Description of problem:
Starting espeak, then piping some text to espeak STDIN. The text is spoken, but espeak dies after the last word of the text has been spoken.

Debug info:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffeceff700 (LWP 21993)]
0x00007ffff78855fe in LookupDict2(Translator*, char const*, char const*, char*, unsigned int*, int, WORD_TAB*) () from /lib64/libespeak.so.1
(gdb) tb
Temporary breakpoint 1 at 0x7ffff78855fe
(gdb) bt
#0  0x00007ffff78855fe in LookupDict2(Translator*, char const*, char const*, char*, unsigned int*, int, WORD_TAB*) () from /lib64/libespeak.so.1
#1  0x00007ffff7885f0e in LookupDictList(Translator*, char**, char*, unsigned int*, int, WORD_TAB*) () from /lib64/libespeak.so.1
#2  0x00007ffff78861cb in Lookup(Translator*, char const*, char*) () from /lib64/libespeak.so.1
#3  0x00007ffff78923c8 in LookupLetter(Translator*, unsigned int, int, char*, int) () from /lib64/libespeak.so.1
#4  0x00007ffff7892872 in TranslateLetter(Translator*, char*, char*, int) () from /lib64/libespeak.so.1
#5  0x00007ffff78992ba in SpeakIndividualLetters(Translator*, char*, char*, int) () from /lib64/libespeak.so.1
#6  0x00007ffff789b1a7 in TranslateWord(Translator*, char*, int, WORD_TAB*, char*) () from /lib64/libespeak.so.1
#7  0x00007ffff789c05e in TranslateWord2(Translator*, char*, WORD_TAB*, int, int) () from /lib64/libespeak.so.1
#8  0x00007ffff789dc90 in TranslateClause(Translator*, _IO_FILE*, void const*, int*, char**) () from /lib64/libespeak.so.1
#9  0x00007ffff7898fe3 in SpeakNextClause(_IO_FILE*, void const*, int) () from /lib64/libespeak.so.1
#10 0x00007ffff787eadb in Synthesize(unsigned int, void const*, int) [clone .isra.1] () from /lib64/libespeak.so.1
#11 0x00007ffff787ef9c in sync_espeak_Synth(unsigned int, void const*, unsigned long, unsigned int, espeak_POSITION_TYPE, unsigned int, unsigned int, void*)
    () from /lib64/libespeak.so.1
#12 0x00007ffff78a9aac in process_espeak_command(t_espeak_command*) () from /lib64/libespeak.so.1
#13 0x00007ffff78aa790 in say_thread(void*) () from /lib64/libespeak.so.1
#14 0x00007ffff6b4bd15 in start_thread () from /lib64/libpthread.so.0
#15 0x00007ffff709b48d in clone () from /lib64/libc.so.6


Version-Release number of selected component (if applicable):


How reproducible:
espeak-1.47.03-1.fc18.x86_64

Steps to Reproduce:
1.

Start espeak inside a gnome-terminal, with no params, and send the following text to STDIN (I read it on http://en.wikipedia.org/wiki/Nasalization, chapter "Other nasal(ized) consonants":), copied with ^C, pasted with Ctrl+Shift+V to
the gnome-terminal:


Nasalized versions of other consonant sounds also exist, though they are much rarer than either nasal occlusives or nasal vowels. Some of the South Arabic languages have phonemic nasalized fricatives, such as /z̃/, which sounds something like a simultaneous [n] and [z]. The sound written ⟨r⟩ in Mandarin has an odd history; for example, it has been borrowed into Japanese as both [z] and [n]. It seems likely that it was once a nasalized fricative, perhaps a palatal [ʝ̃]. In the Hupa velar nasal /ŋ/, the tongue often does not make full contact, resulting in a nasalized approximant, [ɰ̃]. This is cognate with a nasalized palatal approximant [ȷ̃] in other Athabaskan languages. In Umbundu, phonemic /ṽ/ contrasts with the (allophonically) nasalized approximant [w̃], and so is likely to be a true fricative rather than an approximant.[further explanation needed] In Old and Middle Irish, the lenited ⟨m⟩ was a nasalized bilabial fricative.[1] Sundanese has an allophonic nasalized glottal stop [ʔ̃]; nasalized stops can only occur with pharyngeal articulation or lower, or else they'd be simple nasals.[2] Nasal flaps are common allophonically. Many West African languages have a nasal flap [ɾ̃] (or [n̆]) as an allophone of /ɾ/ before a nasal vowel; Pashto, however, has a phonemic nasal retroflex lateral flap.

2. 
3.
  
Actual results:

After the last word "flap" is spoken, espeak crashes

Expected results:
No crash!

Additional info: