\documentclass[11pt]{article}
\usepackage[utf8x]{inputenc}
\usepackage[hebrew,english]{babel}

\newcommand{\heb}[1]{\begin{small}\R{#1}\end{small}}
\newcommand{\hebt}[1]{\begin{scriptsize}\R{#1}\end{scriptsize}}


\begin{document}
\section{\R{הפסקה הבאה מכילה טקסט בעברית}}
The method is based on language-specific rules for constructing a
\emph{similar words} (SW) set for each analysis of a word.  This set is composed
of morphological variations of the word under the given analysis. For example, 
the Hebrew token \heb{ילד} can be analyzed as either a noun (boy) or a verb 
(gave birth).  The \texttt{noun} SW set for this token is composed 
of the definiteness and number inflections \heb{ילד, הילד, הילדים} (the boy, boys, the boys), 
while the \texttt{verb} SW set is composed of gender and tense inflections
\heb{ילדה, ילדו} (she/they gave birth).  The approximated probability of each analysis 
is based on the corpus frequency of its SW set.  For the complete details,
refer to the original paper.

\end{document}