Pernilla Hallonsten Halling

Adverbs from a general perspective

This section gives an introduction to general problems that surround adverbs. It is pri- marily based on the comprehensive account by Himmelmann & Schultze-Berndt (2005, see also section 2.5.2). First, I discuss the many functions of adverbs, and then the criteria for classifying adverbs, with particular attention to semantic classifications. At the end of this section, examples from Icelandic (Indo-European) are taken in order to illustrate some of the problems discussed.
Individual adverbs often occur in several different functions, with a number of interpre- tations. Himmelmann & Schultze-Berndt (2005b) illustrate this with different uses of the English adverb naturally, which allow for various classifications of this adverb (2005b: 5).

(2.1) (a) Elaine spoke naturally.

2. 
   Naturally Elaine spoke.
   
3. 
   Elaine spoke, naturally.
   

In (2.1a), naturally is a predicate-level adverb, which describes the manner in which Elaine spoke Himmelmann & Schultze-Berndt (2005b: 5). The adverb modifies the action of speaking denoted by the verb. In (2.1b), naturally is a sentence-level adverb, which expresses the speaker’s stance towards the proposition, in terms of judging it natural for Elaine’s speaking to happen (2005b: 5).¹ Himmelmann & Schultze-Berndt state that the different functions in (2.1a–b) can be distinguished by the different positions of naturally. The example in (2.1c) is more complex: naturally can be interpreted either as a predicate- level or a sentence-level adverb. Despite this ambiguity, it has a prosodically detached position, and position was used as a criterion for distinguishing the uses of naturally in (2.1a–b). It might be added that naturally in (2.1c) can also be interpreted as an afterthought to the main proposition – perhaps the sentence could even be divided into two clauses, but this option is not discussed by Himmelmann & Schultze-Berndt. The distinction between predicate- and sentence-level adverbs may at first glance appear as a straightforward point of departure for the classification of adverbs. However, with the examples in (2.1), Himmelmann & Schultze-Berndt illustrate that the different positions and interpretations of adverbs complicate the matter. Adding to this the afterthought interpretation of (2.1c), it becomes even more evident that adverbs of the same form are ambiguous between various functions.
In their evaluation of means that are commonly used to classify adverbials, Himmel- mann & Schultze-Berndt discern four different parameters. Although these parameters are independent of each other, they are often combined (2005b: 5-6):

(2.2) (a) internal constituency : simple/lexical adverbs, adjective-derived adverbs, PPs, adverbial clauses

2. 
   morphological marking : English -ly adverbs, case-marked adverbials, adverbials marked with instrumental adpositions
   
3. 
   semantics: manner, location, time, reason/cause, condition
   
4. 
   syntactic distribution and scope: positional variability and syntactic scope
   

Parameters (2.2a–b) are fairly uncontroversial according to Himmelmann & Schultze- Berndt. But the authors do not remark on how the two parameters relate to each other. For instance, English -ly adverbs (morphological marking ) are adjective-derived adverbs (internal constituency ). On the other hand, Himmelmann & Schultze-Berndt argue that parameters (2.2a–b) show little correlation with parameters (2.2c–d) – for instance, En- glish -ly adverbs (morphological marking ) are not restricted to a certain semantic type such as manner (semantics) (2005b: 5). On such grounds, Himmelmann & Schultze- Berndt (2005b) deem internal constituency and morphological marking as less useful when

¹ In spoken language, a yet more fine-grained distinction is available in terms of focus accent: either Elaine speaking, and not anybody else, was a natural thing to happen, or Elaine speaking, and not her doing anything else, was natural. This is not discussed by Himmelmann & Schultze-Berndt (2005b).

