Add COCOA handout and SBL slides

3 files changed, 918 insertions, 5 deletions

diff --git a/resources/pdf/cocoa2022.pdf b/resources/pdf/cocoa2022.pdf
new file mode 100644
index 0000000..b327828
--- /dev/null
+++ b/resources/pdf/cocoa2022.pdf
@@ -0,0 +1,913 @@
+\documentclass[ocgsteps,ocgstepsdraft,larger]{../text/text}
+
+\newcommand{\auth}[3]{%
+	#1 \\[-1ex]
+	{\small \href{https://twitter.com/#2}{@#2}} \\[-1ex]
+	{\small \href{mailto:#3}{#3}}}
+
+\title{\Large Space in the causal chain:\\The perspective from French and Biblical Hebrew}
+\author{
+	\auth{Camil Staps}{SemiCamil}{info@camilstaps.nl}
+	\and
+	\auth{Johan Rooryck}{JohanRooryck}{johan.rooryck@gmail.com}
+}
+\date{COCOA, 9\textsuperscript{th} November 2022}
+
+\usepackage{amssymb}
+\newcommand{\sv}[1]{\ensuremath{\llbracket{#1}\rrbracket}}
+
+\tikzset{
+	>=stealth,
+	every picture/.style={
+		baseline
+	},
+	every tree node/.style={
+		align=center,
+		anchor=north
+	},
+	level 1+/.style={
+		level distance=4.5\baselineskip
+	}
+}
+
+\usepackage{gb4e}
+\let\eachwordone=\it
+\setlist{nosep}
+
+\usepackage{endnotes}
+
+\let\footnote=\endnote
+
+\newcommand{\imp}[1]{{\color{red}#1}}
+
+\setlength\tabcolsep{.2222em} % \>
+
+\DeclareMathOperator{\Agent}{Agent}
+\DeclareMathOperator{\Origin}{Origin}
+\DeclareMathOperator{\Net}{net}
+\DeclareMathOperator{\FROM}{FROM}
+\DeclareMathOperator{\THROUGH}{THROUGH}
+
+\begin{document}
+
+\maketitle
+
+\begin{step}[section]
+	\section{Outline}
+
+	\begin{enumerate}
+		\item
+			Languages with multiple Agent prepositions are problematic for current accounts of Agent PPs
+
+			\begin{enumerate}
+				\item
+					\sloppy
+					These prepositions have interpretive differences,
+						which we can explain with the Space $\Rightarrow$ Causation metaphor~\citep{Croft2012}
+				\item
+					We need a formalism that remains close to conceptual thinking about causation:
+						force dynamics~\citep{CopleyEtAl2015,CopleyEtAl2020}
+				\item
+					This allows a polymorphic denotation of Agent prepositions with which we can predict the interpretive differences based on spatial meaning
+			\end{enumerate}
+
+		\item
+			Space in the causal chain can be interpreted in different ways
+
+			\begin{enumerate}
+				\item
+					In French, greater distance indicates a lack of influence \dots
+				\item
+					\dots\ but in Biblical Hebrew, greater distance indicates more control.
+				\item
+					This is acceptable if we understand that languages can have different {\sc perspectives on the causal chain}
+			\end{enumerate}
+	\end{enumerate}
+\end{step}
+
+\begin{step}[section]
+	\section{Agent PPs}
+
+	\subsection{\texorpdfstring{The Space $\Rightarrow$ Causation metaphor~\citep{Croft2012}}{The Space => Causation metaphor (Croft 2012)}}
+
+	\citet{Croft2012} makes a distinction between roles that precede the Object in the causal chain and roles that follow it:
+
+	\begin{exe}
+		\ex
+		\gls{Sue broke the coconut for Greg with a hammer.}
+		\trailingcitation{\citep[224]{Croft2012}}
+		\begin{tabular}{lllllll}
+			Sue & $\rightarrow$ & hammer & $\rightarrow$ & coconut & $\dashrightarrow$ & Greg \\
+			Subject &           & Antecedent oblique &   & Object  &                   & Subsequent oblique \\
+		\end{tabular}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	Antecedent obliques are commonly marked with ablative/perlative/proximative prepositions (\ref{antecedent}), and subsequent obliques with allative ones (\ref{subsequent}):
+
+	\begin{exe}
+		\ex \label{antecedent}
+		\begin{xlist}
+			\ex Cause: \gls{The rabbit died \imp{from}/\imp{of} thirst.}
+				\trailingcitation{\citep[223]{Croft2012}}
+			\ex Agent: \gls{The cat food was eaten \imp{by} raccoons.}
+			\ex Means: \gls{I went downtown \imp{by} bus.}
+			\ex Instrument: \gls{Sue broke the coconut \imp{with} a hammer.}
+			\ex Comitative: \gls{I went to the park \imp{with} Carol.}
+		\end{xlist}
+
+		\ex \label{subsequent}
+		\begin{xlist}
+			\ex Result: \gls{They smashed the statue \imp{to} pieces.}
+				\trailingcitation{\citep[223]{Croft2012}}
+			\ex Result: \gls{The boy carved the stick \imp{into} a knife.}
+			\ex Beneficiary: \gls{Sue broke the coconut \imp{for} Greg.}
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	This suggests the {\sc Space $\Rightarrow$ Causation metaphor}:
+
+	\begin{exe}
+		\ex
+		\begin{tabular}[t]{lccc}
+			\textit{Causation:} & antecedent role & Object     & subsequent role \\
+													& $\Uparrow$      & $\Uparrow$ & $\Uparrow$ \\
+			\textit{Space:}     & ablative/origin & locative   & allative/goal \\
+		\end{tabular}
+		\trailingcitation{\citep[225]{Croft2012}}
+	\end{exe}
+\end{step}
+
+%\begin{step}[subsection]
+%	\subsection{Formal accounts of Agent PPs}
+%	There are two main theories of Agent PPs.
