Learning in extended and approximate Rational Speech Acts models - PowerPoint PPT Presentation

128 129 set, in consequence, will represent a HIGHER value in the ordering. Subs.ets which are neither AFFIRt-.t(B,rhird chaprer, BEL(B, read(B,third chapter»» 1\ ALT_SENT(read(B,ch3prec_one), read(B,thirdj:hapter), included in, nor include, one another, will be ALTERNATE values in this poser. Consider how 'parts of a dissertation'» the salient ordering in the following exchange mighr be represented: =) SCALAR_IMP(B, A, I read the third, ...,BEL(B, read(B,chapter_one», C i) (214) A: Do you speak: Portuguese? B: My husband does. There are no restrictions on those posers which support scalar implicamre. However, (ar least) one restriction does exist on which posers may be viewed as salient in a given exchange: The inclusion ordering which supports the implicature in 214 might be represented as follows: Above (Section 5.1.6.3) I noted for most metrics that rank utterances. both a given metric 128 129 and irs dual (converse) may be candidares for salience in an exchange. However, no metric (ji which orders values vi and Vj such that a) vi is higher than Vj and b} the truth of Vj entails the truth of vi can supporr scalar implicature -- for the simple reason in such a case, a sentence set, in consequence, will represent a HIGHER value in the ordering. Subs.ets which are neither AFFIRt-.t(B,rhird chaprer, BEL(B, read(B,third chapter»» 1\ Pi ranked higher than a sentence Pj by (ji since then the implicature licensed would be ALT_SENT(read(B,ch3prec_one), read(B,thirdj:hapter), included in, nor include, one another, will be ALTERNATE values in this poser. Consider how 'parts of a dissertation'» inconsistent with the utterance licensing it. In terms of the fonnaIism presented in Chapter 2, the salient ordering in the following exchange mighr be represented: =) SCALAR_IMP(B, A, I read the third, ...,BEL(B, read(B,chapter_one», such a meaning would not be reinforceable. Consider, for example. (212a): C i) (214) A: Do you speak: Portuguese? So, {husband,wife.chiid} wi!! be the highest value in this ordering, with the alternate doubletons (212) A: Are you planning to buy a dog? B: My husband does. There are no restrictions on those posers which support scalar implicamre. However, (ar (husband,wife), (wife.child), and (husband,child) lower values and the alternate values, a. B: A German Shepherd. least) one restriction does exist on which posers may be viewed as salient in a given exchange: The inclusion ordering which supports the implicature in 214 might be represented as follows: b. B: I'm buying a Gennan Shepherd and I'm not buying a dog. {husband}, (wife), and {child} lower values still in this poset. By the scalar implicature Above (Section 5.1.6.3) I noted for most metrics that rank utterances. both a given metric conventions, then, S may affinn. say, (husband.wife) to convey ...,BEL(S. (husband,wile.child)) While one might identify either an ordering defined by 'isa' (i.e .• a Gennan Shepherd isa dog) and irs dual (converse) may be candidares for salience in an exchange. However, no metric (ji as well as -,BEL(S. (husband,child}) and -,BEL(S, (wife.childJ). Note, particularly, that there or by 'subsumes' (i.e., a dog subsumes the subtype Shepherd) as salient in this which orders values vi and Vj such that a) vi is higher than Vj and b} the truth of Vj entails the may be some redundance in scalar implicatures predicted from this representation. Also, any exchange, only the latter permits scalar implicature here. B cannot implicate that she is not truth of vi can supporr scalar implicature -- for the simple reason in such a case, a sentence subsets so represented may be lexica1ized in various ways -- as, the expression (husband.wife) buying a dog vla this response, since buying a Gennan Shepherd entails buying a dog. The Pi ranked higher than a sentence Pj by (ji since then the implicature licensed would be might be lexicalized as • couple' or as 'husband and wife'. The theory presented in this thesis attempted reinforcement of (212b) fails. However, we cannot rule out 'isa' relations as inconsistent with the utterance licensing it. In terms of the fonnaIism presented in Chapter 2, will not distinguish between these. 128 potential supporters of scalar imphcarore: In 213, for example. 8's response might evoke either such a meaning would not be reinforceable. Consider, for example. (212a): an'isa' So, {husband,wife.chiid} wi!! be the highest value in this ordering, with the alternate doubletons (212) A: Are you planning to buy a dog? As noted in Sections 5.1.7, temporal orderings may also be represented as setS ofrernporal (husband,wife), (wife.child), and (husband,child) lower values and the alternate values, a. B: A German Shepherd. (213) A: Would you like a dog? for the analysis of licensed scalar implicacures. So, these orderings too wilt be defined by set b. B: I'm buying a Gennan Shepherd and I'm not buying a dog. 8: I'd like a German Shepherd. {husband}, (wife), and {child} lower values still in this poset. By the scalar implicature inclusion, as: conventions, then, S may affinn. say, (husband.wife) to convey ...,BEL(S. (husband,wile.child)) While one might identify either an ordering defined by 'isa' (i.e .• a Gennan Shepherd isa dog) hierarchy - or irs dual. Apparently, any poser can support scaiar implicature, although other as well as -,BEL(S. (husband,child}) and -,BEL(S, (wife.childJ). Note, particularly, that there or by 'subsumes' (i.e., a dog subsumes the subtype Shepherd) as salient in this tests for conversational implicature may rule out some particuJ3r posers in panicular exchanges. may be some redundance in scalar implicatures predicted from this representation. Also, any exchange, only the latter permits scalar implicature here. B cannot implicate that she is not subsets so represented may be lexica1ized in various ways -- as, the expression (husband.wife) buying a dog vla this response, since buying a Gennan Shepherd entails buying a dog. The 5.3.2.3. Representing Scalar Implicature Orderings as Pos.ets might be lexicalized as • couple' or as 'husband and wife'. The theory presented in this thesis {past, resent} {presenr,future} {past,furure} attempted reinforcement of (212b) fails. However, we cannot rule out 'isa' relations as r---:::-- will not distinguish between these. 