2. 
   Adverbs from a general perspective
   

classifying adverbials. Parameters (2.2c–d) are argued to be even more problematic in themselves, since they each contain very different criteria. This heterogeneity may turn yet more problematic if different parameters are combined. The authors also point out that it is common to mix semantic and syntactic parameters. In conclusion, the predicate- versus sentence-level division, as illustrated in examples (2.1a–b), is judged to be “the best-known scope distinction” for adverbials, although this division too rests on “as much a syntactic as a semantic distinction” (2005b: 6). In the classification of adverbials, as in many other areas of linguistics, the difficulty of treating syntax and semantics as completely separate domains is evident.

Semantic classifications of adverbs are frequent. Himmelmann & Schultze-Berndt point out that such classifications are typically based on semantic function, resulting in classes such as manner, location, time, degree, etc. (2005b: 6). But grammars and theoretically oriented approaches (such as Geuder 2000, see section 2.3.2) use such class labels in different ways. Even when delimiting the scope to the notion of manner, this label can be used to cover quite different meanings. Himmelmann & Schultze-Berndt present one broad and one narrow sense (2005b: 6):

(2.3) (a) Broad sense: ‘an item which usually/potentially conveys something about the manner in which an action is performed’

(b) Narrow sense: ‘an item which actually conveys the manner in which an action is performed, and nothing else’
To illustrate the elusive difference between the broad and narrow senses of manner, the following five examples are used (from Geuder 2000: 29-35, cited in Himmelmann & Schultze-Berndt 2005b: 6):
(2.4) (a) John shouted at them angrily. (pure manner )

2. 
   John answered the question stupidly. (pure manner )
   
3. 
   He angrily broke the door open. (transparent )
   
4. 
   John stupidly answered the question. (agentive)
   
5. 
   They loaded the cart heavily. (resultative)
   

Himmelmann & Schultze-Berndt (2005b: 6) argue that the adverbs in examples (2.4a–b) “convey the manner in which the action is performed and nothing else”: angrily refers to the manner in which John shouts, and stupidly to the way he is answering. This is not argued to be the case with the other examples. (2.4c) points to the agent’s anger while opening the door, while (2.4d) indicates that it was a stupid idea for John to answer the question (2005b: 6). In (2.4e), the cart is being filled with a heavy load. As pointed out by Geuder (2000), these three uses of adverbs are oriented in different ways, something that I will return to in section 2.3.2. It could be argued that these examples are not as clearly interpretable as Himmelmann & Schultze-Berndt (2005b) propose. For instance, in (2.4a), it cannot be excluded that John is angry – in fact, it is quite likely, although arguably not necessary, that a person is angry when performing an action angrily. In (2.4b), the answering is performed in a stupid way, but this also implies a stupid answer, and perhaps

a stupid answerer. These are naturally fine-grained interpretative distinctions, but they illustrate that even for those adverbs that appear to ‘convey the manner in which the action is performed and nothing else’, their meaning is still not very sharply delimited. The main point that the examples illustrate, however, is the clearer and perhaps more important difference between the pure manner ² adverbs in (2.4a–b), on the one hand, and the different senses in (2.4c–e), on the other. In summary, Himmelmann & Schultze- Berndt (2005b) offer a very useful overview of the issues that surround adverbials and their classification.

Although there are many examples of the same adverb occurring in different positions with different functions, it is a well-known fact that semantic classes of adverbs often show preferences for certain positions, as described by, e.g., Jackendoff (1972) and Cinque (1999). A more specific case is that of Icelandic, sketched by Thráinsson (2007), who provides a useful account of the restrictions on positions of adverbs, followed by a clas- sification based on position in combination with semantics (here, the recurrent mixing of criteria pointed to by Himmelmann & Schultze-Berndt 2005b may be recalled). In Icelandic, sentence adverbs such as aldrei ‘never’ and the negator ekki ‘not’ have more or less fixed positions. They are not able to follow the VP, while manner and frequency adverbs may do so.
(2.5) Icelandic (Indo-European) (Thráinsson 2007: 37)

1. 
   Hún
   

she
hafDi
had
lesiD
read
leiDbeiningarnar
instructions-the