+%	In adjunct strategies, Agent PPs are adjuncts to Voice~\citep{Bruening2013,Legate2014,IngasonEtAl2016}:
+%
+%	\begin{exe}
+%		\ex
+%		\begin{xlist}
+%			\ex
+%				\gls{The senator was bribed by the lobbyist.}
+%				\trailingcitation{(\citealp[25]{Bruening2013}, types added)}
+%
+%				{\tikzset{%
+%					level 1+/.style={level distance=3\baselineskip},
+%					level 1/.style={sibling distance=-60pt},
+%					level 2/.style={sibling distance=-30pt},
+%					level 3/.style={sibling distance=-20pt}}
+%				\setstretch{1}
+%				\Tree
+%					[.{Pass \\ $st$}
+%						[.{Pass[S:Voice(S:N)] \\ $(e)st,st$} ]
+%						[.{Voice[S:N] \\ $st$}
+%							[.{Voice[S:N] \\ $est$}
+%								[.{Voice[S:V, S:N] \\ $st,est$} ]
+%								[.{V \\ $st$}
+%									[.{V[S:N] \\ $est$ \\ bribe} ]
+%									[.{N \\ $e$ \\ the senator} ]
+%								]
+%							]
+%							[.{P[S$_a$:Voice(S:N)] \\ $est,st$}
+%								[.{P[S:N, S$_a$:Voice(S:N)] \\ \imp{$e,est,st$} \\ by} ]
+%								[.{N \\ $e$ \\ the lobbyist} ]
+%							]
+%						]
+%					]
+%				}
+%			\ex
+%				$\sv{\mathit{by}} = \lambda x \lambda f_{est} \lambda e . f(e,x)$
+%		\end{xlist}
+%	\end{exe}
+%\end{step}
+
+\begin{step}[subsection]
+	\subsection{A formal account of Agent PPs}
+	In the formalization that we will adopt,
+		Agent PPs merge in the same place as the external argument in the active voice~\citep{AngelopoulosEtAl2020}.
+	This is preferable over an adjunct position because the Agent PP can bind a reflexive pronoun:
+
+	\begin{exe}
+		\ex
+			\gls{The packages were sent by the children$_i$ to themselves$_i$.}
+			\trailingcitation{(\cite[14]{AngelopoulosEtAl2020}, types added)}
+
+			{\tikzset{%
+				level 1+/.style={level distance=3\baselineskip},
+				level 2/.style={sibling distance=30pt},
+				level 3/.style={sibling distance=10pt}}
+			\setstretch{1}
+			\Tree
+				[.{VoiceP \\ $st$}
+					[.{Voice \\ $st,st$} ]
+					[.{\textit{v}P \\ $st$}
+						[.{PP$_i$ \\ $e$}
+							[.{P \\ \imp{$ee$} \\ \gls{by}} ]
+							[.{DP \\ $e$} \edge[roof]; {\gls{the children}$_i$} ]
+						]
+						[.{\textit{v}' \\ $est$}
+							[.{\vphantom{P}\textit{v} \\ $st,est$} ]
+							[.{VP \\ $st$} \edge[roof]; {send the packages to \gls{themselves}$_i$} ]
+						]
+					]
+				]
+			}
+	\end{exe}
+
+	But note that the Agent preposition is seen as a purely functional element with type $ee$,
+		denoting the identity function: $\sv{\textit{by}} = \lambda x.x$.
+	It thus assumes homonymy of the Agent preposition \gls{by} with spatial \gls{by}, as in \gls{the house by the lake}.
+	There is no room for semantic content of the Agent preposition.
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{Multiple Agent prepositions: the case of French}
+	In French passives the Agent can be introduced by both \gls{de} \enquote{from} and \gls{par} \enquote{by, via}.
+	Which prepositions are allowed varies:
+
+	\begin{exe}
+		\ex \label{basic}
+		\begin{xlist}
+			\ex \label{chien}
+				\gls{Le chien est lavé par/*de Marie.}
+				\trailingcitation{\citep[584]{Straub1974}} \\
+				`The dog was washed by Mary.'
+			\ex \label{mois}
+				\gls{Le mois de février est précédé du/*par le mois de janvier.}
+				\trailingcitation{\citep[591]{Straub1974}} \\
+				`February is preceded by January.'
+			\ex \label{etudiantes}
+				\gls{Les étudiantes sont accompagnées par/de leurs familles.}
+				\trailingcitation{(after \citealp[200]{Gaatone1998})} \\
+				`The students are accompanied by their families.'
+			\ex \label{detenu}
+				\gls{Le détenu est accompagné par le/*du policier.} \\
+				`The inmate is accompanied by the policeman.'
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	We will argue that the difference between \gls{de} and \gls{par} has to do with {\sc influence}:
+	In (\ref{detenu}), the inmate is somehow \enquote{influenced} by the policeman,
+		whereas the students in (\ref{etudiantes}) are not necessarily \enquote{influenced} by their parents.
+	Hence \gls{de} marks a lack of influence.
+	This can also be seen in (\ref{chien})--(\ref{mois}).
+\end{step}
+
+\begin{step}
+	Although the French data confirm \citeauthor{Croft2012}'s (\citeyear{Croft2012}) Space $\Rightarrow$ Causation metaphor,
+		his theory cannot explain the difference between \gls{de} and \gls{par}.
+	Both mark an antecedent role, but \citeauthor{Croft2012} does not make further distinctions:
+
+	\begin{quote}
+		Although \imp{one cannot predict which participant roles a specific Oblique case marking will subsume}---%
+			case markers are usually quite polysemous---%
+			one can predict that a specific Oblique case marking will subsume only antecedent roles or only subsequent roles.
+		That is, \imp{one can generally categorize Oblique morphosyntactic markers as either Antecedent or Subsequent}, as in (\ref{antecedent})--(\ref{subsequent}).
+		\trailingcitation{\citep[223, emphasis added]{Croft2012}}
+	\end{quote}
+\end{step}
+
+\begin{step}
+	The formal approach also has trouble handling the French data.
+	If the Agent preposition denotes the identity function,
+		how can we differentiate between \gls{de} and \gls{par}?
+	%Binding data shows that we cannot use an adjunct strategy:
+
+	%\begin{exe}
+	%	\ex
+	%	\begin{xlist}
+	%		\ex
+	%			\gls{Les étudiantes$_i$ sont accompagnées par/de leurs parents$_j$ de leur$_{i\text{/}j}$ propre gré.}
+	%			\trailingcitation{(after \citealp[200]{Gaatone1998})} \\
+	%			`The students were accompanied by their parents out of their free will.'
