Inchmarlo youngster Tom Flaherty, pictured below, is through tothe semi-finals of the Haribo Young Golfer of the Year competition.
The seven-year-old has beaten off hundreds of other younghopefuls to secure his place at the semi-final at Westerwood tomorrow.
With a handicap of 36, Tom hopes to become one of the 10 bestgolfers of the day in order to qualify for the grand final …
Crime, drugs, homelessness, violence, sexual abuse, teen pregnancy, endangered species, conservation, pollution.. the list of issues and concerns generated by fifth graders was long and surprisingly sophisticated. Students discussed and brainstormed their concerns after viewing work by two postmodern artists, Krystoff Wodicizko and Ciel Bergman. This activity was part of a year-long curriculum designed to involve south Georgia fifth-grade students in a district-wide study of postmodern art and related issues. The discipline-based art education (DBAE) curriculum offered opportunities not only to discuss art, but also to investigate the functions of art in contemporary society through …
Manu Ginobli's 19 points wasn't enough as defending Olympic champion Argentina stumbled to a narrow 79-75 defeat to Lithuania in its opening Group A match on Sunday.
In what had been a tight game, the San Antonio Spurs' shooting guard and his teammates were nearly shut down in the third quarter as Lithuania ran up a 51-45 lead. Ginobli and Houston Rockets forward Luis Scola were restricted to just two points each in the period.
Argentina rallied and, with 2:40 left, Ginobli sunk a pair of free throws to cut the deficit to four points. Quickly regaining possession, Ginobli bounced a 3-pointer off the rim that was buried by Toronto Raptors guard Carlos …
On behalf of the Wilts and Berks Canal Trust, we would like toextend our sincere appreciation to everyone who supported us at therecent official opening of the newly-restored Double Bridge nearLacock.
We were delighted to welcome our patron Her Royal Highness theDuchess of Cornwall, together with Canal Partnership president LordLansdowne and all of the county and civic dignitaries.
It was particularly pleasing to see the number of local familiesjoining us to celebrate the occasion.
Enthusiastic members of Wiltshire Youth Canoe Club were the firstpeople to navigate under the bridge in almost a century, paddlingseveral hundred metres along a restored and …
RESEARCHERS INVESTIGATE THE EFFECTS OF REGULATIONS,
MONOPOLY AND MONOPSONY ON AUDIT FEES
For the past 20 years, researchers have viewed the pricing of audit services as a function of supply and demand, much like other commodities or services traded in the marketplace. The audit-pricing model, originally proposed by Dan Simunic (1980), has been used to explore various aspects of auditing, such as the determinants of auditor changes, low-- balling of audit fees and implications of litigation against audit firms. Simunic and Stein provided a comprehensive summary of relevant research literature in these areas in an article titled "The auditing marketplace: exploring the …
COLUMBUS, Ohio (AP) — Terrelle Pryor's career at Ohio State, which started with so much promise and potential, came to an abrupt and scandal-ridden end.
The Ohio State quarterback announced through his attorney Tuesday that he would not play for the Buckeyes this season. He had already been suspended for the first five games for breaking NCAA rules by accepting improper benefits from the owner of a tattoo parlor.
"In the best interests of my teammates, I've made the decision to forgo my senior year of football at The Ohio State University," Pryor said in a statement issued by Columbus lawyer Larry James.
Pryor will most likely make himself available for an NFL …
Officials in Germany say dozens of passengers have been rescued from an overheated train after the air conditioning broke as a heat wave broiled Europe.
The national railway system was quoted by the ddp news agency Sunday as saying 44 people needed medical treatment after they were taken from the train Saturday in the western city of Bielefeld.
Hans-Dieter …
Waterloo, Ont.
Music and readings liberally salted with anecdotes was the fare served up at the Sound in the Land book launch last month in the Conrad Grebel chapel-along with lots of Mennonite food.
Music ranged from "Death Be Not Proud" by John Donne, sung by soprano Stephanie Kramer, to "Aunt Lizzie's Old Order House," written and sung by Rebecca Campbell. The latter was contextualized with a set of artifacts that would appear in a typical Old Order Mennonite house, including a marble roller.
The 2004 Sound in the Land festival and conference held at Conrad Grebel marked the first time that Mennonites, in their 300-plus years of being in North America, gathered …
Television talk show host Elizabeth Hasselbeck has given birth to a son, ABC announced Saturday.
The co-host of "The View" gave birth Friday to a 7-pound, 15-ounce boy at an Arizona hospital, the network said.
Hasselbeck and her husband, NFL quarterback Tim Hasselbeck, did not release the …
(This is not a legal transcript. Bloomberg LP cannot guarantee its accuracy.)
STEPHEN GALLOWAY, AUTHOR, HOLLYWOOD REPORTER, TALKS ABOUT THE ACADEMY AWARDS AND THE ECONOMY AT MIDDAY SURVEILLANCE
FEBRUARY 25, 2011
SPEAKERS: TOM KEENE, MIDDAY SURVEILLANCE HOST
STEPHEN GALLOWAY, AUTHOR, HOLLYWOOD REPORTER
12:41
TOM KEENE, MIDDAY SURVEILLANCE HOST: Folks, as we look at the media, the economics, the financing and investment of Hollywood, it is also about Los Angeles. We're thrilled to have with us Stephen Galloway, the Hollywood Reporter. Stephen, good morning.
STEPHEN GALLOWAY, AUTHOR, HOLLYWOOD REPORTER: Good …
L.A. Lakers at Boston
Sunday, June 8
L.A. Lakers at Boston
Tuesday, June 10
Boston at L.A. Lakers
Thursday, June 12
Boston at L.A. Lakers
…
Comment on proposed revisions to the Reports of Condition and Income is being requested by the FDIC, Federal Reserve Board and the Office of the Comptroller of the Currency. The agencies' proposal, which has been approved by the Federal Financial Institutions Examination Council, contains several types of changes to the call report requirements. These changes relate to the content of the call report itself, the submission deadline for certain banks and the agencies' process for validating and releasing the data that banks report. For the most part, the proposed changes would take effect March 31, 2003.
The agencies encourage bankers to review the proposal and comment on revisions that are of interest. Comments may be sent to any or all of the banking agencies at the addresses listed on the second page of the Federal Register notice. All comments will be shared among the agencies and should be submitted by Jan. 7, 2003.
First, the agencies are proposing several revisions to the content of the call report and one instructional clarification. These revisions are focused on improving the information reported by banks that engage in certain specific activities and generally will be applicable to small percentages of banks rather than to most or all banks. The proposed revisions include:
- Adding five items dealing with accrued fees and finance charges on credit card accounts, allowances for uncollectible accrued fees and finance charges, and charge-offs of such accrued amounts, which would be reported only by banks with a significant volume of credit card activity;
- Splitting the item in the securitization schedule (Schedule RC-S) for seller-provided credit enhancements to the bank's securitization structures (other than creditenhancing interest-only strips) into separate items, one for on-balance sheet assets and another for other enhancements;
- Separating the current income statement (Schedule RI) item for income from insurance activities into two items, one for insurance underwriting income and the other for income from other insurance activities;
- Adding a yes/no question asking whether any of the bank's Internet web sites has transactional capability;
- Extending to banks with less than $100 million in assets the requirement to disclose the fair values of derivative contracts in Schedule RC-L - Derivatives and Off-Balance Sheet Items, because current accounting standards require derivatives to be reported on the balance sheet at fair value;
- Changing where banks report any provisions for allocated transfer risk in the income statement (Schedule RI);
- Clarifying the instructions to describe the limited circumstances in which loans may be reported as held for trading purposes; and
- Creating a supplement to the call report that would enable the agencies to collect a limited amount of data from certain banks in the event of an immediate and critical need for specific information.
Second, the agencies are proposing to reduce the call report filing period for banks with more than one foreign office from 45 to 30 days effective June 30, 2003. In connection with this change, the FDIC would be authorized to contact some of these banks around May 1, 2003, if their March 31 reports have not yet been received in order to obtain certain deposit data needed to estimate insured deposits.
Third, to improve the timeliness with which call report data become available to the public, the agencies would begin posting the reports for individual banks on the FDIC's web site as soon as the agencies' analysis of an individual report has been completed. This change would begin perhaps as early as with the first quarter 2003 reports. At present, all call reports are released to the public simultaneously some 60-75 days after the report date.
Finally, the agencies currently plan to implement a new business model for collecting and validating call reports in March 2004. Under this model, the initial editing of a bank's call report data would take place in conjunction with the bank's submission of the data to the agencies. As part of this change, the agencies are proposing that a bank's call report must pass all validity edits and must include an explanatory comment addressing each quality edit exception identified in the bank's report in order for the agencies to accept the bank's call report submission. Otherwise, the bank's report will not be accepted and the bank will need to make appropriate corrections, add any required explanatory comments, and resubmit its data file by the submission deadline. For most banks, this new validation process will reduce the need for subsequent questions from the banking agencies.
Draft copies of the FFIEC 031 version of the report for banks with foreign offices and the FFIEC 041 report for March 2003 are available on the FFIEC web site, www.ffiec. gov/ffiecreportforms.htm.
For the first time, Rashidi Wheeler's parents publicly saidTuesday they believe the death of their son "could have beenprevented."
The parents also announced the Rev. Jesse Jackson has agreed toact as the family's spokesman. They requested that Jackson and hisRainbow/Push Coalition assist Northwestern in any investigation intothe events leading up to and surrounding the death of the seniorsafety, who died during running drills Friday from an asthma attackas the NU football team took part in conditioning exercises.
"His mother and I and family are looking into trying toinvestigate and understand what happened on the field here atNorthwestern," said George Wheeler, Rashidi's father, after a 90-minute memorial service on the NU campus Tuesday afternoon. "We don'thave all the information. We have asked the Rev. Jackson and hisstaff to work with us and to support us. Our only objective here isto understand what happened to our son."
George Wheeler reiterated what Linda Will, Rashidi's mother, saidMonday--that the family would like to see more safety procedures putin place by the school to help ensure athletes have the best medicalcare available when an emergency situation arises.
The family and Jackson indicated they have obtained informationthat calls into question how quickly emergency medical aid was givento Wheeler once he experienced trouble breathing.
"We do feel that this could have been prevented," George Wheelersaid. "We don't have details, but we have asked for an investigationto look into the death of Rashidi, and we'd like all questionsdirected to the Rev. Jackson and his staff."
Jackson, who gave a rousing eulogy during the service and willattend Wheeler's funeral Monday in Pomona, Calif., was asked whatinformation led to questions about the timing of the treatment givento Wheeler. NU medical staff said he was conscious and coherent forabout 10 minutes before he lost all breathing and suffered heartfailure.
