Istruzioni per luso e linstallazione
Forno da incasso multifunzione con 9 programmi
Instructions for use and installation
Built-in multi-function oven with 9 cooking modes
Mode d'emploi et installation
Inbouwoven met 9 programma's
Four encastrer multifonction 9 programmes
Instructies voor het gebruik en installeren Instrucciones para el uso y la instalacin
Horno empotrable multifuncin con 9 programas
Instrues para o uso e a instalao
Forno de encaixar multifuncional com 9 programas
Kullanmvemontajtalimatlar
9programlokfonksiyonluankastrefrn
AX 97 M DC 97 M DG 96 M DM 97 M

Table of contents

Introduction..... 15 First use.... 15 Advice..... 15 Close-up view..... 16 How to use your appliance.... 17 Suggested cooking table.... 23 Cleaning and oven care.... 24 After-Sales Service..... 25 Safety precautions..... 26 Installation.... 27 Technical Data.... 28

Introduction

Congratulations on choosing a Franke appliance. Before using your new appliance, we recommend you read the instructions and advice provided in this manual. This should help your understanding of how to use the appliance and also extend its life. Always keep this owners manual close to hand, as you may need to refer to it in future.

First use

Before cooking for the first time, you must set the temperature control knob to the highest setting and leave the oven on for at least 40 minutes with the oven door open and the room properly ventilated. The smell that is often detected during this procedure is due to the evaporation of products used to protect the oven during storage and until it is installed. Once this initial step has been taken, the oven is ready for use.

Advice

A good habit to get into immediately is to pre-heat the oven to the desired temperature before placing your dishes inside it. During cooking, never place pans or cookware directly onto the bottom of your oven. These should be placed on the grids or oven trays provided, inserted at one of the 5 rack heights available; failure to observe this particular piece of advice could damage the enamel coating on the oven interior. Never line the oven interior with tin foil, especially the oven bottom. All repairs should be carried out by specialised authorised staff. Contact your nearest after-sales service centre and request the use of original spare parts only. Your Franke oven is coated with a special enamel layer, which is easy to clean. Another good habit to get into is to clean the oven often and to avoid reheating caked on grease spots and food residue from previous uses. All self-cleaning panels (if the oven is fitted with them) should be cleaned with soapy water (see cleaning instructions). N.B.: - before you turn your oven on, read the instructions on using the timer (see page 18). - if the oven is fitted with an automatic cooking programmer or end cooking timer, you will have to select the manual option identified by the symbol to use the oven manually.

Close-up view

Control Panel Selector knob Use this knob to select the oven operating mode depending on the type of cooking required. Set the knob onto any of the cooking modes and the light inside the oven will turn on to indicate that the appliance is on. Timer/Programmer (see page 18) The Franke range of ovens includes various different cooking time control and management systems; page 18 provides detailed instructions for each one; just identify the type of control fitted on your oven and read the relevant section for the correct use of this feature Thermostat indicator light Indicated by the symbol , it informs you that the oven heating elements are in operation. The light turns off when the temperature set has been reached, and is turned back on every time the oven is in a heating phase. It is a useful aid to check that the correct temperature has been reached before you place a dish in the oven. Temperature control knob (Thermostat) This knob allows you to select the right temperature for the dish you wish to cook, keeping it constant during the entire cooking process. To select the desired temperature, turn the knob clockwise until the reference mark coincides with the corresponding number. The maximum temperature is approximately 260C. Oven rack guides These are designed to help in the correct positioning of the grids and oven trays in the 5 rack heights available (from 1 to 5 from the bottom up); the suggested cooking table (on page 23) provides a suggestion for the ideal position of these accessories depending on the dish and cooking mode. Oven tray It is usually used to catch any fat and/or grease dripping from food placed directly on the grid or to cook food directly; when not in use, the oven tray should be removed from the oven for cooking. The oven tray is made of type AA enamelled steel for use with food products. Grids These are used as a support for your cake tins, terrine dishes and any cookware other than the oven tray provided, or when cooking in the grill or ventilated grill modes to grill meat and fish or to toast bread and the like.

How to use your appliance
This oven combines the advantages of traditional convection ovens with those of the more modern fan assisted models. It is a highly versatile appliance that allows you to choose easily and safely between 9 different cooking modes. Use knobs (B) and (E) on the control panel to pick one of the cooking modes available and to set the desired temperature. We recommend you consult the suggested cooking table on page 23 for the best use of your new oven.

Cooling fan

In order to cool down the door, control panel and sides of the appliance, this Franke oven is fitted with a cooling fan which turns on automatically when the oven is hot. When the fan is on, a normal flow of air can be heard exiting between the oven door and the control panel. Special attention has been paid to ensure the speed of the air flow does not disturb you while cooking and to make sure it is as silent as possible. To protect the surrounding furniture cabinets, when cooking is complete, the fan stays on until the oven has cooled down sufficiently.

The oven light

This symbol is turned on when the light inside the oven is turned on, which has no effect whatsoever on the oven heating. It is especially useful for cleaning the oven when the same is turned off as it illuminates the appliance interior to aid you in the cleaning task.