+	%		\ex
+	%			\gls{Les étudiantes$_i$ étaient venues sans leurs parents$_j$ de leur$_{i\text{/*}j}$ propre gré.} \\
+	%			`The students came without their parents out of their free will.'
+	%	\end{xlist}
+	%\end{exe}
+
+	%But in argument strategies the preposition is taken to denote the identity function;
+	%	with type $ee$, it has no access to the eventuality influenced by the complement.
+\end{step}
+
+\begin{step}
+	We need a formalism that remains closer to intuitive thinking about causation.
+	To be able to implement the ideas from \citet{Croft2012} we want a single denotation for each preposition,
+		from which can derive both spatial and causal meanings.
+	Force dynamics~\citep{Talmy1988,CopleyEtAl2015,CopleyEtAl2020} will help us with this.
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{\texorpdfstring{A polymorphic denotation of \gls{de} and \gls{par}}{A polymorphic denotation of de and par}}
+	The use of \gls{de} and \gls{par} outside passives first suggests that force dynamics might be the right approach.
+	\gls{De} is used to name Causes that are situations (\ref{cause-de}),
+		while \gls{par} is used to name Causes that are forces (\ref{cause-par}).
+
+	\begin{exe}
+		\ex \label{cause-de}
+		\begin{xlist}
+			\ex
+				\gls{Jean est mort de/*par} \{\gls{faim / vieillesse / la maladie de Parkinson}\}. \\
+				`Jean died of hunger / old age / Parkinson's disease.'
+			\ex
+				\gls{Marie} \{\gls{s'est écriée / a gémi}\} \gls{de/*par} \{\gls{douleur / admiration / plaisir}\} / ??\gls{de l'impact du ballon}. \\
+				`Marie cried out / groaned from pain / admiration / pleasure / the impact of the ball.'
+		\end{xlist}
+		\ex \label{cause-par}
+			\gls{La fenêtre s'est cassée *de/par} \{\gls{un tremblement de terre / l'impact du ballon}\}. \\
+			`The window broke due to an earthquake / the impact of the ball.'
+	\end{exe}
+
+	This is similar to what \citet[139--142]{CopleyEtAl2015} describe for English \gls{from},
+		which can mark forces but not entities:
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex \gls{The floor broke from the *(weight of the) elephant.}
+				\trailingcitation{\citep[141]{CopleyEtAl2015}}
+			\ex \gls{The window broke from John*('s hitting it).}
+		\end{xlist}
+	\end{exe}
+
+	With a distinction between entities (type $e$), forces (type $f$), and situations (type $s$)
+		we can obtain a formalization that derives the correct facts for (\ref{cause-de})--(\ref{cause-par}),
+		and the behaviour of \gls{de} and \gls{par} in passives as well.
+\end{step}
+
+\begin{step}
+	We will give \gls{de} and \gls{par} an abstract definition based on its basic spatial meaning.
+	The type is $\eta\theta t$, where $\eta$ and $\theta$ can be any type
+		as long as the abstract spatial meaning has a reasonable interpretation for that type~(cf.~\cite[162]{Morrill1994}).
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex $\sv{\mathit{de}}_{\eta\theta t} = \lambda x_\eta \lambda y_\theta . \FROM(x,y)$, for any types $\eta, \theta$
+			\ex $\sv{\mathit{par}}_{\eta\theta t} = \lambda x_\eta \lambda y_\theta . \THROUGH(x,y)$, for any types $\eta, \theta$
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	For instance:
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex
+				\gls{un médecin de Paris} \\
+				`a doctor from Paris'
+			\ex $\sv{\mathit{de}}_{eet} = \lambda x_e \lambda y_e . \FROM(x,y)$, interpreted as ``$y$ comes from $x$''.
+		\end{xlist}
+		\ex
+		\begin{xlist}
+			\ex
+				\gls{le train à Lyon par Dijon} \\
+				`the train to Lyon via Dijon'
+			\ex $\sv{\mathit{par}}_{eet} = \lambda x_e \lambda y_e . \THROUGH(x,y)$, interpreted as ``$y$ goes through $x$''.
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	For the Cause markers \gls{de} and \gls{par} we get:
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex
+				\gls{Jean est mort de/*par faim.} \\
+				`Jean died of hunger.'
+			\ex $\sv{\mathit{de}}_{sst} = \lambda s_0 \lambda s_1 . \FROM(s_0, s_1)$, interpreted as $(\Net(s_0))(s_0) = s_1$.
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}[list]
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex
+				\gls{La fenêtre s'est cassée *de/par un tremblement de terre.} \\
+				`The window broke due to an earthquake.'
+			\ex $\sv{\mathit{par}}_{fst} = \lambda f \lambda s_1 . \THROUGH(f, s_1)$, interpreted as $\exists s_0 : \Net(s_0) = f \land f(s_0) = s_1$.
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	Mapping the concepts of force dynamics onto \citeauthor{Croft2012}'s (\citeyear{Croft2012}) Space $\Rightarrow$ Causation metaphor,
+		we can explain why \gls{de} has type $sst$ but not $fst$,
+		and why \gls{par} has type $fst$ but not $sst$.
