Can computers really mark exams? Benefits of ELT automated assessments

Pearson Languages
Hands typing at a laptop with symbols

Automated assessment, including the use of Artificial Intelligence (AI), is one of the latest education tech solutions. It speeds up exam marking times, removes human biases, and is as accurate and at least as reliable as human examiners. As innovations go, this one is a real game-changer for teachers and students. 

However, it has understandably been met with many questions and sometimes skepticism in the ELT community – can computers really mark speaking and writing exams accurately? 

The answer is a resounding yes. Students from all parts of the world already take AI-graded tests. PTE Academic and Versant tests – for example – provide unbiased, fair and fast automated scoring for speaking and writing exams – irrespective of where the test takers live, or what their accent or gender is. 

This article will explain the main processes involved in AI automated scoring and make the point that AI technologies are built on the foundations of consistent expert human judgments. So, let’s clear up the confusion around automated scoring and AI and look into how it can help teachers and students alike. 

AI versus traditional automated scoring

First of all, let’s distinguish between traditional automated scoring and AI. When we talk about automated scoring, generally, we mean scoring items that are either multiple-choice or cloze items. You may have to reorder sentences, choose from a drop-down list, insert a missing word- that sort of thing. These question types are designed to test particular skills and automated scoring ensures that they can be marked quickly and accurately every time.

While automatically scored items like these can be used to assess receptive skills such as listening and reading comprehension, they cannot mark the productive skills of writing and speaking. Every student's response in writing and speaking items will be different, so how can computers mark them?

This is where AI comes in. 

We hear a lot about how AI is increasingly being used in areas where there is a need to deal with large amounts of unstructured data, effectively and 100% accurately – like in medical diagnostics, for example. In language testing, AI uses specialized computer software to grade written and oral tests. 

How AI is used to score speaking exams

The first step is to build an acoustic model for each language that can recognize speech and convert it into waveforms and text. While this technology used to be very unusual, most of our smartphones can do this now. 

These acoustic models are then trained to score every single prompt or item on a test. We do this by using human expert raters to score the items first, using double marking. They score hundreds of oral responses for each item, and these ‘Standards’ are then used to train the engine. 

Next, we validate the trained engine by feeding in many more human-marked items, and check that the machine scores are very highly correlated to the human scores. If this doesn’t happen for any item, we remove it, as it must match the standard set by human markers. We expect a correlation of between .95-.99. That means that tests will be marked between 95-99% exactly the same as human-marked samples. 

This is incredibly high compared to the reliability of human-marked speaking tests. In essence, we use a group of highly expert human raters to train the AI engine, and then their standard is replicated time after time.  

How AI is used to score writing exams

Our AI writing scoring uses a technology called latent semantic analysis. LSA is a natural language processing technique that can analyze and score writing, based on the meaning behind words – and not just their superficial characteristics. 

Similarly to our speech recognition acoustic models, we first establish a language-specific text recognition model. We feed a large amount of text into the system, and LSA uses artificial intelligence to learn the patterns of how words relate to each other and are used in, for example, the English language. 

Once the language model has been established, we train the engine to score every written item on a test. As in speaking items, we do this by using human expert raters to score the items first, using double marking. They score many hundreds of written responses for each item, and these ‘Standards’ are then used to train the engine. We then validate the trained engine by feeding in many more human-marked items, and check that the machine scores are very highly correlated to the human scores. 

The benchmark is always the expert human scores. If our AI system doesn’t closely match the scores given by human markers, we remove the item, as it is essential to match the standard set by human markers.

AI’s ability to mark multiple traits 

One of the challenges human markers face in scoring speaking and written items is assessing many traits on a single item. For example, when assessing and scoring speaking, they may need to give separate scores for content, fluency and pronunciation. 

In written responses, markers may need to score a piece of writing for vocabulary, style and grammar. Effectively, they may need to mark every single item at least three times, maybe more. However, once we have trained the AI systems on every trait score in speaking and writing, they can then mark items on any number of traits instantaneously – and without error. 

AI’s lack of bias

A fundamental premise for any test is that no advantage or disadvantage should be given to any candidate. In other words, there should be no positive or negative bias. This can be very difficult to achieve in human-marked speaking and written assessments. In fact, candidates often feel they may have received a different score if someone else had heard them or read their work.

Our AI systems eradicate the issue of bias. This is done by ensuring our speaking and writing AI systems are trained on an extensive range of human accents and writing types. 

We don’t want perfect native-speaking accents or writing styles to train our engines. We use representative non-native samples from across the world. When we initially set up our AI systems for speaking and writing scoring, we trialed our items and trained our engines using millions of student responses. We continue to do this now as new items are developed.

