Adverbials, clausal domains and more
3. Major classes of adverbials in the Hungarian clause
3.2 A mini-calculus of clausal domain types
Returning now to Table 1, assuming that different adverbials, as a function of their own meaning, are able to modify constitutents of different semantic types (see Haider 2000, Ernst 2002, Frey 2003, a.o.), the distribution presented in Table 1 suggests that either each of the
three major classes of adverbials in the Hungarian clause is able to modify more than one semantic type of constituents, or else it does not hold that the different syntactic domains (recursively embedded constituents) of the clause (cf. (7)) correspond to distinct semantic types. If it is accepted that the semantic selectional properties of different classes of adverbials are invariant, this would speak in favor of the second solution; however, the “cost”
of this approach is that, in turn, the semantic type of clausal constituents embedded in each other could no longer be taken as being invariant. Certainly, one cannot a priori choose between this view and its opposite (i.e., on which the type, or semantic selectional property, of adverbials is flexible, whereas the semantic type of clausal constituents is constant); the choice needs to be informed by empirical arguments.
For instance, coordination phenomena may provide an argument in favour of the variability of the semantic type of clausal constituents:
(24) a. Tudom, hogy [mindenki hazament] és [ketten ott maradtak]
I.know that everyone home.went and two there stayed
‘I know that everyone went home and two people stayed there.’
b. Azt mondtad, hogy [mindenki hazament] és [csak te maradtál ott]
that you.said that everyone home.went and only you stayed there
‘You said that everyone went home and only you stayed there.’
c. Úgy tudom, hogy [János kezdte el] és [Péter csak befejezte]
so I.know that John started VM and Peter only finished
‘As I know, John started it and Peter just finished it.’
d. Úgy láttam, hogy [bejött valaki] és [ketten felálltak]
so I.saw that VM.came someone and two stood.up
‘I saw as if someone came in and two people stood up.’
(24a) exemplifies the coordination of a DistP and a TopP, (24b) that of a DistP and a FocP, (24c) that of a FocP and a TopP and (24d) that of an AspP and a TopP. These categories are all each other’s alternatives as clausal arguments of a verb (or a complementizer below the verb). Coordination and selectional restrictions show that these clausal categories can be equivalent in terms of semantic types. I will model this type variability with type raising; for example, a “DistP” behaves in the coordination shown in (24a) as being of a “higher” type, viz. a “TopP”.
Let us consider the details of the proposal. Frey (2003) and (Ernst (2002) use partly overlapping inventories of semantic types to describe different parts of the clausal hierarchy;
for instance, the hierarchy of types according to Ernst’s so-called FEO calculus (Fact-Event Object calculus) is comprised by the following semantic types: fact > proposition > event >
specified event. These inventories could be adapted and, if necessary, adapted so as to be applicable to the part of Hungarian clausal hierarchy discussed so far. Nevertheless, as specific semantic types need not be defined precisely to analyze the data presented above, for now I will neutrally designate the types of Hungarian clausal constituents embedded in each other with numerical labels, as shown below:
(25) [3 Topic [3 Topic [3 Quantifier [3 [2 Quantifier [2 Quantifier [2 Focus [2 [1 Verb Modifier... ]]]]]]]]]
Let us consider the Verb Modifier-initial Predicate (VM-Predicate, for short) Type 1. At first, we might be tempted to think that Focus raises this type to Type 2. However, this is contradicted by the grammaticality of clauses containing (real) multiple foci (see É. Kiss 1998c, 2002, Surányi 1999, 2003). If a Focus did raise the type of a clausal constituent from Type 1 to Type 2, then the same type raising potential could no longer be associated with a higher focus. At the same time, it appears that a Focal Predicate is higher than a VM-Predicate, since Low Adverbials may modify only the latter, and not the former. Both of these conditions can be taken into consideration if it is assumed that a Predicate with a Verb Modifier may be raised to Type 2 without adding Focus to the construction, although Type 2 is naturally the type of a Focal Predicate. It must be kept in mind that an adverbial capable of modifying a clausal constituent higher than a Predicate with a Verb Modifier can also modify a Predicate with a Verb Modifier. This means that the type of a Predicate with a Verb Modifier can be raised to higher types without adding a Quantifier or a Topic to the construction; let’s call this type conversion, indicated as n←m.13 Based on the data presented above, type conversion can only raise the type of a clausal constituent (e.g., a Focal Predicate may behave as a DistP with regard to modifiability), but it cannot lower that (i.e., a Focal Predicate may never behave as a Predicate with a Verb Modifier with respect to modifiability). As in view of the facts reviewed above 2←1 conversion as well as 3←2 conversion must exist, there is no need to assume an independent 3←1 conversion.