+
+	\begin{exe}
+		\ex
+			\begin{tikzpicture}
+				\node[circle,draw] (s0) at (0,0) {$s_0$};
+				\node[circle,draw] (s1) at (6,0) {$s_1$};
+				\draw[->] (s0) -- (s1) node[midway,above] {$f$};
+				\node at (0,-1) {Ablative};
+				\node at (0,-1.7) {\gls{de} `from'};
+				\node at (3,-1) {Perlative};
+				\node at (3,-1.7) {\gls{par} `through'};
+				\node at (6,-1) {Object};
+				\node at (6,-1.7) {passive subject};
+			\end{tikzpicture}
+	\end{exe}
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{Formalizing the Agent prepositions}
+	We assume an intermediate projection which we will call OriginP to implement the argument strategy~\citep{AngelopoulosEtAl2020} in the framework of \citet{CopleyEtAl2020}:
+
+	\begin{exe}
+		\ex
+			\gls{the soup was heated by John}
+
+			{\tikzset{level 1+/.style={level distance=3\baselineskip}}
+			\setstretch{1}
+			\Tree
+				[.{VoiceP \\ $fd$}
+					[.{Voice$_{\text{PASS}}$ \\ $efd,fd$} ]
+					[.{OriginP \\ $efd$}
+						[.{PP \\ $ft$}
+							[.{P \\ $eft$ \\ \gls{by}} ]
+							[.{DP \\ $e$} \edge[roof]; {\gls{John}} ]
+						]
+						[.{Origin' \\ $efd$}
+							[.{Origin \\ $fd,efd$} ]
+							[.{\textit{v}P \\ $fd$} \edge[roof]; {\gls{heat the soup}} ]
+						]
+					]
+				]
+			}
+
+		\ex
+		\begin{xlist}
+			\ex \label{denotation-origin}
+				$\sv{\text{Origin}} = \lambda p_{fd} \lambda e \lambda f, \Origin(f,e) . p(f)$
+			\ex \label{denotation-by}
+				$\sv{\textit{by}} = \lambda e \lambda f . \Origin(f,e)$
+			\ex \label{denotation-passive}
+				$\sv{\text{Voice}_{\text{PASS}}} = \lambda p_{efd} \lambda f . \exists x : p(x,f)$
+		\end{xlist}
+
+		\ex
+			$\sv{\text{VoiceP}}$ \\
+			%$= (\lambda p_{efd} \lambda f . \exists x : p(x,f)) (\sv{\text{OriginP}})$
+				%\hfill (\ref{denotation-passive}) \\
+			\null\qquad$= \lambda f . \exists x : \sv{\text{OriginP}}(x,f)$
+				\hfill (\ref{denotation-passive}) \\
+				%\hfill (Function Application) \\
+			\null\qquad$= \lambda f . \exists x, \sv{\text{PP}}(f) : \sv{\text{Origin'}}(x,f)$
+				\hfill (Predicate Restriction) \\
+			%$= \lambda f . \exists x, \sv{\text{PP}}(f) : (\lambda p_{fd} \lambda e \lambda f , \Origin(f,e) . p(f))(\sv{\text{\textit{v}P}})(x,f)$
+				%\hfill (\ref{denotation-origin}) \\
+			\null\qquad$= \lambda f . \exists x, \sv{\text{PP}}(f), \Origin(f,x) . \sv{\text{\textit{v}P}}(f)$
+				\hfill (\ref{denotation-origin}) \\
+				%\hfill (Function Application)
+			\null\qquad$= \lambda f . \exists x, \Origin(f, \sv{\text{John}}), \Origin(f,x) . \sv{\text{\textit{v}P}}(f)$
+				\hfill (\ref{denotation-by}) \\
+			\null\qquad$= \lambda f, \Origin(f, \sv{\text{John}}) . \sv{\text{\textit{v}P}}(f)$
+				\hfill (redundant existential bind)
+	\end{exe}
+\end{step}
+
+\begin{step}
+	So we're looking for an interpretation of \gls{de} and \gls{par} with type $eft$:
+
+	\begin{exe}
+		\exr{etudiantes}
+			\gls{Les étudiantes sont accompagnées par/de leurs familles.}
+			\trailingcitation{(after \citealp[200]{Gaatone1998})} \\
+			`The students are accompanied by their families.'
+		\ex \label{de-passive}
+			$\sv{\mathit{de}}_{eft} = \lambda e \lambda f . \FROM(e, f)$, interpreted as $\Origin(e, f)$.
+		\ex \label{par-passive}
+			$\sv{\mathit{par}}_{eft} = \lambda e \lambda f . \THROUGH(e, f)$, interpreted as $\Agent(e,f)$.
+	\end{exe}
+
+	We hypothesize, based on \citet{Croft2012}, that spatial distance is related to distance to the Patient in the causal chain.
+	A greater distance ($\FROM$) is interpreted as a general $\Origin$.
+	The origin of the causal chain is used not only for Agents,
+		but for anything that can be seen as the ultimate starting point of a force:
+		volitional agents, but also natural phenomena or reasons, for example.
+
+	An entity that is more proximal to the Patient ($\THROUGH$) more likely manipulates the Patient directly.
+	We describe that here with $\Agent$, for lack of a better term.
+
+	This link between proximity to the Patient and direct manipulation is not surprising.
+	We see it also in the distinction between the English Agent marker \gls{by} and Instrument marker \gls{with}.
+	Instruments manipulate the Patient more directly,
+		and \gls{with} is more proximal than \gls{by}: cf.\ \gls{the girl by/with the bike}.
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{Corollaries} 
+
+	The fact that \gls{par} is especially suited to mark direct manipulators (Agents)
+		has an impact on the further pragmatic interpretation of \gls{de} and \gls{par}.
+	This allows us to explain differences in the distribution of \gls{de} and \gls{par} that were previously not well-described.
+	What we called {\sc influence} above can be derived from Proto-Agent (or Proto-Patient) properties (cf.\ \cite{Dowty1991}).
+	Consider (\ref{detenu})--(\ref{detenu-no-influence}):
+	
+	\begin{exe}
+		\exr{detenu}
+			\gls{Le détenu est accompagné par le/*du policier.} \\
+			`The inmate is accompanied by the policeman.'
+		\ex \label{detenu-no-influence}
+			\gls{L'ex-détenu est apparu devant le tribunal, accompagné par le/du policier qui l'avait alors arrêté.} \\
+			`The former inmate appeared in front of the courthouse, accompanied by the policeman that had previously arrested him.'
+	\end{exe}
+
+	These examples establish a different relationship between the policeman and the inmate:
+		only in (\ref{detenu}) do we expect the policeman to restrain the inmate in his movements.
+	This difference can be described in terms of Proto-Agent / Proto-Patient properties:
+
+	\begin{itemize}
+		\item
+			In (\ref{detenu}), the inmate is more \imp{affected},
+				because he is less free to move around.
