Additive particles, prosodic structure and focus sensitivity in Hungarian

2.1 Syntactic distribution

The amount of research on information structure in Hungarian is extensive and the literature dates back to the late 1970s.^[3] One of the most well-known characteristics of the Hungarian language is its discourse-configurational type (É. Kiss 1995; Surányi 2015), where surface order and certain syntactic positions are closely related to the discourse-semantic/pragmatic functions, ‘topic’ and ‘focus’. Topical elements appear in a clause initial position (6a), and the narrow identificational focus of the utterance occupies the immediate preverbal position and triggers inverse order of the verbal particle [vprt] and the verb (6b).

The constituent in the structural focus position (6b) gets the nuclear pitch accent and is assigned an interpretation in terms of exclusion by identification (É. Kiss 1998; Kenesei 2006; Szabolcsi 1981). Placement in the immediate preverbal position is claimed to be an obligatory and default marking strategy for (identificational) narrow focus constructions (e.g., Genzel et al. 2015). Next to this structural focus marking, it is widely assumed that other, syntactically unmarked, focus structures are also available (see, e.g., É. Kiss 1998; Surányi 2011). However, most work on non-structural focus marking concentrates on postverbal focus and the issue of exhaustivity. The possibilities of mere prosodic focus marking in the preverbal field, however, are seriously under-studied or not even acknowledged (see more in Section 2.2).

The Hungarian additive particle is ‘also, too’ is an enclitic:^[4] it is destressed and needs a syntactic host, with which it forms a phonological word. It is positionally bound to a constituent, very often but not exclusively a noun phrase. We will refer to the constituent and the particle together as the ‘is-phrase’. A frequency analysis on the Hungarian National Corpus (http://corpus.nytud.hu/mnsz/ [Oravecz et al. 2014]) shows that in around 64% of all occurrences, it is attached to a noun or pronoun.^[5] It can also be attached to adverbials ( ∼ 8.5%), adjectives ( ∼ 4%), finite verbs and infinitives ( ∼ 3%), the verbal particle ( ∼ 3%), negation ( ∼ 4.5%) and other categories. As shown in Example 7, in its most frequent occurrence, i.e., cliticized to a noun phrase, the is-phrase can either occupy a designated position in the preverbal field or it can occur postverbally, where the relative order of the constituents is free. This is true regardless of the grammatical function of the is-phrase, as illustrated by (7a) and (7b).

As for its surface position, it has been shown that in the preverbal field, the is-phrase cannot occur in the topic field, nor in the immediate focus position (e.g., É. Kiss 2004; Szabolcsi 1997). One argument to support the claim that the is-phrase does not appear in the topic field is given by the linearization constraints of sentence adverbials (e.g., szerencsére ‘fortunately’). In the preverbal field, sentence adverbials cannot enter the comment^[6] but can mark the right edge of the topic field (see, e.g., É. Kiss 2004; Kálmán 2001). Example (8b) illustrates that the adverbial cannot follow the is-phrase, which indicates that the latter is not in the topic field.

Now consider the contrast between (9b) and (9a) which illustrates the claim that the is-phrase in Hungarian is excluded from the immediate preverbal focus position. In (9b), the inverse order of the verb and the verbal particle (vprt) indicates that the is-phrase occupies the immediate preverbal focus position, but this linearization is considered ungrammatical.

This syntactic behavior can straightforwardly be explained on semantic grounds by virtue of the exhaustivity effect assigned to the preverbal focus position^[7] that is incompatible with additivity (see, e.g., É. Kiss 1998; Szabolcsi 1981).

2.2 Focus sensitivity of is

It is generally accepted that additive particles are crosslinguistically focus sensitive. In English, focus marking is primarily prosodic and the focused constituent bears the nuclear pitch accent.^[8] Based on the focus marking, the sentence has different focus-alternatives, by which the additive presupposition is derived (e.g., Krifka 2006; Szabolcsi 2017). A similar semantics is claimed for Hungarian as well (see Szabolcsi 2017), proposing a uniform formal semantic analysis of additive particles. However, in Hungarian sentences containing the additive particle is, focus marking is not directly transparent, at least not structurally (as shown in the previous section). Nevertheless, the Hungarian additive particle is claimed to be focus sensitive, with similar semantics to the corresponding English sentences (see Example 10).