Cooking modes Defrosting

Set the knob to the symbol ( ) to enable the defrosting mode. In this mode, the air circulated by the motor action throughout the interior of the oven affects the food and speeds up the natural thawing process. The thermostat knob (E) should be set to between 50C and Max (260C). The oven is heated both from above (top heating element) and from below (bottom heating element). This is the traditional cooking mode that allows food to be cooked evenly, using one rack height only. The thermostat knob (E) should be set to between 50C and Max (260C). The top and bottom heating elements are turned on, as well as the fan inside the oven. Compared with the previous cooking mode (convection), an even more even distribution of heat is obtained thanks to the fan, and 2 different rack heights can be used at the same time. The thermostat knob (E) should be set to between 50C and Max (260C). The circular heating element and the fan are turned on. The constant even heat cooks and browns food uniformly over its entire surface. With this mode, you can also cook various dishes at the same time, as long as their respective cooking temperatures are the same. This is the most delicate of the ventilated modes available in this Franke oven, and is ideal for baking cakes and pastries in general, as it allows you to cook on three different rack heights simultaneously. The thermostat knob (E) should be set to between 50C and Max (260C). The circular heating element is turned on, accompanied by the partial action of the top and bottom heating elements and the fan. Compared with the previous cooking mode, it is more powerful and quicker to reach the desired temperature; this, together with the uniform distribution of the heat, makes it ideal for cooking meat roasts and fish as well as pasta bakes, using up to 3 different rack heights.

Convection mode

Fan assisted mode

Ventilated mode

Multi power ventilated mode

Quick cooking pizza mode

The thermostat knob (E) should be set to between 50C and Max (260C). The circular heating element, along with the top and bottom heating elements and the fan are turned on. This is the most powerful cooking mode available in your Franke oven, and the heat distribution was especially designed to perform a prevalent action towards the bottom; this makes it ideal for the preparation of pizzas and for cooking frozen food quickly. It can be used to reach the desired temperature quickly, and then select any of the other cooking modes available. The thermostat knob (E) should be set to between 50C and max (260C). The top heating element (ceiling) and the fan will start up. The low power and the fan action enable a delicate and even cooking process,which is ideal for baking cakes and for some types of baked pasta.
"Oven ceiling fan" cooking Grill mode
The thermostat knob (E) should normally be set to the Max setting (260C), however you can choose a lower setting, in which cased the grilling would be slower. The grill heating element situated at the top of the oven is tuned on. In this case, the quick browning of the surface via infrared rays keeps the inside of the meat soft and tender; the grill mode can also be used to toast up to 9 slices of bread very quickly. Your Franke oven is perfectly safe, and has been designed for grilling food with the oven door completely shut. However, during cooking, the oven door can be kept either fully open or ajar, using the half-open position. Warning: the grill heating element becomes extremely hot during operation; try to avoid touching it inadvertently when handling the food to be grilled; however, Franke has designed a special shape of the oven front to protect the users hands as much as possible. The thermostat knob (E) should normally be set to the Max setting (260C), however you can choose a lower setting, in which cased the grilling would be slower. The grill heating element situated at the top of the oven is tuned on, along with the fan. This combination of features increases the effectiveness of the unidirectional thermal radiation of the heating elements through forced circulation of the air throughout the oven. This helps prevent food from burning on the surface, allowing the heat to penetrate right into it. Excellent results are achieved with kebabs made with meat and vegetables, sausages, ribs, lamb chops, spicy chicken, quails, pork chops, etc. The end cooking timer is a practical device that turns the oven off automatically at a set time, ranging from one to 120 minutes. To use the end cooking timer, the ringer must be wound up by turning knob (C) almost one full turn clockwise; then turn it back to the desired time so that the number of minutes on the knob matches the reference mark on the panel. When the set amount of time is up, in addition to the timer ringing, the oven is automatically turned off. N.B. Straight after it is turned off, and for a prolonged period of time, the interior of the oven maintains a temperature close to that set for cooking; so, to avoid overcooking or burning food, we recommend you remove it from the oven as soon as cooking is complete. Warning: to use the oven in manual mode, i.e. without using the end cooking timer, you should align the symbol on the control panel with the reference mark on the control panel. When the oven is turned off, the end cooking timer can be used just like a normal minute minder.

Ventilated grill mode

End Cooking Timer Knob
The electronic cooking programmer
This feature allows you to program the oven or the grill as follows: delayed start for a specific length of time; immediate start for a specific length of time; minute minder. Button Functions: : minute minder with hours and minutes : cooking time : end cooking time : switch to manual mode : set cooking time - backward; : set cooking time - forward.
How to set the digital clock After the appliance has been connected to the power supply, or when the power has gone out, the clock display will automatically reset to 0:00 and begin to blink. 0.00 Press the button and then set the correct time (within four seconds) using the and buttons. Button sets the time forward. Button sets the time backward. Whenever the time needs to be reset, follow this same procedure: Manual operation of the oven Once the correct time has been set, the programmer automatically switches to manual mode. Note: To reset the manual mode after each "Automatic" cooking, press the and buttons. Delayed start with set cooking time The cooking time as well as the time at which the cooking ends must be set. Let us suppose that the display indicates 10:00 a.m. and that we wish to cook something for 30 minutes and we want cooking to end at 13:00. 1. Turn the oven control knob to the desired cooking mode and temperature (for example: convection mode at 200C). 2. Press the button and then set the desired cooking time (within four seconds) using the and buttons. Let us suppose that we set the cooking time to 30 minutes. The display will read as follows:
Release the button and after 4 seconds the correct time will reappear along with the symbol and the word AUTO. 3. Press the button and then use the and buttons to desired end cooking time. Let us suppose this time to be 13:00.
4. After the button has been released, the current time will be displayed after approximately 4 seconds:
The word AUTO will light up reminding you that the cooking time and end cooking time have been programmed in automatic mode. At this point, the oven will turn on automatically at 12:30 and then turn off after 30 minutes. When the oven is turned on, the symbol of a hot pan will be displayed for the entire cooking time. At any time during the cooking process, the length of the cooking time can be displayed by pressing button , and the end cooking time by pressing button. The timer rings when cooking is complete; to turn it off, simply press any of the buttons except the and buttons.

