American political speech more partisan than ever

American political speech more partisan than ever

25/07/2016 0 Di puntoacapo

Questo arti­co­lo è sta­to let­to 4150 volte!

American political speechAmerican political speech is more partisan than ever, new study finds

Amer­i­can polit­i­cal speech — Wide­spread use of talk­ing points and expand­ing role of con­sul­tants, focus groups and polls are like­ly con­tribut­ing to deep­er divi­sions, both in Con­gress and in the broad­er pub­lic, accord­ing to study.

PROVIDENCE, R.I. [Brown Uni­ver­si­ty] —A Demo­c­rat talk­ing about the Afford­able Care Act would like­ly call it “com­pre­hen­sive health reform,” while a Repub­li­can might describe it as a “Wash­ing­ton takeover of health care.” Repub­li­cans often use the term “ille­gal alien,” while Democ­rats tend to speak of “undoc­u­ment­ed work­ers.” The killing of 49 peo­ple at an Orlan­do, Flori­da, night­club in June was described by many Repub­li­cans as an act of “rad­i­cal Islam­ic ter­ror­ism,” while Democ­rats most often called it a “mass shoot­ing.”

A new work­ing paper coau­thored by Jesse Shapiro, the George S. and Nan­cy B. Park­er Pro­fes­sor of Eco­nom­ics at Brown Uni­ver­si­ty, finds that now more than ever, such par­ti­san use of lan­guage is the rule rather than the excep­tion.

We are see­ing evi­dence that, increas­ing­ly, Democ­rats and Repub­li­cans in Con­gress are speak­ing dif­fer­ent lan­guages,We are see­ing evi­dence that, increas­ing­ly, Democ­rats and Repub­li­cans in Con­gress are speak­ing dif­fer­ent lan­guages,” Shapiro said.

This can have an impact out­side of Con­gress. “The fact that par­ti­san lan­guage dif­fus­es wide­ly through media and pub­lic dis­course implies that this could be true not only for con­gress­peo­ple but for the Amer­i­can elec­torate more broad­ly,” the authors write in the study, not­ing that exist­ing research has shown that the way an issue is framed can affect pub­lic opin­ion on mat­ters rang­ing from immi­gra­tion to cli­mate change.

Along with coau­thors Matthew Gentzkow of Stan­ford Uni­ver­si­ty and Matt Tad­dy of the Uni­ver­si­ty of Chica­go, Shapiro crunched data from Con­gres­sion­al speech over the 136-year peri­od from 1873 to 2009.  The results, pub­lished in a work­ing paper titled “Mea­sur­ing Polar­iza­tion in High-Dimen­sion­al Data: Method and Appli­ca­tion to Con­gres­sion­al Speech,” show a sharp trend toward increas­ing­ly divid­ed speech.

“Par­ti­san­ship was low and rough­ly con­stant from 1873 to the ear­ly 1990s,” the authors write, “then increased dra­mat­i­cal­ly in sub­se­quent years.”

Shapiro and his col­leagues devel­oped a machine-learn­ing algo­rithm to ana­lyze tran­scripts of speech­es from the U.S. Con­gres­sion­al Record.

The algo­rithm cap­tured the use of 530,000 unique two-word phras­es — “estate tax,” for exam­ple, or “death tax” — spo­ken by politi­cians in their speech­es. For the peri­od between 1873 to 1990, the algo­rithm had a 54 to 55 per­cent chance of cor­rect­ly guess­ing a speaker’s par­ty based on one minute of speech. In 1994, that changed dra­mat­i­cal­ly.

“Begin­ning with the con­gres­sion­al elec­tion of 1994, par­ti­san­ship turned sharply upward, with the prob­a­bil­i­ty of guess­ing [a speaker’s polit­i­cal par­ty] cor­rect­ly based on a one-minute speech climb­ing to 83 per­cent by the 110th ses­sion (2007–09),” the authors wrote.

The authors point to inno­va­tion in polit­i­cal per­sua­sion begin­ning with the 1994 Con­tract with Amer­i­ca — the doc­u­ment that out­lined leg­is­la­tion that Repub­li­cans wished to enact with­in the first 100 days of the 104th Con­gress — as a water­shed moment that coin­cides with the increase in par­ti­san speech.

The Con­tract marked a shift in polit­i­cal mar­ket­ing tac­tics, as con­sul­tants “applied nov­el focus-group tech­nolo­gies to iden­ti­fy effec­tive lan­guage and dis­sem­i­nate it broad­ly to can­di­dates,” the paper states.

The lan­guage each par­ty uses is devel­oped strate­gi­cal­ly by con­sul­tants who make use of polls and focus groups and dis­sem­i­nate talk­ing points to can­di­dates and politi­cians.

In addi­tion, the years lead­ing up to 1994 “had seen impor­tant changes in the media envi­ron­ment: the intro­duc­tion of tele­vi­sion cam­eras as a per­ma­nent pres­ence in the cham­ber, the live broad­cast of pro­ceed­ings on the C‑SPAN cable chan­nels, and the rise of par­ti­san cable and the 24-hour cable news cycle,” accord­ing to the study.

The authors wrote that their study also draws on pri­or research that “sug­gests that these media changes strength­ened the incen­tive to engi­neer lan­guage and impose par­ty dis­ci­pline on floor speech­es, and made the new atten­tion to lan­guage more effec­tive than it would have been in ear­li­er years.”

That par­ti­san lan­guage, the authors wrote, fil­ters out into media cov­er­age and affects pub­lic dis­course.

“Exper­i­ments and sur­veys show that par­ti­san fram­ing can have large effects on pub­lic opin­ion,” the paper states, “and lan­guage is one of the most basic deter­mi­nants of group iden­ti­ty.”

The very imme­di­ate, dras­tic increase in par­ti­san speech came as a sur­prise to the authors, Gentzkow said in a sto­ry on the Stan­ford Insti­tute for Eco­nom­ic Pol­i­cy Research web­site.

The researchers had pre­dict­ed a strong cor­re­la­tion between leg­isla­tive votes and increas­ing­ly par­ti­san lan­guage in Con­gress, he said. An exist­ing body of research on polit­i­cal polar­iza­tion using roll-call votes as a mea­sure­ment indi­cat­ed a grad­ual increase in par­ti­san lan­guage from the mid-twen­ti­eth cen­tu­ry onward.

But what Shapiro and his coau­thors found instead was an unprece­dent­ed lev­el, or explo­sion, in par­ti­san speech start­ing in 1994. This, they wrote, sug­gests that lan­guage is a dis­tinct ele­ment of par­ty dif­fer­en­ti­a­tion.

The new machine-learn­ing method­ol­o­gy devel­oped by Shapiro, Gentzkow and Tad­dy, and its spe­cif­ic focus on lan­guage, could set a new stan­dard for fur­ther research on par­ti­san­ship. It “can be applied to a broad class of prob­lems in which the goal is to char­ac­ter­ize the polar­iza­tion or seg­re­ga­tion of choic­es,” the authors write.

Their method could shed light on polar­iza­tion by mea­sur­ing res­i­den­tial seg­re­ga­tion in small geo­graph­i­cal areas, polar­iza­tion in web brows­ing or social media behav­ior and dif­fer­ences in the way groups con­sume prod­ucts and infor­ma­tion.