anton-nicole.tex

\chapter{\emph{Irgendein} quantifier and local implicature}      % <=== TITLE HERE
\authorAndAffil{Anton Benz and Nicole Gotzner}{Leibniz-Zentrum Allgemeine Sprachwissenschaft (ZAS)}   % AUTHOR
	%	\addsnippetauthor
		%\snippetauthor{AUTHOR4}{AUTHOR4AFFILIATION}
		%\addsnippetauthor
		%\snippetauthor{AUTHOR5}{AUTHOR5AFFILIATION}
%   
% \usepackage{mathptmx}
% \usepackage{parskip}
% \usepackage{booktabs}


\newcommand{\lineEx}[1]{\emph{#1}}

\newcommand{\spc}{\ensuremath{\:\&\:}}

\newcommand{\utta}[1]{#1}
\newcommand{\uttb}[2]{#1\spc#2}
\newcommand{\uttc}[3]{#1\spc#2\spc#3}



	

This snippet focuses on a puzzle concerning the alternatives associated with the German existential quantifiers \emph{einige} (\emph{some}) and \emph{irgendeine} (\emph{any}) under embedding. In particular, we show that the two quantifiers trigger different implicatures in upward-entailing (UE) and downward-entailing (DE) contexts, see (\ref{ex.A-I.a}) and (\ref{ex.N-I.a}), respectively. In short, \emph{einige} and \lineEx{irgendeine} trigger the same \emph{not all} implicature when embedded under \emph{all} (UE), but behave differently from each other when embedded under \emph{no} (DE). While \emph{einige} triggers a local implicature in DE contexts, \lineEx{irgendeine} is interpreted as \lineEx{none}.
%We consider sentences of the form \ref{ex.A-I} and \ref{ex.N-I}. German \lineEx{irgendeine} can occur in UE contexts, as in \ref{ex.A-I}, unlike English \emph{any}.  The German counterpart to \lineEx{some}, \lineEx{einige}, can also occur in DE contexts \ref{ex.N-I}, although it is slightly marked here. 

 %\citet[p.~255]{Chi13} observed that \lineEx{irgendeine} has \lineEx{an existential FC behavior in positive, modalized contexts, and an NPI behavior in DE contexts.}  \emph{Irgendeine} behaves similarly to English \lineEx{any} in many respects. One difference, however, is that \emph{Irgendeine} can occur in upward entailing (UE) contexts, as in \ref{ex.A-I}. The German counterpart to \lineEx{some}, \lineEx{einige}, can also occur in (DE) contexts, in contrast to English \lineEx{some}, as in \ref{ex.N-I}, although it is slightly marked here. 

\begin{exe}
\ex \label{ex.A-I}
\begin{xlist}
 \ex Jedes der M\"adchen fand \emph{einige/irgendeine} ihrer Murmeln. (\utta{A-E}/\utta{A-I})\label{ex.A-I.a}
 \ex Each of the girls found \emph{some/any} of her marbles.
\ex \({\leadsto}\) Each of the girls found \emph{some but not all} of her marbles.\label{ex.A-I.c}
\end{xlist}		
\ex \label{ex.N-I}
\begin{xlist}
          \ex Keines der M\"adchen fand \emph{einige/irgendeine} ihrer Murmeln. (\utta{N-E}/\utta{N-I})\label{ex.N-I.a}
		  \ex None of the girls found \emph{some/any} of her marbles.
		  \ex \(\stackrel{?}{\leadsto}\) None of the girls found \emph{some but not all} of her marbles.\label{ex.N-I.c}
\end{xlist}
	%\end{xlist}
\end{exe}

%According to the neo--Gricean account \citep{Hor89,Lev83}, the scalar implicature of a sentence like \lineEx{Some of the girls found all of their marbles} (\utta{E-A}) is inferred by first replacing \lineEx{some} with a scalar alternative, and then negating the resulting sentence if it is logically stronger. Hence, \utta{E-A} implicates that not \emph{all} of the girls found all of their marbles (not \utta{A-A}). 