Immediate start with set cooking time By programming the cooking time only (points 1 and 2 of the Delayed start with set cooking time section), cooking will start up immediately. To cancel a scheduled cooking process Press the button. Minute minder function The minute minder allows you to set a given amount of time from which the countdown begins. This feature does not turn the oven on or off; it merely sounds when the time has elapsed. button has been pressed, the display will read as follows: When the
Then set the desired time using the and buttons After the button has been released, the countdown begins and the current time is displayed.
When the time is up, a buzzer sounds, which can be turned off by pressing any of the buttons (except the and buttons) at this point the symbol will also come off. Modifying and cancelling data or The data entered can be changed at any time by pressing the corresponding button and the buttons. When the data for the cooking time is cancelled, the data for the end cooking time is also cancelled automatically, and vice versa. If the oven has been programmed, it will not accept end cooking times which come before the start of the cooking process. N.B. Straight after it is turned off, and for a prolonged period of time, the interior of the oven maintains a temperature close to that set for cooking; so, to avoid overcooking or burning food, we recommend you remove it from the oven as soon as cooking is complete.
Electronic oven temperature control
How to set the digital clock After the appliance has been connected to the power supply, or when the power has gone out, the clock display will automatically reset to 0:00 and begin to blink. 0.00 Press the button and then set the correct time (within four seconds) using the and buttons. Button sets the time forward. Button sets the time backward. Whenever the time needs to be reset, follow this same procedure: Manual operation of the oven Once the correct time has been set, the programmer automatically switches to manual mode. Note: To reset the manual mode after each "Automatic" cooking, press the and buttons.
Manual operation of the oven When a cooking mode is selected, the oven immediately starts the heating process, shown by the symbol of a hot pan on the display. For each cooking mode, the display suggests a recommended cooking temperature (SEE: Suggested cooking table); this temperature may be changed by the user using the knob: turn it anti-clockwise to decrease the temperature and clockwise to increase it. of a hot pan flashing on the display indicates that the oven is in the pre-heating stage, which The symbol means it has not reached the desired temperature yet. When the oven reaches the desired temperature, a "beep" will sound for two seconds and the symbol of the hot pan will stay on permanently, to indicate the cooking stage has begun. Immediate start with set cooking time Let us suppose you want to cook at 200C for 30 minutes. 1. Turn the selector knob, the display will suggest the recommended cooking temperature; use the temperature control knob to change it if necessary. 2. Press the hot pan key and then press the and keys to set the desired cooking time, in this case 30 minutes. Release the key, the hot pan signal will start flashing after 4 seconds to indicate the oven is pre-heating. The countdown, however, will only start once the desired temperature has been reached (except when the grill modes are on, in which case the countdown starts immediately, see B). The information displayed is as follows: A - pre-heating stage

(you can check the time set using the hot pan key and the temperature set using the and keys). When the oven reaches the desired temperature, a "beep" will sound for two seconds and the symbol of the hot pan will stay on permanently, to indicate the cooking stage has begun. B - cooking stage / (remaining cooking time / temperature set - alternating) C - end of cooking You will be informed that cooking has ended by an intermittent beeping (**) lasting one minute, which can be silenced by pressing the key (or even the hot pan and stop keys), and the display will indicate.
The oven suspends heating automatically, the fans will stay on to avoid any heat accumulation and the oven light will stay on so you can check cooking. However, the temperature inside the oven will remain high for button to re-start the heating several minutes. Should you wish to continue cooking manually, press the process. The word AUTO will disappear and the hot pan symbol will reappear. If on the other hand you wish to stop altogether, set the selector knob back to 0. The display will indicate the following for 3 seconds.
After which the display will go back to indicating the time of day (if you want to turn off the display when the oven is off, press the timer and hot pan keys simultaneously and the display will indicate:

then use the

keys to set this parameter to:
the display will be turned off and will only come back on when the oven is on). (**): the intensity of the beeping can be modified while it is emitted and for future beeping by pressing the

key. 21

Delayed start with set cooking time The cooking time as well as the time at which the cooking ends must be set. Let us suppose that it is 8:00 a.m. and that you wish to cook something for 30 minutes that will end at 1 p.m. 1. Turn the selector knob, the display will suggest the recommended cooking temperature; use the temperature control knob to change it if necessary. 2. Press the hot pan key and then press the and keys to set the desired cooking time, in this case 30 minutes. 3. Press the "stop" key and within 4 seconds use the and keys to select the time you want cooking to end, in this case 13:00. The display indicates:

The oven is set, and on stand-by. If you want to check the cooking time set, press the hot pan key To check the temperature set, press the or key, turn knob

; use the

keys to change it.

to change it.

