User-Oriented Design for a User-Oriented Design for a Morphologically-Based Morphologically-Based Transcription Module Transcription Module processing processing Emotionally and Socio- Emotionally and Socio- culturally “Marked” Elements in culturally “Marked” Elements in Spoken Journalistic Texts Spoken Journalistic Texts Dr. Christina Alexandris Dr. Christina Alexandris Lecturer Lecturer , , Linguistics - Computational Linguistics Linguistics - Computational Linguistics National University of Athens National University of Athens email:[email protected]email:[email protected]
56
Embed
User-Oriented Design for a Morphologically-Based Transcription Module processing Emotionally and Socio-culturally “Marked” Elements in Spoken Journalistic.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
User-Oriented Design for a User-Oriented Design for a Morphologically-Based Morphologically-Based Transcription Module Transcription Module
processing processing
Emotionally and Socio-culturally Emotionally and Socio-culturally “Marked” Elements in Spoken “Marked” Elements in Spoken
Journalistic TextsJournalistic Texts
Dr. Christina AlexandrisDr. Christina Alexandris
LecturerLecturer, , Linguistics - Computational Linguistics Linguistics - Computational Linguistics National University of Athens National University of Athens email:[email protected] email:[email protected]
IntroductionIntroduction
Speaker and culture-dependent elements Speaker and culture-dependent elements constituting emotionally and socio-culturally constituting emotionally and socio-culturally
“marked” elements“marked” elements• such as such as assumed informationassumed information (Alexandris, (Alexandris,
2003), 2003), • expressionsexpressions • and the special usage of and the special usage of linguisticlinguistic and and
paralinguistic elementsparalinguistic elements, ,
is essential for the detection and definition of is essential for the detection and definition of speakers’ or journalists’ opinions and/or speakers’ or journalists’ opinions and/or viewpoints, the perception of irony and the viewpoints, the perception of irony and the discovery of additional, often background discovery of additional, often background information. information.
The absence of such information in The absence of such information in processed outputprocessed output
hindershinders the the full understandingfull understanding of of the content of the texts the content of the texts
and in some cases may even create and in some cases may even create a a distancedistance between the between the international public and the news international public and the news related to the foreign language(s) related to the foreign language(s) concernedconcerned
[Bildunterschrift: Ma [Bildunterschrift: Ma Canrong: "Enttäuschung Canrong: "Enttäuschung über einseitige über einseitige Berichterstattung der Berichterstattung der hiesigen Medien" ]hiesigen Medien" ]http://www.heute.de/ZDFheute/http://www.heute.de/ZDFheute/
To suggest that an To suggest that an election will be held in election will be held in 2014 is nothing more than 2014 is nothing more than a sham”a sham” Stephen Smith, Stephen Smith, Australian Foreign MinisterAustralian Foreign Minister By Phil Mercer , BBC News, SydneyBy Phil Mercer , BBC News, Sydney
The present approachThe present approach
concerns the integration of an concerns the integration of an annotation strategyannotation strategy in in transcription tools for spoken transcription tools for spoken journalistic texts. journalistic texts.
The proposed annotation strategy The proposed annotation strategy targets to address re-occurring targets to address re-occurring problems encountered by journalists, problems encountered by journalists, economists and other professionals economists and other professionals working with multilingual spoken working with multilingual spoken journalistic texts available from the journalistic texts available from the media and the web.media and the web.
2. User Requirements and 2. User Requirements and SpecificationsSpecifications
2.1 User Profile and Interests2.1 User Profile and Interests
International public: journalists, International public: journalists, economists and other professionalseconomists and other professionals
Average or even a fairly good fluency of a Average or even a fairly good fluency of a foreign languageforeign language
Non-native speakers and/or often lack the Non-native speakers and/or often lack the necessary exposure to the culture(s) necessary exposure to the culture(s) related to the foreign language(s) related to the foreign language(s) concernedconcerned
Insight in respect to specific issue/event Insight in respect to specific issue/event from an insider’s /outsider’s point of view/ from an insider’s /outsider’s point of view/ from international perspective.from international perspective.
Choose to use or not to use a textChoose to use or not to use a text
2.2 Problems encountered2.2 Problems encountered
Typical indicators (a), (b) and (c) are not reliable Typical indicators (a), (b) and (c) are not reliable and consistent enough to apply in the majority of and consistent enough to apply in the majority of online spoken/written journalistic texts to be used online spoken/written journalistic texts to be used by international professionals. by international professionals.
• Fixed expressions and idioms (a) Fixed expressions and idioms (a) • Paralinguistic elements (b) such as prosody Paralinguistic elements (b) such as prosody
and gestures in spoken texts and multimedia and gestures in spoken texts and multimedia files files
• Journalistic text types (c):Journalistic text types (c): The basic speech The basic speech acts in journalistic texts can be divided into acts in journalistic texts can be divided into four types, namely “Announcement”, four types, namely “Announcement”, “Description”, “Declaration” and “Expression” “Description”, “Declaration” and “Expression” (Alexandris and Fotinea 2004). (Alexandris and Fotinea 2004).