For such a uniform semantic analysis, the crucial question with respect to Hungarian is how to capture the focal status of the host of the particle is. As shown before, structural focus marking is not an option here. It is accepted that focus marking outside of the preverbal focus position also occurs in Hungarian. However most work on this issue concentrates on postverbal focus marking and the expression of two focus types considering the question whether exhaustivity involved (see, e.g., É. Kiss 1998; Surányi 2011).^[9] The relevant distinction here is made by É. Kiss (1998) introducing the notions of identificational focus and information focus. The latter focus type expresses new information focus without identification and exhaustivity. It is not structurally marked but bears prosodic prominence. This notion could be a promising candidate in the analysis of is. However information focus is claimed to only appear postverbally (or “in situ”). This is explicitly stated in a recent characterization by Surányi (2015), referring to previous works by É. Kiss (1998, 2015) and Horváth (2007, 2010): “All and only identificational foci must be fronted to a dedicated left peripheral pre-verbal position, and all and only plain information foci must remain in situ.” (Surányi 2015: 432). Consequently, this distinction cannot account for the focus sensitivity of preverbal occurrences of is.

As shown in the previous section, the is-phrase cannot occur in the preverbal ‘focus position’, nor in the clause-initial ‘topic position’. In the preverbal field, it occupies a designated position between the structural ‘topic’ and ‘focus’ (11).^[10]

This position is not discussed in detail in relation to focus marking and focus types in the relevant literature on Hungarian. In some earlier syntactic and formal semantic accounts, the particle is is analyzed in terms of ‘focus’ (see, e.g., Brasoveanu and Szabolcsi 2013; Brody 1990; Szabolcsi 2017, 2018). However these authors merely stipulate the given focus structure, without discussing how to derive the focal status of the associate. In only a few places (e.g., Kálmán 1985b), has it been stated that the syntactic host of is has a focus accent, albeit without any further discussion on the matter. These works do not investigate the type of accent or the prosodic patterns of the various constructions with is nor do they discuss how it relates to the general issue of focus marking and focus types in Hungarian. With this study we aim to contribute to these questions.

In this article, we support the claim that additive particles are crosslinguistically focus sensitive and that a semantic analysis in terms of focus alternative is the appropriate one across languages. However, this claim needs further justification and grounding in Hungarian,^[11] where focus marking (and information structure in general) behaves significantly differently than in English.

In Hungarian, investigations and analyses of the information structural notions topic and focus, or to be precise sub-types of these, mostly concentrate on the designated structural positions and their syntactic/semantic characterization. Analyses on the prosody-syntax interface (e.g., Kálmán 1985b; Szendrői 2001, 2003; Varga 2002) also mostly concentrate on the prosodic characteristics of the preverbal position. Although flexible focus marking is acknowledged to some degree, the construction with additive particles has not received special attention so far in terms of focus marking and the derivation of the focus structure. In this article, we propose such an analysis and argue for a more prominent role of prosody in the analysis of the Hungarian is ‘also, too’ in the interface of prosody and information structure.

2.3 The semantic associate and various uses

Besides the (syntactic) host (see Section 2.1), we need to refer to the semantic associate (or simply associate) of the particle, which is the element or constituent that is taken as the ‘focus’ and, as such, determines the alternatives and the additive presupposition of the sentence. An important issue is then how to derive the semantic associate, and what linguistic means mark the appropriate focus structure. We argue in this article that the prosodic patterns of sentences containing is determine the focus domain and consequently, contrary to what Surányi (2015) and É. Kiss (1998) assume, that certain focus types distinct from the preverbal, identificational focus can be found in the preverbal field. For a comprehensive analysis of the prosody-information structure interface of Hungarian is-phrases, two aspects are particularly significant: (1) the marking of the range of the focus domain and (2) the various uses of the particle is.

3.1 Stress in Hungarian

Analyses of Hungarian generally make a difference between neutral and non-neutral (focused or corrective) sentences which differ not only in their syntax but also in their characteristic prosodic patterns (see, e.g., Kálmán 1985a, 1985b; Kálmán et al. 1986; Kálmán and Nádasdy 1994; Kenesei 1998).