+		\item
+			In (\ref{detenu}), the policeman is more \imp{goal-oriented},
+				because he has the concrete goal to prevent the inmate from escaping.
+	\end{itemize}
+
+	Example (\ref{detenu}) thus has a more prototypical Agent than (\ref{detenu-no-influence}),
+		which explains why only \gls{par} is allowed.
+\end{step}
+
+\begin{step}
+	This also works for examples with inanimate arguments:
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex \label{village-rainer}
+				\gls{Rainer participe pour la première fois à une course de slalom en montagne. Cette course a eu lieu à Reitnau, petit village situé au coeur de la Suisse et surplombé \imp{par} une chaîne de montagne.}%
+				\footnote{\url{http://www.kueschall.ch/fr/Archiv\_978.aspx}, retrieved November 13, 2010 by \url{http://web.archive.org}.} \\
+				`Rainer takes part in a mountain slalom for the first time. This race took place in Reitnau, a small village located in the heart of Switzerland and overlooked \imp{by} a mountain range.'
+			\ex \label{village-tourism}
+				\gls{La première image offerte aux visiteurs est un village paroi surplombé \imp{de} ses deux tours et de son église auxquels on accède par un réseau de ruelles ou d'escaliers.}%
+				\footnote{\url{https://www.saintmartinlevieil.fr/}, retrieved September 16, 2022.} \\
+				`The first image offered to visitors is a walled village overlooked \imp{by} its two towers and its church which are accessed by a network of alleys or stairs.'
+		\end{xlist}
+	\end{exe}
+
+	In (\ref{village-rainer}) our mental image of the village changes because of the mention of the mountain range:
+		that is what makes it a suitable place for the mountain slalom.
+	This is a more abstract reinterpretation of affectedness and goal-orientedness.
+\end{step}
+
+\begin{step}
+	This can become as abstract as visual dominance:
+
+	\begin{exe}
+		\ex \label{surplomber-de}
+		\begin{xlist}
+			\ex
+				\gls{de \dots\ hauts plateaux désolés surplombés \imp{de} sommets déchiquetés}%
+				\footnote{\url{http://lesdeuxvoyageurs.com/Inde/Ladakh2005/Accueil\_Ladakh/Accueil\_Ladakh2005.html}, retrieved December 8, 2021.} \\
+				`desolate highlands overlooked \imp{by} jagged mountain tops'
+			\ex
+				\gls{un mur d'enceinte surplombé \imp{de} barbelé}%
+				\footnote{\url{http://www.haitiministries.com/www/nouvelles/}, retrieved December 8, 2021.} \\
+				`a compound wall surmounted \imp{by} barbed wire'
+		\end{xlist}
+
+		\ex \label{surplomber-par}
+		\begin{xlist}
+			\ex
+				\gls{un sentier surplombé \imp{par} des rochers de grès rose tout le long du parcours}%
+				\footnote{\url{http://www.netrando.com/fr/direct/PHALDAB012.htm}, captured by Linguee.com, retrieved December 8, 2021.} \\
+				`a small path that winds \imp{at the feet of} pink sandstone cliffs all along the way'
+			\ex
+				\gls{un stade surplombé \imp{par} un tremplin de saut à ski}%
+				\footnote{\url{http://fr.fifa.com/tournaments/archive/tournament=102/edition=6946/news/newsid=88409.html}, captured by Linguee.com, retrieved December 8, 2021.} \\
+				`a [soccer] stadium overlooked \imp{by} a ski jump'
+		\end{xlist}
+	\end{exe}
+
+	In the examples in (\ref{surplomber-de}), neither participant is foregrounded with respect to the other,
+		a balance emphasised by modifiers (\gls{désolés \dots\ déchiquetés}; \gls{d'enceinte}).
+	This is not the case in (\ref{surplomber-par}), where the prepositional object really changes the way we see the subject.
+\end{step}
+
+\begin{step}
+	By focusing on the force/Agent, \gls{par} can pick out a resultative interpretation of a verb
+		when \gls{de} leads to a stative interpretation:
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex
+				\gls{Notre-Dame délaissé \imp{de} ses touristes en plein confinement}%
+				\footnote{\url{https://twitter.com/chouettephoto/status/1344600099113074691}, retrieved March 18, 2022.} \\
+				`the Notre-Dame, abandoned \imp{of} its tourists in full lockdown'
+			\ex
+				\gls{En cas d'enfant délaissé \imp{par} ses parents (art. 501 de la charia), \dots}%
+				\footnote{\url{http://www.arabhumanrights.org/publications/countries/lebanon/crc/lebanon3-05f.pdf}, captured by Linguee.com, retrieved March 18, 2022.} \\
+				`In case of a child neglected \imp{by} its parents (art. 501 of the sharia), \dots'
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	In this way, the spatial meaning of \gls{de} and \gls{par} still resonates in causal uses,
+		and from there it can have various effects, depending on context.
+\end{step}
+
+\begin{step}[section]
+	\section{Different perspectives on the causal chain}
+
+	\subsection{Biblical Hebrew}
+	The Biblical Hebrew prepositions \gls{min} `from, out of' and \gls{b\schwa} `in, by, against'
+		present a similar problem as French \gls{de} and \gls{par}.
+	Both are said to be Agent markers in the grammars, but we look here at (more frequent) other causal functions.
+
+	\gls{B\schwa} is the default preposition for Instruments (\ref{instrument-be}), but \gls{min} occurs as well (\ref{instrument-min}).
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex \label{instrument-be}
+				\heb{רַבִּ֗ים אֲשֶׁר־מֵ֨תוּ֙ בְּאַבְנֵ֣י הַבָּרָ֔ד מֵאֲשֶׁ֥ר הָרְג֛וּ בְּנֵ֥י יִשְׂרָאֵ֖ל בֶּחָֽרֶב׃}
+				\trailingcitation{(Joshua 10:11 ESV)}
+				\gll rabbîm \glot{}ăšer mē\b{t}û \imp{b\schwa{}}=\glot{}a\b{b}nē hab=bārā\b{d} mē=\glot{}ăšer hār\schwa\=gû b\schwa{}nē yiśrā\glot\=el \imp{b\schwa}=ḥāre\b{b} \\
+				many {\sc rel} died {\imp{in}=stones of} the=hail from={\sc rel} killed {sons of} Israel \imp{in}=sword \\
+				\glt `There were more who died \imp{because of} the hailstones than the sons of Israel killed \imp{with} the sword.'