Cooking stage: when it is time for cooking to start, the oven will operate as from point B in the paragraph above, except for the cooking time countdown, which starts immediately and not when the desired temperature has been reached. This is because logically speaking you would have placed the dish in the oven when you set it. The pre-heating time is included in the cooking time.
The end cooking time clock
The figure illustrates the programmer controls; knobs 1 (L/H) and 2 (R/ H) allow for all the necessary adjustments to be made. Knob 1 Use this knob to set the correct time. To do this, pull and turn the knob clockwise until the correct time is displayed.
Knob 2 Use this knob to set the cooking time or to programme the oven for manual use. To do this, turn the knob clockwise. Manual operation of the oven The oven can also be used manually, i.e. without any programming. In this case, turn knob 2 of the programmer clockwise until the symbol corresponds with the marker on the control knob. Semi-automatic operation of the oven Normal cooking start with programmed cooking end: prepare your dish; turn knob 2 clockwise and set the expected minutes for the end cooking time on the dial (Q); turn the selector knob to the chosen setting and the thermostat knob to the desired temperature. The oven starts immediately. When the set amount of time is up, in addition to the timer ringing, the oven is automatically turned off. To silence the buzzer, turn the clock back to 0 or manual mode by turning knob 2. N.B.Straight after it is turned off, and for a prolonged period of time, the interior of the oven maintains a temperature close to that set for cooking; so, to avoid overcooking or burning food, we recommend you remove it from the oven as soon as cooking is complete. Minute minder The clock may be used as a minute minder only if the oven is not in operation. To operate the minute minder, turn knob 2 clockwise until the desired time is displayed on the dial (Q). The maximum amount of time that can be set is 180 minutes. To silence the buzzer, turn knob 2 clockwise until the symbol In this case, turn knob corresponds with the marker on the control knob. Important To use the oven in manual mode, (i.e. without any programming), the following steps are necessary: Turn knob 2 of the programmer clockwise until the symbol corresponds with the marker on the control knob.

Suggested cooking table

4FMFDUPS LOPC TFUUJOH 5ZQF PG GPPE 8FJHIU JO LH
3BDL IFJHIU 1SFIFBUJOH UJNF NJO
5IFSNPTUBU LOPC TFUUJOH $PPLJOH UJNF NJO

$POWFDUJPO NPEF

0ORKROASTS /MELETTES #OD3EABREAM4URBOT 0LAIN"READ 0IES &RESHWATERFISH 'ARNISHED0OLENTA 2ICOTTA0IE 3ADDLEOF0ORK /VENBAKEDRABBIT /VENBAKEDCRAYFISH #OD3EABREAM4URBOT 0LAIN"READ &RUIT#AKES 3AVOURY0IES 0UMPKIN0IE 7HOLEMEAL"READ &LAKY0ASTRY #RACKERS 6EGETABLE0IES $RY#AKES #HEESE3OUFFLS 4ROUT0ARCELS /VENBAKEDRABBIT -EAT0IES 3AVOURY0IES #HICKEN 0ORKROAST #RACKERS &RUIT#AKES 2OAST0OTATOES

SIMULT SIMULT SIMULT SIMULT OROR SIMULT SIMULT SIMULT OR SIMULT SIMULT SIMULT SIMULT SIMULT OR OR OR OR OR OR
5SETHE 1UICK #OOKING FUNCTIONTO CUTDOWNO PREHEATINGTIMES
-AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX -AX

'BO BTTJTUFE NPEF

7FOUJMBUFE NPEF
.VMUJ QPXFS WFOUJMBUFE NPEF

2VJDL DPPLJOH QJ[[B NPEF

&OCACCIA 0LAIN"READ "ROWN"READ 0IZZA 2OAST"EEF 3HIN 4URKEY 4ORTELLINIAUGRATIN 0OTATODUMPLINGS 3WORDFISH -USSELS 3CALLOPS 2OAST0OTATOES #ANNELLONI #HICKEN7INGS #HOPS 3CALLOPS 2OAST0OTATOES &RESHWATERFISH 4OMATOESAUGRATIN /VENBAKED%EL 4URKEYKEBABS #HICKEN4HIGHS 3PARE2IBS 'UINEAFOWL 3EABREAMPARCELS #HICKEN 1UAILS 3AUSAGES 6EGETABLES

0WFO DFJMJOH GBO DPPLJOH

(SJMM.PEF

7FOUJMBUFE (SJMM NPEF

."THESUGGESTIONSINTHETABLEARETHERESULTOFRECIPETRIALSCARRIEDOUTBYATEAMOFPROFESSIONALCHEFS)NANYCASETHE COOKINGTIMESAREAPPROXIMATEANDMAYVARYACCORDINGTOPERSONALTASTE

Cleaning and oven care

Important
Before embarking upon any cleaning and care operations on your oven, you must disconnect it from the electricity supply by unplugging it from the mains or cutting off the power supply on the main switch. The stainless steel and enamelled parts will remain as good as new as long as you clean them with soapy water or with special, purpose-designed products. Make sure you dry them thoroughly afterwards. Avoid using products containing chlorine or diluted chlorine solutions, caustic soda, abrasive detergents, muriatic acid, vinegar, bleach or other products which could scratch or mar the surfaces. Any stains that may appear on the bottom of the oven, of a variety of compositions (sauces, sugar, albumin and grease spots) always originate from splashes or overflowed food. These splashes occur during cooking and are due to the use of an excessively high temperature, whereas the overflowed food is due to the use of cookware which is too small or to an erroneous evaluation of the increase in volume of the dish during cooking. These two inconveniences can be averted by using deep cookware or using the oven tray provided. As for cleaning the lower part of the oven, we recommend you remove any stains when the oven is still lukewarm, as the longer you wait to clean the oven, the tougher the stains are to remove. Clean both the inside and the outside of the oven door when the oven has cooled down completely, using hot water and avoiding the use of rough cloths. For the enamelled parts, as well as the interior of the oven, use hot water and non abrasive detergents. Franke ovens are fitted with two different interior lighting systems: 1. A circular lamp is situated to the top left-hand side of the rear of the oven interior. 2. A rectangular lamp is fitted on either side. Should one of the oven lamps need replacing, proceed as follows: - In every case, and for either lamp type, you should disconnect the appliance from the electricity supply at the omnipolar switch used to connect the oven to the mains, or by unplugging it, if possible; Rear circular lamp: - Unscrew the protective glass cover (A) - Remove the lamp and replace it with one which is resistant to high temperatures (300C) with the following characteristics: Voltage: 220/230V - 50Hz Wattage: 15W Socket: (E14) mignon - Reposition the protective glass cover (A) and reconnect the appliance to the electricity supply. Side lamps: - Remove the glass cover (B) using a flat screwdriver to release it Remove the lamp and replace it with one that is resistant to high temperatures (300C) with the following characteristics: Voltage: 220/230V - 50Hz Wattage: 15W Socket: (E14) mignon

Cleaning the oven door

Oven lamp replacement
Dismounting the oven door
To facilitate your task in cleaning the oven, the door can be dismounted at the hinges as follows: The hinges (A) are fitted with two mobile U-bolts (B). Raise the U-bolt (B) to release the hinge from its seat. Then lift the door upwards and pull it out; to do this, hold the door on either side beside the hinges. To reposition the door, first insert the hinges into the special grooves. Then, before shutting the oven door, remember to turn the mobile U-bolts (B) used to hook the two hinges in place.

After-Sales Service

In the event of a malfunction, please contact your nearest After-Sales Service Centre, as included in the list enclosed. Never contact unauthorised technical servicemen.

Notify us of:

The type of problem; The appliance model (Mod.); The serial number (S/N); This information can be found on the appliance data plate, which can be seen with the appliance door open in the middle of the bottom part of the door.

Safety precautions

To maintain the efficiency and safety of this appliance, we recommend that you do the following: only call authorised After-Sales Service Centres always use original spare parts This appliance is designed for non-professional household use; do not modify any of its characteristics. These instructions are applicable to those countries whose symbols appear in the instruction manual and on the appliance serial number plate. The electrical safety of the appliance is only guaranteed when it has been correctly earthed, as foreseen by the applicable norms in force. Avoid touching the heating elements and some parts of the oven door when the appliance is in use, as these can become very hot. Keep children well away.
Prevent children from playing with:
the controls and the appliance in general; the packaging components (bags, polystyrene, nails, etc.); the appliance itself, during and immediately after use of the oven, due to the heat generated; the appliance itself, when no longer in use (in which case potentially dangerous parts must be made safe). Touching the appliance with wet parts of the body; Using the appliance when barefoot; Pulling on the appliance or the power supply cable to unplug it from the electrical outlet; Improper or dangerous operations; Obstructing the ventilation or heat dissipation slots; Allowing power supply cables of other appliances to come into contact with the hot parts of the appliance; Exposing the appliance to atmospheric agents such as rain, or sunlight; Using the oven for storage purposes; Using flammable liquids in the vicinity of the appliance; Using adapters, multiple sockets and/or extension leads; Installing or repairing the appliance without the assistance of qualified staff. Installation (following the manufacturers instructions); if in doubt about your appliances operation; Replacing the electrical socket when it is not compatible with the appliance plug.

Avoid the following:

Qualified staff must be contacted in the following cases:
Service Centres authorised by the manufacturer must be contacted in the following cases:
If in doubt about the soundness of the appliance after removing it from its packaging; If the power supply cable has been damaged or needs replacing; If the appliance breaks down or malfunctions; ask for original spare parts.
It is a good idea to do the following:
Only use the appliance to cook food and nothing else; Disconnect the appliance from the electrical mains before cleaning and performing maintenance work or if it is malfunctioning; - Use oven gloves to place cookware in the oven or when removing it; - Always grip the oven door handle in the centre as the extremities of the same may be hot due to any hot air leaks; - Make sure the knobs are in the off position when the appliance is not in use; - Disconnect the appliance from the electricity mains when you no longer wish to use it. The manufacturer will not be held responsible for any damage due to: incorrect installation, improper, incorrect and irrational use.

Installation

Important: the appliance must be installed in accordance with the applicable norms and laws in force. The appliance should only be installed by specialised authorised staff.
A special note about the cabinet

m 545 m

595 mm

569 mm

560 mm

595 m m

The components of the cabinet used to house the built-in oven (plastic parts and veneered wood) should be bound using heatproof glues (max 100C): unsuitable materials and glues could cause warping and ungluing. The cabinet must allow for enough clearance for the electrical cables to pass freely. The cabinet used to house the built-in oven should be sturdy enough to bear the weight of the appliance. The oven should be fitted securely to the housing foreseen. The oven can be built into a tall cabinet or underneath the worktop of a modular base unit, as long as sufficient air circulation is ensured.

mm 550 in. m

m 5 mm
Assembly instructions (DC 97 P-DC 97 M-DM 97 M-DG 97 P-DG 96 M)
Measure the width of the back panel (S) of the cabinet, which should usually be 16, 18 or 20 mm. 16 mm back panel: place 2 tabs (T) indicating the number 16 on each fastening bracket (F). 18 mm back panel: place 2 tabs (T) indicating the number 18 on each fastening bracket (F). 20 mm back panel: no tabs are necessary for the 20 mm thick back panel. Position the brackets (F), resting them onto the cabinet back panel and at the bottom on the oven supporting plane. Secure the brackets (F) in place using the screws (V). Fit the oven inside the cabinet, open the oven door completely and secure it to the brackets (F) using the 4 screws (V) and the 4 plastic spacers (D).