"There are concerns about what happened the last hour of his lifeand the quality of assistance," Jackson said. "At this point there isno finger-pointing, no anger, no hostility, but there is a desire toknow what the last hour of their son's life was like. What were thelast minutes like? What happened? Was there appropriate care? Whatcould have been done differently? Was it preventable?
"Those are issues I will investigate on behalf of the family whilethey grieve and work through the trauma of this sudden loss."
Jackson said the family "does not want to be premature into theirconclusions" but has been informed by eyewitnesses that "four or fiveboys fell out at the same time" because of overexertion during thedemanding drills. The family is questioning whether that many worn-out players taxed the six training-staff members present.
Sophomore wide receiver Jason Wright, who sang at the service,told the Los Angeles Times he was one of the players who "collapsedand fell unconscious," which is why he "never saw Rashidi go down,"even though Wheeler was two players ahead of him in line.
"I got up, threw up and was happy as I've ever been that Ipassed," Wright told the Times. "Then I heard someone say, 'Rashidilost his pulse.' "
Jackson also indicated there is reason to believe the medicalstaff on the field had trouble calling for an ambulance.
"The telephone back to the fieldhouse or whatever wasmalfunctionable, and they finally got a rescue [Evanston FireDepartment paramedics] through cell phones or something," he said."But all this is so very preliminary. The family, they're hurt, butthey're not angry or bitter. But their curiosity does need to besatisfied. And the university must, of course, accept responsibilityto be a willing part of that pursuit to those questions.
"And many of those [questions] come from players. So the questionis was [the medical staff] sufficient to deal with the situation. Wasresuscitation given? Was the rescue squad there?"
Kevin Bentley, one of Wheeler's best friends and the player whowill wear Wheeler's No. 30 this season, said he saw "85 to 90percent" of the incident that led to Wheeler's death.
"Maybe we were undermanned [medically], who knows?" Bentley said."I'm not going to speculate, but just hope that some precautions betaken in the future. I just don't want to see another one of myteammates die."
Two AIChE members are showing middle school students how to get in gear for a greener future.
John Weidner and Edward Gatzke - chemical engineering professors at the Univ. of South Carolina (USC) - are among the creatore of the FIRST LEGO League's 2007 "Power Puzzle" Challenge. The challenge, part of an annual technology competition sponsored by the not-for-profit organization FIRST (For Inspiration and Recognition of Science and Technology), introduces tens of tiiousands of young people to concepts in scientific thinking and real world problem-solving.
Founded by inventor Dean Kamen, FIRST'S mission is to inspire students' interest and participation in science and technology. Using LEGO Mindstorms technologies and LEGO play materials, children aged 9-14 work alongside adult mentors to design, build and program robots to perform intricate tasks in an arena competition. More than 10,000 teams from 38 countries are involved in the Power Puzzle, which this year draws attention to energy management and conservation.
Power Puzzle challenges are underway at more man 300 qualifying events. These culminate at the FIRST LEGO League World Festival and Championship, Apr. 17-19, 2008, in Atlanta's Georgia Dome. Eighty-four teams from 27 countries will compete.
The creation of the Power Puzzle called for the expertise of engineers like Weidner and Gatzke - who not only had the necessary technical knowledge, but had also been long-time participants in LEGO League programs.
Since FIRST LEGO League's (FLL) 1998 inception, the South Carolina state tournament has been was one of the nation's largest. Engineers, college faculty and their students have been among the legion of volunteers needed to run the competitions.
In 2002, the South Carolina Dept. of Education asked USCs College of Engineering and Computing to take over the state FLL tournament. John Weidner, whose own children had been participating in the LEGO League, agreed to serve as coordinator of the 2003 "Mission to Mars" challenge. Weidner soon found himself hooked on the concept.
"It was extremely rewarding to watch 9-14-year-old kids handling real engineering issues of teamwork, time constraints, and technical topics - gears, motors and structures," says Weidner. He coordinated the 2004 "No Limits" competition, which involved engineering solutions to accommodate people with disabilities, 2005's "Ocean Odyssey," which asked students to select a sea resource or human activity, and to trace its impact on the ocean's health, biodiversity and productivity, and 2006's "Nano Quest," where students used LEGO elements to visualize engineering tasks conducted on the micro-scale.
During this time, Weidner asked his USC colleague Ed Gatzke, who had been judging local FLL competitions, to act as his head referee. "I've had fun with LEGO toys since I was a small child. I currently have my children playing with LEGOs," says Gatzke.
Gatzke notes that the LEGO League's focus on robotics and construction may not seem like an obvious fit with chemical engineering, but it does relate to his own process control research. "The students typically run everything in open-loop, rarely using sensors," says Gatzke. "But at least they are getting exposure to technology in a fun and creative way."
When FIRST wanted to develop an energy-themed challenge, John and Ed were a logical choice for the design team. "FIRST knew that we were passionate about the program," says Weidner, "and when they heard that Ed and I perform research on fuel cells and hydrogen, they asked us to help design Puzzle Power. Our role was to provide the underlying technical foundation and context for the game. We came up with the main themes - how to create sustainable options to meet our planet's growing energy needs in a way that is good for the environment. Hopefully the challenge is exposing students to many of these options - some simple, like growing biofuels, and others futuristic, such as solar panels in space."
Gatzke adds, "I hope that the puzzle will get people to think about tradeoffs related to energy choices, and expose them to concepts of energy production, conversion, and storage."
Since 2002, Gatzke and Weidner have seen South Carolina LEGO League participation grow from 50 to 140 teams, making it one of the model programs in the U.S. "We want this to be in every elementary and middle school throughout the state," says Weidner.
The tournament experience is also great fun. "FIRST designed this competition as 'Sports for the Mind,' so it is as loud and exciting as any sporting event," says Weidner.
The competitions require lots of volunteer support. 'Tournaments need referees, judges to watch team presentations and examine robots, technical support to assemble the playing fields, and wranglers to get teams in the right place at the right time," says Gatzke. "Before the tournament takes place, volunteers can help build the tournament tables, assemble the challenge, or even coach a local team."
One LEGO League fan is AIChE past president Bill Byers, who has been volunteering as a regional judge in Oregon. "I particularly like FIRST's core value of instilling 'gracious professionalism' in young people," says Byers. "That's a concept that can't be started too early."
The next LEGO League challenge - devoted to the environment - will be announced in September 2008. Like the Power Puzzle, the environmental challenge should provide a good fit for chemical engineering insight.
To learn about FIRST LEGO League competitions and opportunities for involvement, visit www.firstlegoleague.org.
The former U.N. chief and other mediators trying to bring Kenya's warring politicians together found the opposition taking a hard line Tuesday, accusing the government of "crimes against humanity" in a complaint it planned to file at the Hague.
The opposition and President Mwai Kibaki's administration have traded accusations of who is behind the violence stemming from the Dec. 27 election, with both sides accusing the other of "genocide." The death toll has reached 685, the government said Tuesday.
The opposition comments Tuesday were the latest sign the two sides are far from a compromise, and came just hours before former U.N. Secretary-General Kofi Annan was due in Kenya to mediate.
Yoweri Museveni, president of neighboring Uganda, met with Kibaki Tuesday on another mediation mission.
Anyang Nyongo, secretary-general of opposition leader Raila Odinga's Orange Democratic Movement, said the party would file complaint against the government and police with the International Criminal Court about "the abuse of rights by the police in this country." The case would name Kibaki, Cabinet ministers and the police commissioner, he said.
"The complaint states that crimes against humanity and state-sponsored terrorism are being committed by individuals as part of a systematic plan to target selected civilian populations in pursuit of political goals," Nyongo said.
It was not clear the opposition complaint would result in an international investigation. The Hague-based court has looked into information sent to it by scores of groups citing possible abuses in places ranging from Iraq to Ivory Coast, but has not yet opened a formal investigation based on such tips. It also investigates complaints sent to it by the Security Council or countries that signed the treaty creating the court in 2002.
So far the court has launched formal cases in just four countries: Sudan, Congo, Uganda and Central African Republic.
In a statement Tuesday from the Hague on the Kenyan opposition plans, the ICC Office of the Prosecutor noted Kenya was a party to the international statute establishing the court, and that it "considers carefully all information relating to alleged crimes within its jurisdiction committed on the territory of States Parties or by nationals of States Parties, regardless of the individuals or group alleged to have committed the crime."
The Dec. 27 Kenyan election returned Kibaki to power for a second five-year term, with official results putting Odinga second in the closest presidential race in Kenya's history. Odinga accused Kibaki of stealing the vote, and protests exploded into riots and ethnic fighting.
Foreign and local election observers have said the vote count was deeply flawed. Although the electoral chief pronounced Kibaki the victor, he later said he had been pressured to do so and did not know who won.
U.S. Ambassador Michael Ranneberger urged a political settlement.
"The tragedy Kenya is now suffering, and the extremely bitter polarization of Kenyan society, demands that all leaders and institutions speak in a responsible, respectful and dignified tone," Ranneberger said in a statement Tuesday.
The election has tapped into resentments that resurface regularly at election time in Kenya. But never before has the anger been so prolonged or taken so many lives.
On Tuesday, police fired tear gas Tuesday to disperse dozens of Kibaki supporters.
"Kibaki is our president!" the supporters shouted in downtown Nairobi before riot police broke up the gathering.
As Kibaki's power becomes more entrenched each day, the opposition's best hope may rest in working out a power-sharing agreement that could make Odinga prime minister or vice president.
Odinga has called for another "peaceful protest" on Thursday, saying, "let them bring their guns and we will face them."
The protest will take place in defiance of a ban and despite the deaths of at least 24 people in three days of protests last week _ most blamed on police.
Odinga also has urged supporters to boycott companies owned by Kibaki allies, including Brookside Diaries and bus companies Citi Hoppa and Kenya Bus. On Monday, the government condemned the economic boycott as sabotage.
| Seattle | Boston | |||
|---|---|---|---|---|
| ab r h bi | ab r h bi | |||
| ISuzuki rf | 5 1 2 0 | Ellsury cf | 5 1 2 1 | |
| Ryan ss | 4 2 2 5 | Pedroia 2b | 5 1 1 0 | |
| Ackley 2b | 5 0 2 1 | AdGnzl 1b | 5 2 3 2 | |
| Olivo c | 5 1 1 2 | Youkils 3b | 3 2 1 2 | |
| Smoak 1b | 4 0 1 0 | YNavrr 3b | 1 0 0 0 | |
| Carp lf | 4 0 2 0 | D.Ortiz dh | 5 2 2 0 | |
| AKndy 3b | 4 0 0 0 | Crwfrd lf | 4 2 3 2 | |
| Cust dh | 4 1 1 0 | Reddck rf | 5 2 2 1 | |
| FGtrrz cf | 3 3 2 0 | Sltlmch c | 4 0 3 4 | |
| Scutaro ss | 4 0 0 0 | |||
| Totals | 38 8 13 8 | Totals | 41 12 17 12 |
| Seattle | 200 | 010 | 401— 8 |
|---|---|---|---|
| Boston | 500 | 051 | 10x—12 |
E_Ryan (9). LOB_Seattle 5, Boston 7. 2B_Ryan (15), Ackley 2 (7), Smoak (21), Ellsbury (27), Pedroia (24), C.Crawford (13), Reddick (7). HR_Ryan (2), Olivo (14), Youkilis (14). CS_I.Suzuki (5), C.Crawford (5).
| IP | H | R ER BB SO | ||||
| Seattle | ||||||
|---|---|---|---|---|---|---|
| Pineda L,8-7 | 4 1-3 | 8 | 7 | 7 | 1 | 4 |
| Laffey | 0 | 4 | 3 | 3 | 0 | 0 |
| J.Wright | 1 2-3 | 2 | 1 | 1 | 0 | 2 |
| Lueke | 1 | 3 | 1 | 1 | 0 | 0 |
| League | 1 | 0 | 0 | 0 | 1 | 1 |
| Boston | ||||||
| Wakefield W,6-3 | 6 1-3 10 | 7 | 7 | 1 | 4 | |
| Aceves | 2 2-3 | 3 | 1 | 1 | 0 | 1 |
Laffey pitched to 4 batters in the 5th.