Fixed expressions and idioms (a)Fixed expressions and idioms (a)
do not occur often enough in texts to do not occur often enough in texts to signalize a point of view or attitude signalize a point of view or attitude conveyed. conveyed.
are usually easily detected by a non-native are usually easily detected by a non-native speaker with a fair command of the speaker with a fair command of the language concerned. language concerned.
are also easily accessed at specialized are also easily accessed at specialized dictionaries for idiomatic expressions and dictionaries for idiomatic expressions and other printed or online resources.other printed or online resources.
Paralinguistic elements (b) such as Paralinguistic elements (b) such as prosody and gesturesprosody and gestures
may be absent in some multimedia files, for may be absent in some multimedia files, for instance in a cases where the speaker (for, instance in a cases where the speaker (for, example, a correspondent) is not visible.example, a correspondent) is not visible.
in languages such as UK English or Standard in languages such as UK English or Standard German, it is observed that speakers do not German, it is observed that speakers do not always demonstrate obvious changes in prosody always demonstrate obvious changes in prosody when making subtle ironic statements, changes when making subtle ironic statements, changes not easily detected by an international audience.not easily detected by an international audience.
may be misinterpreted or overlooked by an may be misinterpreted or overlooked by an international audience, unless extreme emotions international audience, unless extreme emotions are expressed. are expressed.
Journalistic text types (c) Journalistic text types (c)
speakers or journalists may implicitly speakers or journalists may implicitly express their opinions or even their express their opinions or even their feelings in contexts appearing to be feelings in contexts appearing to be “Announcement” or “Description” “Announcement” or “Description” speech acts. speech acts.
Titles do not always give an idea of Titles do not always give an idea of the degree of neutrality/point of viewthe degree of neutrality/point of view
2.3 Specifications2.3 Specifications
““breadth-first search” decision-making breadth-first search” decision-making process process • to use or not to use an online spoken/written to use or not to use an online spoken/written
journalistic textjournalistic text ““depth-first search” decision-making depth-first search” decision-making
process process • evaluation of the text contentevaluation of the text content
3. Design3. Design The aim of the present user-oriented design is its The aim of the present user-oriented design is its
integration in a transcription program targeting integration in a transcription program targeting (1) to facilitate the user’s decision of actually (1) to facilitate the user’s decision of actually
using the text, depending on its level of using the text, depending on its level of emotionally and socio-culturally “marked” emotionally and socio-culturally “marked” elements, elements,
(2) to allow the user to compare “marked” (2) to allow the user to compare “marked” elements with prosodic and paralinguistic elements with prosodic and paralinguistic features in the respective multimedia files and features in the respective multimedia files and
(3) to indicate the largest possible percentage of (3) to indicate the largest possible percentage of the points in the texts signalizing “marked” the points in the texts signalizing “marked” information. information.
The interface of the transcription The interface of the transcription modulemodule
is designed to (a) annotate and highlight is designed to (a) annotate and highlight all the points possibly containing “marked” all the points possibly containing “marked” information, alerting the user to evaluate information, alerting the user to evaluate the parts of the text containing these the parts of the text containing these expressions and expressions and
(b) to produce a score, according to the (b) to produce a score, according to the number of instances of “marked” number of instances of “marked” information detected. information detected.
The module is designed to be able to be The module is designed to be able to be compatible and to operate with most compatible and to operate with most commercial transcription tools,commercial transcription tools, some of which are available online. some of which are available online.
Example of a transcription tool: Example of a transcription tool: www.visualsubsync.orgwww.visualsubsync.org
““special feature for people who's native special feature for people who's native
language is different from the original videolanguage is different from the original video””
Released on Feb 26th, NYTimes offers an interactive user Released on Feb 26th, NYTimes offers an interactive user interface for analyzing the most recent Democratic Debateinterface for analyzing the most recent Democratic Debate: :
Video Transcript:Video Transcript:
Transcript Analyzer: Transcript Analyzer:
The user activates the proposed module when The user activates the proposed module when
The online journalistic texts are scanned by the The online journalistic texts are scanned by the proposed module and the user is presented with two proposed module and the user is presented with two types of output: types of output: • (1) a score, according to the number of instances of (1) a score, according to the number of instances of
“marked” information detected in the online text “marked” information detected in the online text and and
• (2) the actual online journalistic text with all the (2) the actual online journalistic text with all the instances of “marked” information highlighted with instances of “marked” information highlighted with a different color, format or any other type of a different color, format or any other type of indication. indication.
The score is designed to be presented automatically The score is designed to be presented automatically next to the title of the text or file once the proposed next to the title of the text or file once the proposed module is activated. module is activated.
The user may choose to view or not to view the online The user may choose to view or not to view the online journalistic text with the highlighted “marked” journalistic text with the highlighted “marked” information.information.
ScoreScore
Texts with a low score may be marked Texts with a low score may be marked with a distinct color, for instance, green, with a distinct color, for instance, green, texts with a medium score, with orange, texts with a medium score, with orange, and texts with a high score may be and texts with a high score may be marked with red. marked with red.