Assembly instructions (AX 97 M)
Insert the machine into the space; secure the oven body to the unit with the 4 screws provided using the holes on the sides.
When the oven is built under a hob, turn the hob connection (R) towards the left or right to allow for the correct insertion of the oven.

Technical Data

Inner dimensions of the oven: width mm. 440 depth mm. 415 height mm. 312 Inner Volume of the Oven: 57 litres Voltage and Frequency of Power Supply: 230V ~ 50Hz Total Power: 2.875kW - 10A This appliance conforms to the following European Economic Community directives: 73/23/EEC of 19/02/73 (Low Voltage) and subsequent amendments; 89/336/EEC of 03/05/89 (Electromagnetic Compatibility) and subsequent amendments; 93/68/EEC of 22/07/93 and subsequent amendments.
Consumption (convection mode): to reach 175C: 0.32 kWh to maintain 175C: 0.62 kWh total consumption: 0.94 kWh Consumption (ventilated mode): to reach 200C: 0.31 kWh to maintain 200C: 0.81 kWh total consumption: 1.12 kWh 28
Heating element power: top heating element W 1,000 bottom heating element W 1,250 grill W 2,250 circular W 2000 fan motor W 25 oven lamp W 15 x 1 cooling fan W 20

The Implicit Prosody of Jabberwocky and the Relative Clause Attachment Riddle*

Frank Wijnen

Utrecht University

Abstract

This contribution reports on three sentence processing experiments involving structurally ambiguous Jabberwocky strings. The items are complex NPs, in which a relative clause can be attached to either of two nominal heads (NP1-preposition-NP2-RC). The phonological lengths of the heads, as well as that of the relative clause, are manipulated. The results show modest effects on relative clause attachment preferences of this manipulation, thus lending support to Fodors (1998; 2002) implicit-prosody hypothesis.
1 Introduction One of Sieb Nootebooms relished criticisms of psycholinguistic research is that the distance between the primary data and the processes of interest is big, too big actually. He may be right it often requires an inferential chain of considerable length and complexity to connect something pedestrian like reaction times to something lofty like lexical access or grammatical encoding. To my mind, however, psycholinguistics does not differ very much in this respect from other domains of investigation, including phonetics, definitely also including some of the work done by the distinguished scientist to whom this book is dedicated. Consequently, I consider it quite appropriate to present a study here in which the connection between data and the presumed underlying processes is exquisitely indirect. The hypothesis under scrutiny is the implicit-prosody hypothesis, as put forth by Janet Fodor in several publications in the past few years. In Fodor (2002), the following summary is given: The Implicit-Prosody Hypothesis (IPH): In silent reading, a default prosodic contour is projected onto the stimulus, and it may influence syntactic ambiguity resolution. Other things being equal, the parser favors the syntactic analysis associated with the most natural (default) prosodic contour for the construction. To avoid any misunderstandings, it is useful to point out that prosody (intonation, phrasing and rhythm) can affect sentence processing in the auditory modality (see Cutler, Dahan, & Van Donselaar, 1997 for an extensive review). In particular, various studies have demonstrated that accentuation and phrasing affect the on-line resolution of syntactic ambiguities, if at least the two parses can be associated with different prosodic realizations (e.g. Nagel, Shapiro, Tuller, & Nawy, 1996; Pynte & Prieur, 1996; Schafer, Carter, Clifton, & Frazier, 1996). A point of debate has been whether prosody has an immediate effect on
Thanks are due to Carolien van den Hazelkamp, who ran Experiment III. I am also grateful to Janet Fodor, Jocelyn Ballantyne, Martin Corley, and this volumes editors, for helpful commentary on a first draft. The author was supported by a fellowship of the Royal Netherlands Academy of Arts and Sciences (KNAW).
ambiguity resolution, i.e., affecting the first-pass parse, or whether it is invoked only in cases where the perceiver is garden-pathed (see e.g. Watt & Murray, 1996). Recent neurocognitive evidence suggests that prosodic cues are exploited immediately (Steinhauer, Alter, & Friederici, 1999). The implicit-prosody hypothesis, being a well-behaved psycholinguistic hypothesis, makes reference to an invisible cognitive entity, viz. a phonological (prosodic) representation generated by the reader. Now, how should we go about collecting evidence for this construct? Fodor proposes the following research programme: isolate a factor F that can be manipulated and that measurably affects the prosodic structure of a sentence; show that this prosodic structure resulting from F has an effect on sentence parsing (e.g., ambiguity resolution); demonstrate that F does not affect parsing directly; include F in a reading task. Does the task reveal an effect of F on parsing (just as in an auditory task)? If so, we may infer that the prosody induced by F is in some sense real, even if not manifestly (physically) present.
The present study follows this recipe by exploring the effects of phonological weight (i.e., length in syllables) on the resolution of relative clause attachment ambiguities in a complex noun phrase with two possible relative clause (RC) heads, as illustrated in (1). 1. Iemand schoot op de bediende van de actrice die op het balkon zat. Someone shot the servant of the actress who was on the balcony. Clearly, the relative clause die op het balkon zat who was on the balcony can be attached to either de bediende the servant (NP1) or de actrice the actress (NP2). The implicit-prosody hypothesis is of particular relevance to this ambiguity. Universalist theories of sentence processing (e.g. Frazier & Fodor, 1978; Kimball, 1973) predict a consistent preference for low (i.e., NP2) attachment of the relative clause, on the basis of a structurally defined strategy, late closure. Cuetos & Mitchell (1988) showed, however, that attachment preferences in the construction at hand vary across languages, and numerous studies since then have confirmed and extended this observation (see Ehrlich, Fernndez, Fodor, Stenshoel, & Vinereanu, 1999; Fodor, 2002, for overviews). Fodor has argued that the potential threat to universalist parsing models posed by these observations may be removed, and the cross-linguistic variation better understood at the same time, if we assume that relative-clause attachment in this context is guided by prosodic factors. In line with the implicit-prosody hypothesis, Fodor (1998) assumes that readers engage their phonological processor to chop up the input into manageable chunks or packages. Prosodic packaging in silent reading is fully analogous to what happens in speaking. A phonological factor that leads to the insertion of a prosodic break in speaking will equally likely promote the insertion of a package boundary in reading. Furthermore, in processing written input, the syntactic processor is assumed to refrain from rearranging the packages delivered by the prosodic processor (but see Fodor, 2002). One of the factors that affect prosodic packaging is phonological length (weight). Prosodic boundaries have a tendency to occur at roughly equal distances in the speech stream. A corollary of this tendency is that long, multi-syllable words tend to form their own