HBP_by Wakefield (Ryan). WP_Pineda.
Umpires_Home, Jeff Kellogg; First, Mark Carlson; Second, Eric Cooper; Third, Tim Timmons.
T_3:01. A_37,650 (37,065).
"Every time I watch a chef chop an onion, I learn something aboutthat person," Linda Carucci says.
She's an award-winning cooking teacher; the Julia Child Curatorof Food Arts at Copia, the food museum in Napa, Calif., and, mostrecently, the author of a cookbook that's receiving raves online.
Carucci's chief delight, it seems, is learning and teaching. Forher, it's a natural continuum. Inhale, exhale. Take in newinformation, give it to other people.
Her subject, of course, is cooking, and her first book ("Probablymy only book -- I don't know if I have any words left") is CookingSchool Secrets for Real World Cooks (Chronicle, $22.95). I wassurprised to find that although it's packed with tips,illustrations, recipes, advice, anecdotes and explanations, it's aconvenient-size paperback. This mountain of information is organizedfor accessibility and offered in reader-friendly prose. In ourrecent interview, Carucci praised the editors, designers, recipetesters, chef colleagues and her husband, Allen Rehmke, for theirpart in the project.
Cooking School Secrets makes a great gift for the recent collegegraduate setting up a new apartment and facing a kitchen alone forthe first time. It's also a great refresher course for experiencedcooks. Each recipe is explained so thoroughly that it's almost aclass in itself. Not surprising, since the book is based on theauthor's eight years running her own school, Linda Carucci'sKitchen, in Oakland, Calif.
It's a book that begs to be used, and the reward isn't justrecipes so tasty you forget they're instructional (do try theaccompanying recipes for grilled marinated flank steak au jus andthe savory corn pudding). You also have the pleasure of "meeting"Carucci herself in her writing. Her sense of enjoyment is aninvigorating, illuminating force.
"In my proposal I said I don't want a big, heavy coffee tablebook with a big price tag," she says. "I don't want something that'sgoing to make a dent in your belly when you read it in bed." Butproducing this usable, affordable cookbook was easier said thandone. "On the surface, you wouldn't know there were 116 recipetesters -- that there were home cooks in Burlington, Vt.; Elgin,Ill., and Glendora, Calif., who told me they could find pomegranatemolasses for the muhammara [a Middle Eastern condiment that alsoincludes roasted red peppers and walnuts] or rice noodles for theVietnamese grilled pork salad. Or that on the East Coast they toldme their halibut fillets always come with the skin on." (West Coasthalibut fillets come skinless.)
She realized the originally agreed-upon 150 recipes was just toomuch, and her editor agreed, then slashed the total by a breath-snatching 50. After Carucci had whittled down her "little darlings,"the editor went over the revised recipe list and noted, "You don'thave a chocolate cake in this book. You have to have chocolatecake."
And she had to have it in a week. She started with an idea fromLark Creek Inn in Larkspur, Calif., that sounds a bit quirky: Thischocolate cake calls for beets. (It's true, but your kids won't knowthey're eating a root vegetable unless you tell them.) Then sheexperimented with two kinds of icing: chocolate ganache, using NickMalgieri's technique, and her own adaptation of Hershey's cocoafudge frosting.
Fifty of her testers could turn a recipe around in 48 hours, soshe sent half of them the cake and ganache; the other half, the cakeand fudge frosting. She expected this to determine which icingworked better, but it was a tie. Even her editor couldn't decide, soboth frostings are in the book.
What kept her from writing it sooner? Carucci says it was theprospect of sitting alone at a computer terminal through 100,000words and scads of recipes. She was associate dean of students atOccidental College in Los Angeles when, in 1983, she moved to SanFrancisco to attend the California Culinary Academy. She was one ofits earliest "older students" (she was in her 20s at the time) andlater became its dean. She has been a caterer and, since 1997, hasoperated her own cooking school -- all people-related occupations.
Furthermore, she says, she tested "off the chart" as an extroverton the Myers-Briggs personality profile. That doesn't mean she hasto be the life of every party, but that "I draw my energy from otherpeople." Sit at a computer every day for a year? "What a dull,boring, horrible thing that would be for me." The recipe testersbecame her "lifeline," she says. "Every single morning I woke up andI ran to the computer" to see who'd checked in with triumphs,questions, comments, new problems, possible solutions. "It was likeI was teaching online."
And learning, of course. She gleans information from everyexperience, even breast cancer, which was diagnosed about 15 yearsago. She doesn't refer to herself as a survivor; the experience wasa career-altering fact of her life. She left the academy -- "Theguys couldn't handle it" -- and after free-lancing for a while, shestarted her school in 1997. She was back with her first love,teaching, working directly with people.
"If you're by yourself, how much fun can you have?" she asks. Andenjoyment -- fun -- is one of the priorities lined up in her life."That's one thing breast cancer does for you. I got to live. A lotof my friends in my support group didn't. So I just figure, it's gotto be fun."
For more information about Linda Carucci's Kitchen in Oakland,Calif., visit www.LCKitchen. com. For more on Copia, the AmericanCenter for Wine, Food and the Arts, in Napa, Calif., visitwww.copia.org.
GRILLED MARINATED FLANK STEAK AU JUS
MAKES 6 TO 8 SERVINGS
1 flank steak, about 1 1/2 pounds
2 cloves garlic, crushed
2 tablespoons Dijon mustard, preferably imported
2 tablespoons soy sauce
2 tablespoons Worcestershire sauce
2 tablespoons freshly squeezed lemon juice
1 tablespoon chopped fresh thyme
or 1 teaspoon dried leaf thyme, crumbled between your fingers
1/2 teaspoon freshly ground black pepper, plus more forsprinkling
1/2 cup boldly flavored extra-virgin olive oil
Kosher salt
Trim off any fat and silver skin from the flank steak, and pierceboth sides all over with a meat fork to tenderize the meat and openits pores. Rub both sides of the steak with the crushed garlic.Place the steak and garlic in a 1-gallon zip-top plastic bag.
In a small bowl, whisk together the mustard, soy sauce,Worcestershire sauce, lemon juice, thyme and the 1/2 teaspoonpepper. Slowly drizzle the olive oil into the mixture as you whisksteadily to form an emulsion. Pour the marinade into the bag withthe steak and seal the bag, pressing out as much air as possible.With your fingers, massage the marinade into the steak. Lay the bagflat in the refrigerator and marinate for 2 to 24 hours, turning thebag and massaging the meat periodically.
About 1 hour before serving, remove the bag from the refrigeratorand allow the steak and marinade to come to room temperature.Prepare a hot fire in a charcoal grill, or preheat a gas or stove-top grill to high. Remove the steak from the marinade and blot offexcess marinade with paper towels, removing any pieces of rawgarlic. Season the top side lightly with salt and pepper. Discardthe marinade.
When the grill is very hot (you can hold your hand over it foronly 2 seconds), place the steak on it, seasoned side down. Don'ttry to adjust the position of the steak at this point, as the meatwon't release from the grill until it's sufficiently seared. Cookthe steak on the first side for 3 minutes for rare, or for 4 minutesfor medium-rare. If you want to create cross-hatching, rotate thesteak a quarter turn after 2 minutes, then cook for 1 or 2 moreminutes.
Lightly sprinkle the top of the steak with salt and pepper, andturn to sear the second side. If desired, rotate the steak a quarterturn after 2 minutes to achieve cross-hatching. After 3 minutestotal cooking on the second side, test the internal temperature ofthe steak. Because flank steak is very lean, it's advisable not tocook it beyond rare (125 degrees) or medium-rare (130 degrees).Insert the thermometer on a slant into the center of the steak totest. If the steak is not done, continue grilling and test againafter 1 minute. (If your thermometer has a plastic -- not glass --housing covering the dial, don't leave it in the meat while it's onthe grill, or the plastic will melt.)
When the steak is done, transfer it to a cutting board,preferably one with a trough or well to capture the juices whencarving. Tent it with aluminum foil, or cover completely with aninverted bowl, and let stand for 5 minutes to allow the juices toretract into the meat. Using a carving knife, and holding it at a 45-degree angle, cut the steak across the grain into thin slices. Ifthe steak is more rare than you prefer, place individual slices backon the grill. The residual heat will continue to cook them untilthey reach your desired doneness. Arrange the slices, shingle-style(overlapped close together, see page 179) to retain heat, on awarmed platter. Use a spoon or metal bench scraper to scoop upjuices and drizzle them over the sliced steak.
From Cooking School Secrets for Real World Cooks
by Linda Carucci
Nutrition facts per serving: 371 calories, 29 g fat, 7 gsaturated fat, 48 mg cholesterol, 4 g carbohydrates, 26 g protein,583 mg sodium, 0 g fiber
SAVORY CORN PUDDING
MAKES 6 TO 8 SERVINGS
7 ears sweet corn
1 1/2 cups whole milk, divided (see Note)
2 teaspoons unsalted butter, for baking dish
5 large eggs
1 tablespoon plus 1 teaspoon all-purpose flour
1/4 pound Monterey Jack cheese, coarsely
shredded on the large holes of a box grater to yield 1 cup
2 tablespoons minced fresh chives, divided
1 teaspoon kosher salt
1/4 teaspoon Tabasco sauce (see Note)
Cut off the kernels from the ears of corn and reserve 1 cup.Place the remaining corn and 1 cup of the milk in a blender. Holdthe blender lid in place as you gradually increase and decrease thespeed. Process at the highest speed for a full 3 minutes. Ifnecessary, interrupt blending to scrape down the sides. You shouldhave a thick, smooth puree.