Texts indicated in a red are not only Texts indicated in a red are not only related to high instance of “marked” related to high instance of “marked” elements, but a may also signalize a elements, but a may also signalize a higher probability of particularly sensitive higher probability of particularly sensitive social, political issues and religionsocial, political issues and religion
ExampleExample:: Transcribed Spoken Text with a low Transcribed Spoken Text with a low score score and fragment from textand fragment from text
SCORE: GREEN SCORE: GREEN Joint Press Conference with World Bank Joint Press Conference with World Bank
President, the Hon Robert ZoellickPresident, the Hon Robert Zoellick World Bank Headquarters, Washington World Bank Headquarters, Washington
DC , United States of America DC , United States of America Subject(s): Afghanistan soldiers; G20, Subject(s): Afghanistan soldiers; G20,
Defence Department, Global economyDefence Department, Global economy http://www.pm.gov.au/node/5248http://www.pm.gov.au/node/5248
ExampleExample JOURNALIST:JOURNALIST: Mr Rudd, a domestic question. Joel Mr Rudd, a domestic question. Joel
Fitzgibbon is allegedly being investigated by his own Fitzgibbon is allegedly being investigated by his own department. What do you think about the behaviour department. What do you think about the behaviour if it’s correct and what are you going to do about it?if it’s correct and what are you going to do about it?
PM:PM: I’ve only just become aware of these reports as I’ve only just become aware of these reports as they’ve just emerged in, as I understand, the they’ve just emerged in, as I understand, the morning newspapers in Australia. I’m advised that morning newspapers in Australia. I’m advised that neither the Secretary of the Defence Department nor neither the Secretary of the Defence Department nor the Chief of the Defence Force staff have received the Chief of the Defence Force staff have received any such reports. I’m further advised that the any such reports. I’m further advised that the Secretary of the Defence Department is conducting a Secretary of the Defence Department is conducting a further investigation into this matter. And therefore, I further investigation into this matter. And therefore, I will await the outcome of that investigation […….]will await the outcome of that investigation […….]
JOURNALISTJOURNALIST: There seems to be a difference in : There seems to be a difference in standards between the people on the frontline and standards between the people on the frontline and the (inaudible) in Canberra. Doesn’t that concern the (inaudible) in Canberra. Doesn’t that concern you?you?
ExampleExample:: Transcribed Spoken Text with a Transcribed Spoken Text with a medium score medium score and fragment from textand fragment from text
SCORE: ORANGESCORE: ORANGE Bill Moyers on PBS, The Bill Moyers Bill Moyers on PBS, The Bill Moyers
Journal, Transcript, APRIL 25, Journal, Transcript, APRIL 25, 2007:"Buying the War", 2007:"Buying the War", http://www.pbs.org/moyers/jounral/btw/transcript1.htmlhttp://www.pbs.org/moyers/jounral/btw/transcript1.html
ExampleExample
GEORGE WILL (ABC 10/28/01):GEORGE WILL (ABC 10/28/01): The administration The administration knows he's vowed, Hussein has vowed revenge, he has knows he's vowed, Hussein has vowed revenge, he has anthrax, he loves biological weapons, he has terrorist anthrax, he loves biological weapons, he has terrorist training camps, including 747's to practice on…training camps, including 747's to practice on…
BILL MOYERS:BILL MOYERS: IT WAS PROVING DIFFICULT TO IT WAS PROVING DIFFICULT TO DISTINGUISH THE OPINION OF THE PUNDITS FROM THE DISTINGUISH THE OPINION OF THE PUNDITS FROM THE POLICIES OF THE ADMINISTRATION...BUT AS THE POLICIES OF THE ADMINISTRATION...BUT AS THE HULLABALOO OVER SADDAM GREW IN WASHINGTON, HULLABALOO OVER SADDAM GREW IN WASHINGTON, BOB SIMON OF CBS NEWS 60 MINUTES WAS BOB SIMON OF CBS NEWS 60 MINUTES WAS DUMBFOUNDED. HE IS BASED IN THE MIDDLE EAST. DUMBFOUNDED. HE IS BASED IN THE MIDDLE EAST. [….][….]
BILL MOYERS:BILL MOYERS: Absurdity. The Washington press corps Absurdity. The Washington press corps cannot question an absurdity? cannot question an absurdity?
BILL MOYERS:BILL MOYERS: In a bubble. In a bubble. BOB SIMON:BOB SIMON: Yeah in a bubble. Where as we who've Yeah in a bubble. Where as we who've
spent weeks just walking the streets of Baghdad and in spent weeks just walking the streets of Baghdad and in other situations in Baghdad just were scratching our other situations in Baghdad just were scratching our heads. In ways that perhaps that the Washington press heads. In ways that perhaps that the Washington press corps could not”corps could not”
4. Approach4. Approach
4.1 Error Data4.1 Error Data The proposed transcription module is language-specific and The proposed transcription module is language-specific and
its basic parameters are designed based on translational its basic parameters are designed based on translational error-types retrieved from empirical data concerning the error-types retrieved from empirical data concerning the language pairs English, German and Greek. language pairs English, German and Greek.
Most data involves Most data involves mistranslations of spoken and mistranslations of spoken and written journalistic textswritten journalistic texts (data provided from graduate (data provided from graduate courses for professional translators), as well as courses for professional translators), as well as failure to failure to correctly interpret paralinguistic elements in online correctly interpret paralinguistic elements in online multimedia journalistic filesmultimedia journalistic files by transcribers and by transcribers and professional journalists. professional journalists.