128 J have demonstrated above how part! whole re!arions can be represented. To demonstrate potential supporters of scalar imphcarore: In 213, for example. 8's response might evoke either that L'le other orderings discussed in Section 5.1 are accounted for by a poset condition. I w;i! an'isa' {future} As noted in Sections 5.1.7, temporal orderings may also be represented as setS ofrernporal describe how representative orderings can be accommodated by this condition so mat (213) A: Would you like a dog? for the analysis of licensed scalar implicacures. So, these orderings too wilt be defined by set scalar implicatures are correctly predicted by ImPl_3' 8: I'd like a German Shepherd. inclusion, as: Posers defined by a type! subrype metric, such as that which supports 174, may be hierarchy - or irs dual. Apparently, any poser can support scaiar implicature, although other Rdations defined by ordering the non-null members of the power of some set x by illustrated by me (parrial) classification hierarchy: tests for conversational implicature may rule out some particuJ3r posers in panicular exchanges. set-inclusion allow a poset representation of x and its non-null proper subsets a5 follows: Any non-null proper subset of a set m<lY be nnked as LOWER than the set which it. and 5.3.2.3. Representing Scalar Implicature Orderings as Pos.ets lZ1!Sut see {CorelIa 84, Ka!ita 84) for some approaches to thtS problem. {past, resent} {presenr,future} {past,furure} r---:::-- J have demonstrated above how part! whole re!arions can be represented. To demonstrate that L'le other orderings discussed in Section 5.1 are accounted for by a poset condition. I w;i! {future} describe how representative orderings can be accommodated by this condition so mat scalar implicatures are correctly predicted by ImPl_3' Posers defined by a type! subrype metric, such as that which supports 174, may be Rdations defined by ordering the non-null members of the power of some set x by illustrated by me (parrial) classification hierarchy: set-inclusion allow a poset representation of x and its non-null proper subsets a5 follows: Any non-null proper subset of a set m<lY be nnked as LOWER than the set which it. and lZ1!Sut see {CorelIa 84, Ka!ita 84) for some approaches to thtS problem. A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Partial-order implicature Hirschberg 1985, A Theory of Scalar Implicature 7 / 56

128 129 set, in consequence, will represent a HIGHER value in the ordering. Subs.ets which are neither AFFIRt-.t(B,rhird chaprer, BEL(B, read(B,third chapter»» 1\ ALT_SENT(read(B,ch3prec_one), read(B,thirdj:hapter), included in, nor include, one another, will be ALTERNATE values in this poser. Consider how 'parts of a dissertation'» the salient ordering in the following exchange mighr be represented: =) SCALAR_IMP(B, A, I read the third, ...,BEL(B, read(B,chapter_one», C i) (214) A: Do you speak: Portuguese? B: My husband does. There are no restrictions on those posers which support scalar implicamre. However, (ar least) one restriction does exist on which posers may be viewed as salient in a given exchange: The inclusion ordering which supports the implicature in 214 might be represented as follows: Above (Section 5.1.6.3) I noted for most metrics that rank utterances. both a given metric and irs dual (converse) may be candidares for salience in an exchange. However, no metric (ji which orders values vi and Vj such that a) vi is higher than Vj and b} the truth of Vj entails the truth of vi can supporr scalar implicature -- for the simple reason in such a case, a sentence Pi ranked higher than a sentence Pj by (ji since then the implicature licensed would be inconsistent with the utterance licensing it. In terms of the fonnaIism presented in Chapter 2, such a meaning would not be reinforceable. Consider, for example. (212a): So, {husband,wife.chiid} wi!! be the highest value in this ordering, with the alternate doubletons (212) A: Are you planning to buy a dog? (husband,wife), (wife.child), and (husband,child) lower values and the alternate values, a. B: A German Shepherd. b. B: I'm buying a Gennan Shepherd and I'm not buying a dog. {husband}, (wife), and {child} lower values still in this poset. By the scalar implicature conventions, then, S may affinn. say, (husband.wife) to convey ...,BEL(S. (husband,wile.child)) While one might identify either an ordering defined by 'isa' (i.e .• a Gennan Shepherd isa dog) as well as -,BEL(S. (husband,child}) and -,BEL(S, (wife.childJ). Note, particularly, that there or by 'subsumes' (i.e., a dog subsumes the subtype Shepherd) as salient in this may be some redundance in scalar implicatures predicted from this representation. Also, any exchange, only the latter permits scalar implicature here. B cannot implicate that she is not subsets so represented may be lexica1ized in various ways -- as, the expression (husband.wife) buying a dog vla this response, since buying a Gennan Shepherd entails buying a dog. The might be lexicalized as • couple' or as 'husband and wife'. The theory presented in this thesis attempted reinforcement of (212b) fails. However, we cannot rule out 'isa' relations as will not distinguish between these. 128 potential supporters of scalar imphcarore: In 213, for example. 