Teaching students how to write well by the end of high school will be more important than ever, now that a student’s ability to get into college largely depends on it.

The new SAT, which debuts next March, features both a grammar and essay-writing component for the first time in decades. The change was made possible in large part by advances in technology–and advances in technology also might help educators better prepare their students for success on the new exam. Just how much these new technologies can help, however, is yet to be determined.

In part, the change was brought about by two studies that urged educators and policy makers to increase the emphasis on writing in schools, because it is essential to students’ success in the workplace and for America as a whole.

A 2003 report, called “The Neglected ‘R’: The Need for a Writing Revolution,” argues that writing has been shortchanged and has not received the full attention it deserves in the nation’s schools. Writing weaknesses, it said, cost college campuses up to $1 billion a year to fix–and employers complain about their new employees’ lack of writing skills.

“The Neglected R” was produced by the National Commission on Writing for America’s Families, Schools, and Colleges–a blue-ribbon group made up of university leaders, public school superintendents, and teachers–and was assisted by an advisory panel of writing experts.

A 1990 report, called “Beyond Prediction,” initially planted the seeds for adding a writing component to the SAT.

The College Board, which publishes and administers the SAT, is just now implementing the recommendations made by these reports, because technology has finally advanced enough to help. “The lack of technological capability to transmit millions of student essays to professional readers for scoring” made a writing component unfeasible until now, the College Board explained.

Students won’t be answering the SAT’s essay question via computer, but the internet will play a large role in getting the writing component marked.

“Essays for the new SAT writing section will be scanned and distributed to readers via the web,” said Sandra Riley, associate director of public affairs for the College Board. “Readers will be supervised online by scoring leaders, who are experienced essay readers with special training in online scoring.”

By working with the essay graders via the web, the College Board said, it will be able to recruit more high school and college teachers who have at least three years of classroom experience from across the country. Graders will work from their homes or offices with an online scoring system to allow accurate and effective essay scoring.

“The SAT is changing so that the test is more closely aligned with what students are learning in high school and in college, and to include writing, which is an important skill for success in college and beyond,” the College Board said.

Changes to the exam

The most obvious change to the SAT is the addition of the essay assignment.

Students will be asked to hand-write a 25-minute essay that requires them to take a position on an issue and use reasoning and examples to support their position. Also new is a multiple-choice section that measures a student’s ability to identify sentence errors and improve sentences and paragraphs.

Sixty minutes will be allotted to the writing portion: 25 minutes for the essay and 35 minutes for the multiple choices. Essays will be graded on a one-to-six scale using holistic scoring–an approach that considers a total piece of writing as a whole that is greater than the sum of its parts.

Each essay will take two minutes to score and will be marked by two people. If the two readers’ scores differ by more than one point, a scoring leader will resolve the difference. In all, the essay will be worth one-third of the entire score for the writing section.

Besides the new writing components, “the new SAT has virtually ripped out its vocabulary component,” said Paul Kanarek, president of the Princeton Review’s Southern California operations. Students will no longer have to spend hours drilling words as they did for the old SAT.

Also missing is the analogy section, he said, criticizing it as tricky and confusing for English as a Second Language (ESL) students. It has been replaced with deductive and inductive reading questions.

The math section also features Algebra II questions for the first time. “This is not a big deal,” Kanarek said, as “this will only impact kids who score 650 or higher.”

But the changes have been good for business, Kanarek admits: “As soon as they added Algebra II, they increased the perceived need for our service.”

Preparing students for the new SAT

Test-prep companies contacted by eSchool News, including Bridges Transition Inc., Kaplan Test Prep and Admissions, and the Princeton Review, are well under way in coaching students for the new SAT.

But schools, they say, play the largest part in preparing students to succeed in this high-stakes test. Their advice to schools is simple: Teach more grammar, and practice more writing.

First, teachers should step up their grammar teaching. Over the past 10 years, grammar lessons have declined dramatically nationwide–but a large part of the new SAT requires students to identify grammar errors.

“Grammar is no longer taught in high schools. It needs to be, because two-thirds of the new SAT is grammar,” Kanarek said. “You can no longer ignore grammar and do a fair service to your students.”

Second, teachers should get students to practice timed writing.