The benefits of AI automated assessment

There is nothing wrong with hand-marking homework tests and exams. In fact, it is essential for teachers to get to know their students and provide personal feedback and advice. However, manually correcting hundreds of tests, daily or weekly, can be repetitive, time-consuming, not always reliable and takes time away from working alongside students in the classroom. The use of AI in formative and summative assessments can increase assessed practice time for students and reduce the marking load for teachers.

Language learning takes time, lots of time to progress to high levels of proficiency. The blended use of AI can:

  • address the increasing importance of formative assessment to drive personalized learning and diagnostic assessment feedback 

  • allow students to practice and get instant feedback inside and outside of allocated teaching time

  • address the issue of teacher workload

  • create a virtuous combination between humans and machines, taking advantage of what humans do best and what machines do best. 

  • provide fair, fast and unbiased summative assessment scores in high-stakes testing.

We hope this article has answered a few burning questions about how AI is used to assess speaking and writing in our language tests. An interesting quote from Fei-Fei Li, Chief scientist at Google and Stanford Professor describes AI like this:

“I often tell my students not to be misled by the name ‘artificial intelligence’ — there is nothing artificial about it; A.I. is made by humans, intended to behave [like] humans and, ultimately, to impact human lives and human society.”

AI in formative and summative assessments will never replace the role of teachers. AI will support teachers, provide endless opportunities for students to improve, and provide a solution to slow, unreliable and often unfair high-stakes assessments.

Examples of AI assessments in ELT

At Pearson, we have developed a range of assessments using AI technology.

Versant

The Versant tests are a great tool to help establish language proficiency benchmarks in any school, organization or business. They are specifically designed for placement tests to determine the appropriate level for the learner.

PTE Academic

The Pearson Test of English Academic is aimed at those who need to prove their level of English for a university place, a job or a visa. It uses AI to score tests and results are available within five days. 

Pearson English International Certificate (PEIC)

Pearson English International Certificate (PEIC) also uses automated assessment technology. With a two-hour test available on-demand to take at home or at school (or at a secure test center). Using a combination of advanced speech recognition and exam grading technology and the expertise of professional ELT exam markers worldwide, our patented software can measure English language ability.

Read more about the use of AI in our learning and testing here, or if you're wondering which English test is right for your students make sure to check out our post 'Which exam is right for my students?'.

Explore our courseware and exams

More blogs from Pearson

  • A girl sat at a desk with a laptop and notepad studying and taking notes

    AI scoring vs human scoring for language tests: What's the difference?

    By Charlotte Guest
    Reading time: 6 minutes

    When entering the world of language proficiency tests, test takers are often faced with a dilemma: Should they opt for tests scored by humans or those assessed by artificial intelligence (AI)? The choice might seem trivial at first, but understanding the differences between AI scoring and human language test scoring can significantly impact preparation strategy and, ultimately, determine test outcomes.

    The human touch in language proficiency testing and scoring

    Historically, language tests have been scored by human assessors. This method leverages the nuanced understanding that humans have of language, including idiomatic expressions, cultural references, and the subtleties of tone and even writing style, akin to the capabilities of the human brain. Human scorers can appreciate the creative and original use of language, potentially rewarding test takers for flair and originality in their answers. Scorers are particularly effective at evaluating progress or achievement tests, which are designed to assess a student's language knowledge and progress after completing a particular chapter, unit, or at the end of a course, reflecting how well the language tester is performing in their language learning studies.

    One significant difference between human and AI scoring is how they handle context. Human scorers can understand the significance and implications of a particular word or phrase in a given context, while AI algorithms rely on predetermined rules and datasets.

    The adaptability and learning capabilities of human brains contribute significantly to the effectiveness of scoring in language tests, mirroring how these brains adjust and learn from new information.

    Advantages:

    • Nuanced understanding: Human scorers are adept at interpreting the complexities and nuances of language that AI might miss.
    • Contextual flexibility: Humans can consider context beyond the written or spoken word, understanding cultural and situational implications.

    Disadvantages:

    • Subjectivity and inconsistency: Despite rigorous training, human-based scoring can introduce a level of subjectivity and variability, potentially affecting the fairness and reliability of scores.
    • Time and resource intensive: Human-based scoring is labor-intensive and time-consuming, often resulting in longer waiting times for results.
    • Human bias: Assessors, despite being highly trained and experienced, bring their own perspectives, preferences and preconceptions into the grading process. This can lead to variability in scoring, where two equally competent test takers might receive different scores based on the scorer's subjective judgment.

    The rise of AI in language test scoring

    With advancements in technology, AI-based scoring systems have started to play a significant role in language assessment. These systems utilize algorithms and natural language processing (NLP) techniques to evaluate test responses. AI scoring promises objectivity and efficiency, offering a standardized way to assess language and proficiency level.

    Advantages:

    • Consistency: AI scoring systems provide a consistent scoring method, applying the same criteria across all test takers, thereby reducing the potential for bias.
    • Speed: AI can process and score tests much faster than human scorers can, leading to quicker results turnaround.
    • Great for more nervous testers: Not everyone likes having to take a test in front of a person, so AI removes that extra stress.

    Disadvantages:

    • Lack of nuance recognition: AI may not fully understand subtle nuances, creativity, or complex structures in language the way a human scorer can.
    • Dependence on data: The effectiveness of AI scoring is heavily reliant on the data it has been trained on, which can limit its ability to interpret less common responses accurately.

    Making the choice

    When deciding between tests scored by humans or AI, consider the following factors:

    • Your strengths: If you have a creative flair and excel at expressing original thoughts, human-scored tests might appreciate your unique approach more. Conversely, if you excel in structured language use and clear, concise expression, AI-scored tests could work to your advantage.
    • Your goals: Consider why you're taking the test. Some organizations might prefer one scoring method over the other, so it's worth investigating their preferences.
    • Preparation time: If you're on a tight schedule, the quicker turnaround time of AI-scored tests might be beneficial.

    Ultimately, both scoring methods aim to measure and assess language proficiency accurately. The key is understanding how each approach aligns with your personal strengths and goals.

    The bias factor in language testing

    An often-discussed concern in both AI and human language test scoring is the issue of bias. With AI scoring, biases can be ingrained in the algorithms due to the data they are trained on, but if the system is well designed, bias can be removed and provide fairer scoring.

    Conversely speaking, human scorers, despite their best efforts to remain objective, bring their own subconscious biases to the evaluation process. These biases might be related to a test taker's accent, dialect, or even the content of their responses, which could subtly influence the scorer's perceptions and judgments. Efforts are continually made to mitigate these biases in both approaches to ensure a fair and equitable assessment for all test takers.

    Preparing for success in foreign language proficiency tests

    Regardless of the scoring method, thorough preparation remains, of course, crucial. Familiarize yourself with the test format, practice under timed conditions, and seek feedback on your performance, whether from teachers, peers, or through self-assessment tools.

    The distinctions between AI scoring and human in language tests continue to blur, with many exams now incorporating a mix of both to have students leverage their respective strengths. Understanding and interpreting written language is essential in preparing for language proficiency tests, especially for reading tests. By understanding these differences, test takers can better prepare for their exams, setting themselves up for the best possible outcome.

    Will AI replace human-marked tests?

    The question of whether AI will replace markers in language tests is complex and multifaceted. On one hand, the efficiency, consistency and scalability of AI scoring systems present a compelling case for their increased utilization. These systems can process vast numbers of tests in a fraction of the time it takes markers, providing quick feedback that is invaluable in educational settings. On the other hand, the nuanced understanding, contextual knowledge, flexibility, and ability to appreciate the subtleties of language that human markers bring to the table are qualities that AI has yet to fully replicate.

    Both AI and human-based scoring aim to accurately assess language proficiency levels, such as those defined by the Common European Framework of Reference for Languages or the Global Scale of English, where a level like C2 or 85-90 indicates that a student can understand virtually everything, master the foreign language perfectly, and potentially have superior knowledge compared to a native speaker.

    The integration of AI in language testing is less about replacement and more about complementing and enhancing the existing processes. AI can handle the objective, clear-cut aspects of language testing, freeing markers to focus on the more subjective, nuanced responses that require a human touch. This hybrid approach could lead to a more robust, efficient and fair assessment system, leveraging the strengths of both humans and AI.

    Future developments in AI technology and machine learning may narrow the gap between AI and human grading capabilities. However, the ethical considerations, such as ensuring fairness and addressing bias, along with the desire to maintain a human element in education, suggest that a balanced approach will persist. In conclusion, while AI will increasingly play a significant role in language testing, it is unlikely to completely replace markers. Instead, the future lies in finding the optimal synergy between technological advancements and human judgment to enhance the fairness, accuracy and efficiency of language proficiency assessments.

    Tests to let your language skills shine through

    Explore Pearson's innovative language testing solutions today and discover how we are blending the best of AI technology and our own expertise to offer you reliable, fair and efficient language proficiency assessments. We are committed to offering reliable and credible proficiency tests, ensuring that our certifications are recognized for job applications, university admissions, citizenship applications, and by employers worldwide. Whether you're gearing up for academic, professional, or personal success, our tests are designed to meet your diverse needs and help unlock your full potential.

    Take the next step in your language learning journey with Pearson and experience the difference that a meticulously crafted test can make.