In so--called \emph{grammatical} accounts \citep{Fox07,Chi13}, an embedded occurrence of a scalar element may trigger local calculation of alternatives and their negation. If \lineEx{einige} / \lineEx{irgendeine} trigger a local implicature, then the readings in (\ref{ex.A-I.c}) and (\ref{ex.N-I.c}) result.
%For example, the utterance \lineEx{All of the girls found some of their marbles} (\utta{A-E}) can give rise to the implicature that all found \emph{some, but not all}. %(\utta{A-ENA}). 
What counts as an alternative to these kinds of sentences is a central issue to all theories that emerged from the neo-Gricean account of implicature \citep{Hor89,Sau04}.
 
\citet{GoBe17} have developed an experimental paradigm, in which participants systematically derived the embedded implicature in the case of \utta{A-E}. In an interactive version of this paradigm \citep*{BeGoRa18}, we collected a large corpus of production and interpretation data on German sentences of the form \lineEx{Q of the girls found Q' of their marbles}. Q and Q' could be, among others, \emph{einige (some), alle/jedes (all/every), keines (none),} and \emph{irgendeine (any)} (for details see \texttt{https://osf.io/qs2vj/}). 

In table 1 the percentages of responses indicating a local \emph{some but not all} reading for interpretation (\%int) and production (\%prod) are shown. As can be seen, when \emph{einige} and \emph{irgendeine} are embedded under \emph{alle} (\emph{all}) (\utta{A-E} and \utta{A-I}, respectively), the interpretation that each girl found some but not all marbles arises consistently for both quantifiers. In the case of embedding under \emph{keine} (\emph{no}), however, interpretations diverge:  \emph{einige} (\utta{N-E}) is predominantly interpreted as \emph{some but not all} while \emph{irgendeine} (\utta{N-I}) never gives rise to a local implicature. Instead, the sentence with \emph{irgendeine} is interpreted as no girl found anything.

The data give rise to the following puzzle about alternatives: If one assumes with \citep[and others]{BuHa17} that \emph{all} is an alternative to \emph{irgendeine}, then this explains why \utta{A-E} and \utta{A-I} trigger the same local implicature, but it leaves unexplained why \utta{N-E} and \utta{N-I} behave so differently. If one assumes that \emph{irgendeine} does not activate the \emph{all} alternative, then the observations about \utta{N-E} and \utta{N-I} follow easily, but the embedded implicature of \utta{A-I} remains unexplained. One may argue that the \emph{not all} implicature of \utta{A-I} is the result of \emph{irgendeine} being singular. Note, however, that participants only saw pictures in which girls had either none or at least 2 out of 4 marbles. Hence, the interpretation \emph{all found one} for \utta{A-I} was contextually blocked.

Another suggestion may be that the inference from \emph{einige} to \emph{not all} has become conventionalized, whereas that from \emph{irgendeine} to \emph{not all} is a true implicature that is blocked in DE contexts. 
However, \citet{BeGoRa18} show cases in which \emph{einige} in UE contexts fails to produce the expected implicature. Solving the puzzle on alternatives associated with German \emph{irgendeine} may also require further investigation of its distribution and NPI behaviour, which seems to differ from that of English \emph{any}.

\begin{table}[!htb]
\begin{center}
\begin{tabular}{@{}ccccc@{}}
    \toprule
        & \utta{N-E} & \utta{N-I} & \utta{A-E}  & \utta{A-I} \\
        \midrule
     \#int/\#prod & 35/15 & 98/77 & 76/50 & 4/7 \\
     \%int & 77 & 1 & 91 & 100 \\
     \%prod & 100 & 0 & 96 & 100 \\
     \bottomrule
    \end{tabular}
    \caption*{Table 1\\
      \textbf{\#int}: abs.\ number of interpreted utterance tokens.\\
    \textbf{\#prod}: abs.\ number of produced utterance tokens.
    }\label{res.exp}
\end{center}
\end{table}
\vspace{-1\baselineskip}

\bibliographystyle{linquiry2.bst}
\bibliography{subs/sub5}

\begin{description}
\item[Acknowledgements] This work was supported by the Bundesministerium für Bildung und Forschung (BMBF) (Grant Nr. 01UG1411), and the Deutsche Forschungsgemeinschaft (DFG) (Grant Nr. BE 4348/4-1 and BE 4348/4-2).
\end{description}