The data was collected from graduate courses for journalists (M.A The data was collected from graduate courses for journalists (M.A in Quality Journalism and Digital Technologies, Danube University in Quality Journalism and Digital Technologies, Danube University at Krems, Athena- Research and Innovation Center in Information, at Krems, Athena- Research and Innovation Center in Information, Communication and Knowledge Technologies, Athens, Institution Communication and Knowledge Technologies, Athens, Institution of Promotion of Journalism Ath.Vas. Botsi, Athens) of Promotion of Journalism Ath.Vas. Botsi, Athens)
and from data transcribed from spoken journalistic texts in and from data transcribed from spoken journalistic texts in European Union projects, especially the CIMWOS Project European Union projects, especially the CIMWOS Project
4.2 Processing Speed4.2 Processing Speed For the achievement of speed and efficiency, in For the achievement of speed and efficiency, in
the “breadth-first search” (I) and “depth-first the “breadth-first search” (I) and “depth-first search” (II) decision-making processes, the search” (II) decision-making processes, the transcription module operates on keyword transcription module operates on keyword detection at detection at morpheme-levelmorpheme-level or word-level and or word-level and interaction with a database. interaction with a database.
This is a strategy typically employed in Spoken Dialog This is a strategy typically employed in Spoken Dialog Systems, where processing speed both by the Speech Systems, where processing speed both by the Speech Recognizer and any modules interacting with the speech Recognizer and any modules interacting with the speech recognition process is of crucial importance for task success recognition process is of crucial importance for task success and the overall efficiency of the system. and the overall efficiency of the system.
The present module may also operate with a The present module may also operate with a morphologically based tagger. morphologically based tagger. In transcribed spoken texts, In transcribed spoken texts, the the emotionally and socio-culturally “marked” elements are emotionally and socio-culturally “marked” elements are highlighted in the text component of the transcription tool. highlighted in the text component of the transcription tool.
4.3 The 4.3 The Pragmatic-Morphological Pragmatic-Morphological Approach Approach
The present approach concerns the detection of The present approach concerns the detection of the most (1) commonly occurring and (2) most the most (1) commonly occurring and (2) most efficiently processable emotionally and socio-efficiently processable emotionally and socio-culturally “marked” elements concerning (i) culturally “marked” elements concerning (i) implicit expression of opinions and/or viewpoints, implicit expression of opinions and/or viewpoints, the perception of (ii) irony and (iii) general the perception of (ii) irony and (iii) general context, spirit and background information. context, spirit and background information.
The “marked” elements” are designed to The “marked” elements” are designed to function function as pointers to texts with non-neutral content,as pointers to texts with non-neutral content, even they cannot fully capture all information (i), even they cannot fully capture all information (i), (ii) and (iii) signalizing the intentions and/or (ii) and (iii) signalizing the intentions and/or overall spirit of the speaker (or the author).overall spirit of the speaker (or the author).
The detection-strategiesThe detection-strategies are in the form of are in the form of parallel and parallel and
complementary program functionscomplementary program functions (Functions 1 -5) and are defined within a (Functions 1 -5) and are defined within a Pragmatic framework. Pragmatic framework.
““Marked” elements Marked” elements concerning implicit concerning implicit expression of opinions and/or viewpoints, expression of opinions and/or viewpoints, the perception of irony and general the perception of irony and general context, spirit and background information context, spirit and background information are be related to the flouting of the are be related to the flouting of the Maxims of Quantity, Quality and Maxims of Quantity, Quality and Manner in Grice’s Cooperativity Manner in Grice’s Cooperativity PrinciplePrinciple..
Maxims of Quantity, Quality and Manner Maxims of Quantity, Quality and Manner in Grice’s Cooperativity Principlein Grice’s Cooperativity Principle
QuantityQuantity: Make your contribution as : Make your contribution as informative as is required, informative as is required,
QualityQuality: Do not say what you : Do not say what you believe to be false, Do not say that believe to be false, Do not say that for which you lack adequate for which you lack adequate evidenceevidence
MannerManner: Avoid obscurity of : Avoid obscurity of expression. Avoid ambiguity. Be brief expression. Avoid ambiguity. Be brief (avoid unnecessary prolixity). Be (avoid unnecessary prolixity). Be orderly (Grice, 1975)orderly (Grice, 1975)
Morphological processing on a Morphological processing on a statistical basisstatistical basis
any repetitions of the any repetitions of the same wordsame word in short in short intervalsintervals (i.e. in each sentence or in every other (i.e. in each sentence or in every other sentence) (sentence) (Function 1Function 1) )
any high any high frequency of adjectives and frequency of adjectives and adverbialsadverbials, containing semantic features related , containing semantic features related to (i) to (i) descriptivedescriptive features (ii) features (ii) modemode (iii) (iii) malignant/benignmalignant/benign action or (iv) action or (iv) emotional/ethicalemotional/ethical gravity (gravity (Function 2Function 2))
Function 2 may result to a slight overgeneration of Function 2 may result to a slight overgeneration of “marked” adjectives and adverbs, however is helps “trap” “marked” adjectives and adverbs, however is helps “trap” and sustain passages in texts with understatements and and sustain passages in texts with understatements and subtle puns that are difficult to process by typical subtle puns that are difficult to process by typical systematic strategies. systematic strategies.