+			\ex \label{instrument-min}
+				\heb{וְעַתָּ֗ה הוֹאֵל֙ וּבָרֵךְ֙ אֶת־בֵּ֣ית עַבְדְּךָ֔ ... וּמִבִּרְכָ֣תְךָ֔ יְבֹרַ֥ךְ בֵּֽית־עַבְדְּךָ֖ לְעוֹלָֽם׃}
+				\trailingcitation{(2 Samuel 7:29 ESV)}
+				\gll w\schwa=\phar{}attâ hô\glot\=el û=\b{b}ārē\b{k} \glot{}e\b{t} bê\b{t} \phar{}a\b{b}d-\schwa\b{k}ā \dots{} û=\imp{mib}=bir\b{k}ā\b{t}-\schwa\b{k}ā y\schwa\b{b}ōra\b{k} bêt \phar{}a\b{b}d-\schwa\b{k}ā l\schwa=\phar{}ôlām \\
+				and=now want and=bless {\sc obj} {house of} servant-your \dots{} and=\imp{from}=blessing-your {will be blessed} house servant-your to=eternity \\
+				\glt `Now therefore may it please you to bless the house of your servant, \dots, and \imp{with} your blessing shall the house of your servant be blessed forever.'
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	Conversely, \gls{min} is the default preposition for Reasons (\ref{reason-min}), but \gls{b\schwa} occurs as well (\ref{reason-be}).
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex \label{reason-min}
+				\heb{וַיְמַהֵ֣ר שָׁא֗וּל וַיִּפֹּ֤ל מְלֹא־קֽוֹמָתוֹ֙ אַ֔רְצָה וַיִּרָ֥א מְאֹ֖ד מִדִּבְרֵ֣י שְׁמוּאֵ֑ל}
+				\trailingcitation{(1 Samuel 28:20 ESV)}
+				\gll way\schwa{}mahēr šā\glot{}ûl wayyippōl m\schwa{}lō\glot{} qômā\b{t}-ô \glot{}arṣ-â wayyīrā\glot{} m\schwa\glot{}ō\b{d} \imp{mid}=di\b{b}rê š\schwa{}mû\glot{}ēl \\
+				{and hastened} Saul {and fell} {full of} height-his earth-wards {and he feared} very {\imp{from}=words of} Samuel \\
+				\glt `Then Saul fell at once full length on the ground, filled with fear \imp{because of} the words of Samuel.'
+			\ex \label{reason-be}
+				\heb{וְלֹֽא־תִכָּרֵ֥ת הָאָ֖רֶץ בָּרָעָֽב׃}
+				\trailingcitation{(Gen 41:36 ESV)}
+				\gll w\schwa=lō\glot{} \b{t}ikkārē\b{t} hā=\glot{}āreṣ \imp{b}=ā=rā\phar{}ā\b{b} \\
+				and=not {shall be cut off ({\sc mid})} the=land \imp{in}=the=famine \\
+				\glt `(That food shall be a reserve \dots), so that the land may not perish \imp{through} the famine.'
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	We argue that \gls{min} `from' marks {\sc dominance} or {\sc full control},
+		while \gls{b\schwa} `in, by, against' is used for entities that are not in full control:
+
+	\begin{itemize}
+		\item In (\ref{instrument-be}), the stones / sword are Instruments of other entities (God and the Israelites): no full control.
+		\item In (\ref{instrument-min}), the blessing is so powerful that it lasts forever, hence it is dominant.
+		\item In (\ref{reason-min}), ``at once'', ``full length'', and ``very'' indicate the complete fear that overcomes Saul: full control.
+		\item In (\ref{reason-be}), the famine is not a dominant factor since the land has built up reserves.
+	\end{itemize}
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{Biblical Hebrew: more examples}
+
+	When a reason for joy is given, \gls{b\schwa} `in' is often used,
+		while \gls{min} is used for reasons for fear.
+	This is understandable, since fear is usually something you are overcome by,
+		while joy is something that is more easily controlled.
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex
+				\heb{שִׂמְחוּ֙ בַּאֲבִימֶ֔לֶךְ וְיִשְׂמַ֥ח גַּם־ה֖וּא בָּכֶֽם׃}
+				\trailingcitation{(Judges 9:19 ESV)}
+				\gll śimḥû \imp{ba}=\glot\upa\b{b}îmele\b{k} w\schwa=yiśmaḥ gam hû\glot{} \imp{bā}=\b{k}em \\
+				rejoice \imp{in}=Abimelech and={let rejoice} also he \imp{in}=you \\
+				\glt `rejoice \imp{in} Abimelech, and let him also rejoice \imp{in} you.'
+			\ex
+				\heb{וְלִשְׂמֹ֖חַ בַּעֲמָל֑וֹ זֹ֕ה מַתַּ֥ת אֱלֹהִ֖ים הִֽיא׃}
+				\trailingcitation{(Ecclesiastes 5:19 ESV)}
+				\gll w\schwa=li=śmōaḥ \imp{ba}=\phar\upa{}māl-ô zô mattat \glot\upe{}lōhîm hî\glot{} \\
+				and=to=rejoice \imp{in}=toil-his this {gift of} God she \\
+				\glt `(\dots) and [to] rejoice \imp{in} his toil---this is the gift of God.'
+		\end{xlist}
+
+		\ex
+		\begin{xlist}
+			\ex
+				\heb{וַיַּסְתֵּ֤ר מֹשֶׁה֙ פָּנָ֔יו כִּ֣י יָרֵ֔א מֵהַבִּ֖יט אֶל־הָאֱלֹהִֽים׃}
+				\trailingcitation{(Exodus 3:6 ESV)}
+				\gll wayyastēr mōš\=ę pān-āw kî yārē\glot{} \imp{mē}=habbîṭ \glot{}el hā=\glot\upe{}lōhîm \\
+				{and hid} Moses face-his because feared \imp{from}=look.{\sc inf} to the=God \\
+				\glt `And Moses hid his face, for he was afraid to look at God.'