8's response might evoke either an'isa' As noted in Sections 5.1.7, temporal orderings may also be represented as setS ofrernporal (213) A: Would you like a dog? for the analysis of licensed scalar implicacures. So, these orderings too wilt be defined by set 8: I'd like a German Shepherd. inclusion, as: hierarchy - or irs dual. Apparently, any poser can support scaiar implicature, although other tests for conversational implicature may rule out some particuJ3r posers in panicular exchanges. 5.3.2.3. Representing Scalar Implicature Orderings as Pos.ets {past, resent} {presenr,future} {past,furure} r---:::-- J have demonstrated above how part! whole re!arions can be represented. To demonstrate that L'le other orderings discussed in Section 5.1 are accounted for by a poset condition. I w;i! {future} describe how representative orderings can be accommodated by this condition so mat scalar implicatures are correctly predicted by ImPl_3' Posers defined by a type! subrype metric, such as that which supports 174, may be Rdations defined by ordering the non-null members of the power of some set x by illustrated by me (parrial) classification hierarchy: set-inclusion allow a poset representation of x and its non-null proper subsets a5 follows: Any non-null proper subset of a set m<lY be nnked as LOWER than the set which it. and lZ1!Sut see {CorelIa 84, Ka!ita 84) for some approaches to thtS problem. 128 129 A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects set, in consequence, will represent a HIGHER value in the ordering. Subs.ets which are neither AFFIRt-.t(B,rhird chaprer, BEL(B, read(B,third chapter»» 1\ ALT_SENT(read(B,ch3prec_one), read(B,thirdj:hapter), included in, nor include, one another, will be ALTERNATE values in this poser. Consider how 'parts of a dissertation'» the salient ordering in the following exchange mighr be represented: Partial-order implicature =) SCALAR_IMP(B, A, I read the third, ...,BEL(B, read(B,chapter_one», C i) (214) A: Do you speak: Portuguese? B: My husband does. There are no restrictions on those posers which support scalar implicamre. However, (ar least) one restriction does exist on which posers may be viewed as salient in a given exchange: The inclusion ordering which supports the implicature in 214 might be represented as follows: Above (Section 5.1.6.3) I noted for most metrics that rank utterances. both a given metric A: Do you speak and irs dual (converse) may be candidares for salience in an exchange. However, no metric (ji German? which orders values vi and Vj such that a) vi is higher than Vj and b} the truth of Vj entails the truth of vi can supporr scalar implicature -- for the simple reason in such a case, a sentence Pi ranked higher than a sentence Pj by (ji since then the implicature licensed would be B: My husband does. inconsistent with the utterance licensing it. In terms of the fonnaIism presented in Chapter 2, such a meaning would not be reinforceable. Consider, for example. (212a): So, {husband,wife.chiid} wi!! be the highest value in this ordering, with the alternate doubletons (212) A: Are you planning to buy a dog? (husband,wife), (wife.child), and (husband,child) lower values and the alternate values, a. B: A German Shepherd. b. B: I'm buying a Gennan Shepherd and I'm not buying a dog. {husband}, (wife), and {child} lower values still in this poset. By the scalar implicature conventions, then, S may affinn. say, (husband.wife) to convey ...,BEL(S. (husband,wile.child)) While one might identify either an ordering defined by 'isa' (i.e .• a Gennan Shepherd isa dog) as well as -,BEL(S. (husband,child}) and -,BEL(S, (wife.childJ). Note, particularly, that there or by 'subsumes' (i.e., a dog subsumes the subtype Shepherd) as salient in this may be some redundance in scalar implicatures predicted from this representation. Also, any exchange, only the latter permits scalar implicature here. B cannot implicate that she is not subsets so represented may be lexica1ized in various ways -- as, the expression (husband.wife) buying a dog vla this response, since buying a Gennan Shepherd entails buying a dog. The might be lexicalized as • couple' or as 'husband and wife'. The theory presented in this thesis attempted reinforcement of (212b) fails. However, we cannot rule out 'isa' relations as will not distinguish between these. 128 potential supporters of scalar imphcarore: In 213, for example. 8's response might evoke either an'isa' As noted in Sections 5.1.7, temporal orderings may also be represented as setS ofrernporal (213) A: Would you like a dog? for the analysis of licensed scalar implicacures. So, these orderings too wilt be defined by set 8: I'd like a German Shepherd. inclusion, as: Apparently, any poser can support scaiar implicature, although other hierarchy - or irs dual. tests for conversational implicature may rule out some particuJ3r posers in panicular exchanges. 5.3.2.3. Representing Scalar Implicature Orderings as Pos.ets {past, resent} {presenr,future} {past,furure} r---:::-- J have demonstrated above how part! whole re!arions can be represented. To demonstrate Hirschberg 1985, A Theory of Scalar Implicature that L'le other orderings discussed in Section 5.1 are accounted for by a poset condition. I w;i! {future} describe how representative orderings can be accommodated by this condition so mat scalar implicatures are correctly predicted by ImPl_3' Posers defined by a type! subrype metric, such as that which supports 174, may be Rdations defined by ordering the non-null members of the power of some set x by illustrated by me (parrial) classification hierarchy: 7 / 56 set-inclusion allow a poset representation of x and its non-null proper subsets a5 follows: Any non-null proper subset of a set m<lY be nnked as LOWER than the set which it. and lZ1!Sut see {CorelIa 84, Ka!ita 84) for some approaches to thtS problem.