Morphological processing on a Morphological processing on a statistical basisstatistical basis
detection of detection of elements with a typically elements with a typically pragmaticpragmatic functionfunction (for example, emphasis), namely of (for example, emphasis), namely of specific sets of particles, adverbials and specific sets of particles, adverbials and exclamationsexclamations are included in the operations of are included in the operations of the morphological processing task (the morphological processing task (Function 3Function 3).).
Typical examples of such elements are the expressions Typical examples of such elements are the expressions
“merely”, “utterly”, “surely”, “doch”, “eben”, “gee”, “oops” “merely”, “utterly”, “surely”, “doch”, “eben”, “gee”, “oops” and “oh”. and “oh”.
Speaker: Kathryn Serkes, Assn. of American Speaker: Kathryn Serkes, Assn. of American Physicians and SurgeonsPhysicians and Surgeons
Lou Dobbs Tonight, CNN , Is universal health care Lou Dobbs Tonight, CNN , Is universal health care the solution for the United States health care the solution for the United States health care system?: CNN, Added On August 19, 2009system?: CNN, Added On August 19, 2009 , , http://www.cnn.com
(Average time of speaker: 2: 00 mins)(Average time of speaker: 2: 00 mins)
ExampleExample
……““laessiglaessig” (casual, cool), “” (casual, cool), “bequembequem” ” (comfortable, convenient), “(comfortable, convenient), “purpur” (pure, a word of ” (pure, a word of foreign origin in German),”foreign origin in German),”frischenfrischen“ (fresh)“ (fresh), , ““pompoesenpompoesen” (pompous)...” (pompous)...
Speaker: Female Correspondent (name not Speaker: Female Correspondent (name not available)available), Spiegel Online, Spiegel Online
Merkels Wahlkampf auf Russisch: Medwedew, Merkels Wahlkampf auf Russisch: Medwedew, Machismo und MenschenrechteMachismo und Menschenrechte
Morphological processingMorphological processing operates on a operates on a word-stem basisword-stem basis involving involving
the detection of (1) the detection of (1) verb-stemsverb-stems from from verbsverbs containing semantic features containing semantic features (including implied connotations in (including implied connotations in language use) related to (i) language use) related to (i) modemode (ii) (ii) malignant/benign actionmalignant/benign action or (iii) or (iii) emotional/ethicalemotional/ethical gravity gravity ((Function 4Function 4))
These verb stems are located low in an ontology, These verb stems are located low in an ontology, approachable with Selectional Restrictions of the approachable with Selectional Restrictions of the modifying type (and/or Wordnets). Other verb-modifying type (and/or Wordnets). Other verb-stems, such as verb-stems of primitive verbs are stems, such as verb-stems of primitive verbs are discardeddiscarded
Morphological processingMorphological processing also operates on a also operates on a suffix basissuffix basis involving the involving the
detection of (2) detection of (2) suffixessuffixes producing producing diminutives diminutives and (3) a predefined set of derivational and (3) a predefined set of derivational suffixessuffixes resulting to a resulting to a nominalization of verbsnominalization of verbs (derivational suffixes producing participles and (derivational suffixes producing participles and actor thematic roles are excluded), actor thematic roles are excluded), verbalization or adjectivization of proper verbalization or adjectivization of proper nounsnouns ( (Function 5Function 5). ).
Thus, a specified and finite set of "marked" suffixes such as Thus, a specified and finite set of "marked" suffixes such as (“-istas”, in American English, “-erei” in German) is defined, (“-istas”, in American English, “-erei” in German) is defined, as well as a defined group of "marked" verb-stems, such as as well as a defined group of "marked" verb-stems, such as “swear”,“struggle”, “schuften” (partially equivalent to “to “swear”,“struggle”, “schuften” (partially equivalent to “to slave away”) (German).slave away”) (German).
ExampleExample GEORGE WILL (ABC 10/28/01):GEORGE WILL (ABC 10/28/01): The administration The administration
knows he's knows he's vowedvowed, Hussein has , Hussein has vowedvowed revenge, he revenge, he has anthrax, he has anthrax, he lovesloves biological weapons, he has biological weapons, he has terrorist training camps, including 747's to practice terrorist training camps, including 747's to practice on... on...
BILL MOYERS:BILL MOYERS: IT WAS PROVING DIFFICULT TO IT WAS PROVING DIFFICULT TO DISTINGUISH THE OPINION OF THE PUNDITS FROM THE DISTINGUISH THE OPINION OF THE PUNDITS FROM THE POLICIES OF THE ADMINISTRATION...BUT AS THE POLICIES OF THE ADMINISTRATION...BUT AS THE HULLABALOOHULLABALOO OVER SADDAM GREW IN WASHINGTON, OVER SADDAM GREW IN WASHINGTON, BOB SIMON OF CBS NEWS 60 MINUTES WAS BOB SIMON OF CBS NEWS 60 MINUTES WAS DUMBFOUNDEDDUMBFOUNDED. HE IS BASED IN THE MIDDLE EAST.. HE IS BASED IN THE MIDDLE EAST.[[…….]…….]
BILL MOYERS:BILL MOYERS: AbsurdityAbsurdity. The Washington press . The Washington press corps cannot question an corps cannot question an absurdityabsurdity? ?