Therefore, the basic type of a Focal Predicate is Type 2. Since Focus is a recursive constituent (as has already been mentioned, Hungarian does indeed have real multiple foci;
see also Chapter 3), one Focus constituent does not raise the type of the clause (i.e., it is 2←2). Similarly, the semantic type of a Type 2 constituent does not change either if a Quantifier is added to the construction: a Quantifier can make a Type 2 constituent out of another Type 2 constituent (i.e., it is also 2←2; otherwise the type raising potential of quantifiers could not be considered as identical, and it could hardly be explained why every adverbial appearing just below a Quantifier can also appear just above it). At this point, the question arises as to how it is possible that Foci and Quantifiers are both 2←2. As a matter of fact, the mini-calculus just presented above is not expected to reflect and capture every semantic difference between preverbal elements, only those that the selectional properties – and thus the distribution – of adverbials are sensitive to. Since every adverbial appearing just above Focus may also appear just above a Quantifier, and Middle Adverbials may appear both just above and just below Focus, it is necessary from the perspective of the calculus of adverbials that the constituent that FocP dominates (=AspP), the Focal Predicate (FocP), and the DistP constituent of the clause all be of the same type. This falls into place in the context of the results of Chapter 3, where it was argued that id-focus (=Focus) creates an identificational predication structure, thereby giving rise to a new proposition. It was argued further that both id-focus and Quantifier raise out of and adjoin to full propositional units (complete with a time variable in the verbal predicate). This type of unit is labelled as Type 2 in the mini-calculus above.
Thus far we have been looking at what goes on inside the logical predicate, or comment. At a certain point the border of this domain is reached (in (25) this happesn after merging in not one but two Quantifiers). This is the stage where 3←2 conversion happens, which closes off, and thus produces, the logical predicate by raising the type of the clausal domain to 3. Recursive topics do not raise the type of the domain they attacht to (i.e., they are 3←3). As for the Quantifiers appearing above High Adverbials (see (11)), I suggested in Chapter 2 that they are part of the logical subject (i.e., they are Topics) instead of the logical predicate (see, for example, Surányi 2003; a Quantifier in Topic position may precede another, nominal phrase Topic as well). A Quantifier at the beginning of the predicate may function as a short answer to a yes/no question. The Quantifier in (26a) can be found below a High Adverbial, the one in (26b) is below a Middle Adverbial, while the one in (26c) is above a Middle Adverbial. However, the Quantifier above a High Adverbial cannot function
as an answer to a yes/no question (in cases when it is able to appear in a yes/no question), as can be seen in (27):
(26) a. A: Általában mindenki eljön? B: Mindenki.
usually everyone comes everyone
‘Does everyone usually come?’ ‘Yes.’
b. A: Még mindenki itt van? B: Mindenki.
still everyone here is everyone
‘Is everyone still here?’ ‘Yes.’
c. A: Mindenki megint eljött? B: Mindenki.
everyone again came everyone
‘Has everyone come again?’ ‘Yes.’
(27) a. A: Mindenki általában késın érkezik? B1: *Mindenki. B2: Késın.
everyone usually late arrives everyone late
‘Does everyone usually arrive late?’ ‘Yes.’
b. A: Mindenkit szerinted mindenhova meghívnak?
Everyone-ACC according.to.you everywhere VM.invite-3RD-PL
‘Do you think that everyone is invited everywhere?’
B1: *Mindenkit.
everyone-ACC
B2: Mindenhova.
everywhere
‘Yes.’
These Quantifiers will behave as Topics from the perspective of our mini-calculus as well (3←3).