Neutral sentences, i.e., level prosody sentences (Kálmán et al. 1986), do not have a single prominent stress or pitch accent, but equal main stresses on every lexical word in the sentence. The intensity of these main stresses declines during the sentence, which is called downdrift (in Hungarian, see Kenesei 1998; Varga 2002). At the beginning of a new intonational phrase the intensity of the main stresses steps up again, but does not reach the same height as the first main stress of the utterance (Varga 2002). By the notion ‘neutral sentence’ it is common to refer to sentences without a preverbal focus constituent. Figure 1 shows Kálmán et al.’s (1986) schematic representation of the prosodic pattern of a neutral sentence. They divide the sentence into a preparatory section containing the topic and certain adjuncts and an essential section containing the ‘hocus’ position,^[17] the verb and its postverbal arguments.

Figure 1:

Prosodic scheme of a neutral sentence (Kálmán et al. 1986: 131).

In non-neutral sentences, one or more lexical words are main-stressed, while the others are deaccented. Figure 2 shows the schematic representation of the prosodic pattern of a non-neutral sentence with a single pitch accent on the focus position. Again, the sentence is divided into a preparatory section and an essential section starting with the pitch accent.

Figure 2:

Prosodic scheme of a non-neutral sentence (Kálmán et al. 1986: 130).

It is widely accepted that non-neutral sentences with a single main stress have a narrow (identificational) focus interpretation of the stressed element (see, e.g., Varga 2002). For example, in (21) the main stress is on Mari-t, which occupies the preverbal focus position interpreted as contrastive/identificational focus.^[18]

In these kind of sentences, the intensity of the pitch accent does not have to be stronger than a normal main stress, but the lexical words after the pitch accent are deaccented.^[19] Even though the single pitch accent most often falls on the designated focus position in front of the verb, Kálmán (1985b) claims that this pre-verbal main stress can also fall, for example, on a quantifier or an is-phrase that cannot occupy this position. According to Kenesei (1998), elements preceding the focus (e.g., topic) keep their unreduced stresses because they form their own phonological phrase while elements between the focus and the verb (including the verb) are fully destressed.

Non-neutral sentences with multiple main stresses can either express that multiple constituents are contrasted or that they are new information and thus the sentence gets a broad focus interpretation (Kálmán 1985b; Varga 2002). According to Varga (2002) and Kenesei (1998), the only exception to this rule is that verbs are deaccented if the focus position is filled even if they are new information and part of the focus domain. As with the non-neutral sentences with a single main stress, the differences in intensity of the main accents do not play a role. Example (22c) illustrates multiple main stresses in a non-neutral sentence. The only difference between (22a), (22b) and (22c) is that mindig ‘always’ is new information in (22c) and thus accented.

3.2 Syntax-prosody mapping in Hungarian

Building upon previous work by Reinhart (1995, 2006) and Szendrői (2001, 2003), Hamlaoui and Szendrői (2015) propose a crosslinguistically applicable approach to the mapping of syntax and prosody. As widely accepted in contemporary phonological theory (e.g., Nespor and Vogel 1986; Selkirk 2011), speech is considered to be built up from hierarchically organized phonological domains: phonological words (ω), phonological phrases (φ) and intonational phrases (ι). Hamlaoui and Szendrői (2015) argue that while it is widely accepted that phonological words and phonological phrases are based on clear syntactic domains (mostly lexical XPs), identifying the basis for intonational phrases has been rather problematic. For the clause-level mapping of syntax and phonology, Hamlaoui and Szendrői (2015) propose six constraints, four of which determine the mapping from syntax to prosody and two of which the mapping from prosody to syntax. Based on these constraints, they predict that elements in a specifier position with accompanying verb movement will be within the core ι (23a), while elements occurring in a left-peripheral position without verb movement will prosodically be outside the core ι (23b) but still inside an outer ι containing the core ι. Hence, recursive intonational phrases are possible. The phrasing in (23a) occurs, for example, in Hungarian preverbal focus constructions, English wh-questions, as well as in German V2 clauses. The phrasing in (23b) occurs, for example, with the Hungarian topic position.

Beside the mapping rules, ranked stress-alignment constraints are also defined as given in (24). In Hungarian, the ranking of these constraints is Stress-ι ≫ EndRule-L ≫ EndRule-R.

Given these constraints together with the ‘Stress-Focus Correspondence Principle’ (25), the phrasing of neutral and non-neutral sentences in Hungarian can be derived uniformly.