BILL MOYERS:BILL MOYERS: In aIn a bubblebubble. . BOB SIMON:BOB SIMON: Yeah Yeah in a bubblein a bubble. Where as we who've . Where as we who've
spent weeks spent weeks justjust walking the streets of Baghdad and walking the streets of Baghdad and in other situations in Baghdad just were in other situations in Baghdad just were scratchingscratching our headsour heads. . In ways that perhaps that the Washington In ways that perhaps that the Washington press corps could not. press corps could not.
ExampleExample
…….“(Rekordverschuldung) .“(Rekordverschuldung) egalegal” ((record of debt), ” ((record of debt), who cares about that/ that’s (totally) indifferent), who cares about that/ that’s (totally) indifferent), ““paperlapappaperlapap” (= nonsence), “(ein) ” (= nonsence), “(ein) schoenes schoenes (Wahlversprechen)” (“that’s some (ironic: great) (Wahlversprechen)” (“that’s some (ironic: great) pre-election promise), “(die Chefin will es so und pre-election promise), “(die Chefin will es so und damit) damit) bastabasta” (that’s what the boss wants and ” (that’s what the boss wants and that’s final), “that’s final), “zurueckegerudertzurueckegerudert (= to row back) (= to row back)….. …..
Speaker: Male Correspondent,Authors/ Speaker: Male Correspondent,Authors/ Correspondents: Martin Heller, Martin SuemeringCorrespondents: Martin Heller, Martin Suemering
CDU/CSU-Wahlprogramm: Wer glaubt an CDU/CSU-Wahlprogramm: Wer glaubt an Steuergeschenke? Steuergeschenke? 29.06.200929.06.2009 (02:05 mins) (02:05 mins)
Online written journalistic text with a low score Online written journalistic text with a low score and fragment from textand fragment from text
SCORE: GREENSCORE: GREEN Pakistan detains 'Indian spies’Pakistan detains 'Indian spies’ Rathor said that the suspected spies Rathor said that the suspected spies
were also paid to carry out violence were also paid to carry out violence in Pakistanin Pakistan
ALJAZEERA. NET ALJAZEERA. NET UPDATED ON: THURSDAY, JANUARY 29, 2009UPDATED ON: THURSDAY, JANUARY 29, 2009
Online written journalistic text with a low score Online written journalistic text with a low score and fragment from textand fragment from text
““Pakistan has arrested three men on suspicion of Pakistan has arrested three men on suspicion of spying for India and being involved in a bombing spying for India and being involved in a bombing in 2006.Perviz Rathor, the police chief in the in 2006.Perviz Rathor, the police chief in the eastern city of Lahore, said the three men, all eastern city of Lahore, said the three men, all Pakistani, were suspected to be on the payroll of Pakistani, were suspected to be on the payroll of the Research and Analysis Wing (Raw), India's the Research and Analysis Wing (Raw), India's foreign intelligence agency." They are Pakistani foreign intelligence agency." They are Pakistani nationals, they were spying for Indian nationals, they were spying for Indian intelligence," Rathor said on Thursday." They intelligence," Rathor said on Thursday." They were working for Raw and were spying for them were working for Raw and were spying for them since 2005. They were also involved in a bomb since 2005. They were also involved in a bomb blast in Lahore in 2006, in which 16 people were blast in Lahore in 2006, in which 16 people were injured.“injured.“
Online written journalistic text with a medium Online written journalistic text with a medium score score and fragment from textand fragment from text
SCORE: ORANGESCORE: ORANGE Nordic nerves Denmark and the euroNordic nerves Denmark and the euro Jan 29th 2009 | COPENHAGEN, From Jan 29th 2009 | COPENHAGEN, From
The EconomistThe Economist print edition print edition Danes ponder joining the euro, Danes ponder joining the euro,
Icelanders wish they couldIcelanders wish they could
Online written journalistic text with a medium Online written journalistic text with a medium score score and fragment from textand fragment from text
DENMARK’S story has an Old Testament resonance DENMARK’S story has an Old Testament resonance just just now. now. Seven years of Seven years of plentyplenty since Anders Fogh Rasmussen became since Anders Fogh Rasmussen became prime minister in 2001 are now being followed by prime minister in 2001 are now being followed by leanerleaner times. times. Danes Danes hardlyhardly face seven years of famine, but jobs are being face seven years of famine, but jobs are being axedaxed, property prices are , property prices are plummetingplummeting and belts are being and belts are being tightened everywhere. The government has already produced two tightened everywhere. The government has already produced two bank rescue packages and it is now pondering the notion of a bank rescue packages and it is now pondering the notion of a much largermuch larger fiscal stimulus. fiscal stimulus.
……Yet the Yet the gloomgloom has not, so far, has not, so far, envelopedenveloped the prime minister….. the prime minister….. …“…“It is It is wellwell known that the international financial crisis came known that the international financial crisis came
washing over us, as it did a lot of other countries,” Mr Fogh washing over us, as it did a lot of other countries,” Mr Fogh Rasmussen said this week. He added that, thanks to Rasmussen said this week. He added that, thanks to prudentprudent husbandry, Denmark is in a husbandry, Denmark is in a goodgood position …. position ….
……The prime minister’s belief that his government “behaved The prime minister’s belief that his government “behaved sensiblysensibly” and is accordingly without ” and is accordingly without blameblame for the country’s for the country’s economic economic woeswoes is not universally accepted. is not universally accepted.
