Speeches of US politicians 'have the reading age of a 13-year-old'

Congressional speeches made by US politicians have become simpler since the 1970s and only require the reading age of a 13-year-old to be followed, study found.

Computer scientists from Kansas State University analysed two million congressional speeches from Republican and Democrat politicians made between 1873 and 2010. 

Text analysis algorithms were used to examine how congressional speeches changed in terms of complexity, emotion and divisiveness over 138 years. 

More recent speeches use a smaller vocabulary, simpler language and talk about ‘the other party’ more than speeches made even a decade ago, the authors found. 

Researchers put the drop in the reading level down to the rise of broadcast media in congress that started in the mid-1970s – with politicians ‘playing to the camera’.   

The study is published as Joe Biden officially becomes the Democratic nominee for President amid dozens of speeches including from former president Bill Clinton who called out President Trump for ‘spending hours a day watching TV’.  

Computer scientists from Kansas State University analysed two million congressional speeches from Republican and Democrat politicians made between 1873 and 2010

Computer scientists from Kansas State University analysed two million congressional speeches from Republican and Democrat politicians made between 1873 and 2010

This chart assigns a school grade reading ability to the speeches of US politicians and shows a decline in the required level to understand the speeches over the past 40 years

This chart assigns a school grade reading ability to the speeches of US politicians and shows a decline in the required level to understand the speeches over the past 40 years

Speech numbers in Congress varies, but researchers found they are on an upward trend – with nearly double as many in the 2000s as there were in the 1990s. 

The reading comprehension level of the speeches changed significantly over the years – increasing in complexity first then getting less complex since the 1970s. 

The analysis measured the Coleman-Liau readability index, which estimates the reading level of a certain text and associates it with the appropriate school grade. 

The analysis showed that the reading level of congressional speeches made by both Republican and Democratic legislators increased consistently from the eighth-grade reading level in the 19th century, to the 10th-grade level in the 1970s. 

But since 1976 the reading level of political speeches has been declining consistently, and as of the 21st century, it is below the ninth-grade reading level. 

The same trend was also observed with the vocabulary used by congressional members in speeches, which had been increasing consistently until the early 1970s, and then started to decline—and it is still declining, co-author Lior Shamir said. 

President Barack Obama

President George W. Bush

During the George W. Bush administration, speeches of Democratic legislators expressed more negative sentiments compared to their Republican counterparts and it flipped when President Barack Obama took office

According to the study, the decline in reading level and vocabulary of the speeches can be related to the increasing presence of media.

This includes live radio and TV coverage of Congress beginning in the 1970s. 

Members of Congress started to gradually adjust their speech styles, addressing the public through the media rather than addressing their fellow legislators. 

As part of the study into congressional speeches, the team, including students Ethan Tucker and Colton Capps and professor Shamir, also looked at sentiment.

The algorithms measured aspects of the speeches such as the vocabulary, the reading level and the positive or negative sentiments expressed in the speeches.  

‘Based on that analysis, the algorithm determines whether a piece of text is positive, very positive, negative, very negative or neutral,’ Shamir said.  

The algorithms also measured the frequency in which different topics were discussed to find out how often ideas repeat themselves using the thousands of speeches made in Congress every yer.  

The research showed that the frequency of words related to women’s identity – such as she, her, hers, woman, women – has been increasing consistently since the early 1980s, while the frequency of words that identify men have been decreasing. 

The frequency of words related to women’s identity in the 21st century is five times higher compared to the 1950s, but lower than words related to men’s identity. 

Since the 1990s, terms related to women’s identity are more frequent in speeches made by Democrats compared to speeches made by Republican legislators.

Researchers found that the number of speeches spiked in the 2000s, with almost double the number in the previous decade

Researchers found that the number of speeches spiked in the 2000s, with almost double the number in the previous decade

‘For most of the 20th century, however, there were no substantial differences between women’s identity in Democratic and Republican speeches, and expressions of women’s identity were about 10 times less frequent than expressions of men’s identity by legislators from both parties,’ Shamir said.

The researchers’ analysis of the speeches also showed that more recent congressional speeches express more positive and negative sentiments than the speeches made in Congress during the 19th century and early 20th century. 

The sentiments in political speeches became gradually more positive and peaked in the 1960s, but declined sharply during the 1970s. 

Since the 1970s the sentiments expressed in congressional speeches have been becoming more positive again, the authors found.

Text analysis algorithms were used to examine how congressional speeches changed in terms of complexity, emotion and divisiveness over 138 years

Text analysis algorithms were used to examine how congressional speeches changed in terms of complexity, emotion and divisiveness over 138 years

Another aspect reflected through the analysis was the partisan split, Shamir said. 

Starting in the mid-1990s, Republican and Democratic speeches became increasingly different and correlated with the party in the White House.

During the George W. Bush administration, speeches of Democratic legislators expressed more negative sentiments compared to their Republican counterparts. 

That difference flipped immediately after 2008, with the beginning of the Obama administration, during which Republican speeches became more negative.

The study only looked at speeches up to 2010 so doesn’t include any references to Donald Trump or the response to the highly divisive president from legislators.  

The findings have been published in the journal Heliyon. 

HOW ARTIFICIAL INTELLIGENCES LEARN USING NEURAL NETWORKS

AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn.

ANNs can be trained to recognise patterns in information – including speech, text data, or visual images – and are the basis for a large number of the developments in AI over recent years.

Conventional AI uses input to ‘teach’ an algorithm about a particular subject by feeding it massive amounts of information.   

AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn. ANNs can be trained to recognise patterns in information - including speech, text data, or visual images

AI systems rely on artificial neural networks (ANNs), which try to simulate the way the brain works in order to learn. ANNs can be trained to recognise patterns in information – including speech, text data, or visual images

Practical applications include Google’s language translation services, Facebook’s facial recognition software and Snapchat’s image altering live filters.

The process of inputting this data can be extremely time consuming, and is limited to one type of knowledge. 

A new breed of ANNs called Adversarial Neural Networks pits the wits of two AI bots against each other, which allows them to learn from each other. 

This approach is designed to speed up the process of learning, as well as refining the output created by AI systems. 

source: dailymail.co.uk