1 Introduction*
This paper discusses the semantics and pragmatics of the Bavarian discourse particle fei, which does not have a direct counterpart in standard German. Etymologically, fei is related to Latin finis and French fin ‘end, border’ and has entered Bavarian in the 12th century (Schlieben-Lange 1979; Glaser 1999). As usual for discourse particles (cf. Jacobs (this volume); Dörre & Trotzke (this volume)) the exact meaning of fei is hard to pin down. Traditionally, it is taken to be an emphatic particle that conveys the speaker’s assumption that the content of the sentence containing fei (henceforth: the prejacent of fei) is surprising for the hearer (Schlieben-Lange 1979; Glaser 1999). In Thoma (2009), which is (to the best of my knowledge) the only explicit discussion of fei in modern, theoretical linguistic terms, fei is analysed as an item that is not only a discourse particle, but evokes at the same time polarity focus. Thoma (2009) proposes that the meaning of a sentence of the schematic form fei p can (roughly) be paraphrased as follows: The speaker believes that the hearer believes that not p. Additionally, she claims that no matter which part of the prejacent of fei receives the main accent, the sentence can only be understood as expressing polarity focus, i.e. focus on the truth value of the respective sentence (Höhle 1992).
In this paper, I will argue that while Thoma (2009) captures important aspects of the meaning of fei, there is a crucial feature setting fei apart from other discourse particles expressing some form of contrast such as aber and doch, which is not addressed by her analysis at all: fei not only makes its contribution at the level of non-at-issue content (Potts 2005)—which is standard for discourse particles; see Gutzmann (2015) and the references therein—, but also relates to non-at-issue content exclusively. Consequently, fei can only be added felicitously to a sentence denoting the proposition p in cases where the speaker’s assumption that the hearer believes not p is not based on the hearer’s prior assertion of a proposition entailing not p. Rather, that assumption has to be based on the hearer’s non-verbal behavior in the respective situation, on general background knowledge, on conventional or conversational implicatures of sentences previously uttered by the hearer or on the presuppositions of such sentences (i.e. on propositions which the hearer, erroneously, assumes to be part of the Common Ground in the sense of Stalnaker (1978), but which the speaker in fact takes to be false).
Building on the idea that the at-issue content of an assertion has a special status insofar as it does not enter the Common Ground directly, but rather is explicitly put on the table by the speaker in a first step and has to be explicitly or implicitly accepted by the addressee before entering the Common Ground (AnderBois, Brasoveanu & Henderson 2015; Murray 2017), the fact that fei relates to non-at-issue content exclusively can be explained along the following lines: By adding fei, the speaker draws the hearer’s attention to a conflict between her own beliefs and the hearer’s beliefs that is not salient at the point where the respective sentence is uttered. In a case where the proposition q believed by the hearer that contradicts the (propositional content of the) prejacent of fei is the at-issue content of a sentence previously uttered by the hearer, the question of whether q is true is highly salient, however. Uttering a sentence that contradicts q therefore makes the conflict between the speaker’s and the hearer’s beliefs automatically salient as well. The addition of fei to such a sentence would thus be superfluous.
The paper is structured as follows. Section 2 gives some general background on discourse particles and summarizes Thoma’s 2009 analysis of fei. Section 3 presents the novel data that motivate the account argued for in this paper. My own analysis of fei is presented in Section 4, and Section 5 briefly compares fei with another discourse particle that expresses some form of contrast, namely doch. Section 6 is the conclusion.
2 Thoma’s (2009) Analysis of fei
2.1 General Background
Standard German as well as its dialectal variants has a large number of discourse particles. Discourse particles are characterized by a number of syntactic and semantic features: They are always optional, they can only occur in the so-called middle field, they cannot receive the main accent of the respective clause,1 they cannot be questioned, they cannot be negated and they do not contribute to the truth conditions of sentences containing them (Weydt 1969; Thurmair 1989; Jacobs 1991; Ormelius-Sandblom 1997; Karagjosova 2004; Coniglio 2011; Zimmermann 2008, 2012; Gutzmann 2015; see the papers in Bayer & Struckmeier 2017 for a recent overview). Rather, they serve to link the proposition denoted by the respective sentence to the context by indicating how it is related to the speaker’s or the hearer’s epistemic states or the Common Ground at the point where the sentence is uttered.
The sentences in (1b) and (1c), for example, have the same propositional content as the one in (1a) and are thus true in exactly the same circumstances. But whereas (1b) is felicitous in a context where it is an established fact that the speaker is going to Hamburg on the day following the day where the sentence is uttered and where the speaker considers it at least possible that the hearers knows that (Lindner 1991; Kratzer 1999), (1c) is infelicitous in such a context. Rather, it is (very roughly) felicitous in a context where the speaker assumes the hearer to have known previously that she is going to Hamburg, but to be temporarily unaware of that fact; see Zeevat (2003); Karagjosova (2004a, 2009); Egg (2013); Grosz (2014); Rojas-Esponda (2014) for detailed analyses built on or related to that idea. The sentence in (1b), in contrast, would be infelicitous in such a context. Finally, (1a) is felicitous in both kinds of contexts.
(1) a. Ich fahre morgen nach Hamburg.
b. Ich fahre ja morgen nach Hamburg.
c.
Ich
fahre
doch
morgen
nach
Hamburg.
I
drive
(PART)
tomorrow
to
Hamburg
‘I’m going to Hamburg tomorrow.’
Based on the observation that discourse particles do not alter the truth conditions of sentences containing them, but rather impose certain felicity conditions on the contexts in which they are uttered, Kratzer (1999) has proposed that discourse particles do not have truth conditions, but rather use conditions. This analysis has been formally worked out in a multi-dimensional semantics inspired by the framework of Potts (2005) in Gutzmann (2015). The basic idea behind this approach is that sentences containing discourse particles make two contributions at the same time: First, they have regular truth-conditional content that is unaffected by the respective discourse particle. At the same time, they have (non-trivial) use-conditional content that is interpreted at a level distinct from the level where truth-conditional content is interpreted. Technically, discourse particles take the entire proposition denoted by their prejacent as their argument and return a proposition whose success is not defined in terms of truth, but rather in terms of felicitous use. A sentence such as (1c), for example, is thus (a) true iff the speaker goes to Hamburg on the day following the day where the sentence is uttered, and (b) used felicitously iff the hearer (roughly) knew that before but has temporarily forgotten it at the time where the sentence is uttered.
2.2 Thoma’s Analysis of fei
Thoma (2009) is, to the best of my knowldegde, the only existing analysis of the Bavarian discourse particle fei in modern, theoretical linguistic terms. Thoma assumes that fei is not only a discourse particle, but also encodes polarity focus (Höhle 1992), i.e. focus on the truth value of the respective sentence. She does not give a formal implementation of her analysis, but rather offers a paraphrase along the following lines: A sentence containing fei is felicitous in a context in which the speaker believes that the hearer believes the propositional content of the respective sentence to be false. This assumption is based on examples such as the one in (2), which can be uttered felicitously in a situation where the speaker observes the addressee to return from the restroom with his fly unzipped:
(2)
Dei
Hos’ntiarl
is
fei
auf.