…….Mr Fogh Rasmussen will .Mr Fogh Rasmussen will surelysurely then have won enough kudos to then have won enough kudos to secure his place in posterity—and perhaps a secure his place in posterity—and perhaps a plumplum foreign job foreign job such as NATO secretary-general or, if the Lisbon treaty is ratified, such as NATO secretary-general or, if the Lisbon treaty is ratified, the EU’s first permanent president. At least he would not then the EU’s first permanent president. At least he would not then have to worry about seven have to worry about seven leanlean years.years.
Fragment from a text with a medium score Fragment from a text with a medium score and and
explicitly stated point of viewexplicitly stated point of view SCORE: ORANGESCORE: ORANGE Britain must tackle its energy gap, Britain must tackle its energy gap, By Andrew Duff By Andrew Duff Published: February 4 2009 19:39 | Last updated: Published: February 4 2009 19:39 | Last updated:
February 4 2009 19:39February 4 2009 19:39
Copyright The Financial Times Limited 2009Copyright The Financial Times Limited 2009
Fragment from a text with a medium score Fragment from a text with a medium score and and
explicitly stated point of viewexplicitly stated point of view
At a time when confidence is At a time when confidence is soso lowlow, when , when investment is investment is soso desperatelydesperately needed and when needed and when finance is finance is soso challengingchallenging, commentators seem , commentators seem to to demonisedemonise even even modestmodest investor returns in investor returns in the the most most strategically strategically importantimportant part of the part of the economy. What is needed to stimulate the economy. What is needed to stimulate the extraordinaryextraordinary levels of investment required is levels of investment required is simplysimply encouragement. We have a encouragement. We have a realreal desire to desire to work in partnership with government and work in partnership with government and communities to deliver billions of pounds of communities to deliver billions of pounds of infrastructure, along with the jobs and infrastructure, along with the jobs and technologies that will technologies that will surelysurely follow”.follow”.
Copyright The Financial Times Limited 2009Copyright The Financial Times Limited 2009
As an optional componentAs an optional component to the transcription module, an interaction of the to the transcription module, an interaction of the
module with a module with a lexiconlexicon involves the mapping of involves the mapping of noun-stems or noun-stem + verb-stem pairs in noun-stems or noun-stem + verb-stem pairs in respect to a lexicon containing respect to a lexicon containing expressionsexpressions or a or a database constructed from available specialized database constructed from available specialized dictionaries for expressions (dictionaries for expressions (Function 6Function 6). ).
These are usually stems of words, not directly related to These are usually stems of words, not directly related to political, financial and journalistic sublanguages, such as political, financial and journalistic sublanguages, such as “run-gauntlet” and “Kakao-ziehen” (partially equivalent to “run-gauntlet” and “Kakao-ziehen” (partially equivalent to “to run somebody down”). “to run somebody down”).
However, the optional component may entail more However, the optional component may entail more processing time and/or computational cost.processing time and/or computational cost.
5. Evaluation5. Evaluation The initial implementation and test-versions (core The initial implementation and test-versions (core
module and “toy” program) of the proposed module and “toy” program) of the proposed module, along with user-tests, will give the first module, along with user-tests, will give the first concrete indicators of the usability and fine-concrete indicators of the usability and fine-tuning of the module’s design. tuning of the module’s design.
However, there are three factors to be taken into However, there are three factors to be taken into account for the module’s evaluation, namely account for the module’s evaluation, namely
(1) speed, (1) speed, (2) correctness and (2) correctness and (3) avoidance of overgeneration(3) avoidance of overgeneration..
In other words, effectiveness and In other words, effectiveness and
usability of the module is achievedusability of the module is achieved
when the when the speed speed of the output provided of the output provided allows the user to check a allows the user to check a multiple multiple online journalistic textsonline journalistic texts within a within a short short time spantime span
and when in the average online journalistic and when in the average online journalistic text the indicated “text the indicated “marked” elements marked” elements and the parts of the text containingand the parts of the text containing them are them are correctlycorrectly selected but do selected but do not not exceed 50%exceed 50% of the elements in the text. of the elements in the text.
EvaluationEvaluation
However, since emotionally and socio-culturally However, since emotionally and socio-culturally “marked” elements most often do not occur in “marked” elements most often do not occur in every sentence and are usually every sentence and are usually disperseddispersed throughout the text, in most cases, a relatively throughout the text, in most cases, a relatively small number of them may be enough to small number of them may be enough to signalize speaker’s or author’s intentions. signalize speaker’s or author’s intentions.
The morphologically-based approach may not The morphologically-based approach may not provide a 100% coverage of all implied provide a 100% coverage of all implied information, however, the above-described information, however, the above-described categories of “marked” elements show to be able categories of “marked” elements show to be able to function as to function as indicators indicators of texts with a non of texts with a non neutral, emotionally and socially “marked” neutral, emotionally and socially “marked” content.content.
6. Conclusions and Further 6. Conclusions and Further ResearchResearch
Further research will investigate two basic issues. Further research will investigate two basic issues. The first issue to be investigated is (I) the extent The first issue to be investigated is (I) the extent to which approaches involving “to which approaches involving “marked” marked” elements elements related to the flouting of the Maxims in related to the flouting of the Maxims in Grice’s Cooperativity Principle can apply to Grice’s Cooperativity Principle can apply to other other languages and/or language familieslanguages and/or language families with with similar or different morphological parameters. similar or different morphological parameters.