“The more English teachers can help students practice and to know what timed writing feels like, the better they are going to be,” said Jennifer Karan, director of SAT and ACT programs at Kaplan Test Prep and Admissions.

Practicing writing by hand also is important. “Because the test is written by hand, it’s important for students to become accustomed to writing by hand,” Karan said. In Kaplan’s new SAT prep course, for example, students will hand-write the essay absolutely every time, she said.

The Princeton Review’s Kanarek agrees: “How [often] is a kid asked to write for 25 minutes using [his or her] hand? It’s a dead art. We’ve got to get them used to writing.”

Karan also says teachers should familiarize students with a two-pronged approach to grading, giving them a holistic grade as well as a technical grade.

The general nature of holistic scoring means first impressions are crucial. Longer essays that are well organized and have strong introductions and conclusions will naturally get higher scores. In holistic grading, “you focus more on style than on substance,” Kanarek said.

Software solutions for writing improvement

Although it’s clear that giving students more opportunities to practice writing is essential to helping them excel on the writing components of high-stakes tests, teachers hardly have the time to assign and mark more essays than they currently do.

Several new software programs aim to help (see “Writing practice software,” page 38). Some of these programs give students step-by-step guidance in the essay-writing process as well as ample opportunities for practice, though teachers still do the correcting and provide feedback. But a growing number of programs now employ sophisticated artificial-intelligence technology that purports to grade essays in seconds, giving already overburdened teachers much-needed relief.

Major essay assessment engines on the market include Educational Testing Service’s e-Rater, Pearson Education’s Intelligent Essay Assessor, and Vantage Learning’s IntelliMetric. Different developers created each of these computerized essay-grading systems, but they all work on the same basic principle.

Typically, the assessment engine first learns the English language by studying and analyzing vast quantities of text from sources such as student writing or books. Then, for each essay assignment, or “prompt,” the engine is fed two to three hundred essays written by students and marked by expert graders. These essays range in quality from poor to perfect.

As the engine analyzes the text and the score assigned to each essay, it tags and identifies elements of the English language and creates mathematical relations, or algorithms, between those elements.

At the end of this process, the engine has taught itself what a “one” or a “six” essay looks like and has been “trained,” or tuned, to mark a particular essay prompt according to the marking scheme or rubric used.

To train Pearson’s Intelligent Essay Assessor, its developers literally dumped billions of words from electronic versions of school textbooks and library books into the engine for it to study.

“We just feed it in, and it munches on it for a couple of hours and then it’s all done,” said Tom Landauer, president and CEO of Knowledge Analysis Technologies LLC, which Pearson Education acquired in June. As a division of Pearson, the company is now called Pearson Knowledge Analysis Technologies (Pearson KAT).

The engine creates a detailed algebraic matrix as well as a statistical number of measures of what’s going on in an essay. “What it’s trying to understand is how words are used and how they can substitute for each other,” Landauer said.

The machine is basically learning to read and write all at once. When learning to read, “you read an enormous amount of text. When you are done, you don’t know how your brain did it–but when you read the words ‘doctor’ and ‘physician,’ you knew they meant relatively the same thing,” he said.

Next, to train the assessment engine to mark essays on each prompt, its developers fed in about 200 essays scored by human experts for each one.

Similarly, Vantage Learning’s IntelliMetric engine reviewed hundreds of student papers scored by expert graders. At the end, it created a semantic map of the English language of 400 characteristics and attributes of writing.

“We’re trying to understand how a human goes through the grading process, and we try to digitize that,” said Harry N. Barfoot III, vice president of sales and marketing for Vantage Learning.

One company, however, has evolved its assessment engine to use an alternative method to grade papers. ETS’s early versions of its assessment engine, e-Rater, used the method described above, but the company’s latest version now takes a more simplified approach.

ETS has identified 12 linguistic principles that it uses to grade essays. Now, the e-Rater gives grades based on the proportion, or percentage, of errors of each linguistic principle–including grammar, style, usage, and others.

The weight associated with each of the variables is different, said Richard Swartz, executive director of technology products and services at ETS.

Swartz says the way e-Rater works is no secret. On the ETS web site, “we list the 12 variables and deliberately describe the research basis for those variables,” he said.