Your
pant.door
is
fei
open
‘Your fly is down.’ (Thoma 2009: 140)
In such a situation, it is plausible for the speaker to assume that the hearer believes the proposition that his fly is unzipped to be false, i.e. to believe that his fly is zipped. Intuitively, such a belief can plausibly be attributed to the hearer by the speaker for the following reasons: First, people normally (i.e. in standard social situations) don’t want others to see them with their flies unzipped. Second, it is not uncommon to forget zipping one’s fly after using the restroom. Further examples discussed by Thoma are the ones in (3) and (4):
(3) A: I glaub’ de Sechzga san recht beliebt in Minga.
‘I believe that 1860 (a soccer club) is pretty popular in Munich.’
B:
De
meist’n
Muenchna
san
fei
Bayernfans.
The
most
Munichers
are
fei
fans.of.Bayern
‘Most Munichers are fans of Bayern München.’ (Thoma 2009: 145)
(4) A: Can I open the window?
B:
Mi
friats
fei.
Me
is.cold
fei
‘I am cold.’ (Thoma 2009: 146)
In the case of (3), the use of fei in B’s answer is licensed as follows: A’s assertion in combination with the general background knowledge that nobody is both a fan of 1860 and a fan of Bayern München entails that most Munichers are not fans of Bayern München. B thus has good reasons to believe that A believes that proposition, which is the negation of the propositional content of her utterance. In the case of B’s assertion in (4), the use of fei is licensed since it is at least not implausible to interpret A’s question as a politeness question rather than as a genuine question, i.e. in such a way that A assumes that the speaker is not cold and thus will not object to her opening the window, but just wants to be polite. Otherwise, i.e. if A’s question was a genuine question, the use of fei would be unexpected on Thoma’s account, contrary to what Thoma herself assumes—she takes A’s question to conventionally implicate that A believes that the addressee is not cold. In my view, this assumption is misguided: Asking for B’s permission to open the window would be rather pointless if A already believed that the most likely reason for why B would deny her permission—namely that she is cold—can be disregarded.
Thoma’s assumption that fei is not only a discourse particle with the felicity conditions stated above, but also encodes polarity focus, is based on the following claim: A sentence such as (5a), with focal accent on the direct object Sepp, is necessarily understood as introducing the polar opposite of the proposition it denotes, i.e. the proposition that Hans has not invited Sepp, as its only focal alternative. According to Thoma, the retorts in (5c) and (5d) are thus infelicitous, in contrast to the one in (5b), in a context in which Vroni and Maria are talking about Hans’ upcoming party and in which Sepp is Hans’s best friend.
(5)
a.Vroni:
Da
Hans
hod
fei
an
SEPP
ei’glon.
The
Hans
has
fei
the-ACC
Sepp
invited
‘Hans has invited Sepp.’
b.Maria:
Ja
wos
denn
sunst?
Häd’an
NED
eilon
soin?
PRT
what
PRT
else
have.he.him
not
invite
should
‘What else? Souldn’t he have invited him?’
c.Maria:
Ja
wos
denn
sunst?
#
Häd’a
an
BÄDA
eilon
soin?
PRT
what
PRT
else
Have.he
the-ACC
Peter
invite
should?
‘What else? Should he have invited Peter?’
d.Maria:
Ja
wos
denn
sunst?
#
Häd’a
an
Sepp
BUSSLN
soin?
PRT
what
PRT
else
have.he
the-ACC
Sepp
kiss
should?
‘What else? Should he have kissed Sepp?’ (Thoma 2009: 144–145)
Thoma’s arguments for the claim that fei encodes polarity focus are not convincing. First of all, neither me nor the other three native speakers of Bavarian I consulted share the intuition that (5b) is infelicitous as a retort to Maria’s claim in (5a). Rather, (5b) is perfectly fine in a situation in which Vroni (wrongly) assumes Maria to believe that Hans has invited some other person instead of Peter. Second, while (5d) is indeed infelicitous as a retort to (5a), this is simply due to stress placement in (5a): If the main accent falls an the verb, as in the variant of (5a) given in (6), (5b) is in principle acceptable as a retort. It is admittedly a bit hard to think of a suitable context, i.e. one in which Vroni (wrongly) assumes Maria to believe that Hans did not invite Sepp to his party, but rather did something else to Sepp that counts as a suitable alternative to inviting him.
(6)
Vroni:
Da
Hans
hod
an
Sepp
fei
EI’GLON.
The
Hans
has
the-ACC
Sepp
fei
invited
I therefore conclude that the focal alternatives introduced by a sentence containing fei are not restricted in the way Thoma claims them to be, and that there is thus no need to complicate the semantics of fei by assuming that it is not only a discourse particle with the felicity conditions stated above, but additionally encodes polarity focus.
Thoma points out that the infelicity of fei both in direct answers to questions and in questions themselves (see 7–9) is directly predicted by her account: Concerning direct answers to questions, it would make little sense to add fei, since that would indicate that the person answering the question assumes the person who has asked the question to believe the polar opposite of the respective answer. Asking a (non-rhetorical) question, however, commits the speaker to not already believe one of its possible answers to be true. Someone who adds fei to her answer would thus assume the person who asked the question to have acted irrationally. Concerning questions themselves, the addition of fei would likewise make little sense, since fei is defined to operate on propositions, not sets of propositions (examples from Thoma 2009: 148).
(7) A: Schneibts draussn?
‘Is it snowing outside?’
B: #Ja, es schneibt fei.
‘Yes it snows fei.’
(8) A: Wer mog an Kafä?
‘Who wants coffee?’
B:
#
I
mog
fei
an
Kafä.
I
want
fei
a-ACC
coffee
(9)
a.
#
Schneibts
fei
draussn?
snows.it
fei
outside
b.
#
Wer
mog
fei
an
Kafä?
who
wants
fei
a-ACC
coffee
Additionally, Thoma points out that the unacceptability of fei in infinitival commands such as (10b), as opposed to its acceptability in ‘real’ imperatives such as (10a), which project a CP layer, follows from the status of fei as a discourse particle: As such, it needs to be hosted in the CP-layer at LF (see Zimmermann 2012 and the references therein for discussion; see also Gärtner 2013; 2014a,b for additional discussion).2 Apart from that basic observation, Thoma does not say anything about the conditions under which fei is acceptable in imperatives. I will return to that point in Section 3.
(10)
a.
Bass
fei
auf!
watch
fei
out
b.
*
Fei
aufbass’n!
fei
out.watch
Thoma ends her paper by a brief comparison of fei and doch. As already mentioned in Section 2.1, doch is (roughly) felicitous iff the speaker assumes that the hearer previously believed the propositional content of the prejacent of doch, but has either forgotten it at the time where that sentence is uttered or believes it to be false. It is thus predicted that replacing fei by doch in (2), repeated here as (11a), leads to infelicity (in the absence of a special context), as shown in (11b): While it is plausible for the speaker to assume that the hearer is unaware of his fly being down (and hence believes it to be not down) when he returns from the toilet, it is implausible to assume that he was previously aware of his fly being down and has temporarily forgotten it.
(11)
a.
Dei
Hos’ntiarl
is
fei
auf.
your
pant.door
is
fei
open
‘Your fly is down.’ (Thoma 2009: 150)
b.
#
Dei
Hos’ntiarl
is
doch
auf.
your
pant.door
is
doch
open.
I will turn to a more detailed comparison between fei and doch in section 5.