In the meantime, position a rack in the center of the oven andpreheat to 325 degrees. Prepare a bain-marie (hot-water bath):Choose a roasting pan large enough to hold a 9-inch round or squarebaking dish with 2-inch sides. Line the bottom of the roasting panwith a paper towel (to prevent the pudding from sliding around) andset aside. Butter the bottom and sides of the baking dish and setaside. Bring a kettle of water to a boil. Set aside.
In a bowl, whisk the eggs lightly. Add the blended corn mixtureand whisk to combine. Sprinkle with the flour and whisk to blendthoroughly. Add the remaining 1/2 cup milk, the cheese, all butabout 1 teaspoon of the chives, the salt, the Tabasco, and thereserved corn kernels and stir to combine. Pour into the preparedbaking dish and sprinkle with the remaining 1 teaspoon chives. (Thepudding can be prepared to this point up to 24 hours in advance,covered, and refrigerated. Bring to room temperature before baking.)
Pull out the oven rack halfway. Place the baking dish on thepaper towel in the bain-marie and place in the center of the ovenrack. Carefully pour the hot water from the kettle into the bain-marie to reach halfway up the sides of the baking dish. Gently slidethe oven rack into place. Bake the pudding until the blade of aparing knife inserted in the center comes out almost clean, 45 to 60minutes, depending on the depth of the baking dish. The top of thepudding should be firm and pale and the edges should begin to pullaway from the sides of the baking dish.
Remove the bain-marie from the oven. With oven mitts and/or along, wide offset spatula, carefully transfer the baking dish fromthe bain-marie to a cooling rack. Let cool for 5 minutes for puddingto set up a bit before serving.
Note: To make an even richer pudding, substitute half-and-half orheavy (whipping) cream for the milk. Likewise, to cut back on someof the fat, substitute a 12-ounce can evaporated low-fat milk forthe whole milk. Evaporated milk gives baked custards a very creamytexture.
The Tabasco sauce here is just enough to add a bit of complexity,without overpowering the sweet corn flavor. If you prefer,substitute 1/2 teaspoon cayenne pepper.
From Cooking School Secrets for Real World Cooks
Nutrition facts per serving: 268 calories, 15 g fat, 7 gsaturated fat, 238 mg cholesterol, 21 g carbohydrates, 15 g protein,522 mg sodium, 2 g fiber
Meaty issues: From Cooking School Secrets for Real World Cooks
Here are Linda Carucci's answers and comments on frequently askedquestions concerning meat.
Q. What's that iridescent membrane I sometimes see on the surfaceof pork tenderloin, turkey breast and some cuts of beef?
A. Occasionally (when the butcher hasn't removed it first),you'll find a layer of silver skin, a shiny whitish membrane thatseparates muscle groups, running along the surface of certainmuscles of meat. Silver skin is tough to chew and it shrinks andcurls up when it cooks, so it's critical that you remove it beforemarinating or cooking. Use a boning knife to trim off the silverskin by slicing along the grain, parallel to the meat.
Q. Why does my roast sometimes come out with a solid red core andgray edges?
A. The roast with the red core was too cold when it went into theoven. For even cooking, be sure meats are at room temperature beforecooking. This helps to achieve restaurant-quality roasts with acaramelized exterior and an interior that's evenly pink -- or medium-rare -- throughout.
Q. How do restaurants manage to serve hot, yet rare, prime rib?
A. This is a wonderful trick I learned when I worked in arestaurant that served prime rib. The whole roast would come out ofthe oven just before dinner. Then, as dinner service progressed, thechef would carve off pieces to order.
To reheat, he'd put a slab of meat on an aluminum pie plate,cover it with a large outer leaf of romaine lettuce, and pop itunder the broiler. In this example, the pan gets hot and transfersheat to the meat, and the lettuce keeps the meat both red and moistunder the direct heat of the broiler. By the time the lettuce startsto shrivel, the meat is hot.
Carucci's comments about flank steak and marinating:
I wouldn't even consider cooking a flank steak that hasn't beenmarinated to break down some of its fibers and boost its flavor.While it's not typical to include an acid, such as lemon juice, inan overnight marinade, the tough protein fibers of flank steakbenefit from the acid's tenderizing effect without the steakbecoming mushy.
You also could use this marinade for chicken legs and thighs,pork tenderloin or lamb chops. For the best results with flank teak,marinate overnight, cook rare and cut across the grain on thediagonal into one-quarter-inch-thick slices.
You'll notice that beyond the sodium in the soy andWorcestershire sauces, no additional salt is added to this marinade.By all means, do season with salt -- and pepper -- before grillingto create an integrated, flavorful crust. Seasoning after grillingwould result in less flavorful meat, with discernible flecks of saltand pepper that are not as pleasing to the palate.
Abstract Three experiments examined the effects of semantic characteristics of word pairs on memory using the encoding specificity paradigm. The paradigm involved four phases: (a) an encoding phase to relate cues and targets, (b) a phase in which words were generated to new cues, (c) a phase for recognition of generated targets, and (d) a cued-recall phase using the original encoding cues. Encoding pairs were classified a priori as either semantically similar (e.g., alluringPRETTY), semantically contrasting (e.g., drab-PRETTY), or semantically unrelated (e.g., sore-PRETTY). Generation pairs were classified a priori as either semantically similar (e.g., beautiful-PRETTY) or semantically contrasting (e.g., ugly-PRETTY). For recall, the results showed that both the semantic relations between the encoding cue and target and the reprovision of the encoding cue at retrieval were important factors. In the case of recognition, however, both the semantic congruence between the encoding and generation contexts and the amount of semantic elaboration provided by the encoding context were important factors.
The concept of encoding specificity was initially proposed by Endel Tulving and his colleagues to give an account of when and how retrieval cues are effective for episodic memory. "Specific encoding operations performed on what is perceived determine what is stored, and what is stored determines what retrieval cues are effective in providing access to what is stored" (Tulving & Thomson, 1973, p. 369). The essential idea is that a retrieval cue is effective only if information in the cue was incorporated in the memory trace of the target event at the time of its original encoding. In a later discussion of the concept, Tulving (1983) commented that "It is no more just an answer to the question concerning effectiveness of retrieval cues. It is now a theory about the relationship between encoding and retrieval conditions that is necessary for the recollection of an event to occur. We could now say that recollection of an event, or a certain aspect of it, occurs if and only if properties of the trace of the event are sufficiently similar to the properties of the retrieval information" (p. 223). This similarity between encoded information and information provided at retrieval (by cues or context) was also stressed by proponents of the concept of transfer-appropriate processing (Morris, Bransford, & Franks, 1977; Roediger, Weldon, & Challis, 1989) and was the central idea in the concept of repetition of operations proposed by Kolers (1973). It is difficult to see why this commonsense notion should be at all controversial (yet see Tulving, 1983, pp. 223-299).
The present study investigated the role of semantics in episodic memory for words. Tulving and Thomson (1973) drew the distinction between the semantic characteristics of words as lexical units and words as to-beremembered events. A semantically related cue should be effective in retrieving a word presented on a specific occasion only to the extent that the semantic information in question was encoded in the trace of the target word. Thus the word BRIDGE encoded as "an engineering structure" will not be easily recalled by the later cue "a card game." More subtly, the word WATER encoded in the pair whisky-WATER is not well recalled by the cue lake because whisky and lake emphasize rather different aspects of the word WATER Most dramatically, the same word re-presented in a recognition test may not be recognized because the word's context at encoding differs from its context at the time of recognition. This outcome was demonstrated in an ingenious paradigm devised by Tulving and Thomson (1973). They cast doubt on the influential generationrecognition model of retrieval by showing that when participants are explicitly induced to generate candidate responses and then given the opportunity to recognize target words from those generated, performance is quite poor.
The paradigm used by Tulving and Thomson to demonstrate this effect involved four major phases: (a) an encoding phase, (b) a generation phase, (c) a recognition phase for the words generated, and finally (d) a cued recall phase. During the encoding phase, weakly associated word pairs (e.g., whisky-WATER) designed to bias the capitalized words toward specific meanings, were presented. Participants were instructed not only to learn the capitalized words (e.g., WATER) but also to note that the preceding words (e.g., whisky) may help during retrieval. Thus the instructions influenced the encoding of the capitalized words to be semantically biased toward specific meanings (e.g., WATER as something you drink). During the generation phase, generation cues (e.g., lake) were presented. These cues were strongly associated to the capitalized words but they emphasized a different semantic meaning from that emphasized during encoding (e.g., WATER as something you swim in). Participants looked at each generation cue (e.g., lake) and then generated six related words (e.g., water, swim, cup, blue, fish, boat). During the recognition test that followed the generation phase, participants scanned their lists of generated words and circled those words that they believed were the capitalized words (e.g., WATER) previously shown in the encoding phase. Finally, participants completed a cued recall task by responding to the original cues (e.g., whisky) that were presented with the capitalized words at encoding. Tulving and Thomson observed a strong superiority of cued recall (63%) over recognition (24% of those targets generated in the generation phase) (Tulving & Thomson, 1973, Experiment 1). That is, cued recall was substantially higher than recognition memory, presumably because pairs such as whiskyWATER were encoded in a specific manner that made whisky an effective cue, but that made WATER itself, generated as a response to lake, relatively ineffective.
Tulving and Thomson's (1973) results thus questioned the generation-recognition account of retrieval, and favoured the encoding specificity principle. There was, however, an unexpected finding in their study. This finding was that certain target words were consistently recognized more frequently than others. Some words, such as CUT, BUG, WASH, STUPID, QUEEN, and BEING, were recognized more than 50% of the time, whereas other words, such as COLD, LIGHT, HIGH, SHORT, and WATER, were never recognized, even though each word was generated by anywhere from 13 to 21 participants. One plausible explanation for this puzzling result is that it again demonstrates the concept of encoding specificity. That is, high levels of recognition will be observed in the Tulving and Thomson paradigm when encoding processes and generation processes are semantically congruent, whereas recognition failure will occur when these two sets of processes are semantically incongruent. For example, the target word WEAK was well recognized in the Tulving and Thomson studies when its encoding context was brave-WEAK and the cue presented in the generation phase was strong. The senses of WEAK evoked by brave-WEAK and strong-WEAK are semantically congruent, and therefore the target word was well recognized. Conversely, it seems possible that the target words resulting in 0% recognition lacked this congruency relation. For example, somewhat different senses of COLD may be evoked by the encoding pair groundCOLD and the generation pair hot-COLD, and different senses of LIGHT may be evoked by the pairs headLIGHT and dark-LIGHT.