+			\ex
+				\heb{וְהִנֵּ֥ה קָרַ֖ן ע֣וֹר פָּנָ֑יו וַיִּֽירְא֖וּ מִגֶּ֥שֶׁת אֵלָֽיו׃}
+				\trailingcitation{(Exodus 34:30 ESV)}
+				\gll w\schwa=hinnēh qāran \phar{}ôr pān-āw wayyîr\schwa\glot{}û \imp{mig}=geše\b{t} \glot\=el-āw \\
+				and=look shone skin face-his {and they feared} \imp{from}=approach.{\sc inf} to-him \\
+				\glt `and behold, the skin of [Moses's] face shone, and they were afraid to come near him.'
+		\end{xlist}
+	\end{exe}
+
+	Only \gls{min} `from' can render something impossible.
+	This is understandable if only \gls{min} has enough control to completely rule something out:
+
+	\begin{exe}
+		\ex
+			\heb{הַרְבָּ֥ה אַרְבֶּ֖ה אֶת־זַרְעֵ֑ךְ וְלֹ֥א יִסָּפֵ֖ר מֵרֹֽב׃}
+			\trailingcitation{(Genesis 16:10 ESV)}
+			\gll harbâ \glot{}arb\=ę \glot{}e\b{t} zar\phar\=e-\b{k} w\schwa=lō\glot{} yissā\=pēr \imp{mē}=rō\b{b} \\
+			multiply.{\sc inf-abs} {I will multiply} {\sc obj} seed-yours and=not {it will be countable} \imp{from}=multitude \\
+			\glt `I will surely multiply your offspring so that they cannot be numbered \imp{for} multitude.'
+	\end{exe}
+
+	And in Isaiah 28:7, \gls{b\schwa} and \gls{min} are used,
+		together with \gls{yayin} `wine' / \gls{šē\b{k}ār} `strong drink' and various verbs,
+		to build a climax:
+
+	\begin{exe}
+		\ex
+		\begin{xlist}
+			\ex \gls{ŠGH \imp{b\schwa{}} yayin} `go astray in wine';
+			\ex \gls{T\phar{}H \imp{b\schwa{}} šē\b{k}ār} `stagger in strong drink';
+			\ex \gls{ŠGH \imp{b\schwa{}} šē\b{k}ār} `go astray in strong drink';
+			\ex \gls{BL\phar{} \imp{min} yayin} `be numbed from wine';
+			\ex \gls{T\phar{}H \imp{min} šē\b{k}ār} `stagger from strong drink'.
+		\end{xlist}
+	\end{exe}
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{Comparison with French}
+
+	At first, these results seem to conflict with what we said for French.
+	In Hebrew the more distal preposition marks higher dominance / control,
+		while in French the more distal preposition marked less influence:
+
+	\begin{exe}
+		\ex
+			Biblical Hebrew:
+
+			\begin{tabular}[t]{llllll}
+				\emph{Causation:} & Dominance / full control & $\rightarrow$ & Less control & $\rightarrow$ & Patient\\
+				                  & \gls{min} `from' &                       & \gls{b\schwa} `in, by, against' \\
+				\emph{Space:}     & Ablative &                               & Locative/Proximative \\
+			\end{tabular}
+	\end{exe}
+\end{step}
+
+\begin{step}[list]
+	\begin{exe}
+		\ex
+			French:
+
+			\begin{tabular}[t]{llllll}
+				\emph{Causation:} & No influence & $\rightarrow$ & Influence & $\rightarrow$ & Patient\\
+				                  & \gls{de} `from' &            & \gls{par} `through' \\
+				\emph{Space:}     & Ablative &                   & Perlative \\
+			\end{tabular}
+	\end{exe}
+\end{step}
+
+\begin{step}
+	Still, both languages can be understood using \citeauthor{Croft2012}'s (\citeyear{Croft2012}) Space $\Rightarrow$ Causation metaphor.
+	It is logical if the Ablative marks more control,
+		because the entity at the start of the causal chain is not controlled itself (Biblical Hebrew).
+	But it is also logical if the Ablative marks less influence,
+		because the entity is not necessarily in direct contact with the Patient (French).
+\end{step}
+
+\begin{step}
+	We can explain these seemingly contradictory results with different perspectives on the causal chain:
+
+	\begin{itemize}
+		\item
+			French has a {\sc Patient-oriented perspective}:
+				in this language, \gls{de} and \gls{par} developed causal semantics
+				through recycling the distance they express to the Patient.
+			It is not crucial that \gls{de} marks the absolute beginning of the causal chain,
+				but the relative proximity to the Patient is important.
+
+		\item
+			Biblical Hebrew has a {\sc Origin-oriented perspective}:
+				\gls{b\schwa} and \gls{min} developed causal semantics
+				based on the distance to the beginning of the causal chain, rather than the Patient.
+			It is crucial that \gls{min} points to the absolute beginning of the causal chain
+				for it to express dominance or full control.
+	\end{itemize}
+\end{step}
+
+\begin{step}[section]
+	\section{Conclusion}
+
+	\subsection{Formalizing Agent prepositions}
+	Common formal theories of Agent prepositions take the Agent preposition as a purely functional element~(e.g.\ \cite{AngelopoulosEtAl2020}).
+	They have trouble accounting for languages with multiple Agent prepositions,
+		where the spatial meaning of these prepositions has not been lost entirely~(cf.\ \cite{Croft2012}).
+
+	Force dynamics~\citep{CopleyEtAl2020} provides a formalism that remains closer to intuitive thinking about causation.
+	The distinctions it makes between entity, force, and situation are useful to formalize differences between Agent prepositions.
+	We have proposed a way to implement Agent prepositions à la \citet{AngelopoulosEtAl2020} in the framework of \citet{CopleyEtAl2020}.