Neutral sentences, i.e., without preverbal (identificational) focus, can be V-initial, expressing all-new or out-of-blue content, or they can contain left-peripherical topic constituents. This position is often referred to as the ‘topic position’ in Hungarian. The topic position can also be iterated.

In neutral sentences, all phonological phrases (φs) are stressed, carrying pitch accents. The verbal particle and the verb form one morphological unit, its own phonological phrase (φ), and the unit bears φ-level accent. Following Szendrői (2001), the verbal complex (vprt-V) carries the main stress of the sentence.^[20] The φ of the verbal complex is located at the left edge of the intonational phrase (ι).

Non-neutral sentences with a filled preverbal focus position have a distinguished prosodic pattern, as illustrated in (27a).

(27)

A	maláj	lány	Eleonórához	menekül	el	Emília	elöl.
the	Malay	girl	Eleonora-to	flees	away	Emilia	from
‘The Malay girl escapes from Emilia to Eleonora.’
( ≈ It is Eleonora to whom the Malay girl flees from Emilia.)

a.

(Hamlaoui and Szendrői 2015: 87)

b.

([A	maláj	lány k	([Eleonórá-hoz j	menekül i	[el	[t i t j t k
the	Malay	girl	Eleonora-to	flees	away
	Emília		elöl] VP ] PrP ] FP ] TP ) ι ) ι
	Emilia		from
(following Hamlaoui and Szendrői 2015)

In non-neutral sentences with preverbal focus, the main stress falls on the focused constituent and post focal elements undergo stress reduction, indicating given information (see also Varga 2002). As the syntax-prosody mapping in (27b) shows, the focused constituent, and hence the main accent, occurs at the left-edge of the intonational phrase.

As illustrated above, in narrow focus constructions (with identificational semantics), the focused constituent bears an eradicating stress (Kálmán et al. 1986), where everything behind is destressed indicating given information status. Focus projection with the preverbal focus construction is only available with a distinguished intonation pattern, as shown by Kenesei (2006), with accents appearing after the constituent in the preverbal focus position. Based on these observations (and the syntactic properties discussed before), we hypothesize that focus marking in sentences with is ‘also, too’ is prosodic, where (a) the main accent falls on the syntactic host of the particle and (b) the prosodic pattern after the is-phrase indicates the range of focus (narrow, predicate or sentence). In the following section, we discuss our prosodic study of various uses of the particle is and investigate whether our hypothesis is correct and appropriate for all cases.

4.1 Plain additive use

First we discuss the use of is as a plain additive particle (=also, too). Consider Example (28), presented together with its local discourse context.

The local discourse context contains the information that the frog is angry, hence the context suggests a narrow semantic associate of the particle. The additive presupposition/requirement that ‘someone different from the boy was angry’ is directly fulfilled. According to our hypothesis, we predict that with this interpretation the syntactic host of the particle bears the main stress and everything following it is destressed. This prediction is borne out:

(29)

prosodic pattern of the target sentence in (28):

As predicted, the prosodic pattern above shows that the host (a kisfiú ‘the boy’) of the additive particle bears the main accent (H*) of the sentence while none of the constituents after the is-phrase are accented.

Now take another example, presented in the clauses in (30a) and (30b) together with their immediate context.

The context suggests a broad focus interpretation in both (30a) and (30b), with a semantic associate that includes the whole sentence/the whole proposition. The additive presupposition/requirement of the broad interpretation is satisfied in both sentences, i.e., something different from ‘the drummer scratched his head’ happened before (for 30a) and something different from ‘the trumpeter was looking to see what happened’ (for 30b) was the case before: the context provides the information that ‘that saxofonist shook his instrument’. This is considered as an element of an elaboration list of the reactions of musicians after the frog jumped into the saxophone. The context does not contain the information that ‘someone different from the drummer is scratching his head’ or ‘someone different from the trumpeter is looking to see what happened’, which would license a narrow interpretation.

By our hypothesis, we expect that this broad semantic associate in (30a) and (30b) is marked prosodically, by accent placement after the is-phrase as well. Again, this prediction is borne out:

(31)

a.

prosodic pattern of (30a):

b.

prosodic pattern of (30b):

In Examples in (30a) and (30b), similarly to the Example (28), the main accent falls on the syntactic host of the particle: a dobos ‘the drummer’ and a trombitás ‘the trumpeter’ respectively. However, the stress pattern of the constituents following the is-phrase is different from that of a narrow focus reading. In (30a) every element after the is-phrase bears neutral stress. This indicates that the content of these elements is considered new information, hence not presupposed. Example (30b) is prosodically somewhat less clear-cut, since it contains a subordinate clause. However, the verb is accented here, which indicates again that it is not presupposed.