The second issue under investigation is a more The second issue under investigation is a more rigorous statistical analysis related to the extent rigorous statistical analysis related to the extent to which the to which the processable “marked” elements processable “marked” elements functioning as pointers to non-neutral content can functioning as pointers to non-neutral content can (II) (II) compensate for and/or balance all compensate for and/or balance all unprocessable unprocessable pragmatic informationpragmatic information signalizing the intentions and/or overall spirit of signalizing the intentions and/or overall spirit of the speaker or the authorthe speaker or the author. .
ReferencesReferences Alexandris, C., Fotinea, S-E. (2004): “Reusing Language Resources for Speech Alexandris, C., Fotinea, S-E. (2004): “Reusing Language Resources for Speech
Applications involving Emotion”, in: Applications involving Emotion”, in: Proceedings of the 4th LREC ConferenceProceedings of the 4th LREC Conference, , Lisbon, Portugal, Volume IX, 1383-1386.Lisbon, Portugal, Volume IX, 1383-1386.
AlexandrisAlexandris, , CC. (2003): Translational Issues in the Sublanguage of Written and . (2003): Translational Issues in the Sublanguage of Written and Spoken Journalistic TextsSpoken Journalistic Texts in Modern Greek, in Modern Greek, in: in: Proceedings of the International Proceedings of the International Conference on Choice and Difference in Translation, Conference on Choice and Difference in Translation, Athens 2003, 287-307.Athens 2003, 287-307.
Cockburn, A. (2001): Cockburn, A. (2001): Writing Effective Use CasesWriting Effective Use Cases. Boston, MA: Addison-. Boston, MA: Addison-Wesley.Wesley.
Constantie L.L., Lockwood, L.A.D. (1999): Constantie L.L., Lockwood, L.A.D. (1999): Software for Use: A Practical Guide Software for Use: A Practical Guide to the Models and Methods of Usage-Centered Designto the Models and Methods of Usage-Centered Design. Reading, MA: Addison-. Reading, MA: Addison-Wesley.Wesley.
De Silva, Robert, (2006): Explicitation in Media Translation: English Translation De Silva, Robert, (2006): Explicitation in Media Translation: English Translation of Japanese Journalistic Texts, in: of Japanese Journalistic Texts, in: Proceedings of the 1st Portuguese Proceedings of the 1st Portuguese Translation Conference,Translation Conference, Caparica, Portugal, 2006. Caparica, Portugal, 2006.
Grice,Grice, H.P. (1989): H.P. (1989): Studies in the Way of WordsStudies in the Way of Words. Cambridge, Mass: Harvard . Cambridge, Mass: Harvard University Press.University Press.
Grice, H.P. (1975): Logic and conversation, in: Cole, P. and Morgan, J. (eds.) Grice, H.P. (1975): Logic and conversation, in: Cole, P. and Morgan, J. (eds.) Syntax andSyntax and SemanticsSemantics, Volume 3. New York: Academic Press., Volume 3. New York: Academic Press.
Hatim, B. (1997): Hatim, B. (1997): Communication Across Cultures: Translation Theory and Communication Across Cultures: Translation Theory and Contrastive Text LinguisticsContrastive Text Linguistics, Exeter: University of Exeter Press., Exeter: University of Exeter Press.
Nielsen, J. (2000): Nielsen, J. (2000): Designing Web UsabilityDesigning Web Usability. Indianapolis, IN: New Riders . Indianapolis, IN: New Riders Publishing.Publishing.
BCBC IAEA Director General Mohamed ElBaradei shares IAEA Director General Mohamed ElBaradei shares
a laugh with Tim Sebastian, former BBC Hardtalk a laugh with Tim Sebastian, former BBC Hardtalk anchorman, before an interview. Vienna, Austria, anchorman, before an interview. Vienna, Austria, 28 September 200428 September 2004
BBC newsreader George Alagiah BBC newsreader George Alagiah News anchorman, Chikushi Tetsuya, TBS news News anchorman, Chikushi Tetsuya, TBS news
program “News 23″program “News 23″ Huw Edwards, BBC News (Huw Edwards, BBC News (The real Huw EdwardsThe real Huw Edwards, ,
By Lucy Wallis, BBC News Online )By Lucy Wallis, BBC News Online ) Sir Michael Parkinson, BBC NewsSir Michael Parkinson, BBC News Patrick Poivre d'Arvor , TF1 FrancePatrick Poivre d'Arvor , TF1 France
ToolsTools
Interactive User interface by the New Interactive User interface by the New York Times: York Times: for analyzing the most recent Democfor analyzing the most recent Democratic Debateratic Debate:: http://www.nytimes.com/interactive/2008/02/26/us/politics/http://www.nytimes.com/interactive/2008/02/26/us/politics/20080226_DEBATE_GRAPHIC.html#transcript20080226_DEBATE_GRAPHIC.html#transcript
Subtitling Services by Visualsubsync: Subtitling Services by Visualsubsync: http://www.visualsubsync.org/http://www.visualsubsync.org/