3 New Data
In the previous section we have seen that Thoma (2009) analyzes fei as a discourse particle encoding the speaker’s belief that the hearer believes the polar opposite of the respective sentence’s propositional content. In this section I will show that while Thoma’s analysis captures an important part of fei’s felicity conditions, it misses another crucial aspect: fei is infelicitous iff the hearer has previously uttered a sentence whose at-issue content entails the polar opposite of the proposition denoted by the sentence containing fei (cf. Schwenter 2005 for related discussion concerning the use of negation in Brazilian Portugese).
Consider the sentences in (12) and (13b). The second sentence in (12) is felicitous in a situation in which the speaker can plausibly assume the hearer to have a generally positive attitude towards the movies of Sofia Coppola, and thus to believe that her new movie is good as well (or at least not bad). Similarly, (13b) is felicitous in a context in which Maria can plausibly assume that Tom does not like twelve tone operas and hence did not know that Moses and Aron is a twelve tone opera when he (implicitly) suggested to watch it on TV by uttering (13a).
(12)
Gesta’n
hob
i
den
nei’n
Film
von
da
Sofia
Coppola
g’seng.
Des
is
fei
a
richtiga
Schmarr’n.
Yesterday
have
I
the
new
movie
by
the
Sofia
Coppola
seen
That
is
fei
a
real
nonsense
‘Yesterday I have seen the new movie by Sofia Coppola. It’s real nonsense.’
(13)
a.Tom:
Heit
o’mnd
kimt
Moses und Aron
im
Feanseh’n.
Today
evening
comes
Moses and Aron
in.the
TV
‘Moses and Aron is on TV tonight.’
b.Maria:
Des
is
fei
a
Zwölftonoper.
That
is
fei
a
twelve-tone opera
‘It’s a twelve-tone opera.’
In these two examples as well as in Thoma’s (2009) examples of felicitous uses of fei discussed in Section 2.1, the speaker’s belief that the hearer believes the polar opposite of (the propositional content of) fei’s prejacent is always an inference based on (a) the hearer’s behavior in combination with general background assumptions ((2)), (b) a previous utterance of the hearer in combination with general background assumptions ((3), (4), (13)), or (c) general background assumptions about the hearer exclusively ((12)). Now, Thoma’s analysis makes a clear prediction: A sentence containing fei should be perfectly fine in a situation in which the hearer has previously uttered a sentence entailing the negation of the propositional content of that sentence, or a sentence entailing that she believes the negation of the propositional content of that sentence. After all, the fact that someone has uttered a sentence entailing not p or the proposition that she believes that not p is an excellent reason to believe that she believes that not p (in fact, the best I can think of). Consequently, the second sentence in (12) and the one in (13b) should both be fine in a context in which the hearer has previously uttered a sentence entailing that she does not believe the new movie by Sofia Coppola to be complete nonsense, and that she does not believe Moses und Aron to be a twelve-tone opera, respectively. This prediction is not borne out, however: The two sentences are extremely awkward in the respective contexts, as shown in (14) and (15).
(14) a.Maria: Da neie Film von da Sofia Coppola is supa.
‘The new movie by Sofia Copploa is great.’
b.Tom:
(Quatsch!)
Des
is
(#fei)
a
richtiga
Schmarrn.
Bullshit
That
is
fei
a
real
nonsense
‘Bullshit! It’s complete nonsense.’
(15) a.Maria: Moses und Aron is a Belcanto Oper.
‘Moses and Aron is a belcanto opera.’
b.Tom: (Quatsch!) Des is (#fei) a Zwölftonoper.
‘Bullshit! It’s a twelve-tone opera.’
Likewise, the sentence with fei in (3) is infelicitous as a reaction to an utterance entailing the negation of its propositional content, as shown in (16):
(16) A: Koa Muenchna is a Bayernfan.
‘No Municher is a fan of Bayern München.’
B:
(Quatsch!)
De
meist’n
Muenchna
san
(#fei)
Bayernfans.
Bullshit
The
most
Munichers
are
fei
fans.of.Bayern
‘Bullshit! Most Munichers are fans of Bayern München.’
Consider next the contrast between the infelicity of the sentence with fei as a reaction to A’s utterance in (17), and its felicity as a reaction to A’s utterance in (18).
(17) A: Da Bäda is a Depp.
‘Peter is an idiot.’
B: (Quatsch!) Da Bäda is (#fei) koa Depp.
‘(Bullshit!) Peter is no idiot.’
(18)
A:
I
frog
an
Bäda
ob’a
mei
oids
Radl
kafa
mecht.
I
mecht
mindestens
200
Euro
dafür
ho’m.
I
ask
the-ACC
Peter
if.he
my
old
bicycle
buy
wants
I
want
at.least
200
euros
for.it
have
‘I’ll ask Peter if he wants to buy my old bicycle. I would like to have at least 200 Euros for it.’
B:
Da
Bäda
is
fei
koa
Depp.
the
Peter
is
fei
no
idiot
‘Peter is no idiot.’
In (17) B asserts the negation of the sentence previously uttered by A. This seems to be incompatible with the felicity conditions of fei. In (18), in contrast, B asserts the negation of a proposition that she can plausibly assume A to believe, given a suitable context—namely a context in which A’s old bicycle is known to be worth far less than 200 Euros and in which it would consequently be irrational for Peter to pay 200 Euros or more for it. Again, since B only infers that A believes the negation of the propositional content of B’s utterance on the basis of a previous assertion by A in combination with background knowledge, the addition of fei is perfectly fine. A similar contrast obtains between (19) and (20): Adding fei is fine in (20), where B infers from A’s behavior that A believes the negation of the propositional content of her utterance (i.e. that it is not cold), but infelicitous in (19), where A has previously asserted the corresponding proposition.
(19)
A:
S’is
goa
ned
koit
drauss’n.
It.is
at.all
not
cold
outside
‘It’s not cold at all outside.’
B:
(Quatsch!)
S’is
(#fei)
saukoit
drauss’n.
Bullshit
It.is
fei
terribly.cold
outside
‘(Bullshit!) It’s terribly cold outside.’
(20) [A is about to leave the house wearing only a shirt.]
B:
S’is
fei
saukoit
drauss’n.
It.is
fei
terribly.cold
outside
Finally, consider the contrast between the infelicity of fei in B’s reaction to A’s utterance in (21), and it’s felicity in B’s reaction to A’s utterance in (22). The only difference between (21) and (22) is that in the case of (21), A has asserted the negation of the proposition asserted by B, while in the case of (22) A has asserted a proposition that conversationally implicates the negation of that proposition.
(21)
A:
Da
Otto
hot
den
ganz’n
Kuacha
gess’n.
The
Otto
has
the
whole
cake
eaten
‘Otto has eaten the whole cake.’
B:
(Quatsch!)
Da
Otto
hot
den
Kuacha
(#fei)
ned
gess’n.
Des
woa
d’Maria.
(Bullshit)
The
Otto
has
the
cake
fei
not
eaten
That
was
the.Maria
‘(Bullshit!) Otto has not eaten the cake. It was Maria.’
(22)
A:
Da
Otto
is
in
da
Kich
g’wen
und
da
Kuacha
is
weg.
The
Otto
is
in
the
kitchen
been
and
the
cake
is
gone
‘Otto was in the kitchen and the cake is gone.’