In our data, we found a significant number of examples with a broad focus reading, where all elements after the is-phrase, including the verb, bear accents. This finding is contrary to the claims that even in broad focus readings the verb is deaccented (Varga 2002) and that elements between the verb and the focus are destressed (Kenesei 1998). To settle this particular issue we need further prosodic investigations on a larger amount of data.

The narrow and broad semantic associate reading is also available in sentences where the additive particle is attached to the verb. The narrow interpretation corresponds to a focused verb with deaccented postverbal elements. In a broad focus interpretation (predicate focus or sentence focus), on the other hand, postverbal constituents receive stress, as before. Consider Example (32).

(32)

local discourse context: [the turtle causes all kinds of trouble] A kisfiú sem tudta hirtelen, mi történt. De aztán, amikor meglátta, hogy mi történt, hogy megint csak a teknősbéka az,
‘The boy did not know what had happened either. But when he saw what had happened, that it was the turtle again,’
és	bele	is	húzta	a	kutyá-t	a	víz-be,
and	vprt	also	pulled	the	dog-acc	the	water-ill
‘and he(=turtle) also pulled the dog into the water’

Based on the prosodic patterns in our data we can generalize that the focus in is-constructions bears the main accent of the sentence. When the semantic associate is narrow, the elements after the is-phrase are deaccented, indicating that their information is given, and with a broad semantic associate they receive an accent, indicating that their information is new (or non-given). There are, however, some examples in our data that seem to show a mismatch between the prosody and the context. Take, for example, sentence (33).

(33)

local discourse context: A kutya szegény alig tudott menni a szomorúságtól,
‘The poor dog could hardly walk from sadness,’
a	béka	is	csak	kullogott	utánuk.
the	frog	also	only	lag	behind.3pl
‘the frog, too, was just lagging behind them.’

At first sight, the context contains the information that someone other than the dog is lagging, given the semantic similarities of the verb lag and the expression walk sadly. Thus, only looking at the context, one could derive a reading with a narrow semantic associate. However, the prosody clearly indicates a broad associate, since the verb and the adverb behind the is-phrase both bear an accent. This can be explained if we look at the lexical semantics of the two verbs. The verb kullog ‘lag’ includes a manner component, which is missing in the expression in the preceding sentence. This meaning component is non-given, therefore we can argue that the verb kullog ‘lag’ expresses new information, hence confirming a broad focus reading. Another example of a possible mismatch is given in (34):

(34)

local discourse context: A béka rémülten úszott utánuk, hogy vajon most mi történik, és aztán kimászva a szárazföldre mindenki úgy gondolta, hogy na most végre megkönnyebbülhetünk, és pihenhetünk együtt.
‘The frog was swimming frightened after them, wondering what was happening now, and when the others climbed out onto the shore, they were all thinking: well, finally we can be relieved and rest together.’
A	béka	is	ki-mászott,
the	frog	also	vprt-climbed
‘The frog, too, climbed out’

In this example, the context indicates even more strongly that ‘someone different from the frog climbed out’, given that the same verb is used in the target sentence and in the immediately preceding sentence. However, as before, the prosodic pattern shows a broad focus interpretation: the verbal complex ki-mászott ‘vprt _out -climbed (= climbed out)’ is stressed in the target clause. A possible explanation for this example is that for the correct interpretation, which is indicated in the prosody, we need to look over the clause boundary in the target sentence see (35).

(35)

A	béka	is	[ki-mászott,	és	próbált	csatlakoz-ni	a	csapat-hoz]SA.
the	frog	also	vprt-climbed	and	tried	join-inf	the	group-all
‘The frog, too, climbed out and tried to join the group.’

The predication about the frog is more complex than simply ‘climbed out’, it should be taken as ‘climbed out and by this tried to join the others’, where the reason relation is crucial. This complex predication is taken as the semantic associate of the additive particle, so we can argue for a broad focus reading again, which is marked in the prosody.