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{\texorpdfstring{Making the Space $\Rightarrow$ Causation metaphor more fine-grained}{Making the Space => Causation metaphor more fine-grained}}
+	\citeauthor{Croft2012} only distinguishes Antecedent and Subsequent Obliques,
+		and actively warns against going any further:
+
+	\begin{quote}
+		Although \imp{one cannot predict which participant roles a specific Oblique case marking will subsume}---%
+			case markers are usually quite polysemous---%
+			one can predict that a specific Oblique case marking will subsume only antecedent roles or only subsequent roles.
+		That is, \imp{one can generally categorize Oblique morphosyntactic markers as either Antecedent or Subsequent}, as in (\ref{antecedent})--(\ref{subsequent}).
+		\trailingcitation{\citep[223, emphasis added]{Croft2012}}
+	\end{quote}
+
+	There is indeed no universal mapping between spatial categories and causal concepts.
+	However, we \emph{can} make the metaphor more fine-grained when looking at a concrete language:
+		\imp{relative position in space} mirrors \imp{relative position in the causal chain}.
+	Thus we can describe the system of French as ``proximity to Patient is influence'',
+		and the system of Biblical Hebrew as ``proximity to Origin is control''.
+\end{step}
+
+\begin{step}
+	We can do the same for English:
+
+	\begin{exe}
+		\exr{antecedent}
+		\begin{xlist}
+			\ex Cause: \gls{The rabbit died \imp{from}/\imp{of} thirst.}
+				\trailingcitation{\citep[223]{Croft2012}}
+			\ex Agent: \gls{The cat food was eaten \imp{by} raccoons.}
+			\ex Means: \gls{I went downtown \imp{by} bus.}
+			\ex Instrument: \gls{Sue broke the coconut \imp{with} a hammer.}
+			\ex Comitative: \gls{I went to the park \imp{with} Carol.}
+		\end{xlist}
+
+		\exr{subsequent}
+		\begin{xlist}
+			\ex Result: \gls{They smashed the statue \imp{to} pieces.}
+				\trailingcitation{\citep[223]{Croft2012}}
+			\ex Result: \gls{The boy carved the stick \imp{into} a knife.}
+			\ex Beneficiary: \gls{Sue broke the coconut \imp{for} Greg.}
+		\end{xlist}
+
+		\ex
+		\begin{tabular}[t]{llllll}
+			Cause $\>\rightarrow$ & Agent $\>\rightarrow$ & Comitative/Instrument $\>\rightarrow$ & Patient $\>\rightarrow$ & Result $\>\rightarrow$ & Beneficiary \\
+			\gls{of, from}        & \gls{by}              & \gls{by, with}                        &                         & \gls{to, into}         & \gls{for} \\
+			Ablative              & Proximative           & Closer proximative                    &                         & Allative               & Closer allative \\
+		\end{tabular}
+	\end{exe}
+
+	{\footnotesize
+		(For \enquote{closer proximative}, consider the difference between \gls{the house by/with the lake}.
+		For \enquote{closer allative}, consider that \gls{(be)for(e)} is the endpoint of the direction expressed by \gls{to}.)}
+\end{step}
+
+\begin{step}[subsection]
+	\subsection{Perspectives on the causal chain}
+	Furthermore, we suggested that languages may have different perspectives on the causal chain.
+	We identified the Patient-oriented perspective of French and the Origin-oriented perspective of Biblical Hebrew.
+	This makes \citeauthor{Croft2012}'s~(\citeyear{Croft2012}) theory more predictive while remaining falsifiable.
+\end{step}
+
+\begin{step}
+	Natural follow-up questions are:
+
+	\begin{enumerate}
+		\item Are there other perspectives besides Origin-oriented and Patient-oriented?
+		\item Is the perspective a language-wide parameter or can different prepositions in the same language take a different perspective?
+		\item What determines the perspective of a language (or of a preposition)?
+	\end{enumerate}
+\end{step}
+
+\clearpage
+\begin{step}[section]
+	\vspace{.8em}
+	\setstretch{1}
+	\def\enotesize{\normalsize}
+	\theendnotes
+
+	\printbibliography
+\end{step}
+
+\end{document}
diff --git a/resources/pdf/sbl2022.pdf b/resources/pdf/sbl2022.pdf
new file mode 100644
index 0000000..3d34125
--- /dev/null
+++ b/resources/pdf/sbl2022.pdf
diff --git a/resources/pug/finals/index.pug b/resources/pug/finals/index.pug
index 0a3fb10..92aa446 100644
--- a/resources/pug/finals/index.pug
+++ b/resources/pug/finals/index.pug
@@ -166,12 +166,8 @@ block content
 					h4 Upcoming
 			tr
 				td 2022
-				td Nov 9, #[a(href="https://bcopley.com/cocoa", title="Converging On Causal Ontology Analyses") COCOA]
-				td With Johan Rooryck. Space in the causal chain: The perspective from French and Biblical Hebrew.
-			tr
-				td
 				td Nov 21, #[a(href="https://www.sbl-site.org/meetings/AnnualMeeting.aspx", title="Society of Biblical Literature") SBL Annual]
-				td With Martijn Beukenhorst. Semantic properties of prepositions: the distinction between causal מן and ב.
+				td With Martijn Beukenhorst. Semantic properties of prepositions: the distinction between causal מן and ב (#[a(href="/assets/pdf/sbl2022.pdf") slides]).
 			tr
 				td
 				td Dec 9, #[a(href="https://benecla.com/", title="CogLing Days") CogLing Days]
@@ -182,6 +178,10 @@ block content
 					h4 Past
 			tr
 				td 2022
+				td Nov 9, #[a(href="https://bcopley.com/cocoa", title="Converging On Causal Ontology Analyses") COCOA]
+				td With Johan Rooryck. Space in the causal chain: The perspective from French and Biblical Hebrew (#[a(href="/assets/pdf/cocoa2022.pdf") handout]).
+			tr
+				td
 				td Aug 25, #[a(href="https://societaslinguistica.eu/sle2022/", title="Societas Linguistica Europaea") SLE2022]
 				td With Johan Rooryck. On the demonstratives nature of finite complementizers (#[a(href="/assets/pdf/sle2022.pdf") slides]; #[a(href="/assets/pdf/sle2022-handout.pdf") handout]).
 			tr