phonological phrases, whereas short words are more likely to be chunked together. Phonological length can affect the resolution of relative-clause attachment ambiguities of the type exemplified in (1) in various ways. First, the length of the relative clause itself matters: long (heavy) relative clauses tend to attach high, whereas short (light) ones tend to attach low, a phenomenon known as the anti-gravity effect. The explanation Fodor advances is that a short RC is more easily packaged with the second of the two noun phrases, whereas a long RC promotes the insertion of a prosodic break right before it, which blocks its attachment to the second of the two NPs. The lengths of the potential head nouns (or NPs) are predicted to have an effect as well (Fodor, 1998). If, in the construction NP1-prepositionNP2-RC, NP1 is long and NP2 is short, there will be a tendency to insert a prosodic boundary after NP1. This boundary will render attachment of the relative clause to NP1 difficult, and consequently NP2-attachment will be the preferred option. If, however, NP1 is short and NP2 is long, a boundary is likely to be inserted after NP2, rendering the attachment of the relative clause to NP2 difficult, resulting in an NP1 attachment bias. To my knowledge, there is as yet no direct evidence in support of this prediction, either from auditory or visual language processing. Colonna & Pynte (2001) have reported data that seem to go against this prediction (from French), but their materials are not without problems. The aim of the present study is to provide further evidence for the effects of phonological length on the resolution of relative-clause attachment ambiguities in the structure exemplified in (1). In an attempt to neutralize as much as possible the influence of lexical, semantic and pragmatic factors on attachment decisions, the three experiments reported here make use of Jabberwocky prose, i.e., strings in which all content words have been replaced by nonexisting (but phonologically legal) elements. Such strings pose no problem for the human parser. Recent ERP evidence shows that they engage the syntactic processing system just like normal sentences do (Hahne & Jescheniak, 2001). They cannot be interpreted, however, at least not in the normal sense of the word, as they prevent (normal) access to the lexicon and the semantic system. The first experiment is a questionnaire study looking at the effect of relative phonological weights of the attachment sites (NP1 vs. NP2). Experiment II is a replication of Experiment I with relative clauses that are longer than those in Experiment I. These two off-line experiments involve globally ambiguous materials. This means that readers cannot be garden-pathed, i.e., pursue an analysis that is proven incorrect downstream in the input by lexical or structural cues. The main aim of Experiment III is to determine whether phonological length of the potential heads of a relative clause can have an effect in an on-line task (continuous acceptability judgment) as well. 2 Experiment I 2.1 Method Participants. Twenty-two volunteers (10 female) took part, the majority of them being faculty and students at Utrecht University. Materials. Thirty-two items were constructed according to the following template: NP1preposition-NP2-Relative Clause. These 32 items were divided into four sets of eight, each of which corresponded to one of the four conditions that resulted from systematically varying the lengths of the two critical nouns, as exemplified in (2). 2. long-short: de kalambulo van de fup die verstritst was

long-long: de knilpatsiera van de astrublankor die verdrimd werd short-long: de slos van de prefrastiaan die bedrept was short-short: de vrink van de orcht die betrind werd Sentence fragments (i.e., complex NPs) were used instead of complete sentences, in order to suppress as much as possible the influence of structural position and its (prosodic) correlates in terms of e.g. the presupposition~focus contrast. Both possible heads of the relative clause were always non-neuter nouns (as indicated by the determiner de), so that the relative pronoun (die) did not disambiguate the structure through grammatical gender. All relative clauses were passives (which could either be construed as adjectival or verbal), and consisted of three words (four syllables): the relative pronoun, a past participle (marked by an appropriate prefix) and one of the auxiliary verbs werd was or was was. Procedure. The participants completed a questionnaire sent by electronic mail. The e-mail message contained the experimental items in a quasi-random order, and the participants were instructed to read the items one by one at a leisurely pace, and to immediately indicate whether they felt the relative clause modified the first or the second noun, by replying with either 1 or 2. They were explicitly requested not to read the stimuli aloud, and were instructed to utter a nonsense phrase after completing each item (the suggestion given was abracadabra simsalabim). Both of these recommendations were made in an attempt to counteract the emergence of response biases (or response priming). 2.2 Results and discussion A total of 704 responses (22 participants * 32 items) were collected. Eight participants showed hardly any variation in their responses (they selected NP1 in merely one or two cases); their data were discarded, leaving 448 valid data points generated by 14 participants.