4.2 Scalar additive use

Sentences with a scalar additive reading exhibit similar prosodic patterns with regard to the relation between the range of focus (i.e., the focus domain) and (de)accenting of the constituents after the is-phrase. Various instances in our data support the claim that with the scalar additive reading (= even reading) the broad semantic associate is also available. Take, for example, sentence (36) below.

(36)

local discourse context: Egy hirtelen mozdulattal a béka újra lerúgja a kisbékát a tutajról, aki egy nagy placcsal beleesik a vízbe.
‘With a sudden movement, the frog kicks the little frog off the raft again, who falls into the water with a big splash.’
És	a	béka	még	a	nyelv-é-t	is	ki-nyújtja,
and	the	frog	yet	the	tongue-3sg.ps-acc	also	vprt-streches
‘And the frog even stretches his tongue out,’

Both the prosodic pattern and the context of (36) indicate a broad focus interpretation. In the context, it is described that the frog is annoyed that there is a new little frog in the family. Different actions of the frog against the little frog are mentioned and it is not presupposed that anything else was stretched out. As shown in the figure above, the verbal complex after the is-phrase also receives an accent, indicating its non-presupposed status. The relation of the prosodic pattern and the reach of the focus domain as illustrated above can also be shown in multi-word expression (or metaphoric expressions). Take, for example, the sentence in (37).

(37)

local discourse context: [the frog ate a bee instead of a mosquito] Azt hiszi, hogy egy jó kis szúnyog, ami vacsorára vagy ebédre neki nagyon jól fog esni, de aztán nagyon meglepődik, amikor valami hatalmas dolog történik belül,
‘He(=frog) thinks that it is a mosquito, that will be great for lunch or dinner, but he is very surprised, when something happens inside his mouth,’
mert	még	[a	szem-e	is	ki-gúvadt] SA
and	yet	the	eye-3sg.ps	also	vprt-goggled
‘because he was even looking with goggled eyes.’

Example (37) contains the multi-word/metaphoric expression kigúvad a szeme ‘is very surprised [lit. his eyes are goggled]’. Given that the noun and the verb form a multi-word expression, the sentence cannot be interpreted with a narrow associate of the additive particle. This is confirmed by the prosodic pattern, suggesting that the verbal complex is part of the focus domain. The sentence with deaccenting after the particle is odd.

In Sections 4.1 and 4.2, we provided prosodic evidence for a focus sensitive analysis of two basic uses of the particle is: the plain additive and the scalar additive use. We found that the syntactic host of the additive particle generally bears the main accent of the clause, hence, following Szendrői (2001, 2003) and Hamlaoui and Szendrői (2015), it marks the left edge of the focus domain of the sentence. We show that differences in the prosodic pattern after the is-phrase are closely related to the range of the focus domain, suggesting either a narrow or a broad focus interpretation. In narrow focus readings, the constituents after the is-phrase are deaccented, indicating that the information they carry is presupposed. On the other hand, in the broad focus interpretation, the constituents after the is-phrase retain their accents, indicating that the information is not presupposed. In the coming sections, we discuss further uses of the particle is and investigate whether and how the prosody of these cases supports a uniform focus-based analysis.

4.3 Coordination and listing

The next constructions we look at in this section are the iterated is-phrase expressing coordination (38) and listing (39).

These constructions are unfortunately much less frequent in our data,^[21] so further prosodic investigations are needed that concentrate on these particular uses. The two constructions are grouped together here, based on a common property of their prosodic patterns, namely that they have a special intonation contour with boundary tones (H-H%) after each item, which is a characteristic pattern for listing.

(40)

prosodic pattern of (39)/(14d):

Although we group the two readings together here, there are important structural differences between them that provide a reasonable ground for a distinction. In the coordination construction (e.g., Example 38), the use of és ‘and’ or meg ‘plus’ is optional, while the use of is ‘also’ is obligatory at each conjunct. In the listing construction (e.g., Example 39) on the other hand, the use of és ‘and’ is not optional and is ‘also’ can only appear at the last element of the list.^[22] A list can also have two elements (such examples occur more in our data), which is very similar to the coordination use, but based on the above criteria we can make the necessary distinction.

