Feature 2. Position of objects (O arguments)

found in question(s): 5a, 5b, 5c, 5d

Definition and illustration

This feature addresses the relative position of adverbs with respect to the object and the verb. Unlike for oblique elements (feature 3), there are comparatively few cross-linguistic studies of the placement of adverbs (e.g., Hölzl 2025). Question 5 addresses the position of a single adverb, differentiating between frequentative (question 5a), locative (question 5b), manner (question 5c), and temporal adverbs (question 5d). Each question allows six potential word order patterns as answer: O-Adv-V / Adv-O-V / V-O-Adv / V-Adv-O / Adv-V-O / O-V-Adv. For instance, in the Gur language Bwamu spoken in Burkina Faso, the placement of adverbs allows two different possibilities, V-Adv-O and V-O-Adv. The adverb intervening between the verb and the object is the preferred option. The same pattern holds for all types of adverbs. Other orders are not grammatical (VO being ungrammatical also without adverbs).

(1)

Questions 16a and 17a address the placement of two adverbs in a neutral, wide-focus context. For simplicity, these questions addressed the relative position of the adverbs and the verb, allowing six possible answers. For temporal and locative adverbs, these are: V-Temp-Loc / V-Loc-Temp / Temp-Loc-V / Loc-Temp-V / Temp-V-Loc / Loc-V-Temp. In many cases, however, additional examples were provided, depending on the position of the object. In Bwamu, if there are multiple adverbs, the adverbs usually follow the verb and object. The preferred order is V-O-Temp-Loc. However, either the temporal or the locative adverb can intervene between the verb and the object while the other follows the object.

(2)

In the context of a question, however, the questioned adverb obligatorily stands in the initial position. The following examples are the only possible answers to questions about 'when', 'where':

(3)

This illustrates the necessity to test information structure (see questions 16b, 16c, 17b, 17c). See also question 10 for other tests relating to the placement of adverbs.

Correlations

There are previous claims in the literature about a connection between the placement of adverbs and basic word order, although these were usually based on few languages or biased samples (e.g., Tomlin 1986: 84-88). For instance, Haider (2020: 349) claims: “In VO, adverbs either precede the VP or follow the verb plus its nominal arguments, but they do not intervene.” (Haider 2020: 349). In other words, if there is adverbial intervention (-Adv-), this should take the form O-Adv-V, but not V-Adv-O. As a testable prediction (about surface orders), this observation could take the following form:

(12) if -Adv- then OV

However, unlike oblique elements, the placement of adverbs has not been studied with the help of large cross-linguistic samples. Table 1 shows the numbers from two available medium-sized samples using dominant word order and adverbs (Hölzl 2025). The first is an unpublished global sample of 94 languages provided by Matthew Dryer (p.c. 2022). The second is a sample of 70 creole languages in Haspelmath et al. (2013). This latter sample only includes frequency adverbs and mostly VO languages (numbers in parentheses refer to languages that also exhibit other orders).

Table 1: Order of verb, object, and adverb in two different samples (Dryer, p.c. 2022; Haspelmath et al. 2013)

According to Dryer (p.c. 2022), VAdvO is overall dispreferred but still occurs as the dominant order in 11 languages of his sample, many of which are Austronesian (Zulu, Nkore-Kiga, Mod-ern Greek, Khasi, Tukang Besi, Nuaulu, Tigak, Kara (Papua New Guinea), Lenakel, Woleaian, Boumaa Fijian). A few other languages are said to have either VOAdv or VAdvO, depending on the type of adverb. Haspelmath et al. (2013) mention one additional language with VAdvO order for frequency adverbs (Reunion Creole) while 16 more languages exhibit this as well as other word order types. Thus, VAdvO seems to be less strongly dispreferred than VXO (Dryer, p.c. 2022, see feature 3). The data, although limited, are sufficient to show that the claimed absence of V-Adv-O order needs to be qualified but could well remain valid as a statistical tendency.

The data in Table 1 can be interpreted in different ways. Tables 2 and 3 focus on the position of the adverb (six types) and the relative order of adverb and noun (four types), respectively.

Table 2: Matrix showing the order of OV/VO and adverbs (Global / Creoles)

Table 3: Matrix showing the order of OV/VO and OAdv/AdvO (Global / Creoles)

The preliminary samples indicate the following testable unidirectional correlations:

(13) if OV then -Adv- / OAdv

(14) if -Adv then VO

(15) if AdvO / Adv- then VO

Note that these predictions do not differentiate between different types of adverbs. Furthermore, the samples did not check for potential confounding factors, such as V2 in German (which leads to V-Adv-O order). The extensive background information provided for every language in this database allows to test for these alternative explanations.

There do not seem to be any cross-linguistic studies on the relative placement of more than one adverb.

References