Possible Transcript Items

The following table gives a rough overview about possible labels and tags contained in the orthographic transcript^8.8. You may review the following tags and decide which of them might be useful for your special needs. In the third column you see an example of how the items may be tagged. Of course you may also use an XML-style tagging instead.

Assume for the following list of tags that a dialogue between two or more speakers is transcribed turn by turn by listening to the signals.

Item	Remarks	Example
Lexical unit	Standardized spelling/character coding. Define a lexical unit: words, interjections, neologisms? Lexical units are usually not tagged.	station
Spelling	Spelling of a word or abbreviation letter by letter.	$U $S $A
Acronyms	Official substitutes for words or phrases, spelled like a word	OPEC
Proper names	All names that cannot be translated into another language: People's names, street names, restaurants etc.	~Peter ~Marine+World
Numbers	Numerals, combinations of numbers and ordinal numbers. All number written as words	#three #twenty #hundred
Neologism	Word that has been made up by the speaker	*deliverator
Foreign Words	Words that are from another language and have not been officially adopted by the main language	$<$*IT$>$saluti
Off-Talk	person is speaking to himself or herself and not to the partner(s) of the dialogue	what$<$OOT$>$ do$<$OOT$>$ I$<$OOT$>$
Read Off-Talk	Off-talk caused by reading aloud	seven$<$ROT$>$
Command Words	Words to operate a dialogue system	!KEYComputer
Lengthening	Markup of sounds within an item that are lengthened	so$<$L$>$rry
Garbage	words completely or partly incomprehensible	$<$%$>$ three%
Truncation	Item is truncated for several reasons (technical, stutter etc.)	so the que= by hel= $<$*T$>$
Interruption	Items may be interrupted for several reasons: pauses, breathing, hesitations etc.	trans_ $<$A$>$ _lation
Missing signal	Missing parts of the signal for technical reasons have to be marked in the transcript.	see

Unusual Pronunciation	Slang, dialect, contractions, assimilations or mispronunciations. May either be marked in orthographic or phonetic transcript. It is important to keep the `correct' orthographic form to allow lexical mapping.	no $<$!1 nope$>$ going to $<$!2 gonna$>$
Repetition	Stutter of parts or complete items.	like +/to/+ to see
False start	Breaking off an utterance and starting a new one.	-/this evening/- tomorrow
Breathing	Clearly audible breathing	$<$A$>$
Filled Pauses	Pauses filled with vocalization or nasalization or a combination or other articulatory noises with the same intention	$<$uh$>$ $<$hm$>$ $<$uhm$>$ $<$hes$>$
Empty Pause	Temporary unfilled gaps in the speech. Usually not marked at the beginning or the end of a recording.	$<$P$>$
Articulatory Noise	Noise produced by the articulatory system of a speaker but no filled pause.	$<$noise$>$ $<$cough$>$ $<$laugh$>$ $<$smack$>$ ...
Other noise	Noise caused by background events, by touching the microphone, by the recording equipment etc.	$<$#$>$ $<$#microtouch$>$ $<$#knock$>$ $<$#hum$>$ ...
Cross talk	Overlapping speech caused by speakers interrupting each other. If information is needed about who is interrupting whom and where, this can be rather complicated.	see
Overlay	Overlay of noise or crosstalk may be marked by using a bracket system. Recommendation: tag each overlayed word individually.	here $<$:$<$#$>$ you:$>$ are
Prosody	Prosodic events like emphasis, main and secondary phrase accent and boundaries may be marked up in the transcript.	[PA] [NA] [B3] [B9]