B: Da Otto hot den Kuacha fei ned gess’n. Des woa d’Maria.
Now, there is a plausible pragmatic explanation for the data considered in this section so far that allows us to stick with the felicity conditions of fei stated by Thoma (2009): Whenever an interlocuter A in a conversation utters Not p, I believe that not p or a sentence entailing one of those propositions, the other interlocuters automatically know that A believes that not p (at least in a standard setting where A is assumed to be sincere). Since know is stronger than believe, the addition of fei, which only requires the speaker to believe that the hearer believes that not p, may be blocked. The reason for the contrast between (21) and (22), for example, would thus be that in (21) B knows that A believes that Otto has eaten the cake since A has publicly committed to believing the corresponding proposition, while in (22) B only believes that A believes that Otto has eaten the cake: After all, A has only asserted a proposition that conversationally implicates that Otto has eaten the cake, and since it is one of the defining characteristics of conversational implicatures that they are defeasible (Grice 1989), A could have continued her first sentence in (22) by uttering (the Bavarian equivalent of) … but he has not eaten the cake without contradicting herself. Consequently, in (22) B only believes, but crucially does not know that A believes that Otto has eaten the cake, and similarly for all other cases considered so far where fei is felicitous.
But now consider the contrast between (23) and (24): While the infelicity of fei in B’s reply in (24) is directly predicted by Thoma’s (2009) analysis augmented by the pragmatic reasoning just sketched, the felicity of fei in B’s reply in (23) is entirely unexpected.
(23) A: Da Kini von Frankreich is a Depp.
‘The king of France is an idiot.’
B:
In
Frankreich
gibt’s
fei
koan
Kini
nemma.
Des
is
a
Republik.
In
France
there.is
fei
no
king
anymore
That
is
a
rebublic
‘In France there is no king anymore. It’s a republic.’
(24)
A:
In
Frankreich
gibt’s
imma
no
an
Kini.
in
France
there.is
always
still
a
king
‘In France there is still a king.’
B:
(So
a
Schmarr’n).
In
Frankreich
gibt’s
(#fei)
koan
Kini
nemma.
such
a
nonsense
in
France
there.is
fei
no
king
anymore
‘(What nonsense!) In France there is no king anymore. It’s a republic.’
It is generally assumed, following Strawson (1950), that the definite article presupposes the existence and uniqueness (possibly with respect to a contextually restricted domain) of an individual satisfying the predicate denoted by its NP-complement. This is shown by the fact that the negation of a sentence such as The king of France is bald, i.e. The king of France is not bald, is not simply true in virtue of France’s being a republic. Rather, the non-existence of a king of France seems to prevent the negated as well as the non-negated sentence from getting a truth value. Not being affected by semantic operations such as negating or questioning is one of the defining features of presupposed as opposed to asserted content: Consequently, a question such as Is the king of France bald? cannot be understood as asking for the existence of an individual that has the property of being the unique king of France and of being bald. Rather, a person who asks such a question can only be understood as taking the existence of a unique king of France for granted and asking whether that individual is bald.
By uttering the sentence in (23), A therefore does not assert, but rather presupposes, i.e. takes for granted the existence of a (unique) king of France. Consequently, she (wrongly) assumes the proposition that there is a (unique) king of France to be included in A’s and B’s Common Ground (Stalnaker 1978), i.e. to be one of their mutually shared assumptions. This, however, entails that A believes that there is a king of France, and B consequently knows as soon as A has uttered the first sentence in (23) that she believes there to be a king of France. Consequently, given the pragmatic reasoning just outlined, the use of fei in B’s reply should be just as infelicitous in (23) as it is in (24), contrary to fact. The contrast between the felicitous and the infelicitous uses of fei can thus not be derived from the contrast between the respective speaker’s believing vs. knowing that the hearer believes the negation of the proposition denoted by the sentence containing fei. Rather, what seems to be crucial is whether the negation of that proposition has entered the context via being asserted by the hearer or not: If it has, using fei is infelicitous. This is confirmed by the following contrast:
(25)
A:
Gesta’n
bin
i
mi’m
Bäda
im
Wirtshaus
g’wen.
Dea
hot
mi’m
Raucha
aufgheat.
Yesterday
am
I
with.the
Peter
in.the
pub
been
He
has
with.the
smoking
stopped
‘Yesterday I was in the pub with Peter. He has stopped smoking.’
B:
Da
Bäda
hot
fei
no
nia
g’raucht.
The
Peter
has
fei
as.yet
never
smoked
‘Peter never smoked.’
(26)
A:
Da
Bäda
raucht.
The
Peter
smokes
‘Peter smokes.’
B:
(So
a
Schmarr’n).
Da
Bäda
hat
(#fei)
no
nia
g’aucht.
Such
a
nonsense
the
Peter
has
fei
as.yet
never
smoked
‘(What nonsense!) Peter never smoked.’
Just like the definite article presupposes the existence and uniqueness of an individual satisfying the predicate denoted by its NP-complement, the verb stop presupposes that its subject used to be engaged in the activity denoted by the verb it combines with (see Beaver 1997 and the references cited therein). Consequently, A in (25) (wrongly) assumes the proposition that Peter used to smoke to be in the Common Ground of A and B, which entails that A believes that Peter used to smoke. B therefore knows in both (25) and (26) that A believes that Peter used to smoke as soon as A has uttered the respective first sentence. Nevertheless, the use of fei is only infelicitous in (26), where A has asserted the corresponding proposition.
Apart from presuppositions, there is a second kind of linguistic content that is not affected by semantic operations such as negation and questioning—namely so-called conventional implicatures (in the sense of Potts (2005)) such as nominal appositives, appositive relative clauses and expressives. The sentences in (27a)–(27b) and (28a)–(28b), for example, are both not understood as negating or questioning that Peter is a great dancer/an idiot, respectively.
(27) a. Peter, (who is) a great dancer, did not come to the party.
b. Did Peter, (who is) a great dancer, come to the party?
(28) a. That idiot Peter did not come to the party.
b. Did that idiot Peter come to the party?
Consider now the contrast between (29) and (30), on the one hand, and (31) and (32), on the other: In both (29) and (30) B knows after A’s utterance that A believes Chomsky to be a sociologist, and in both (31) and (32) B knows after A’s utterance that A considers Peter to be an idiot. Nevertheless, fei can felicitously be added to a sentence entailing the negation of the respective proposition believed by A in (29) and (31), where that proposition is a conventional implicature. In (30) and (32), in contrast, where that proposition is the at-issue content of A’s utterance, the addition of fei is infelicitous.
(29)
A:
Da
Chomsky,
a
berühmta
Soziologe,
is
a
Anarchist.
the
Chomsky
a
famous
sociologist
is
an
anarchist
B:
Da
Chomsky
is
fei
koa
Soziologe.
the
Chomsky
is
fei
no
sociologist
(30)
A:
Da
Chomsky
is
a
berühmta
Soziologe.
the
Chomsky
is
a
famous
sociologist
B:
(So
a
Schmarr’n).
Da
Chomsky
is
(#
fei)
koa
Soziologe.
such
a
nonsense
the
Chomsky
is
fei
no
sociologist
‘What nonsense! Chomsky is no sociologist!’
(31)
A:
Dea
Saudepp
Bäda
is
a
Anarchist.
That
pig.idiot
Peter
is
a
anarchist
‘That idiot Peter is an anarchist.’
B:
Da
Bäda
is
fei
koa
Depp.
the
Peter
is
fei
no
idiot
(32)
A:
Da
Bäda
is
a
Saudepp.
the
Peter
is
a
pig.idiot
B:
(So
a
Schmarr’n!)
Da
Bäda
is
(#
fei)
koa
Depp.
such
a
nonsense
the
Peter
is
fei
no
idiot
‘What nonsense! Peter is no idiot!’
The fact that fei can be added felicitously to sentences entailing the negation of a proposition that is either presupposed or conventionally implicated by a previous utterance of the interlocutor shows that the pragmatic account sketched above cannot be maintained: In both cases the speaker can be as certain that the hearer believes the respective proposition as she could be if that proposition had been the at-issue content of a previous assertion by the hearer. In Section 4 I will thus present an analysis of fei that retains the insights of Thoma (2009) but can at the same time account for the full range of data considered in this section. But before I will turn to that analysis, I will briefly return to an issue that I have briefly mentioned at the end of Section 2: The fact that fei can be added to imperatives.
Thoma (2009) points out that the contrast between (10a) and (10b), repeated here as (33a)–(33b), directly follows from the status of fei as a discourse particle which as such requires a C-projection. Since (10a)/(33a), but not (10b)/(33b), is a full CP, fei can only be added to the former, but not the latter. Apart from the remark that the use of fei in imperatives such as (10a)/(33a) is quite common in Bavarian, Thoma (2009) does not say anything more about the distribution of fei in imperatives.
(33)
a.
Bass
fei
auf!
watch
fei
out
‘Watch out!’
b.
*
Fei
aufbass’n!
fei
out.watch
Let us now turn to the question of whether the licensing conditions of fei in imperatives are identical or at least related to those in assertions. There is an obvious complication: Since declarative sentences clearly denote propositions, it is unproblematic to assume that by adding fei to a declarative sentence 
the speaker indicates that she does not believe the hearer to believe the proposition p denoted by 
. Concerning imperatives, in contrast, the question of what semantic objects they denote is rather controversial. Kaufmann (2012) assumes imperatives to denote implicitly modalized propositions and therefore to be essentially equivalent to performatively used sentences with overt deontic modals. Setting aside the contribution of fei, (33a) is thus equivalent to a sentence such as You must/should be careful, on the performative use. Portner (2007), in contrast, assumes imperatives to denote properties that get added to the addressee’s To-Do-List, where To-Do-Lists are sets of properties that commit an individual to act in such a way that those properties apply to her in the future. Ignoring the contribution of fei again, uttering an imperative such as (33a) on Portner’s (2007) account thus has the effect of committing the addressee to act in such a way that the property of being careful applies to her in the future.
Developing a unified analysis that accounts for the behaviour of fei in both declaratives and imperatives is straightforward if one adopts an analysis along the lines of Kaufmann (2012), but more complicated if one assumes with Portner (2007) that imperatives denote properties rather than propositions. Since the analysis of imperatives is not my main concern in this paper, I will thus without further argument simply adopt Kaufmann’s (2012) analysis of imperatives. This allows us to retain the assumption that fei can only felicitously be added to a sentence 
if the speaker believes that the addressee believes the negation of the propositional content of 
.
Concerning (33a), its felicity is predicted since it is quite plausible to assume that when uttering a warning the speaker assumes the hearer to be not aware of the need to be careful, i.e. to believe that she does not have to be careful. Likewise, the command in (34) is predicted to be felicitous in a context like the following one: The speaker, who is the mother of the addressee, is about to leave the house and, based on previous experience, assumes that her teenage son does not see the need to clean up his messy room, i.e. to believe that he does not have to clean up his room. This prediction seems to me to be correct.
(34)
Reim
fei
dei
Zimmer
auf!
Clean
fei
your
room
up
‘Clean up your room!’
Concerning the second crucial felicity condition of fei, namely that it cannot be added to sentences entailing the negation of the at-issue content of a previous utterance by the hearer, this is a little harder to test for imperatives, since it is not easy to construe natural scenarios where that condition is violated. Nevertheless, the following observation provides preliminary evidence that with respect to the second felicity condition, too, fei behaves identically in declaratives and imperatives: If the variant of the first sentence in (35) with fei is uttered in the context of (35), it sounds rather awkward. At the same time, the sentence is felicitous in a context where the speaker assumes on the basis of general background knowledge or the addressee’s non-verbal behavior that she might be interested in drinking another glass of beer, but is unsure of whether she is allowed to do so.
(35)
A:
I
deaf
ma
bestimmt
koa
Bia
mea
nemma,
sunst
langt’s
nimma
für
d’Maria.
I
may
myself
certainly
no
beer
anymore
take
otherwise
it.suffices
no.more
for
the.Maria
‘I’m certainly not allowed to take another glass of beer, otherwise there won’t be enough left for Maria.’
B:
Nimm
da
(#
fei)
no
a
Bia!
S’is
no
gnuag
do.
Take
yourself
fei
still
a
beer
It.is
still
enough
there
‘Take another glass of beer! There’s still enough left.’
Consider finally the following contrast, which has been pointed out to me by Josef Bayer: The sentence in (36) is felicitous in a context where the speaker assumes on the basis of general background knowledge concerning the addressee’s violent behavior in the past that he is about to beat up his son. At the same time, it would be extremely awkward for the speaker to utter the same sentence in a situation where she is directly observing that the addressee is beating up his son.
(36)
Hau
fei
dein
Bua’m
ned!
Beat.up
fei
your
boy-ACC
not
‘Don’t beat up your son!’
Intuitively, this contrast is related to the constraint preventing the usage of fei in cases where a sentence entailing the negation of the (propositional content of the) prejacent of fei has been asserted by the addressee: Performing an action in the utterance situation that brings about a state of affairs which is the direct opposite of the state of affairs that the speaker wants the addressee to bring about by uttering an imperative seems to have the same status as the assertion of a contradicting proposition by the addressee in the case of declaratives. That this only holds for imperatives is shown by the felicity of (37) in the same situation in which (36) is infelicitous.
(37)
Du
deafst
fei
dein
Bua’m
ned
hau’n!
you
may
fei
your
boy-ACC
not
beat.up
‘You must not beat up your son!’
Since the analysis of imperatives is not my main concern in this paper, I will leave a deeper investigation of that contrast and its consequences as a topic for future research.
4 The Analysis
In the preceding section we have seen that fei not only indicates the speaker’s belief that the hearer believes the negation of the propositional content p of the speaker’s utterance (Thoma 2009), but is subject to an additional felicity condition: The negation of p may not be entailed by the at-issue content of a previous assertion by the hearer. Consequently, the speaker’s belief that the hearer believes not p either has to be inferred on the basis of extralinguistic contextual information or general background assumptions (or a combination thereof), or not p is presupposed or conventionally or conversationally implicated by a previous utterance of the addressee. An adequate analysis of fei has to account not only for the first, but also for the second felicity condition.
Concerning the first felicity condition, there is a subtle complication that I have set aside so far. Consider the sentence uttered by B in (38) in the context of the one uttered by A.
(38)
A:
I
woaß
ned
ob
i
ma
des
neie
Buach
üba
d’Münchna
Räterepublik
kafa
soi.