  • Woman standing outside with a coffee and headphones

    Using language learning as a form of self-care for wellbeing

    By Charlotte Guest
    Reading time: 6.5 minuts

    In today’s fast-paced world, finding time for self-care is more important than ever. Among a range of traditional self-care practices, learning a language emerges as an unexpected but incredibly rewarding approach. Learning a foreign language is a key aspect of personal development and can help your mental health, offering benefits like improved career opportunities, enhanced creativity, and the ability to connect with people from diverse cultures.

  • Friends walking outdoors chatting to eachother

    Understanding dialects in the English language

    By Charlotte Guest
    Reading time: 7 minutes

    Language reflects the diversity of human culture and society. Among its most fascinating parts are dialects, regional or social varieties of a language distinguished by pronunciation, grammar and vocabulary. Dialects are the heartbeat of a language, pulsing with the rich stories, traditions and identities of those who speak them.

    Understanding a language and dialect, and its significance, can enrich the learning experience for language learners, offering a deeper appreciation of a language and its speakers. Dialects are not just variations within a language; they are often considered separate entities, each with its own rich history and cultural significance, highlighting the complexity and diversity of linguistic expression.

    What exactly is a regional dialect?

    At its core, a dialect is a variation of a language spoken by a particular group of people. However, the distinction between dialects and a different language can often be subjective. These variations can occur due to geographical, social class, ethnic, or historical reasons. While all speakers of a language share the same basic grammar rules and vocabulary, those speaking different dialects might use unique words and slang or have distinct pronunciations, highlighting the lack of an objective difference between dialects and languages.

    For instance, British and American English are two dialects of the English language that are mutually intelligible, meaning speakers of either dialect can understand, and be understood, by the other. They share the same foundational grammar and most of the core vocabulary but differ in pronunciation, spelling, and some aspects of vocabulary and idioms. Similarly, within Britain or the United States, there are numerous regional dialects (e.g., Yorkshire English, Southern American English) that further showcase the diversity within a single language. Some of these dialects are considered by their speakers to be distinct languages, emphasizing the complex nature of linguistic identity and classification.

    What is an example of a dialect?

    An example of dialect variation can be seen in the Italian language, which boasts a wide range of regional dialects, showcasing linguistic diversity with multiple dialects spoken across Italy.

    For example, the Tuscan dialect has historically been recognized as the basis for standard Italian, largely due to its use in influential literature. However, other dialects from regions like Sicily or Lombardy vary significantly from Tuscan Italian in terms of pronunciation, vocabulary and syntax, reflecting the diverse cultural landscapes and histories of Italy’s regions.

    Another example of this variation of dialect within a single language is found in the United Kingdom. For instance, the Cockney dialect, originating from London’s East End, is renowned for its rhyming slang and distinct vowel sounds, serving as a prime example of spoken dialects that emphasize the importance of oral tradition. Contrastingly, the Geordie dialect, native to Newcastle and the surrounding areas, boasts an entirely different set of vocabulary, pronunciation patterns, and even grammatical structures, further highlighting the diverse range of spoken dialects within the standard English language.

    What is the difference between a dialect and an accent?

    The distinction between a dialect and an accent is subtle yet significant. An accent relates solely to differences in pronunciation - the distinct manner in which people say words, often influenced by unique speech patterns that can vary significantly across different languages and regions.

    In contrast, a dialect encompasses not only pronunciation and accent but also includes specific grammar and vocabulary. Accents can be a component of a dialect, but dialects offer a broader spectrum of linguistic variety, including lexical and grammatical differences.

    For instance, someone might speak English with a Scottish accent but use the same grammatical structures and vocabulary as an English speaker from London; however, Scots, a variety spoken in Scotland, is considered a dialect (or even a separate, distinct language, by some) because it possesses unique grammar, vocabulary and pronunciation.

    Why are different dialects important?

    Dialects are more than just linguistic variations; they are windows into communities' cultural and social fabric. They carry with them histories, traditions and the identity of their speakers. Some dialects are even considered 'distinct languages' by their speakers, highlighting the deep cultural significance of these linguistic forms. Learning about dialects, including regional dialects, can thus offer insights into:

    • Cultural contexts: Understanding the dialects of a language, especially regional dialects, can provide language learners with a richer cultural understanding and a more nuanced perspective of the language’s speakers. This exploration into regional dialects reveals the arbitrary distinction between 'standard' and 'nonstandard' dialects, which is often based on social, political, cultural, or historical considerations.
    • Social dynamics: Dialects can reflect social distinctions, historical migrations and contact with other languages, offering clues about social hierarchies, historical conflicts and integrations.
    • Language evolution: Studying dialects reveals how languages change over time, adapting to societies' needs, migrations and innovations.