called Standard Average European languages. As the examples from the three Finno-Ugric languages suggest, genetic affiliation of languages does not imply identical structural behaviour. This applies to Indo-European languages too, which mostly adhere to identity in Europe whereas their relatives in Asia are divided between differentiation and identity (Stolz 1996). Nevertheless, identity is a relatively frequent pattern, although both the comitative and the instrumental can also combine syncretistically with quite different categories (Stolz 2001a, 2001b, Stassen 2000 and Chapter 63 below).