7. Common Ground

7.1 What is Common Ground?

One (of the many) things we take for granted when we use language is our effortless use of appropriate expressions that our hearer will understand. How is it that we know to select appropriately from among the available names and descriptions for referring to particular entities (e.g., "Lauder", "a town in the Borders", "a village in Scotland")? The selection we make depends on context, namely the context of who's listening and what they know or what we estimate that they know.

Of course there are times that we underestimate or overestimate what our hearers know, but those moments in conversation are notable because they do not predominate. We notice them precisely because most of the time our descriptions correspond well to what our hearer knows or else they are able and willing to accommodate. The two examples below are from the 'Call Home' corpus (a collection of transcribed phone conversations in which people were paid to call someone they know and have the conversation recorded; available through the LDC https://catalog.ldc.upenn.edu/LDC97S42) and they highlight how speakers signal to each other what they know and what they expect to be shared knowledge (and how they can misestimate).

(1) A: And one of her students showed her how to get into the X-500 directories.
B: Which are?
A: Hm?
B: What are the X-500 directories?
A: Oh um where you put- your um- How c- How can you not know?

(2) B: Last weekend Lida and Irv Teisher were in town.
A: Oh.
B: They uh- they have a son who is the uh- I guess he buys books for something called Borders, which is a bookstore that-
A: Yes, we have it here too.

What's notable about examples (1-2) is that they involve infelicitous referring expressions, in which the speaker attributes either too much or too little knowledge to the addressee. This is something that is surprisingly rare in our daily conversation. It is rare because usually speakers know what the other person knows (or can use various heuristics to approximate that knowledge) and such errors reveal the complexity of successful reference generation.

The adjustment of our language for different listeners is called Audience Design. It is based on an assumption that individuals interact based on beliefs about their common ground, which informs decisions about when and how to refer to objects and entities and how to understand such references. As evidence of this kind of adjustment, there are examples like "motherese" (child-directed speech) and "elderspeak" (speech directed to the elderly) in which the speaker adjusts properties of their speech like volume and lexicon and speech rate to accommodate perceived (though not always accurate) deficits on the part of the listener. "Motherese" and "elderspeak" are examples of adjustments based on global characteristics of the category of the listener ("when I'm with a 2-year-old, I don't talk about postmodernism"). But we also make fine-tuned adjustments based on knowledge of particular individuals. To do this well requires keeping track of a huge amount of information ("was this listener at the party on Friday? does he know the host? was he there when the host danced on the table? what's in our common ground?!").

COMMON GROUND: the set of knowledge and beliefs taken as shared between interlocutors (things that are known and known to be known)

Based on the definition above, there is a set of propositions (facts about the world, beliefs, opinions, etc.) which are shared and which a conversation depends on. Consider examples (3-5).

(3) A: Do you like the Pragmatics class?
B: Is the Pope Catholic?

(4) A: It's hard to see the lecture slides in this room.
B: Can you reach the blinds on those windows?

(5) A: I love the Pragmatics class.
B: I hate the homeworks.
A: Well, true, the homeworks are a pain and I wish Hannah didn't use so many lame examples in class and the readings are sometimes confusing.
B: I thought you said you loved the class unlike me.

• Examples (3-5) highlight the inclusion of what types of knowledge in common ground? [click to reveal]
  The Pope being Catholic is real-world knowledge and that shared knowledge must be in the common ground for that rhetorical question to work as a response to A's question in (3). Example (4) shows that knowledge about the physical environment is part of common ground. Lastly, example (5) shows that the record of what has been said during the conversation is part of common ground. Thus, common ground includes what is known via real-world knowledge, what is salient in the context, and who has said what. In addition, when speakers contribute to the conversation by making statements about their own beliefs about the world (e.g., the fact that it's hard to see the lecture slides or that SpeakerA in (5) wishes Hannah didn't use as many lame examples), the content of those statements also enters the common ground (if the content is not contested).

The common ground depends on community membership and thus includes information that is part of the interlocutors' shared socio-cultural background and real-world knowledge. It also depends on physical co-presence, which is information that is in the shared physical or perceptual environment of the interlocutors. Lastly, there's linguistic co-presence, which is information that can be derived from past and present conversations between interlocutors. In this view, the common ground can be conceived of as a set of propositions. For the last utterances in (3-5) to be felicitous, the common ground must look (minimally) as follows in (6):

(6)
CG = [Pope is Catholic;
This room has windows;
A said "I love the Pragmatics class";
B said "I hate the homeworks"]

Alternatively, the common ground can be represented as sets of different speakers' commitments as in (7):

(7)
Speaker A's commitments = [Pope is Catholic; This room has windows; I love the Pragmatics class]

Speaker B's commitments = [Pope is Catholic; This room has windows; I hate the Pragmatics homeworks]

So you may be saying, okay, so that's common ground, great. But what would we possibly do with that and why would it matter if we use a particular representation like (6) or (7)? What follows is an example of how the representation of common ground in (7) provides a framework for talking about something specific speakers do when they make statements and ask questions. It turns out that being able to separate the commitment sets for the Speaker and Addressee allows us to identify distinctions in meaning along dimensions of utterance form and intonation.

What's at stake is how an utterance updates the status of a conversation, specifically in American English. This work was done by Christine Gunlogson for her 2001 PhD research. Formally we can say that when you add an utterance (utt) to a context (C), you end up with a new context (C'):

C + utt = C'

Depending on the syntactic form (interrogative vs declarative) and the intonation (falling vs rising), there is different information attributed to the commitment set (cs) of the Speaker and Hearer. Click on each of the utterances below to see the commitment updates for the Speaker and Hearer, and see if you can figure out why they make sense

	falling intonation	rising intonation
declarative form	"It's raining" csspeaker(C') = csspeaker(C) + SDecl csaddressee(C') = csaddressee(C)	"It's raining?" csspeaker(C') = csspeaker(C) csaddressee(C') = csaddressee(C) + SDecl
interrogative form	?? None. We don't have a falling interrogative form in American English.	"Is it raining?" csX(C') = csX(C)

The table above captures the distinction between a speaker who updates her own commitment set (by uttering a falling declarative) and one who attributes a commitment to the addressee (by uttering a rising declarative). With the falling declarative, the speaker adds to her own commitment set (cs_speaker) the content of the declarative sentence (S_{Decl). With a rising declarative, the commitments are reversed: the speaker is in effect adding a commitment to the addressee's commitment set (csAddressee), which the Addressee woud have to go out of his way to contest.}

• What context might invite a rising declarative?
  One can imagine a speaker seeing someone come in with a dripping wet umbrella and asking "It's raining?", in effect asking the addressee for confirmation of the non-verbal message that it's wet outside.

• What context might invite a rising declarative?
  One can imagine a speaker seeing someone come in with a dripping wet umbrella and asking "It's raining?", in effect asking the addressee for confirmation of the non-verbal message that it's wet outside.

KEY POINT: A characterisation of common ground that maintains separate commitment sets for speaker and addressee allows for an analysis of the way that form and intonation influence interpretation (Gunlogson 2001).

---

To go on to section 7.2 "Shared experience", click here.