From Discourse to Logic

Overview

Book: From Discourse to Logic, Student Edition, first printing (paperback)
ISBN: 0-7923-1028-4

Official book website (no errata page at present)

Status: I emailed Hans Kamp with a copy of this errata list on 2003-09-11, but to date I have not received any reply.

Administrivia

This page uses MathML to display equations. If you can't see them, you will need to install a suitable viewer. I personally use MathPlayer.

This page also uses SVG. If you can't see the figures below, you will need to install a suitable viewer, such as the Adobe plug-in.

Errata

Page 69:

In (1.20), it should say "Gen = male" rather than "Gen = -hum"

Page 93:

Footnote 8 refers to [Lewis 1986], but that entry doesn't appear in the bibliography.

Page 95:

M₃ doesn't actually extend M₂ (as claimed on p96), because owns_M₃ doesn't include the tuple <c,f>. fascinates_M₃ is a superset of fascinates_M₂, but this would be clearer if <d,a> appeared between <c,a> and <c,b>, i.e. if all the elements of fascinates_M₂ came first followed by the new elements.
N.B. This doesn't affect the truth of the DRSs in the models.

Page 96:

Does it actually make sense to say "Pred_M'(Q) ∩ U_M"? This seems ok for unary predicates, but I would think that it has problems for n-ary predicates when n>=2. For instance, if you take the intersection of owns_M₂ and U_M₁, you are comparing pairs (e.g. <a,d>) to individual elements (e.g. a), so I would expect the result to be either unspecified or an empty set. I would phrase the definition differently - something like: Pred_M(Q) = {<x₁, x₂, ..., x_n> | <x₁, x₂, ..., x_n> ∈ Pred_M'(Q) ∧ x₁ ∈ U_M ∧ x₂ ∈ U_M ∧ ... ∧ x_n ∈ U_M}

Page 132:

Minor typo: in the final paragraph, "DRS-conditons" should be "DRS-conditions".

p163:

In question 4, "him to Bill" should say "he to Bill".

Page 202:

It seems that the specialised variant of the construction rule "CR.OR" for NP is unnecessary, since the standard rule will do the job perfectly well.

In case (i), they are equivalent. You can verify this by using example (2.129): "Smith or Jones loves Lady Hermione".

I also think that part (ii) of the NP variant is incorrect. Consider the sentence "Lady Hermione loves Smith or Jones", which we would like to split up into "Lady Hermione loves Smith or Lady Hermione loves Jones". The initial tree would be:

Sentence structure (SVG) (in case it's not visible above).

So, NP₁ =

Noun phrase structure (SVG) (in case it's not visible above).

and NP₂ =

Noun phrase structure (SVG) (in case it's not visible above).

This would match condition (ii) of CR.OR(NP), where γ =

Verb phrase structure (SVG) (in case it's not visible above).

and $\bar{γ} =$

Sentence structure (SVG) (in case it's not visible above).

So, following the bottom line of the construction rule, we would replace $\bar{γ}$ by:

Sentence structure (SVG) (in case it's not visible above).

In other words, we would get "Lady Hermione Smith or Lady Hermione Jones", which isn't what we want.

As mentioned above, the original version of CR.OR (on p197) would be correct here, since it would substitute in NP₁ and NP₂ for the second NP in the original sentence.

Page 203:

Since CR.OR(NP) is redundant (as explained above), that means that CR.OR (≠ NP) is also redundant, as it is identical to the original CR.OR construction rule on page 197 (aside from the X ≠ NP line).

"(2.130)" should say "(2.131)"

This page was last updated on 2003-12-30 by John C. Kirk