128 129 set, in consequence, will represent a HIGHER value in the ordering. Subs.ets which are neither AFFIRt-.t(B,rhird chaprer, BEL(B, read(B,third chapter»» 1\ ALT_SENT(read(B,ch3prec_one), read(B,thirdj:hapter), included in, nor include, one another, will be ALTERNATE values in this poser. Consider how 'parts of a dissertation'» the salient ordering in the following exchange mighr be represented: =) SCALAR_IMP(B, A, I read the third, ...,BEL(B, read(B,chapter_one», C i) (214) A: Do you speak: Portuguese? B: My husband does. There are no restrictions on those posers which support scalar implicamre. However, (ar least) one restriction does exist on which posers may be viewed as salient in a given exchange: The inclusion ordering which supports the implicature in 214 might be represented as follows: Above (Section 5.1.6.3) I noted for most metrics that rank utterances. both a given metric 128 129 and irs dual (converse) may be candidares for salience in an exchange. However, no metric (ji which orders values vi and Vj such that a) vi is higher than Vj and b} the truth of Vj entails the A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects truth of vi can supporr scalar implicature -- for the simple reason in such a case, a sentence set, in consequence, will represent a HIGHER value in the ordering. Subs.ets which are neither AFFIRt-.t(B,rhird chaprer, BEL(B, read(B,third chapter»» 1\ Pi ranked higher than a sentence Pj by (ji since then the implicature licensed would be ALT_SENT(read(B,ch3prec_one), read(B,thirdj:hapter), included in, nor include, one another, will be ALTERNATE values in this poser. Consider how 'parts of a dissertation'» inconsistent with the utterance licensing it. In terms of the fonnaIism presented in Chapter 2, the salient ordering in the following exchange mighr be represented: Partial-order implicature =) SCALAR_IMP(B, A, I read the third, ...,BEL(B, read(B,chapter_one», such a meaning would not be reinforceable. Consider, for example. (212a): C i) (214) A: Do you speak: Portuguese? So, {husband,wife.chiid} wi!! be the highest value in this ordering, with the alternate doubletons (212) A: Are you planning to buy a dog? B: My husband does. There are no restrictions on those posers which support scalar implicamre. However, (ar (husband,wife), (wife.child), and (husband,child) lower values and the alternate values, a. B: A German Shepherd. least) one restriction does exist on which posers may be viewed as salient in a given exchange: The inclusion ordering which supports the implicature in 214 might be represented as follows: b. B: I'm buying a Gennan Shepherd and I'm not buying a dog. {husband}, (wife), and {child} lower values still in this poset. By the scalar implicature Above (Section 5.1.6.3) I noted for most metrics that rank utterances. both a given metric A: Do you speak conventions, then, S may affinn. say, (husband.wife) to convey ...,BEL(S. (husband,wile.child)) While one might identify either an ordering defined by 'isa' (i.e .• a Gennan Shepherd isa dog) and irs dual (converse) may be candidares for salience in an exchange. However, no metric (ji as well as -,BEL(S. (husband,child}) and -,BEL(S, (wife.childJ). Note, particularly, that there or by 'subsumes' (i.e., a dog subsumes the subtype Shepherd) as salient in this German? which orders values vi and Vj such that a) vi is higher than Vj and b} the truth of Vj entails the may be some redundance in scalar implicatures predicted from this representation. Also, any exchange, only the latter permits scalar implicature here. B cannot implicate that she is not truth of vi can supporr scalar implicature -- for the simple reason in such a case, a sentence subsets so represented may be lexica1ized in various ways -- as, the expression (husband.wife) buying a dog vla this response, since buying a Gennan Shepherd entails buying a dog. The Pi ranked higher than a sentence Pj by (ji since then the implicature licensed would be B: My husband does. might be lexicalized as • couple' or as 'husband and wife'. The theory presented in this thesis attempted reinforcement of (212b) fails. However, we cannot rule out 'isa' relations as inconsistent with the utterance licensing it. In terms of the fonnaIism presented in Chapter 2, will not distinguish between these. 128 potential supporters of scalar imphcarore: In 213, for example. 