The interpretation of the construction expressing coordination is discussed by Brasoveanu and Szabolcsi (2013) and Szabolcsi (2017, 2018) in a crosslinguistic setting. They propose a dynamic semantic approach, based on Kobuchi-Philip’s (2009) analysis where the two particles mutually satisfy their additive presupposition and Brasoveanu’s (2013) notion of ‘post-supposition’. What is relevant for our discussion is that their analysis is based on the claim that the particle is is uniformly focus sensitive, and both syntactic hosts in this construction are focal. Without showing the prosodic patterns, Szabolcsi (2018) posits that both syntactic hosts are focal, and bear a focus accent. The analysis of the prosodic patterns of this construction in our data confirms that each syntactic host bears a pitch accent.

(41)

prosodic pattern of (38a):

prosodic pattern of (38b):

In our proposal, however, not only the accent of the syntactic host but also the prosodic pattern after the is-phrase determines the focus. In this use of is, a broad semantic associate is possible, similarly to the plain and scalar additive uses.

In our data, the variation of accenting/deaccenting after the is-phrases in the coordination use is neutralized, and is generally pronounced with the neutral stress pattern for both narrow and broad associates. This is problematic for our basic proposal on focus marking in this particular reading. We assume that the presence of the list intonation pattern plays a crucial role of the prosodic pattern here. However, we have too few sentences of this construction in our data, therefore we cannot draw hard conclusions on this particular issue. Further, directed experiments are necessary with regard to the prosody of this particular use of is.

To sum up our findings on the coordination use and listing, we found that in both constructions, (i) the syntactic host of the particle bears a pitch accent (high or low) and (ii) there is a special intonation contour specific to listing. Based on our findings, we can argue that the prosodic patterns partly support the semantic analysis by Brasoveanu and Szabolcsi (2013) and Szabolcsi (2017). The accent of the syntactic host can be related to focus marking, and the list intonation pattern can be related to the conjunction involved in the interpretation. This reflects the view of Szabolcsi (2017, 2018) in the interpretation of the coordination use, namely that is is responsible for the additive presupposition based on the focus semantics of its syntactic host, but not responsible for the conjunction. We see a similar division in the prosody, where accent on the host marks the focus as before, and the contribution of coordination, special for this construction, is marked by an additional list intonation contour, also only occurring in this particular use. The unresolved issue of the prosody after the is-phrase in this particular construction needs further investigation.

4.4 The ‘indeed’-reading

The last construction we discuss is the one that (to the best of our knowledge) has not been discussed before in the literature. As illustrated in (14e), the additive particle is can also be associated with a special interpretation, we call the ‘indeed-reading’, where additivity targets the intention or goal of the participant. With this reading, the semantic associate is broad, a full proposition, and the presupposition is that the proposition is true in an alternative world, representing the goal, wish or intention of an individual. Consider Example (14e) repeated here as (43):

The semantic associate of is in (43) is the whole proposition ‘frog arrives in the flat’. This broad associate is reflected in the prosodic structure, the elements after the is-phrase retain their accents.

(44)

prosodic pattern of (43):

The syntactic structures of the sentences with the plain additive reading and the ‘indeed-reading’ are the same (see also Tables 1 and 2 before). We also found similar prosodic patterns to those in the plain/scalar additive use, hence the prosodic structure does not disambiguate between the ‘indeed’-reading and the others. The prosody only gives the information that the semantic associate must be broad (a sentence). The correct reading is derived based on the context that determines which possible presupposition is fulfilled.

As shown in Section 2.3.2, this special ‘indeed-reading’ is also available with a different linearization, licensed by a context where identificational focus is already present. Such examples were not present in our data set, so we recorded some constructed examples to investigate the prosodic structure in this construction, see (45), for example.

In the target sentence, the inverse order of the verb and the particle suggests that the particle is could associate with the structural focus, and that the whole is-phrase is in the immediate preverbal focus position.

(46)

prosodic structure of (45):

We argue, however, that the above structural analysis is incorrect. This construction only occurs with the special ‘indeed’-reading within a local discourse context where the preverbal/identificational focus (id-foc) is already present. The additive particle in (45) takes the sentence containing the preverbal/identificational focus as its semantic associate, not merely the constituent in the preverbal focus position (47). See also the discussion of Example (18) in Section 2.3.2.

Based on its interpretation, we argue that it is not the whole is-phrase that is in the focus position, which needs to be represented in the syntactic structure. This challenging issue is left for further work.