The following three experiments examined this hypothesis directly. To specify semantic congruency and incongruency more precisely, we created word pairs for the encoding and generation phases that exemplified either similar or contrasting semantic relations between the cue and target words. Word pairs classified as semantically similar have meanings that greatly overlap, for example, female-WOMAN, infantBABY, beautiful-PRETTY (e.g., Chaffin & Herrmann, 1984). Conversely, word pairs classified as semantically contrasting contain words whose meanings are opposed or contradictory in nature, for example, manWOMAN, white-BLACK, ugly-PRETTY Congruency between encoding and generation pairs was then achieved by using (for example) similar word pairs (e.g., female-WOMAN and lady-WOMAM, whereas an incongruent relation between encoding and generation was achieved by using (for example) a similar pair at encoding and a contrasting pair at generation (e.g., female-WOMAN and man-WOMAN, respectively). We used high-frequency noun pairs in Experiment 1, and high-frequency adjective pairs in Experiments 2 and 3. High-frequency noun and adjective pairs were chosen to avoid potential confounds with previous research investigating the effects of word type differences and word frequency on encoding specificity (e.g., Bartling, 1992; Reder, Anderson, & Bjork, 1974).
Our basic experimental design involved the presentation of weakly related cue-target pairs to be learned in Phase 1, presentation of a strongly related generation cue in Phase 2, recognition of generated targets in Phase 3, and finally cued recall of target words using the original weakly related cues in Phase 4. The encoding-target pairs and generation-target pairs were themselves rated as either similar, contrasting, or unrelated by independent judges. In the encoding phase (Phase 1), semantically similar (e.g., milkmaid WOMAN, contrasting (e.g., groom-WOMAN), or unrelated word pairs (e.g., floor-WOMAN) were presented, and these three encoding conditions were crossed with similar (e.g., female) or contrasting generation cues (e.g., man) in Phase 2. The major predictions were that target recognition would be high in the two cases of semantic congruency (similar/similar and contrasting/contrasting pairs) between encoding and generation, and that recognition would be low in the two cases of semantic incongruency (similar/contrasting and contrasting/similar pairs) between these phases. It was assumed that the conditions unrelated/similar and unrelated/contrasting would yield intermediate levels of recognition memory.
Experiment 1
METHOD
Participants. There were 62 participants who performed the main experiment: 37 were undergraduate psychology students from the University of Toronto who participated in the experiment for course credit, and 25 were students who were paid $5 for their participation. All participants were fluent in English and were tested individually. Data from two participants were not analyzed as they treated the final phase as free recall, rather than using the cues. In addition, a group of 39 participants generated free associates for norming purposes, and a further group of 20 participants rated semantic relations. None of these additional participants took part in the main experiments.
Encoding specificity task. The encoding specificity task was identical to that used by Tulving and Thomson (1973) with the exception that participants generated four words instead of six during the generation phase. This reduction in the number of words was not deemed problematic because Tulving and Thomson observed that 70% of their targeted words were generated as first responses. During the encoding phase, word pairs (e.g., milkmaid-WOMAN were presented in the centre of a computer screen and participants learned these word pairs for a subsequent memory test. In the generation phase, cues (e.g., man) were presented and participants wrote the first four words that came to mind (e.g., woman, bride, female, boy). During the recognition phase, participants scanned their generated words and circled those words that they believed were the capitalized words (e.g., WOMAN) previously shown during encoding. Finally, participants completed a cued recall task for the original targets by responding to the original encoding cues (e.g., milkmaid- ?).
Materials. Ali target words were high-frequency nouns with a Kucera and Francis (1967) average rating of 123 and a Thorndyke and Lorge (1944) rating of AA or A. The only exceptions were the targets medicine, basement, robber, and loser, which had Thorndyke and Lorge ratings of 46, 8, 27, and 2, respectively.
As prescribed by Tulving and Thomson (1973), the word pairs used in the encoding phase were only weakly related. To measure the associative strength for the pairs used, we had 39 participants, who were naive with respect to the purpose of the study, generate one free association to each encoding cue. In this preliminary part of the study, participants were given a sheet containing the encoding cues and were asked to write down the first word that came to mind after reading each cue. The target words used in the experiment were generated in 12%, 3%, and 2% of the trials to encoding cues that were similar, contrasting, and unrelated, respectively.
The word pairs in the encoding phase were classified a priori as semantically similar, contrasting, or unrelated, as described previously. These classifications were checked by having 20 norming participants rate the strength of the semantic relations between words in each pair on a scale running from 0 to 4, where ratings of 0.0 to 1.0 were defined as strongly opposite in meaning, ratings from 1.5 to 2.5 were defined as unrelated, and ratings from 3.0 to 4.0 were highly similar in meaning. As Table 1 shows, the experimenters' judgments were supported, in that contrasting pairs received a lower mean rating than both unrelated and similar pairs, and unrelated pairs received a lower mean rating than similar pairs. The semantic relations between the generation cues and target words were also classified a priori as similar or contrasting. The same 20 judges verified these classifications on the same scale of semantic associations. In this case, contrasting pairs were rated lower than similar pairs. Finally, the associative relations between the generation cues and targets, as well as those between the generation cues and encoding cues, followed the criteria set forth by Tulving and Thomson (1973). That is, all generation cues and targets had moderate to strong associative relations (36% for similar generation cues, and 27% for contrasting generation cues) and all generation cues and encoding cues had very weak associative relations (on average 0%). Associative relation norms were determined by having 40 norming participants generate one free association to each generation cue.
We also checked how the senses of the targets varied from encoding to generation. Twenty norming participants compared the senses of targets at encoding (e.g., milkmaid-WOMAN) to their respective senses at generation (e.g., female-WOMAN) on a scale running from 0 to 10, where a rating of 0 was defined as absolutely no similarity, a rating of 5 was defined as quite similar yet many differences too, and a rating of 10 was identical. As Table 1 shows, the average ratings for the congruent similar/similar condition yielded higher ratings than all other conditions, and the congruent contrasting/contrasting condition yielded higher ratings than all of the incongruent conditions. The pattern of these ratings suggests that semantic congruency between encoding and generation is more likely to evoke similar senses in the target words than semantic incongruency between encoding and generation.
Design and procedure. Prior to presentation of the critical list, two practice lists composed of 30 word pairs each were presented. To avoid proactive interference with the critical list, the words in the practice lists (e.g., ostrich-AUSTRALIA, bluejay-GREENLAND) were unrelated to the words in the critical list. There was no generation phase for these practice lists because their purpose was to encourage participants to relate the words in the word pairs. Therefore each practice list was followed by only a cued recall task, given after a 30-s arithmetic distractor task.
During the encoding of the critical list, 30 word pairs (3 primacy, 8 similar, 8 contrasting, 8 unrelated, and 3 recency) were presented in the centre of a computer screen for a duration of 3 s each. The primacy and recency pairs were presented in a fixed order, whereas the 24 critical stimuli were presented in a different random order for each participant. Each target was presented in upper-case directly below the encoding cue, which was presented in lower case letters. Participants were instructed not only to learn the capitalized words but to also note that the preceding words may help during subsequent retrieval. After the presentation of the critical list, participants performed a 30-s distractor task, namely, counting backwards from 100 by 3 s.
At this point, they were informed that the memory task, would come later, but that they would now perform a generation task, namely writing down four associates to a series of words. In this generation phase, the cues were presented on the computer screen at a self-paced rate. The generation cues were related to the targets in either a similar or contrasting fashion and each semantic generation type was divided equally among the semantic encoding types. For example, four of the eight similar word pairs from the encoding phase were matched with similar generation cues while the other four similar word pairs were matched with contrasting generation cues. Thus there were four word pairs in each of the six encoding/generation combinations: similar/similar, similar/contrasting, contrasting/similar, contrasting/contrasting, unrelated/similar, and unrelated/contrasting. All target words were counterbalanced across these six combinations.
During the recognition phase, participants were given as much time as they needed to review their generated words and circle those words that they believed were the capitalized words presented during the encoding phase. In the subsequent cued recall phase, participants were given up to 4.5 min to review the encoding cues and respond with the original targets. All recall cues were identical to the lowercase words presented during the encoding phase.
RESULTS
All ps reported are less than .05, unless otherwise reported. The proportions of targets generated to similar generation cues were 0.72, 0.63, and 0.58 for the conditions similar encoding, contrasting encoding, and unrelated encoding, respectively. The corresponding proportions for contrasting generation cues were 0.52, 0.48, and 0.39, respectively. Thus the similar generation cues were more effective than contrasting generation cues in generating target words.
Table 2 shows the proportions of generated target words recognized in the six conditions, conditionalized on generation of the targets. The main prediction was that these recognition proportions would be relatively large for the congruent conditions (similar/similar and contrasting/contrasting) and relatively small for the semantically incongruent conditions (similar/contrasting and contrasting/similar). Table 2 shows that whereas recognition proportions for the congruent conditions were large as predicted (0.46 and 0.53), and the incongruent contrasting/similar condition yielded a smaller recognition proportion, as predicted (0.32), the similar/contrasting recognition proportion was substantially higher than predicted (0.46). For both the similar and contrasting generation cues, words encoded with an unrelated encoding cue were least well recognized (0.27 and 0.36, respectively).
The predicted interaction between similar/contrasting encoding and similar/contrasting generation was found, F(1, 59) = 6.88, MSE = .01.(1) However, subsequent t-tests showed that whereas the similar/similar (0.46) and contrasting/contrasting (0.53) conditions yielded reliably higher recognition proportions than the contrasting/similar (0.32) condition, t(59) = 2.76, SE = .05 and t(59) = 3.69, SE = .05 respectively, neither the similar/similar nor contrasting/contrasting proportions reliably exceeded the value for the similar/contrasting (0.46) condition, t(59) = .06, SE = .05 and t(59) = 1.08, SE = .05, respectively, both ps > .28. The incongruent similar/contrasting condition thus yielded a recognition proportion that was higher than predicted. Table 2 also shows the results of the final cued recall phase. An ANOVA showed a highly significant effect of encoding condition, F(2,118) = 43.03, MSE = .03, and all pairwise t-tests were significant. The minimum t-value was t(59) = 4.21, SE = .03.
DISCUSSION
The incongruent similar/contrasting condition yielded a recognition proportion that was comparable to the recognition proportions for the congruent similar/simifar and contrasting/contrasting conditions. Therefore, the recognition results only partly supported our main prediction that recognition proportions would be relatively large for the congruent conditions and relatively small for the semantically incongruent conditions. One way of accounting for the unexpected high recognition proportion in the similar/contrasting condition is to suggest that a second factor besides semantic congruency is operating. For example, recognition proportions may be high following similar encoding conditions, regardless of the generation condition. Alternatively, recognition may be high following contrasting generation, regardless of the encoding condition. By this type of account, recognition would be high for those encoding and generation types that are either congruent or involve the second factor. Of the two possibilities, perhaps the factor of similar encoding relations is the more plausible as it may be associated with greater amounts of elaboration (cf. Craik & Tulving, 1975). On the other hand, the unrelated/contrasting condition (0.36) gave a higher recognition proportion than the unrelated/similar condition (0.27), so perhaps the generation of a contrasting target leads to better recognition. Further discussion is deferred until the results of Experiment 2 are considered.
A second interesting result was that for the final cued recall phase there were large differences among the three encoding conditions despite the fact that the original encoding cues were reprovided in all cases. This result may simply reflect the greater associative strength inherent in similar (12%) than in contrasting (3%) or unrelated (2%) pairs, although the difference between the second two types remains puzzling if associative strength is the sole explanation for the differences among the three encoding conditions. To test the robustness of both the recognition and cued recall results, a second experiment was carried out using adjective pairs as stimuli instead of nouns.
Experiment 2
METHOD
Participants. There were 62 participants who performed the main experiment: Fifty-two were undergraduate psychology students from the University of Toronto who participated in this experiment for course credit, and 10 were students who were paid $5 for their participation. All participants were fluent in English and were tested individually. Data from two participants were excluded from further analysis: one because of lack of compliance with instructions and one because of experimenter error. In addition, a group of 40 participants generated free associates for norming purposes, and a further group of 20 participants rated semantic relations. None of these additional participants took part in the main experiment.
Materials. The target words in this experiment were common adjectives with an average rating of 173 in the Kucera and Francis (1967) norms and Thorndyke and Lorge (1994) ratings of AA or A. Encoding cues and generation cues were also adjectives. As in Experiment 1, the associative relations between encoding cues and targets were determined by having 40 norming participants generate one free associate to each encoding cue. Participants generated the target word on 14%, 0%, and 2% of occasions for similar, contrasting, and unrelated encoding cues, respectively. The relations between generation cues and the target words were measured in the same way. In this case, 41 participants generated the target words on 47% of occasions for similar generation cues, and on 31% of occasions for contrasting cues.
As in Experiment 1, 20 norming participants rated the encoding cue-target pairs a priori as similar, contrasting, or unrelated using a scale ranging from 0.0 to 4.0. As Table I shows, the experimenters' judgements were supported in that the contrasting pairs received a lower mean rating than both unrelated and similar pairs, and unrelated pairs received a lower mean rating than similar pairs. The semantic relations between the generation cues and target words were also classified a priori as similar or contrasting. The same 20 judges verified these classifications on the same scale of semantic associations. In this case contrasting pairs were rated lower than similar pairs.
As in Experiment 1, 20 norming participants also evaluated how the senses of the targets varied from encoding to generation by comparing the senses of targets at encoding (e.g., alluring-PRETTY to their respective senses at generation (e.g., beautiful-PRETTY) on a scale running from 0 to 10. As Table 1 shows, the average ratings for the congruent similar/similar condition yielded higher ratings than all other conditions, and the congruent contrasting/contrasting condition yielded higher ratings than all of the incongruent conditions. The pattern of these ratings suggests that semantic congruency between encoding and generation is more likely to evoke similar senses in the target words than semantic incongruency between encoding and generation.
Design and procedure. The procedure was identical to that used in the first experiment except that in Experiment 1 the presentation duration for word pairs during the encoding phase was 3 s each, whereas in Experiment 2 the presentation duration was 8 s each. The longer presentation duration was used because pilot testing had shown that it was needed to yield appropriate levels of performance.
RESULTS
In the generation phase of Experiment 2, an average proportion of 0.79 target words were generated to similar generation cues, and an average proportion of 0.52 target words were generated to contrasting generation cues. In the case of similar generation cues, the overall rate of 0.79 was made up of individual rates of 0.80, 0.80, and 0.76 for similar, contrasting, and unrelated encoding cues, respectively. The corresponding proportions for the contrasting generation cues were 0.45, 0.59, and 0.51, respectively.
Table 2 shows the proportions recognized, conditionalized on target generation. The original prediction was that congruent encoding/generation conditions (i.e., similar/similar and contrasting/contrasting) would yield comparatively high levels of recognition, whereas the incongruent conditions (i.e., similar/contrasting and contrasting/similar) would yield low levels, with the unrelated encoding conditions yielding intermediate levels. This pattern was obtained in the case of the similar generation cues (0.40, 0.20, and 0.28 for similar, contrasting, and unrelated encoding cues, respectively), but not for the contrasting generation cues, where the corresponding proportions were 0.43, 0.42, and 0,26, respectively. That is, as in Experiment 1, the semantically incongruent similar-encoding/contrasting-generation condition yielded an anomalously high proportion - in this case the highest value of recognition. As in Experiment 1, these findings were supported by a significant Encoding x Generation interaction F(2, 118) = 6.07, MSE = .066.(2) A series of t-tests among the recognition proportions revealed the presence of two clusters. The values 0.40, 0.43, and 0.42 did not differ statistically among themselves, maximum t-value was t(59) = .63, p > .53, and the values 0.20, 0.28, and 0.26 did not differ among themselves, maximum t-value was t(59) = 1.29, p > .20. However, all pairs of proportions across the clusters were significantly different, minim u m t-value was t(59) = 2.83.
Table 2 also shows the values for cued recall. As in Experiment 1, recall levels decreased from similar to contrasting to unrelated encoding conditions. An ANOVA showed a reliable effect of encoding condition, F(2,118) = 33.82, MSE = .032, with all proportions differing statistically as shown by subsequent t-tests, minimum t-value was t(59) = 2.98.
DISCUSSION
The recognition results of Experiment 2 thus replicated the essential findings from Experiment 1. That is, recognition was comparatively high both in the cases of semantic congruence between encoding and generation, and in the case of similar encoding followed by contrasting generation. The other possible interpretation from Experiment 1, that good recognition performance is associated with contrasting generation, is essentially ruled out here by the finding of a low recognition score for the unrelated/contrasting case (0.26). Rather, it seems that a similarity relation at encoding ensures a comparatively high proportion of subsequent recognition regardless of how the target word is generated. A contrasting relation at encoding yields good recognition only if the generation cue is also contrasting, and the unrelated encoding conditions yield low levels of recognition in all cases.
The cued recall results of Experiment 2 also replicated the essential findings from Experiment 1. That is, despite the use of the original encoding cues as retrieval cues, recall performance varied substantially as a function of the semantic relations between cues and targets. As in Experiment 1, this result may simply reflect the greater associative strength inherent in similar (14%) than in contrasting (0%) or unrelated pairs (2%), although by this account the difference between the contrasting and unrelated pairs cannot be explained by using an associative strength argument.
Thus, it appears that the pattern of results for recognition and cued recall in Experiment 2 is identical to that of Experiment 1. Our explanation for this pattern of results is that another factor, in addition to semantic congruency, influences recognition and recall. This factor is that the three types of semantic encoding have varying levels of semantic elaboration. Under standard encoding conditions, similar encoding yields rich semantic elaboration, contrasting encoding yields moderate semantic elaboration, and unrelated encoding yields relatively impoverished semantic elaboration. By this account, recognition should be high for those encoding and generation types that have either rich elaboration at encoding or semantic congruency between encoding and generation. In contrast, recognition for moderate to poor levels of elaboration at encoding should be high only if there is congruence in semantics between encoding and generation types. If there is no semantic congruence between encoding and generation types then recognition should be low. In respect to cued recall, recall scores should decline as the amount of semantic elaboration at encoding decreases - from extensive to moderate to restricted.
To test this hypothesis, in Experiment 3 we controlled semantic elaboration at encoding. The participants were presented word pairs that were either read (erg., alluring-PRETTY), generated easily (e.g., alluring PRETT -), or generated with effort (e.g., alluring-PRE - ). Because words that are generated have higher recognition levels than those read (e.g., Slamecka & Graf, 1978) we anticipated that those requiring minimal generation (i.e., generated easily) should yield intermediate recognition levels between read and effortful. Indeed, pilot research in our lab showed that hard-generate, easy-generate, and read types of encoding yielded three distinct levels of recognition (0.76, 0.58, 0.32, respectively) and cued recall (0.81, 0.66, 0.40, respectively). To add further control to our experiment, we used only the semantically similar word pairs from Experiment 2.
As in Experiments 1 and 2, our basic experimental design involved encoding, generation (similar, contrasting), recognition, and cued recall phases. However, during the encoding phase participants read, generated easily, or generated with effort the critical targets. The major predictions were that target recognition would be high in the three conditions that included either effortful generation at encoding and/or semantic congruency between encoding and generation, and recognition would be low in the three conditions that included either easier generation/reading at encoding and/or semantic incongruency between encoding and generation. For cued recall, we predicted that performance levels would decline from hard generate to easy generate to read encoding conditions.
Experiment 3
METHOD
Participants. The 36 participants were students from the University of Toronto who were paid $5 for their participation. All participants were fluent in English and were tested individually.
Encoding specificity task. The encoding specificity task was identical to that used in Experiments 1 and 2, with the exception that the encoding phase was changed from an intentional task to an incidental one in which participants either read (e.g., alluring-PRETTY or generated the critical targets (e.g., alluring-PRETT, or alluringPRE---).
Materials. The word pairs used in the encoding phase consisted of 66 pairs: 24 critical, 36 filler, 3 primacy, and 3 recency. The 24 critical pairs were the semantically similar ones taken from Experiment 2. Semantically similar pairs were chosen in order to yield reasonably high levels of recognition, given the switch from intentional learning in Experiments 1 and 2 to incidental learning in this experiment. Three versions of each critical word pair were created: hard-generate, easy-generate, and read. Hard-generate pairs consisted of a lowercase cue and a capitalized word stem with more than one letter missing (e.g., alluring-PRE - - ). Easygenerate pairs consisted of a lowercase cue and a capitalized word stem with the last letter missing (e.g., alluring-PRETT-). Read pairs consisted of a lowercase cue and a capitalized word (e.g., alluring-PRETTY).
The 36 filler pairs were not semantically related to the critical pairs. Moreover, 12 were read, 12 were easy-generate, and 12 were hard-generate. Six of the filler pairs were semantically similar (e.g., expandINCREASE) and 30 were semantically contrasting (eg., good-EVIL). The semantically contrasting pairs were included to prevent participants from adopting a "similar relation strategy" for generating words.
Design and procedure. As in Experiments 1 and 2, two practice lists were presented. However, prior to the retrieval phase of the second practice list, participants were informed that there was another task for them to do and that this task was designed to interfere with their remembering of the words they had just learned. This "interference" task was our incidental task that contained the critical target words. In this incidental task, each target word was presented in upper case directly below the encoding cue, which was presented in lower case letters. Participants were told that when they see word pairs on the computer screen (e.g., tie SHIRT) they should read them out loud. They were also told that when they see word pairs with letters missing (e.g., polite-RUD - or keyboard-PI - - -) they should read the first word out loud and then complete the word stem out loud with a word related to the first word (e.g., RUDE, PIANO).
During the encoding phase of the incidental task, 66 word pairs (3 primacy, 24 critical: 8 read, 8 easy-generate, 8 hard-generate; 36 filler: 12 read, 12 easy-generate, 12 hard-generate, and 3 recency) were presented in the centre of a computer screen at a self-paced rate. The primacy and recency pairs were presented in a fixed order, whereas the 24 critical and 36 filler stimuli were presented in a different random order for each participant. Participants were given 10 seconds to generate words. Once 10 seconds had elapsed the experimenter provided the appropriate word. Occasionally (less than 2% of the time) participants generated words that were not the critical targets. In instances such as this, the experimenter recorded the anomaly and the word was removed from the critical target list for that participant.
As in Experiments 1 and 2, participants completed a recognition phase in which they free-associated words to recognition-generation cues and then circled those free associates that they believed were the capitalized words that they either read or generated during the encoding phase of the incidental task. The two types of generation cues for the recognition phase (i.e., similar, contrasting) were identical to those used in Experiment 2, and each of the semantic generation types for recognition were divided equally among the three generation types for encoding. Participants also completed a cued recall task containing recall cues that were identical to the lower case words presented during the encoding phase of the incidental task. Thus, there were four word pairs in each of the six encoding/generation combinations. All target words were counterbalanced across these six combinations.
RESULTS
In the recognition-generation phase of Experiment 3, an average proportion of 0.77 target words were generated to similar generation cues, and an average proportion of 0.50 target words were generated to contrasting generation cues. In the case of similar generation cues, the overall rate of 0.77 was made up of individual rates of 0.80, 0.71, and 0.79 for similar, contrasting, and unrelated encoding cues, respectively. The corresponding proportions for the contrasting generation cues were 0.48, 0.52, and 0.51, respectively.
Table 2 shows the proportions recognized, conditionalized on target generation. Our prediction was that conditions that had either effortful generation at encoding or were semantically congruent would yield comparatively high levels of recognition, whereas conditions that had easier generation/reading at encoding or were semantically incongruent would yield low levels of recognition. As Table 2 shows this prediction was confirmed. Conditions that had either effortful generation at encoding or were semantically congruent yielded higher recognition levels (i.e., .50, .51, and .50) than those that had easier generation/reading at encoding or were semantically incongruent (e.g., .32, .23, and .27). This pattern of results was supported by a marginal Encoding x Generation interaction, F(2, 70) = 2.92, MSE = .09, p < .06. A series of t-tests among the recognition proportions revealed the presence of two clusters. The values .50, .51, and .50 did not differ statistically among themselves, maximum t-value was t(35) = .19, p > .85 and the values .32, .23, .27 did not differ among themselves, maximum t-value was t(35) = 1.65, p > .10. However, all pairs of proportions across the clusters were significantly different, minimum t-value was t(35) = 2.69.
Table 2 also shows the results of the final cued recall phase. Cued recall scores were highest in the hard-generate condition, intermediate in the easy-generate condition, and lowest in the read encoding condition, despite the facts that the original cues were reprovided in all cases and that all the critical word pairs were semantically related with the same associative relation. An ANOVA showed a highly significant effect of encoding condition, F(2, 70) = 46.51, MSE = .03 and all pairwise t-tests were significant. The minimum tvalue was t(35) = 3.51. These findings clearly confirm that recall increases as the difficulty level of generation at encoding increases (Jacoby, 1978; Slamecka & Graf, 1978).
DISCUSSION
The patterns of results for recognition and cued recall in Experiment 3 lend support to our suggestion that another factor besides semantic congruency influences recognition and recall. This factor is the extent of elaboration at encoding. This account is discussed further in the next section. With regard to cued recall, similar beneficial effects of effortful or difficult initial processing on subsequent memory performance have been found in experiments on interference effects with initial stimulus identification (e.g., Hirshman, Trembath, & Mulligan, 1994; Mulligan, 1996) and in cases in which textual coherence is established with effort during initial reading (Duffy, Shinjo, & Myers, 1990; Myers, Shinjo, & Duffy, 1987). In all cases, the higher memory performance appears to be attributable to the enhanced processing of higher-order perceptual or semantic representations at the time of initial encoding (Craik & Lockhart, 1972).
General Discussion
The starting point of our study was the observation by Tulving and Thomson (1973) that recognition rates in their generate-recognize paradigm were quite variable, depending on the particular word pairs examined. Our suggested explanation was that recognition would be high in cases where the encoding cue-target relation and the generation cue-target relation were semantically congruent, and that recognition of generated targets would be low when these relations were semantically incongruent. However, the observed pattern of results in Experiments I and 2 conformed to this prediction only partly. In both experiments recognition levels were comparatively high in cases of semantic congruence (similar encoding/similar generation, and contrasting encoding/contrasting generation) in line with our explanation, but recognition was also high in the incongruent combinations of similar encoding/contrasting generation. Recognition levels were comparatively low in the contrasting encoding/similar generation combination, and in combinations involving unrelated encoding cue-target pairs. Our modified account of these results is that apparently word recognition can be good either if the encoding/generation contexts are semantically congruent, or if the encoding cue-target relation is one of richer elaboration as in the case of semantic similarity. It is also clear from Table 2 that recognition performance is no higher when both of these factors hold (i.e., in the case of similar encoding/similar generation).
The original suggestion that semantic congruency between encoding cue-target pairs and generation cuetarget pairs would be associated with comparatively high levels of target recognition is parallel with the notions of encoding specificity and transfer-appropriate processing. That is, the encoding cue modifies and specifies the exact sense of the target word, and when the generation cue produces the target word with the same semantic nuance, it will be recognized comparatively well because the encoded and generated targets share many of the same semantic features. By this account, recognition levels should be low when encoding and generation induce somewhat different senses of the target word, and we argue that this is the case for the contrasting/similar condition, and for the cases involving unrelated encoding cues.
But why should similar encoding yield good recognition regardless of how the recognition target was generated? Our explanation is that, under standard encoding conditions, the similar encoding cue elaborates and enriches the target word in much the same way as positive semantic orienting task questions do in the levels of processing paradigm (Craik & Tulving, 1975). In this latter paradigm, it is well established that question-target pairs that yield positive responses (e.g., "type of furniture?-SETTEE') are associated with higher levels of subsequent target word recognition than pairs that yield negative responses (e.g., "a jungle animal?POTATO"). Craik and Tulving suggested that recognition superiority was a function of the greater ease of forming a richly elaborated trace in the case of positive question-target pairs. Consistent with this suggestion, the results in Experiment 3 showed that richly elaborated targets may be recognized well despite a shift in semantic context from encoding to retrieval.
Our suggestion that similar encoding cues yield comparatively well-elaborated traces is supported by the further finding in Experiments 1 and 2 that cued recall is significantly higher for similar encodings than for contrasting or unrelated encodings. The notion that, in general, recognition performance is less affected by contextual change than recall, for example, is supported by findings from the state-dependent learning literature. In several studies of this type, researchers have reported large changes in recall performance, but relatively small changes in recognition performance, as a function of state or context changes between encoding and retrieval (e.g., Baddeley, 1976; Eich, 1980; Godden & Baddeley, 1975). The present data suggest the interesting possibility that as the encoded trace is made richer and more elaborate, it becomes progressively less dependent on reinstatement of the original encoding context for later recognition to occur successfully.
Finally, our suggestion that well-elaborated memory traces may be recognized well despite a shift in semantic context from encoding to retrieval is supported by our results in Experiment 3. In this experiment, we observed that targets that had richer elaboration at encoding (i.e., hard-generate condition) yielded high recognition levels even when there was a shift from similar semantic encoding to contrasting semantic generation. However, the results in Experiment 3 also suggested that when semantic elaboration is only moderate at encoding, as in the easy-generate condition, recognition is strongly affected by a shift from similar-semantic encoding to contrasting-semantic generation.
Other possible factors affecting recognition variability, such as word frequency or word type differences, are essentially ruled out by the present results. For example, it is well known that low-frequency words are better recognized than high-frequency words (Gregg, 1976), and consistent with this finding, Reder, Anderson, and Bjork (1974) observed this same result using the encoding specificity task. Nevertheless, in the present study, recognition levels varied considerably despite the fact that only high-frequency targets were used. Similarly, it has been suggested that recognition failure occurs when adjectives are the target words and nouns are the encoding cues and not when just noun-noun (milkmaid-WOMAN) and adjectiveadjective (alluring-PRETTY pairs are used (Bartling, 1992; Bartling & Thompson, 1977; Nilsson & Gardiner, 1993). However, the present results demonstrated that recognition levels can be comparatively high or low, regardless of whether the encoding pairs are adjectives or nouns.
The cued recall findings are more straightforward. In both Experiments 1 and 2, cued recall levels were highest for words encoded with (and retrieved by) a cue that bore a similar relation to the target word, and recall was significantly lower for contrasting cues and lowest for unrelated cues. These variations in cued recall occurred despite the fact that the encoding cue was reprovided at retrieval in all cases, and thus in a sense the conditions for the encoding specificity principle to work positively were always present. The associative norms collected in the course of this study seemed to support the argument of stronger associative relations between cues and targets for similar cues than for the other two types (means of 13% compared with 2%, respectively). However, the contrasting and unrelated associative norms had the same generation rate yet still resulted in significantly different levels of cued recall. Moreover, when associative relations were held constant by using only semantically similar pairs (as in Experiment 3), cued recall still varied as a function of elaboration at encoding. Thus, it seems that the cued recall performances in all three experiments are accounted for by the amount of elaboration afforded by the semantic relation between cue and target (e.g., similar > contrasting > unrelated) rather than by preexisting associative relations.
One possible problem with the cued recall results is that cued recall always followed recognition in this paradigm, so it is possible that the pattern of recall was influenced, or even caused, by prior recognition. The potential seriousness of this point is underlined by the fact that the mean recognition probabilities in all three experiments are ordered similar > contrasting > unrelated (or hard-generate > easy-generate > read for Experiment 3), in the same way recall is ordered. Further analysis reduced this concern, however. For each condition, we multiplied the generation probability by the conditional probability of recognition to find the proportion of encoded items that were recognized in that condition. For example, in the similar encoding condition in Experiment 1, the two generation probabilities were 0.72 and 0.52 for similar and contrasting generation, respectively. When these generation probabilities are multiplied by the conditional recognition probabilities given in Table 2 (0.46 and 0.46, respectively), we observed 0.72 x 0.46 = 0.33 and 0.52 x 0.46 = 0.24. The average of 0.33 and 0.24 is 0.29; thus for the similar encoding condition in Experiment 1, 0.29 of the original items were generated and recognized. The corresponding recognition probabilities for the contrasting and unrelated conditions were 0.23 and 0.15, respectively. If we make the extreme assumption that all recognized items were subsequently recalled, we can simply subtract these "contaminated" items from the cued recall results to yield a measure of recall that is uninfluenced by prior recognition. After subtraction, the cued recall results for Experiment 1 were 0.35, 0.27, and 0.19 for the similar, contrasting, and unrelated conditions, respectively. The corresponding proportions for the other experiments were 0.21, 0.15, and 0.02 for Experiment 2, and 0.30, 0.22, and 0.09 for Experiment 3. The finding that the pattern of cued recall is maintained in all three experiments after removing the effects of prior recognition argues strongly in favour of cued recall reflecting the degree of elaboration achieved in the initial encoding phase.