I
know
not
if
I
myself
the
new
book
about
the.of.Munich
Soviet.republic
buy
should
‘I don’t know if I should buy the new book about the Soviet republic of Munich.’
B:
Des
is
fei
interessant.
that
is
fei
interesting
‘It’s interesting.’
Having uttered the sentence in (38), it is quite clear that A does not believe the new book about the Soviet republic of Munich to be uninteresting. Rather, she is presumably unsure about whether it is interesting or not (i.e. in some worlds that are compatible with ber beliefs it is interesting, while in others it is not) and by uttering the sentence in (38) indirectly asks for B’s opinion. Therefore, it would be rather unreasonable for B to believe that A believes the new book about the Soviet republic of Munich to be uninteresting. But the sentence in (38) sounds perfectly reasonable in the context of (38). What B seems to indicate by the addition of fei is not that she believes that A believes the book to be uninteresting, but rather that she believes that B considers it a likely option that the book is uninteresting.
Thoma (2009) states her analysis of fei in informal terms. In this section, I will propose an implementation in a possible worlds framework along the lines of Hintikka (1969) for concreteness. Hintikka (1969) analyzes the propositional attitude verb believe as a universal quantifier over worlds that are compatible with what the (individual referred to by the) respective subject believes. Consequently, for a sentence 
with fei to be felicitous it would be required that it is true in all worlds w′ compatible with what the speaker believes in the actual world that the negation of 
’s propositional content p, i.e. not p, is true in all worlds w′′ that are compatible with what the addressee believes in w′. While this is unproblematic for many of the examples discussed above, it is clearly too strong for cases such as (38) where it does not have to be the case that the speaker assumes not p to be true in all of the addressee’s belief worlds. Rather, it is sufficient that she believes not p to be true in some of the addressee’s belief worlds. What we thus need is not universal, but existential quantification over the addressee’s belief worlds. This gives the right result for the cases where it is plausible for the speaker to believe that not p is true in all of the addressee’s belief worlds as well, since existential quantification is then guaranteed to be satisfied as well. At the same time, completely unrestricted existential quantification over the addressee’s belief worlds is probably too weak, since it predicts that the use of fei is felicitous whenever the speaker believes that the hearer does not completely exclude the possibility that not p is true. Intuitively, what we want is fei to be felicitous in cases where the speaker believes that the addressee at least considers not p to be a plausible or likely option. I will thus assume existential quantification not over the entire set of the addressee’s belief worlds, but over the subset containing only those worlds that correspond to the addressee’s assumptions about what is stereotypically the case; cf. Kratzer’s (1981) analysis of modal verbs.
Let us now turn to the second felicity condition, which, recall, precludes the use of fei in cases where the negation of the respective proposition is entailed by the at-issue content of a previous utterance of the addressee, but allows it when the same content is presupposed or conversationally or conventionally implicated by a previous utterance of the addressee. To state this condition precisely, we first need to define the notion of at-issue content as opposed to secondary (i.e. non-at-issue) content. Currently, there are two lines of analysis on the market. The first one takes as its starting point the assumption that a sentence in an oral or written discourse always answers an implicit or explicit question, the so-called question under discussion (QUD) (Roberts 1996; 2012; see also Klein & von Stutterheim 1987 and van Kuppevelt 1995 for similar views and Onea & Zimmermann 2019 for an overview over the discussion). In cases where there is no explicit QUD, the implicit QUD is reconstructed on the basis of the focus-background structure: Roughly speaking, the focal part of a sentence replaces the wh-term contained in the question the sentence answers, thus picking one from the set of possible answers. The given or inferable material contained in the respective sentence, the background, in contrast, corresponds to the remaining part of the question. Simons, Tonhauser, Beaver & Roberts (2010) (see also Beaver, Roberts, Simons & Tonhauser 2017) now assume that the at-issue content of a sentence is that part which addresses the current QUD. Evidence for this assumption is provided by contrasts like the one between (39) and (40a), where only the matrix clause, but not the appositive relative clause can answer the respective question.
(39) A: Where did Martin buy his new suit?
B: Martin, who lives in Cologne, bought it at a store in Düsseldorf.
(40) A: Where does Martin live?
B: #Martin, who lives in Cologne, bought his new suit at a store in Düsseldorf.
B′: Martin, who bought his new suit at a store in Düsseldorf, lives in Cologne.
As observed by Gutzmann (2017), however, there are cases like B’s answer in (41B) to A’s question in (41A) where the appositive clearly contributes to answering the respective question.
(41) A: Why is Alex so sad recently?
B: Alex is in love with Chris, who doesn’t love her back.
Interestingly, even in such cases A can react to B’s utterance by uttering a proposition which entails the negation of the appositive’s content, but not the content of the main clause. Consequently, the sentence in (42A) would be felicitous as a reaction of A to (the Bavarian version of) of B’s utterance in (41B), but not the one in (42B).
(42)
A:
Da
Chris
liebt
die
Alex
fei
aa.
The
Chris
loves
the
Alex
fei
also
‘Chris is in love with Alex, too.’
B:
#
Die
Alex
liebt
an
Chris
fei
goa
net.
The
Alex
loves
the
Chris
fei
at.all
not
‘Alex isn’t in love with Chris at all.’
This shows that the behaviour of fei with respect to the distinction between at-issue and secondary content cannot be accounted for in terms of the QUD-approach, i.e. it is not the case that fei cannot be added to a sentence whose propositional content entails the negation of a proposition that addresses the current QUD.
The second line of analysis assumes that at-issue content differs from secondary content in the way in which it enters the Common Ground (AnderBois, Brasoveanu & Henderson 2015; Murray 2017). It is only the at-issue content that is asserted, where for a proposition to be asserted means that the speaker explicitly proposes to add it to the Common Ground. It is only after the addressee has implicitly or explicitly accepted the proposal that the respective proposition actually is added to the Common Ground. Consequently, the addressee can always directly react to the at-issue content of an utterance by denying it or questioning it. Conventional implicatures, in contrast, are not proposed, but rather “imposed”, i.e. they enter the Common Ground directly, without an intermediate step. Consequently, denying or questioning the content of a conventional implicature requires some extra effort. Similarly, since presupposed content is treated by the speaker as if it already was part of the Common Ground, the addressee likewise cannot deny or question it directly, but only via special means that interrupt the flow of the conversation such as saying “Hey, wait a minute!” first (Shanon 1976; von Fintel 2004). Finally, concerning conversational implicatures, the fact that they can be cancelled by the speaker without contradicting herself shows that they are likewise not asserted.
As already mentioned in the introduction, the assumption that only the at-issue content of an utterance is asserted offers an attractive explanation for the observation that fei relates to non-at-issue content exclusively: By adding fei, the speaker draws the hearer’s attention to a conflict between her own beliefs and the addressee’s beliefs that is not salient at the point where the respective sentence is uttered. But now consider what happens in a case where the proposition q believed by the addressee that contradicts the (propositional content of the) prejacent of fei has been asserted by the addressee: The addressee has put q on the table explicitly, and the question of whether q is true is consequently highly salient. Uttering a sentence that contradicts q therefore makes the conflict between the speaker’s and the hearer’s beliefs automatically salient as well. The addition of fei to such a sentence would thus be superfluous. Putting everything together, the use conditions of fei can thus be stated informally as given in (43), and more formally as given in (44).
(43) fei can be added felicitously to a sentence
denoting the proposition p in a context C iff
(i) all of the speaker’s belief worlds contain at least one world that is compatible with as many of the addressee’s assumptions about what is stereotypically the case as possible where the negation of p is true (i.e. the speaker believes that the addressee considers the negation of p at least a likely option),
(ii) there is no recent assertion by the addressee of a proposition q such that in all worlds where q is true, the negation of p is true (i.e. q entails the negation of p), where recent just means that the addressee has not asserted q in the current conversation.
(44) fei can be added felicitously to a sentence
denoting the proposition p in context C iff
(i)
′
″
″
, where
SP is the speaker in C,
ADR is the addressee in C,
is the world of C,
is the set of worlds compatible with what SP believes in
, and
is the function mapping a set of worlds to the subset that makes as many of ADR’s assumptions about what is stereotypically the case true as possible.
(ii)
, where
is the run time of e.
5 A Comparison of fei and doch
As already mentioned in Section 2, Thoma (2009) observes that fei differs from doch insofar as doch is only felicitous if the speaker assumes that the hearer previously believed the propositional content of the sentence containing doch, but has either forgotten it at the time where that sentence is uttered or believes it to be false. Replacing fei in (11a), repeated here as (45a), by doch thus leads to infelicity, as shown in (11b), repeated here as (45b): While it is plausible for the speaker to assume that the hearer is unaware of his fly being down (and hence believes it to be not down) when he returns from the toilet, it is implausible to assume that he was previously aware of his fly being down and has temporarily forgotten it.
(45)
a.
Dei
Hos’ntiarl
is
fei
auf.
Your
pant.door
is
fei
open
‘Your fly is down.’
b.
#
Dei
Hos’ntiarl
is
doch
auf.
Your
pant.door
is
doch
open
(Thoma 2009: 150)
Interestingly, doch differs from fei in another respect as well: In contrast to fei, doch is not sensitive to the distinction between at-issue and secondary content, i.e. doch can felicitously be used in cases where the proposition contradicting the propositional content of the sentence containing it was previously asserted by the addressee. To see this, compare the doch- and fei-variants of the sentences in (15), (17) and (24), for example, given here as (46b), (47) and (48), respectively. The variant of (46b) with doch, for example, is perfectly fine: First, given Maria’s previous utterance it is extremely plausible that Tom assumes her to be unaware of Moses und Aron being a twelve-tone opera at the utterance time. Second, it is at least not implausible for him to assume that Maria knew that before, but has temporarily forgotten it. A similar reasoning applies to the variants with doch in (47) and (48).
(46) a.Maria: Moses und Aron is a Belcanto Oper.
‘Moses and Aron is a belcanto opera.’
b.Tom: (Quatsch!) Des is (#fei)/doch a Zwölftonoper.
‘Bullshit! It’s a twelve-tone opera.’
(47) A: Da Bäda is a Depp.
‘Peter is an idiot.’
B: (Quatsch!) Da Bäda is (#fei)/doch koa Depp.
‘(Bullshit!) Peter is no idiot.’
(48)
A:
In
Frankreich
gibt’s
imma
no
an
Kini.
In
France
there.is
always
still
a
king
‘In France there is still a king.’
B:
(So
a
Schmarr’n).
In
Frankreich
gibt’s
(#
fei)/doch
koan
Kini
nemma.
Such
a
nonsense
In
France
there.is
fei/doch
no
king
anymore
‘(What nonsense!) In France there is no king anymore.’
6 Conclusion
In this paper I have shown that the German discourse particle fei is special among discourse particles insofar as it relates to non-at-issue content exclusively. It indicates the speaker’s belief that the addressee believes the negation of the (propositional content of the) prejacent of fei, or at least considers the negation of the (propositional content of the) prejacent of fei to be a likely option (recall the discussion of (38)). Crucially, the speaker’s belief may not have been caused by a previous utterance of the addressee whose at-issue content entails the propositional content of the prejacent of fei. Rather, it has to be based on contextual information, general background assumptions or on the presuppositions or conventional or conversational implicatures of the addressee’s utterances. I have argued that fei’s resistance against at-issue content is a consequence of its discourse function: fei draws the addressee’s attention to a conflict between her beliefs and the speaker’s beliefs that is not salient at the point where the sentence with fei is uttered. In virtue of having been explicitly put on the table for the interlocutor(s) to accept or reject, however, the content of an assertion is by definition highly salient.
Given that fei relates to non-at-issue content exclusively, I would like to end this paper by suggesting a new test that could be an additional diagnostic for distinguishing at-issue from non-at-issue content: Whenever you want to know whether some part of a sentence denotes at-issue or non-at-issue content, translate the sentence containing it into Bavarian and check with a native speaker whether fei can felicitously be added to a sentence negating the respective proposition as a reaction to an utterance of the original sentence.
I would like to thank Josef Bayer and an anonymous reviewer for their detailed and insightful comments on an earlier version of this paper. Parts of this material were presented at the University of Cologne, at the workshop Secondary information and linguistic encoding, which was part of the 39th Annual Conference of the German Linguistic Society, at Sinn und Bedeutung 22, and at the workshop Transmodal perspectives on secondary meaning, which was part of the Twelfth International Tbilisi Symposium on Language, Logic and Computation. I would like to thank the audiences at those events for comments and discussion, in particular Fabian Bross, Daniel Gutzmann, Andreas Haida, Daniel Hole, Katja Jasinskaja, Radek Šimík, Henk Zeevat and Sarah Zobel. The research reported in this paper was supported by the DFG grant for the project Discourse Referents as Perspectival Centers within the Collaborative Research Center Prominence in Language.
This criterion is problematic, however, in light of the fact that some discourse particles such as doch and schon have stressed variants.
As pointed out to me by Josef Bayer, fei is at least marginally acceptable in more elaborate infinitival commands such as und dann beim Gebetleitn (owa) fei wieder dahoam sei! (and then be home again when the prayer bell rings).
References
AnderBois, Scott, Adrian Brasoveanu & Robert Henderson (2015). At-issue proposals and appositive impositions in discourse. Journal of Semantics 32 (1), 93–138.
Bayer, Josef & Volker Struckmeier, eds. (2017). Discourse particles. Formal approaches to their syntax and semantics. Berlin.
Beaver, David I. (1997). Presupposition. In Johan van Bethem & Alice ter Meulen eds., The handbook of logic and language, 939–1008. Oxford: Elsevier.
Beaver, David I., Craige Roberts, Mandy Simons & Judith Tonhauser (2017). Questions under discussion: Where information structure meets projective content. Annual Review of Linguistics 3 (1), 265–284.
Coniglio, Marco (2011). Die Syntax der deutschen Modalpartikeln. Ihre Distribution und Lizenzierung in Haupt- und Nebensätzen. (Studia grammatica 73). Berlin: Akademie Verlag.
Dörre, Laura & Andreas Trotzke (this volume). The processing of secondary meaning: An experimental comparison of focus and modal particles in wh-questions.
Egg, Markus (2013). Discourse particles, common ground, and felicity conditions. In Daniel Gutzmann & Hans-Martin Gärtner eds., Beyond Expressives. Explorations in use-conditional meaning, 125–149. Leiden: Brill.
