LBS Prüfungsprotokol ()
Wie immer wird am Anfang gesagt es geht darum eine Diskussion zu führen. Bei Fragen oder Unklarheiten ist es besser was zu sagen als zu schweigen, etc. Im wesentlichen das gleiche wie bei den anderen Protokollen.
Ich bitte darum die Prüfung auf Englisch zu belegen. Es wird gestattet.
- What is the method of fragments
- I explain that language is difficult, it seems, and that fragments are an approach to translate parts of natural language into some formal language
- What does a fragment consist of. I mention grammar and that seems to have been enough
- Give an example for a grammar
- Syntax tree for the sentence "every woman sleeps"
- \(s(np(det(\mathtt{every}), n(\mathtt{woman})), vp(v(\mathtt{sleeps})))\)
- Talking about the language pipeline
- Give an example for a quasi-logical formula (\(\mathrm{every}' \mathrm{woman}' \mathrm{sleep}'\))
- Define \(\mathrm{every}'\) as \(\lambda N \ldotp \lambda V \ldotp \forall x \ldotp N(x) \implies V(x)\)
- Apply a beta reduction (replacing \(N\) with \(\mathrm{woman'}\))
- Translate this into a FOL formula \(\forall x \ldotp \mathrm{woman}' x \implies \mathrm{sleep}' x\) (I actually wrote this down before the quasi-logical form)
- Some talk about compositionality, why it is interesting
- What to do with a sentence like "peter kicks the dog in the bathroom
at midnight"
- Mention events (forgot the name "Neo-Davidsonian")
- Translate this into a kind-of-FOL formula \(\forall e \ldotp \mathrm{event}(e) \land \mathrm{peterKickDog}(e) \land \mathrm{time}(e, \mathrm{midnight}) \land \dots\)
- Issue with dealing with a lot of different logics, what do we do? MMT
- Formalise propositional logic on paper (I declare
proplogas a theory, write downprop,not,and,orwith a definition) - Formalise first order logic on paper (I declare
foland includeproplog, write downindivandforall. Don't have to doeveryandequal) - We talk about how to formalise the "Neo-Davidsonian" stuff into
MMT. I suggest creating a predicate for events, Kohlhase suggests
a type. I claim that in that case we cannot use
forallfromfol, so we need something of the same form just with an event to omicron function instead of iota to omicron.
- Formalise propositional logic on paper (I declare
- We jump from logic back to language
- I am given the sentence "a man sleeps. he snores" and have to mention DRT
- What is a discourse (confused me a bit but I apparently said the right thing with "a sequence of sentences")
- Show how to process the aforementioned sentence: \((\delta M \ldotp
\mathrm{sleep}(M)) \otimes (\delta X \ldotp \mathrm{snores}(X))
\leadsto (\delta M, X \ldotp \mathrm{sleep}(M) \land
\mathrm{snores}(X) \land M = X)\)
- Mentioning that \(M = X\) is anaphora resolution, give more examples if the first sentence were something like "A man and his dog" or "A man and his friend"
- How would the direct semantics be defined? I mention that an interpretation \(\mathcal{I}\) would be necessary, and try to define the interpretation of a condition. Kohlhase says that would be too easy, and that the interpretation of a DRS is more interesting. I write down \[\mathcal{I}(\delta \dots \ldotp C) = \forall \dots \implies \exists\] but admit that I don't know the details. It is ok, as this was just a bonus question.
Werde darum gebeten kurz vor die Tür zu treten, und werde nach weniger als einer Minute wieder hineingelassen. War sehr gut (1.0), auch wenn ich an ein paar stellen etwas gestockt habe. Ich bedanke mich und verlasse das Büro. Ich darf mein Blatt mitnehmen, was mir hilft mein Protokoll zu verfassen (nudge nudge).
Zur Vorbereitung: Habe an der Vorlesung aktiv teilgenommen, die Übungen aus Zeitgründen mehr oder weniger gut gemacht. Vor der Klausur habe ich mich vier mal zu Lerngruppen getroffen gehabt, aber hatte sonst keine wirkliche Motivation zu lernen (hatte daher also eher Glück mit der Note). Ein mal sind wird auch zum Kohlhase gegangen und haben ihm unsere Fragen gestellt, was sehr hilfreich war.