8's response might evoke either such a meaning would not be reinforceable. Consider, for example. (212a): an'isa' So, {husband,wife.chiid} wi!! be the highest value in this ordering, with the alternate doubletons (212) A: Are you planning to buy a dog? As noted in Sections 5.1.7, temporal orderings may also be represented as setS ofrernporal (husband,wife), (wife.child), and (husband,child) lower values and the alternate values, a. B: A German Shepherd. (213) A: Would you like a dog? for the analysis of licensed scalar implicacures. So, these orderings too wilt be defined by set b. B: I'm buying a Gennan Shepherd and I'm not buying a dog. 8: I'd like a German Shepherd. {husband}, (wife), and {child} lower values still in this poset. By the scalar implicature inclusion, as: conventions, then, S may affinn. say, (husband.wife) to convey ...,BEL(S. (husband,wile.child)) While one might identify either an ordering defined by 'isa' (i.e .• a Gennan Shepherd isa dog) hierarchy - or irs dual. Apparently, any poser can support scaiar implicature, although other as well as -,BEL(S. (husband,child}) and -,BEL(S, (wife.childJ). Note, particularly, that there or by 'subsumes' (i.e., a dog subsumes the subtype Shepherd) as salient in this tests for conversational implicature may rule out some particuJ3r posers in panicular exchanges. A: Are you on your may be some redundance in scalar implicatures predicted from this representation. Also, any exchange, only the latter permits scalar implicature here. B cannot implicate that she is not subsets so represented may be lexica1ized in various ways -- as, the expression (husband.wife) buying a dog vla this response, since buying a Gennan Shepherd entails buying a dog. The 5.3.2.3. Representing Scalar Implicature Orderings as Pos.ets honeymoon? might be lexicalized as • couple' or as 'husband and wife'. The theory presented in this thesis {past, resent} {presenr,future} {past,furure} attempted reinforcement of (212b) fails. However, we cannot rule out 'isa' relations as r---:::-- will not distinguish between these. 128 J have demonstrated above how part! whole re!arions can be represented. To demonstrate potential supporters of scalar imphcarore: In 213, for example. 8's response might evoke either B: Well, I was. that L'le other orderings discussed in Section 5.1 are accounted for by a poset condition. I w;i! an'isa' {future} As noted in Sections 5.1.7, temporal orderings may also be represented as setS ofrernporal describe how representative orderings can be accommodated by this condition so mat (213) A: Would you like a dog? for the analysis of licensed scalar implicacures. So, these orderings too wilt be defined by set scalar implicatures are correctly predicted by ImPl_3' 8: I'd like a German Shepherd. inclusion, as: Posers defined by a type! subrype metric, such as that which supports 174, may be Apparently, any poser can support scaiar implicature, although other hierarchy - or irs dual. Rdations defined by ordering the non-null members of the power of some set x by illustrated by me (parrial) classification hierarchy: tests for conversational implicature may rule out some particuJ3r posers in panicular exchanges. set-inclusion allow a poset representation of x and its non-null proper subsets a5 follows: Any non-null proper subset of a set m<lY be nnked as LOWER than the set which it. and 5.3.2.3. Representing Scalar Implicature Orderings as Pos.ets lZ1!Sut see {CorelIa 84, Ka!ita 84) for some approaches to thtS problem. {past, resent} {presenr,future} {past,furure} r---:::-- J have demonstrated above how part! whole re!arions can be represented. To demonstrate Hirschberg 1985, A Theory of Scalar Implicature that L'le other orderings discussed in Section 5.1 are accounted for by a poset condition. I w;i! {future} describe how representative orderings can be accommodated by this condition so mat scalar implicatures are correctly predicted by ImPl_3' Posers defined by a type! subrype metric, such as that which supports 174, may be Rdations defined by ordering the non-null members of the power of some set x by illustrated by me (parrial) classification hierarchy: 7 / 56 set-inclusion allow a poset representation of x and its non-null proper subsets a5 follows: Any non-null proper subset of a set m<lY be nnked as LOWER than the set which it. and lZ1!Sut see {CorelIa 84, Ka!ita 84) for some approaches to thtS problem.