Gärtner, Hans-Martin (2013). Infinite Hauptsatzstrukturen. In Jörg Meibauer, Markus Steinbach & Hans Altmann eds., Satztypen des Deutschen, 202–231. Berlin/New York: de Gruyter Mouton.
Gärtner, Hans-Martin (2014a). On covert modality in German root infinitives. Proceeding of WCCFL 31, 199–206.
Gärtner, Hans-Martin (2014b). Überlegungen zur versteckten Modalität infiniter Hauptsatzstrukturen. Linguistische Berichte 237, 81–92. http://www.ingentaconnect.com/content/buske/lingber/2014/00002014/00000237/art00005.
Glaser, Elvira (1999). Funktion und Verbreitung der Partikel fäi. In H. Tatzreiter, M. Hornung & P. Ernst eds., Erträge der Dialektologie und Lexikographie. Festgabe für Werner Bauer zum 6o. Geburtstag, 165–190. Wien.
Grice, H. Paul (1989). Studies in the way of words. Cambridge, MA: Harvard University Press.
Grosz, Patrick (2014). German doch: an element that triggers a contrast presupposition. Proceedings of CLS 46, 163–177.
Gutzmann, Daniel (2015). Use-conditional meaning. Studies in multidimensional semantics. Oxford: Oxford University Press.
Gutzmann, Daniel (2017). How to do multiple things with words? Paper presented at the workshop Speech Acts, Leibniz-Zentrum für Allgemeine Sprachwissenschaft, May 30, 2017.
Hintikka, Jaakko (1969). Semantics for propositional attitudes. In J.W. Davis & D.J. Hockney eds., Philosophical logic, 21–45. Dordrecht: Reidel.
Höhle, Tilman N. (1992). Über Verum-Fokus im Deutschen. In Joachim Jacobs ed., Informationsstruktur und Grammatik. Vol. 4. (Linguistische Berichte Sonderheft), 112–141. Opladen: Westdeutscher Verlag.
Jacobs, Joachim (1991). On the semantics of modal particles. In Werner Abraham ed., Discourse particles, 141–162. Amsterdam: Benjamins.
Jacobs, Joachim (this volume). Why the meaning of discourse particles is separated from focus-background structure.
Karagjosova, Elena (2004a). German “doch” as a marker of given information. Sprache und Datenverarbeitung 28, 71–78.
Karagjosova, Elena (2004b). The meaning and function of German modal particles. PhD thesis. Saarbrücken: Universität des Saarlands.
Karagjosova, Elena (2009). A unified DRT-based account of accented and unaccented middle field doch. Sprache und Datenverarbeitung 33, 77–93.
Kaufmann, Magdalena (2012). Interpreting imperatives. Springer.
Klein, Wolfgang & Christiane von Stutterheim (1987). Quaestio und die referentielle Bewegung in Erzählungen. Linguistische Berichte 109, 163–185.
Kratzer, Angelika (1981). The notional category of modality. In Hans J. Eikemeyer & Hannes Rieser eds., Words, worlds, and contexts, 38–74. Berlin: de Gruyter.
Kratzer, Angelika (1999). “Beyond ouch and oops. How descriptive and expressive meaning interact”. In Cornell Conference on Theories of Context Dependency. March 26, 1999. Cornell University. Ithaca, NY. http://semanticsarchive.net/Archive/WEwNGUyO/.
Lindner, Karin (1991). ‘Wir sind ja doch alte Bekannte’—the use of German ja and doch as modal particles. In Werner Abraham ed., Discourse particles, 303–328. Amsterdam: Benjamins.
Murray, Sarah E. (2017). The semantics of evidentials. Oxford: Oxford University Press.
Onea, Edgar & Malte Zimmermann (2019). Questions in discourse: an overview. In Klaus von Heusinger, Malte Zimmermann & Edgar Onea eds., Questions in discourse. Volume 1: Semantics, 5–117. Leiden: Brill.
Ormelius-Sandblom, Elisabet (1997). Die Modalpartikeln ja, doch, schon. Zu ihrer Syntax, Semantik und Pragmatik. (Lunder Germanistische Forschungen 61). Stockholm: Almqvist & Wiksell.
Portner, Paul (2007). Imperatives and modals. Natural Language Semantics 15, 351–383.
Potts, Christopher (2005). The logic of conventional implicature. (Oxford Studies in Theoretical Linguistics 7). Oxford: Oxford University Press.
Roberts, C. (1996). Information structure in discourse: Towards an integrated formal theory of pragmatics. Working Papers in Linguistics—Ohio State University Department of Linguistics, 91–136.
Roberts, Craige (2012). Information structure in discourse: Towards an integrated formal theory of pragmatics. Semantics & Pragmatics 5. http://dx.doi.org/10.3765/sp.5.6.
Rojas-Esponda, Tania (2014). A QUD account of German doch. Proceedings of Sinn und Bedeutung 18, 359–376.
Schlieben-Lange, Brigitte (1979). Bairisch eh-halt-fei. In Harald Weydt ed., Die Partikeln der deutschen Sprache, 307–317. Berlin: de Gruyter.
Schwenter, Scott (2005). The pragmatics of negation in Brazilian Portuguese. Lingua 115, 1427–1456.
Shanon, Benny (1976). On the two kinds of presuppositions in natural language. Foundations of Language (14), 247–249.
Simons, Mandy, Judith Tonhauser, David I. Beaver & Craige Roberts (2010). What projects and why. Proceedings of SALT 20, 309–327. http://elanguage.net/journals/index.php/salt/article/viewFile/20.309/1326.
Stalnaker, Robert (1978). Assertion. In Peter Cole ed., Syntax and Semantics 9. Pragmatics, 315–332. New York: Academic Press.
Strawson, Peter F. (1950). On referring. Mind 59, 320–344.
Thoma, Sonja (2009). To p or to ¬p—The Bavarian particle fei as polarity discourse particle. Sprache und Datenverarbeitung 33, 139–152.
Thurmair, Maria (1989). Modalpartikeln und ihre Kombinationen. Tübingen: Niemeyer.
Van Kuppevelt, Jan (1995). Discourse structure, topicality and questioning. Journal of Linguistics 31, 109–147.
Von Fintel, Kai (2004). Would you believe it? The king of France is back! Presuppositions and truth-value intuitions. In Marga Reimer & Anne Bezuidenhout eds., Descriptions and beyond, 269–296. Oxford: Oxford University Press.
Weydt, Harald (1969). Abtönungspartikel. Die deutschen Modalwörter und ihre französischen Entsprechungen. Bad Homburg: Gehlen.
Zeevat, Henk (2003). Particles: presupposition triggers, context markers or speech act markers. In Reinhard Blutner & Henk Zeevat eds., Optimality theory and pragmatics, 91–112. Basingstoke: Palgrave MacMillan.
Zimmermann, Malte (2008). Discourse particles in the left periphery. In Benjamin Shaer, Philippa Cook, Werner Frey & Claudia Maienborn eds., Dislocated elements in discourse: Syntactic, semantic, and pragmatic perspectives, 200–231. London: Routledge.
Zimmermann, Malte (2012). Discourse particles. In Paul Portner, Claudia Maienborn & Klaus von Heusinger eds., Semantics. An international handbook of natural language meaning, 2012–2038. Berlin/New York: de Gruyter.