Even though the calculus outlined above is likely to be overly simplistic, it is sufficient for our present objectives.14 In terms of this calculus, the basic distribution of the three main classes of adverbials in Hungarian is determined by the types of clausal domains selected by these classes, as specified in (28) below. (28) can be readily verified on the basis of Table 1 and the schema presented in (25). (29) shows the distributions outlined in Table 1 integrated into the schema displayed in (25).
(28) The types selected by Hungarian Classes of Adverbials:
a. Low Adverbials (LA): 1 b. Middle Adverbials (MA): 2 c. High Adverbials (HA): 3
(29) a. HA [3 Topic HA [3 Topic HA [3 Quantifier HA [3 [2 Quantifier …]]]]]
b. …[3 MA [2 Quantifier MA [2 Quantifier MA [2 Focus MA [2 [1 Verb Modifier…]]]]]]
c. …[3 [2 Quantifier [2 Quantifier [2 Focus [2 LA [1 Verb Modifier…]]]]]]
The type raising potential of certain adverbials is not included in this calculus. Even though adverbials – being adjuncts – do not typically raise the syntactic type of the modified constituent, as it was emphasized in the Introduction to this chapter, the hierarchical order of adverbials within any one of the main classes of adverbials is far from being free. According to Cinque (1999), each class of adverbials categorized here in terms of three major types constitutes a class of its own, and their hierarchy results in a complete linear ordering within the structure of the clause (e.g., the internal order of the examples in (11), (12) and (15) all follow Cinque’s hierarchy).
Nevertheless, there exist pairs of adverbials in Hungarian which can appear preverbally in either order. (30) illustrates this fact with Middle Adverbials, while (31) does so with High Adverbials. (32a) and (32b) present two permutations of a High and a Middle Adverbial, respectively. In each pair, the first version conforms to the canonical hierarchy.
(30) a. Feltétlenül megint el akar jönni / Megint feltétlenül el akar jönni definitely again VM wants come-INF again definitely VM wants come-INF
‘He definitely wants to come again.’
b. Gyakran szándékosan otthon marad / Szándékosan gyakran otthon marad often deliberately home stays deliberately often home stays
‘(S)he often stays at home deliberately.’
(31) a. Szerinted állítólag otthon van / Állítólag szerinted otthon van according.to.you allegedly home is allegedly according.to.you home is
‘In your opinion, (s)he is allegedly at home.’
b. Szerencsére általában igaza van / Általában szerencsére igaza van
fortunately usually truth-POSS is usually fortunately truth-POSS is
‘Fortunately, (s)he is allegedly at home.’
(32) a. Szükségszerően gyakran/ Gyakran szükségszerően leváltanak valakit necessarily often often necessarily VM.replace-3-PL someone-ACC
‘Necessarily, someone is often replaced.’
b. Valószínőleg már / Már valószínőleg mindenki ismeri probably already already probably everyone knows-OBJ
‘Probably everyone has already met him/her.’
Each order in the examples above is associated with a different meaning, even if this difference is not always completely clear: adverbials are interpreted in their surface position.
There is no difference in (32b). Notice also that both versions in (31a) are ambiguous: if the second adverbial in either version is realized with a parenthetical prosody, “scope” relations get inverted. Parenthetical use, which is not always trivial to detect in prosody, is a regular possibility (especially in the case of High Adverbials), and an independent factor in the free order of adverbials. Adverbials of parenthetical use should be taken as a separate case both preverbally and postverbally. As for the preverbal permutability of adverbials and their restricted freedom of selectional properties in the light of (28), consult Surányi (2008: Section 6). 15
In this section, I have examined what kind of classification of adverbials is necessary, and what kind of pattern they follow when appearing in certain positions of the preverbal field. I have argued that it is necessary to distinguish three semantic types of clausal domains, which the three major classes of adverbials select for (see (28)). Flexibility in the pre-verbal distribution of adverbials stems from two main sources. (i) Elements selecting for the same semantic type of clausal domain – including different members of the same adverbial class, as well as other elements of the clause, like Quantifer or Topics – may attach to it in a relatively free order. (ii) The type of clausal domains is partially flexible, as summed up in (29).
The remainder of the chapter investigates the post-verbal behaviour of adverbials, as well as the possible analogy with Quantifiers.