The new method allows ETS to provide students with better feedback on what they did wrong by combining holistic scores with analytical feedback relevant to their essay, he said.

“When you provide somebody with a holistic score, like a four, naturally you want to know how to get from a four to a six,” Swartz said. “The diagnostic feedback we provide is much more refined.”

The other major difference that sets ETS apart is that teachers and students are not limited to the essay topics, or prompts, that come with the product.

“We can allow an infinite number of prompts,” Swartz said. “Teachers can enter their own questions.”

This is possible because ETS trained its assessment engine not only for each prompt, but also for competencies associated with each grade level. A teacher simply has to pick a grade, a mode for the essay (such as descriptive, persuasive, or cause and effect), and either choose a pre-existing essay prompt or enter one of his or her own.

When teachers enter their own prompt, Swartz said, they must phrase their prompt to match the essay’s mode. For example, if the prompt says “The school board wants to adopt a longer school year. Write a letter to the board in favor or against the proposal,” a teacher could swap “dress code” for “longer school year.”

In addition, if students are ahead or behind their grade level, teachers can choose to grade their essays accordingly, because ETS claims it can grade essays based on the competency common to each grade level.

Uses of intelligent essay-scoring engines

ETS and Vantage Learning have integrated their assessment engines into programs aimed at helping students practice their writing skills, with feedback generated by the computer.

Pearson’s Intelligent Essay Assessor is not currently integrated into any writing practice software, but it’s used by several major textbook publishers–including Holt, Rinehart and Winston and Prentice Hall–to allow students to answer the end-of-chapter questions found in many textbooks. “We score about 1,000 of those a day,” Landauer said.

Plans are in the works to integrate the Intelligent Essay Assessor into SAT practice courses that teachers could administer in class, the library, or a computer lab. “The distribution channels have not been settled on yet,” Landauer said.

Currently, Pearson’s assessment engine contains roughly 100 prompts suitable for SAT practice. The Intelligent Essay Assessor, like the SAT, returns a one- to six-point score, as well as diagnostic information such as egregious errors, run-on sentences, irrelevant sentences, or mechanics and style.

Vantage Learning’s marking scheme is based on national writing standards, not specifically the SAT’s rubric, but is still useful for practicing for tests like the SAT, Barfoot said.

In addition, Vantage Learning’s automated essay scoring engine, IntelliMetric, is used to grade the long-answer portion of the American Board Certification of Teacher Excellence (ABCTE) exam. Also, it is currently used by thousands of schools and students as part of the company’s My Access 5.0 writing practice software, and it’s embedded in other applications such as the College Board’s WritePlacer Program, CTB/McGraw-Hill’s Writing Roadmap, and others.

ETS e-Rater, which has been implemented statewide in Indiana to mark end-of-course tests, also has not been calibrated specifically for the new SAT. “The essay questions we have in the system for 11th and 12th grade, students are going to find them very similar to [the Texas Assessment of Knowledge and Skills] and the SAT,” Swartz said.

With the exception of ETS, teachers are limited to the essay topics provided by the software. “It costs too much on a particular prompt for schools to do it on their own,” Landauer explained.

But one enterprising school district in California decided it was worth the cost to customize an essay assessment engine to the rubric for the writing component of the state’s standardized test.

Not only would the Pomona Unified School District have a custom product tailored to its needs, but the district would save nearly $500,000 in yearly licensing fees charged by the commercial products it tested, including those of Vantage Learning and ETS.

To do this, the district issued a Request for Proposals for someone to build a custom engine tuned to the state’s rubric that the district would own and could use in perpetuity.

Professor Steven Donahue, who teaches English as a Second Language (ESL) at Miami Dade College and had developed an online writing critique program called WriteNowABC, agreed to build the district a customized K-12 version for a one-time fee of $140,000.

The district provided language experts and curriculum specialists to help create the software, called the California Electronic Writer, which should be ready for initial implementation this month.

“This tool is seen by the district as a strategy to develop specific content. It really helps guide and inform the writing process,” said Jerry Livesey, a technology consultant for the district, which has a high population of ESL students.

Students will use the California Electronic Writer to practice writing for classroomassignments as well as the state’s standardized test.