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Highly particularized implicature R1 R2 R3 “glasses” 8 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Reference games Frank, Gómez, Peloquin, Goodman, and Potts 2016, 10 experiments, each N ≈ 600 (4,651 participants). The summary picture: https://github.com/langcog/pragmods 9 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Reference games Frank, Gómez, Peloquin, Goodman, and Potts 2016, 10 experiments, each N ≈ 600 (4,651 participants). The summary picture: https://github.com/langcog/pragmods 9 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Reference games Frank, Gómez, Peloquin, Goodman, and Potts 2016, 10 experiments, each N ≈ 600 (4,651 participants). The summary picture: betting forced_choice likert 1.00 Normalized measure mean 0.75 0.50 0.25 0.00 foil target logical foil target logical foil target logical Target https://github.com/langcog/pragmods 9 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat boat motorboat sailboat canoe kayak 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat 2. “boat or canoe” 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat 2. “boat or canoe” 3. Kim is in France. (in Paris) 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat 2. “boat or canoe” 3. Kim is in France. (in Paris) 4. “The nuptials will take place in either France or Paris.” 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat 2. “boat or canoe” 3. Kim is in France. (in Paris) 4. “The nuptials will take place in either France or Paris.” 5. I hit the button and it started. (causation) 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects I-implicature Levinson: “what is simply described is stereotypically exemplified”. 1. At a busy marina in water-skiing country: “boat” interpreted as motorboat 2. “boat or canoe” 3. Kim is in France. (in Paris) 4. “The nuptials will take place in either France or Paris.” 5. I hit the button and it started. (causation) 6. Sandy finished the book. (reading) 10 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects M-implicature Levinson: “What’s said in an abnormal way isn’t normal.” 1. a. T urn on the car. b. Get the car to turn on. 2. a. Stop the car. b. Cause the car to stop. 11 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Sociolinguistic variables 12 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Sociolinguistic variables Generalization Where two forms are in salient contrast, the choice of one will lead to inferences about the other. 12 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Sociolinguistic variables Generalization Where two forms are in salient contrast, the choice of one will lead to inferences about the other. • Community: Community members adopt a speech style that is easily distinguished from the mainstream, enhancing solidarity. 12 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Sociolinguistic variables Generalization Where two forms are in salient contrast, the choice of one will lead to inferences about the other. • Community: Community members adopt a speech style that is easily distinguished from the mainstream, enhancing solidarity. • Individual: An individual systematically varies their speech style by context to construct different personae. 12 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects 1. Meaning from a communicative tension 2. The Rational Speech Acts (RSA) model 3. Learning in the Rational Speech Acts Model 4. Neural RSA 5. Language and action 13 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension • Lewis 1969: signaling systems 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension • Lewis 1969: signaling systems • Rabin 1990: recursive strategic signaling 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension • Lewis 1969: signaling systems • Rabin 1990: recursive strategic signaling • Camerer and Ho 2004: cognitive hierarchy models for games of conflict and coordination 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension • Lewis 1969: signaling systems • Rabin 1990: recursive strategic signaling • Camerer and Ho 2004: cognitive hierarchy models for games of conflict and coordination • Michael Franke and Gerhard Jäger: iterated best response 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension • Lewis 1969: signaling systems • Rabin 1990: recursive strategic signaling • Camerer and Ho 2004: cognitive hierarchy models for games of conflict and coordination • Michael Franke and Gerhard Jäger: iterated best response • Golland, Liang, and Klein 2010 ( EMNLP ): pragmatic listeners and probabilistic compositionality 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Origin story • Rosenberg and Cohen 1964: early Bayesian model of production and comprehension • Lewis 1969: signaling systems • Rabin 1990: recursive strategic signaling • Camerer and Ho 2004: cognitive hierarchy models for games of conflict and coordination • Michael Franke and Gerhard Jäger: iterated best response • Golland, Liang, and Klein 2010 ( EMNLP ): pragmatic listeners and probabilistic compositionality • Frank and Goodman 2012 ( Science ): very sophisticated pragmatic agents and a new Bayesian foundation 14 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic listeners 15 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic listeners Literal listener l 0 ( w | msg, Lex ) ∝ Lex ( msg, w ) P ( w ) 15 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic listeners Pragmatic speaker s 1 ( msg | w, Lex ) ∝ exp λ ( log l 0 ( w | msg, Lex ) − C ( msg )) Literal listener l 0 ( w | msg, Lex ) ∝ Lex ( msg, w ) P ( w ) 15 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic listeners Pragmatic listener l 1 ( w | msg, Lex ) ∝ s 1 ( msg | w, Lex ) P ( w ) Pragmatic speaker s 1 ( msg | w, Lex ) ∝ exp λ ( log l 0 ( w | msg, Lex ) − C ( msg )) Literal listener l 0 ( w | msg, Lex ) ∝ Lex ( msg, w ) P ( w ) 15 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic listeners Pragmatic listener l 1 ( w | msg, Lex ) = pragmatic speaker × state prior Pragmatic speaker s 1 ( msg | w, Lex ) = literal listener − message costs Literal listener l 0 ( w | msg, Lex ) = lexicon × state prior 15 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA listener example l 1 1 0 0 beard s 1 1 1 0 glasses l 0 Lex 0 1 1 tie 16 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA listener example 1 l 1 0 0 beard s 1 .5 .5 0 glasses l 0 Lex 0 .5 .5 tie 16 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA listener example beard glasses tie .67 .33 0 l 1 s 1 1 l 0 0 0 Lex 0 1 0 16 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA listener example 1 l 1 0 0 beard s 1 .25 .75 0 glasses l 0 Lex 1 0 0 tie 16 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic speakers 17 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic speakers Literal speaker s 0 ( msg | w, Lex ) ∝ exp λ ( log Lex ( msg, w ) − C ( msg )) 17 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic speakers Pragmatic listener l 1 ( w | msg, Lex ) ∝ s 0 ( msg | w, Lex ) P ( w ) Literal speaker s 0 ( msg | w, Lex ) ∝ exp λ ( log Lex ( msg, w ) − C ( msg )) 17 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic speakers Pragmatic speaker s 1 ( msg | w, Lex ) ∝ exp λ ( log l 1 ( w | msg, Lex ) − C ( msg )) Pragmatic listener l 1 ( w | msg, Lex ) ∝ s 0 ( msg | w, Lex ) P ( w ) Literal speaker s 0 ( msg | w, Lex ) ∝ exp λ ( log Lex ( msg, w ) − C ( msg )) 17 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Pragmatic speakers Pragmatic speaker s 1 ( msg | w, Lex ) = pragmatic listener − message costs Pragmatic listener l 1 ( w | msg, Lex ) = literal speaker × state prior Literal speaker s 0 ( msg | w, Lex ) = lexicon − message costs 17 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA speaker example beard glasses tie 1 1 0 s 1 l 1 s 0 0 1 1 Lex 0 0 1 18 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA speaker example beard glasses tie .5 .5 0 s 1 l 1 s 0 0 .5 .5 Lex 0 1 0 18 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA speaker example 1 s 1 0 0 beard l 1 .5 .5 0 glasses s 0 Lex .33 .67 0 tie 18 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects RSA speaker example beard glasses tie .67 .33 0 s 1 l 1 .6 s 0 0 .4 Lex 0 1 0 18 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Joint reasoning L ( w, Context | msg ) ∝ P ( w ) P C ( Context ) s 1 ( msg | w, Context ) 19 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Joint reasoning L ( w, Context | msg ) ∝ P ( w ) P C ( Context ) s 1 ( msg | w, Context ) � L ( w | msg ) ∝ P ( w ) P C ( Context ) s 1 ( msg | w, Context ) Context ∈ C 19 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Achievements 20 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Achievements • M-implicatures Bergen, Levy, Goodman, ‘Pragmatic reasoning through semantic inference’ 20 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Achievements • M-implicatures Bergen, Levy, Goodman, ‘Pragmatic reasoning through semantic inference’ • I-implicatures and implicature blocking Potts and Levy, ‘Negotiating lexical uncertainty and speaker expertise with disjunction’ 20 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Achievements • M-implicatures Bergen, Levy, Goodman, ‘Pragmatic reasoning through semantic inference’ • I-implicatures and implicature blocking Potts and Levy, ‘Negotiating lexical uncertainty and speaker expertise with disjunction’ • Implicatures and compositionality Potts, Lassiter, Levy, Frank, ‘Embedded implicatures as pragmatic inferences under compositional lexical uncertainty’ 20 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Achievements • M-implicatures Bergen, Levy, Goodman, ‘Pragmatic reasoning through semantic inference’ • I-implicatures and implicature blocking Potts and Levy, ‘Negotiating lexical uncertainty and speaker expertise with disjunction’ • Implicatures and compositionality Potts, Lassiter, Levy, Frank, ‘Embedded implicatures as pragmatic inferences under compositional lexical uncertainty’ • Hyperbole Kao, Wu, Bergen, Goodman, ‘Nonliteral understanding of number words’ 20 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Achievements • M-implicatures Bergen, Levy, Goodman, ‘Pragmatic reasoning through semantic inference’ • I-implicatures and implicature blocking Potts and Levy, ‘Negotiating lexical uncertainty and speaker expertise with disjunction’ • Implicatures and compositionality Potts, Lassiter, Levy, Frank, ‘Embedded implicatures as pragmatic inferences under compositional lexical uncertainty’ • Hyperbole Kao, Wu, Bergen, Goodman, ‘Nonliteral understanding of number words’ • Metaphor Kao, Bergen, Goodman, ‘Formalizing the pragmatics of metaphor understanding’ 20 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Limitations • Hand-specified lexicon • High-bias model; few chances to learn from data • Cognitive demands limit speaker rationality • Speaker preferences • Scalability 21 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects 1. Meaning from a communicative tension 2. The Rational Speech Acts (RSA) model 3. Learning in the Rational Speech Acts Model 4. Neural RSA 5. Language and action Will Monroe 22 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects TUNA furniture example 23 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects TUNA furniture example Utterance: “blue fan small” 23 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects TUNA furniture example colour:green colour:green colour:red orientation:left orientation:left orientation:back size:small size:small size:large type:fan type:sofa type:fan x-dimension:1 x-dimension:1 x-dimension:1 y-dimension:1 y-dimension:2 y-dimension:3 colour:red colour:blue orientation:back orientation:left size:large size:large type:sofa type:fan x-dimension:2 x-dimension:2 y-dimension:1 y-dimension:2 colour:blue colour:blue orientation:left orientation:left size:large size:small type:sofa type:fan x-dimension:3 x-dimension:3 y-dimension:1 y-dimension:3 Utterance: “blue fan small” Utterance attributes: [ colour:blue ] ; [ size:small ] ; [ type:fan ] 24 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects TUNA people example 25 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects TUNA people example Utterance: “The bald man with a beard” 25 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects TUNA people example age:young age:old hairColour:dark hairColour:light hasBeard:0 hasBeard:1 hasGlasses:0 hasGlasses:0 hasHair:1 hasHair:0 hasShirt:1 hasShirt:1 hasSuit:0 hasSuit:0 hasTie:0 hasTie:0 type:person type:person age:young age:young hairColour:dark hairColour:dark hasBeard:1 hasBeard:1 hasGlasses:0 hasGlasses:0 hasHair:1 hasHair:1 hasShirt:0 hasShirt:1 hasSuit:0 hasSuit:1 hasTie:1 hasTie:1 type:person type:person age:young age:young age:young hairColour:dark hairColour:dark hairColour:dark hasBeard:0 hasBeard:1 hasBeard:0 hasGlasses:0 hasGlasses:0 hasGlasses:0 hasHair:1 hasHair:1 hasHair:1 hasShirt:0 hasShirt:1 hasShirt:0 hasSuit:1 hasSuit:0 hasSuit:1 hasTie:1 hasTie:0 hasTie:1 type:person type:person type:person Utterance: “The bald man with a beard” Utterance attributes: [ hasBeard:1 ] ; [ hasHair:0 ] ; [ type:person ] 26 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Feature representations   colour:blue orientation:left [ colour:blue ]  size:small ,  [ size:small ] type:fan   [ type:fan ] x-dimension:3 y-dimension:3 27 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Feature representations   colour:blue orientation:left [ colour:blue ]  size:small ,  [ size:small ] type:fan   [ type:fan ] x-dimension:3 y-dimension:3 Cross-product features colour:blue ∧ [ colour:blue ] colour:blue ∧ [ size:small ] colour:blue ∧ [ type:fan ] orientation:left ∧ [ colour:blue ] orientation:left ∧ [ size:small ] . . . 27 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Feature representations   colour:blue orientation:left [ colour:blue ]  size:small ,  [ size:small ] type:fan   [ type:fan ] x-dimension:3 y-dimension:3 Cross-product features Generation features colour:blue ∧ [ colour:blue ] color colour:blue ∧ [ size:small ] type + color colour:blue ∧ [ type:fan ] color + ¬ size orientation:left ∧ [ colour:blue ] attribute-count = 3 . orientation:left ∧ [ size:small ] . . . . . 27 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Feature representations   colour:blue orientation:left [ colour:blue ]  size:small ,  [ size:small ] type:fan   [ type:fan ] x-dimension:3 y-dimension:3 Cross-product features Generation features colour:blue ∧ [ colour:blue ] color colour:blue ∧ [ size:small ] type + color colour:blue ∧ [ type:fan ] color + ¬ size orientation:left ∧ [ colour:blue ] attribute-count = 3 . orientation:left ∧ [ size:small ] . . . . . type ≫ orientation ≫ color ≫ size 27 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Model definition S 1 ( m | t , θ)∝ L 1 ( t | m , θ) L 1 ( t | m , θ)∝ S 0 ( m | t , θ) T ϕ( t ,m )] S 0 ( m | t , θ)∝ exp [θ “beard” “guy with the beard” ⊙ “guy with glasses” ϕ θ ... 28 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Optimization “guy with the beard” ∂ ∂θ log S 1 ( m | t , θ) “beard” “guy with the beard” ⊙ “guy with glasses” ϕ θ ... 29 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production Features like color 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production Features like color Learn attribute hierarchies 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production Features like color Learn attribute hierarchies Features like color + ¬ size 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production Features like color Learn attribute hierarchies Features like color + ¬ size Learn message costs 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production Features like color Learn attribute hierarchies Features like color + ¬ size Learn message costs Length features and others 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Addressing the drawbacks of RSA Goal Features Avoid hand-built lexicon Cross-product features Learn quirks of production Features like color Learn attribute hierarchies Features like color + ¬ size Learn message costs Length features and others Cognitive and linguistic insights combined with learning 30 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects Example Train [ person ] [ person ] [ glasses ] [ beard ] T est 31 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects ∅ [ person ] [ glasses ] [ beard ] [ person ] ; [ glasses ] [ person ] ; [ beard ] [ glasses ] ; [ beard ] [ all ] 32 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects ∅ .08 .25 [ person ] .08 .25 [ glasses ] .17 .00 .08 .25 [ beard ] [ person ] ; [ glasses ] .17 .00 [ person ] ; [ beard ] .08 .25 [ glasses ] ; [ beard ] .17 .00 .17 .00 [ all ] RSA 32 / 56

A Gricean ideal Implicatures RSA Learned RSA Neural RSA Language and action Prospects ∅ .08 .25 .03 .00 [ person ] .08 .25 .22 .10 [ glasses ] .17 .00 .03 .00 .08 .25 .03 .04 [ beard ] [ person ] ; [ glasses ] .17 .00 .22 .01 [ person ] ; [ beard ] .08 .25 .22 .74 [ glasses ] ; [ beard ] .17 .00 .03 .00 .17 .00 .22 .10 [ all ] RSA Learned S 0 32 / 56