The most striking finding in the original Tulving and Thomson (1973) study was that recognition levels were considerably lower than cued recall levels. In the present study, recognition was lower than recall in approximately half of the conditions. The important point is not the comparative levels of recognition and recall, but the fact that somewhat different factors affect these two measures of memory. This finding is consistent with the findings of Wiseman and Tulving (1976), who demonstrated that regardless of whether the overall level of recognition was higher or lower than the level of cued recall, substantial amounts of recognition failure were observed in all cases.
The main contribution of the present study is the demonstration that levels of recognition memory and cued recall can vary considerably between conditions in the Tulving and Thomson paradigm (recognition ranged from 0.20 to 0.53 and recall ranged for 0.20 to 0.64) and that somewhat different factors are important in the two cases. For recall, the crucial factors appear to be the depth and semantic richness of the initial encoding, the strength of the semantic relation between the encoding cue (or context) and the target, and the reprovision of the cue or context at the time of retrieval (Fisher & Craik, 1977; Morris et al., 1977; Roediger et al., 1989). In the case of recognition, the present results show that the semantic congruence between the encoding and retrieval contexts is an important factor as well as the amount of semantic elaboration provided by the encoding context. These results suggest that a semantically rich initial encoding may be associated with high levels of subsequent recognition regardless of the compatibility of the encoding and retrieval contexts.
We thank M. Masson and an anonymous reviewer for their helpful comments on an earlier draft of this paper. Correspondence concerning this article should be addressed to Brenda Hannon, University of Toronto, Mississauga, Ontario, L5L 1C6 (E-mail: brendah@psych.utoronto.ca).
[Reference]
References
[Reference]
Baddeley, A.D. (1976) The psychology of memory. New York: Basic Books
Bartling, C.A. (1992) List subset effects and the Tulving-- Wiseman function. Bulletin of the Psychonomic Society, 30, 131 134.
[Reference]
Bartling, C. A., & Thompson, C. P. (1977). Encoding specificity: Retrieval asymmetry in the recognition failure paradigm. Journal of Experimental Psychology: Human Learning and Memory, 3, 690-700.
Chaffin, R., & Herrmann, D. J. (1984). The similarity and diversity of semantic relations. Memory ear Cognition, 12, 134141.
[Reference]
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671-684.
Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology: General, 104, 268-294.
Duffy, S. A., Shinjo, M., & Myers, J. L. (1990). The effect of encoding task on memory for sentence pairs varying in causal relatedness. Journal ofMemory and Language, 29, 27-42.
Eich, J. E. (1980). The cue-dependent nature of state-dependent retention. Memory er Cognition, 8, 157-173.
Fernald, J. C. (Ed.) (1947). Funk and Wagnalls standard handbook of synonyms, antonyms, and prepositions. New York: Funk and Wagnall's Company, Inc.
Fisher, R. P., & Craik, F. I. M. (1977). Interaction between encoding and retrieval operations in cued recall. Journal of Experimental Psychology: Learning, Memory and Cognition, 3, 701711.
[Reference]
Godden, D.R., & Baddeley, A.D. (1975) Context-dependent memory in two natural environments: On land and underwater. British Journal of Psychology, 66, 325-331.
Gregg, V. (1976). Word frequency, recognition and recall. In J. Brown (Ed.), Recall and recognition (pp. 183-216). London: Wiley.
[Reference]
Hayakawa, S., & the Funk & Wagnall's Dictionary Staff. (1968). Modern guide to synonyms and related words. New York: Funk and Wagnall's.
Hirshman, E., Trembath, D., & Mulligan, N. W. (1994). Theoretical implications of the mnemonic benefits of perceptual interference. Journal of Experimental Psychology: Learning Memor and Cognition, 20, 608-620.
Jacoby, L. L. (1978). On interpreting the effects of repetition: Solving a problem versus remembering a solution. journal of Verbal Learning and Verbal Behavior, 17, 649-667.
Kolers, P. (1973). Remembering operations. Memory ear Cognition, 1, 347-355.
Kucera, H., & Francis, W. N. (1970). Computational analysis of present day American English, Providence, Ri: Brown University Press.
[Reference]
Lewis, N. (Ed.) (1980). The new Roget thesaurus in dictionary form. New York: Berkeley Books.
Morris, C. D., Bransford, J. D, & Franks, J. J. (1977). Levels of processing versus test-appropriate strategies. Journal of Verbal Learning and Verbal Behavior, 16, 519-533.
Mulligan, N. W. (1996). The effects of perceptual interference at encoding on implicit memory, explicit memory, and memory for source. Journal of Experimental Psychology: Learning, Memor and Cognition, 22, 1067-1087.
Myers, J. L., Shinjo, M., & Duffy, S. A. (1987). Degree of causal relatedness and memory. Journal of Memory and Language, 26, 453-465.
Nilsson, L. G., & Gardiner J. M. (1993). Identifying exceptions in a database of recognition failure studies from 1973 to 1992. Memory ear Cognition, 21, 397-410.
[Reference]
Reder, L. M., Anderson, J. R., & Bjork, R. A. (1974). A semantic interpretation of encoding specificity. Journal of Experimental Psychology, 102, 648-656.
Roediger, H. L., Weldon, M. S., & Chalis, B. A. (1989). Explaining dissociations between implicit and explicit measures of retention: A processing account. In H. L. Roediger III and F. I. M. Craik (Eds.), Variations of memory and consciousness: Essays in honour of Endel Tulving (pp. 3-41). Hillsdale, NJ: Erlbaum.
[Reference]
Slamecka, NJ., & Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental Psychology: Human Learning and Memory, 4, 592-604.
Thompson, D. (Ed.) (1993). The Oxford dictionary of current English: New Edition. Toronto, ON: Oxford University Press. Thorndyke, E., & forge, I. (1944). The teachers word book of 30,000 words. New York: Bureau of Publications; Teacher's College, Columbia University.
Tulving, E. (1983). Elements of episodic memory. New York: Oxford University Press.
Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80, 352-373.
Wiseman, S., & Tulving, E. (1976). Encoding specificity: Relation between recall superiority and recognition failure. Journal of Experimental Psychology: Human Learning and Memory, 2, 349-361.
Date of acceptance: October 2, 2000
[Reference]
Sommaire
[Reference]
Selon le principe de l'encodage specifique, " les operations specifiques d'encodage de ce qui est percu determinent ce qui est emmagasine, et ce qui est emmagasine determine quel sont les indices de recouvrement dormant acces a ce qui est emmagasine. - (Tulving & Thomson, 1973, p. 369, traduction libre). En d'autres mots, le rappel indice est possible seulement si (information contenue dans le rappel est incorporee a la trace mnesique relative a l'evenement cible au moment de fencodage initial. Ce principe a ete illustre a l'aide d'un ingenieux modele en quatre phases, mis au point par Tulving et Thompson (1973), qui se resume ainsi a) une phase d'encodage de couples de mots a correspondance faible (p. ex., whisky-EAU) constitues de facon a induire un biais dans le sens du mot ecrit en lettres majuscules; b) une phase d'encodage basee sur la categorie, au cours de laquelle on donne aux participants des indices categoriels (p. ex., lac) laissant prevoir le rappel du mot-cible encode (p. ex., EAU)
[Reference]
mais possedant un sens legerement different; c) une phase de reconnaissance au cours de laquelle on demande aux participants de reconnaitre tous les mots produits par les cibles encodees; et d) une phase de rappel indice pendant laquelle le mot indice est presente de nouveau. Les resultats ont demontre qu'a la troisieme phase la reconnaissance des mots produits etait considerablement plus faible que le rappel indice a la quatrieme phase. La discordance presente dans l'information semantique au test de reconnaissance a donne lieu a un rendement mnesique faible.
De facon inattendue, certains mots-cibles ont ete reconnus plutot frequemment a la troisieme phase, tandis que d'autres ne Pont ete par aucun participant. Le but de la presente recherche etait justement d'examiner ces ecarts. Dans les etudes de ce type, les couples de mots relies par association presentent parfois des significations apparentees (p. ex., neige-BLANC) et d'autres fois, des sens opposes (p. ex., noir-BLANC). Nous
[Reference]
avons predit que la reconnaissance des mots produits serait elevee si le couple de mots a encoder et le couple de mots produits etaient similaires du point de vue semantique (p. ex., couple de mots au sens apparente ou couple de mots au sens oppose). Dans l'ensemble, les experiences 1 et 2 ont permis de confirmer cette provision, au moyen de couples de noms, dans la premiere experience, et d'adjectifs, dans la deuxieme. Nous avons cependant releve, au cours des deux experiences, un resultat anormal: la reconnaissance etait bonne lorsque des couples de mots au sens apparente etaient utilises pour l'encodage, qu'ils proviennent ou non d'une categorie apparentee a leur sens. Nous avons a ce sujet emis l'hypothese que les couples de mots semblables etaient encodees de facon tres elaboree et qu'un tel degre d'encodage en preservait des effets de legers changements de sens durant les phases subsequentes de production de mots et de
[Reference]
reconnaissance. L'experience 3, au cours de laquelle variait le degre d'elaboration de l'encodage au moment de la presentation des categories, a confirme cette hypothese. Dans ce cas, les mots encodes avec difficulte etaient reconnus apres la presentation, a la phase 2, de mots faisant ou non partie d'une categorie apparentee; les mots encodes sans grande difficulte etaient aisement reconnus uniquement apres la presentation de mots de leur categorie d'appartenance et les mots simplement lus, sans difficulte aucune a l'encodage, etaient difficilement reconnus. Nous pouvons ainsi affirmer que le niveau de rappel indice releve a la quatrieme phase correspond a celui de l' elaboration semantique effectuee a la premiere phase. Dans (ensemble, notre etude met en evidence l'importance de la compatibility semantique entre la tache d'encodage et la tache de rappel pour arriver a des niveaux eleves de rendement run-m6monique.
[Author Affiliation]
BRENDA HANNON and FERGUS I. M. CRAIK, University of Toronto
We thank M. Masson and an anonymous reviewer for their helpful comments on an earlier draft of this paper. Correspondence concerning this article should be addressed to Brenda Hannon, University of Toronto, Mississauga, Ontario, L5L 1C6 (E-mail: brendah@psych.utoronto.ca).