% NP1 attachment

LS LL SL SS
Lengths of candidate heads
Figure 1. Experiment I: Percentage of NP1 attachment responses as a function of phonological length variation in the potential NP heads of the relative clause. LS: NP1 long, NP2 short; LL: both NPs long; SL: NP1 short, NP2 long; SS: both NPs short. The first noun was indicated as the head of the relative clause 154 times (34%), reflecting a relatively strong low attachment preference, which is at odds with previous results on relative-clause attachment in (real) Dutch (Brysbaert & Mitchell, 1996). The percentages of NP1 attachments per condition are displayed in Figure 1.
The distributions of NP1 and NP2 responses across conditions turn out to be significantly different (2 = 13.48, df = 3, p =.004). The conditions for which the implicit-phonology hypothesis provides clear predictions are LS (long NP1, short NP2) and SL (short NP1, long NP2). The percentage of NP1 responses in the latter (SL) is almost twice as high as in the former (LS): 48% vs. 25%. This difference is significant (2 = 6.08, df = 1, p =.014). The first conclusion that can be drawn from these results is that phonological length of the potential nominal heads affects relative clause attachment in reading. It should be stressed that the manipulated variable (i.e., length in syllables) is, indeed, of a purely phonological nature. Moreover, in this particular instance, we may assume the effect is unconfounded by lexico-semantic or pragmatic factors. Therefore, the result can be plausibly ascribed to the phrasing contrast in implicit prosody that results from the length manipulations. It is also important to reiterate that the present experiment involved globally ambiguous items only. Reanalysis on the basis of lexical or grammatical disambiguation can therefore not have occurred. Consequently, the length contrasts, i.e., their effects on implicit prosody, can not be associated with a last resort strategy, applied only in the case of parsing breakdown. 3 Experiment II 3.1 Method This experiment was conducted in exactly the same way as Experiment I. Eighteen new volunteers (from the same population as in Experiment I; 9 female) participated. The sentence fragments used here were adapted from those in Experiment I by inserting a prepositional phrase before the past participle in the relative clause, consisting of a preposition, a definite determiner and a trisyllabic pseudo-noun, as illustrated in (3) below. 3. long-short: de kalambulo van de fup die in de strindosnees verstritst was long-long: de knilpatsiera van de astrublankor die van de spatterplons verdrimd werd short-long: de slos van de prefrastiaan die voor de elkodrator bedrept was short-short: de vrink van de orcht die door een kaskadeur betrind werd 3.2 Results and discussion Two of the participants displayed virtually no variation in attachment decisions; their data were discarded, leaving 512 valid responses. A chi-squared analysis indicates that the distributions of attachment responses across conditions are not reliably different (2 = 1.71, df = 3, p =.64). Across all conditions and participants, NP1 attachment was chosen in 209 cases (40.8%). This is a conspicuously higher percentage than in Experiment I; the difference between the response distributions in Experiments I and II is significant (2 = 4.22, df = 1, p =.039), in agreement with the anti-gravity effect. This effect is explained as the result of a tendency to insert a prosodic break before a long RC, blocking its attachment to NP2. Apparently, this tendency outweighs the effects of the two potential heads lengths. This issue, obviously, deserves further scrutiny, but it should be clear that, taken together, the results of experiments I and II agree with the implicit-prosody hypothesis. 4 Experiment III Up to this point, we have considered Dutch native speakers relative-clause attachment preferences in an unconstrained, off-line judgment situation. The next step in our exploration is to determine whether the effect of phonological length found in Experiment I generalizes to a more rigorously controlled experimental task. The task used in the present experiment,

continuous acceptability judgment, has a hybrid character. On the one hand, it requires participants to deliberately evaluate the acceptability of stimuli, which renders it similar to off-line judgment tasks. On the other hand, the stimuli are presented segment by segment, emulating normal reading in some respects (notably: sequential, incremental processing). The continuous acceptability judgment task has been shown to be sensitive to readers on-line parsing preferences. Crucially, readers tend to judge a sentence segment as unacceptable not only if it is incorrect in some objective way (e.g. when it is ungrammatical), but also when it contradicts an ambiguity resolution preference. In addition, segments that cause a parsing problem but are nonetheless accepted, yield longer average reaction times than those of unproblematic segments. In contrast to Experiments I and II, the relative-clause attachment in the items used here is disambiguated, by means of grammatical number agreement of the auxiliary verb in the relative clause with one of the two potential head NPs. In all stimuli, the potential head NPs differ in length. The prediction is that LS items (long NP1, short NP2) yield a low (NP2) attachment preference, whereas SL items will promote high (NP1) attachment. Thus, forced NP1 attachment (by the auxiliary) will negatively affect the acceptance rate and average reading times in the LS items, as compared to the SL items. The mirror image of this pattern should be observed for forced NP2 attachment. 4.1 Method Participants. Fifty-five paid volunteers, all of them students at Utrecht University, took part (18-40 years of age; 11 male). None of them had participated in experiments I or II. Materials. Thirty-two items were adapted from those used in Experiment I. Each item contained two RC-attachment sites, one containing a long noun (four or five syllables) and the other a monosyllabic noun. Four variants of each item were made by varying the order of the long and the short noun and the attachment disambiguation (NP1 vs. NP2; see example 4). All long nouns had a plural affix (-s), whereas the short nouns were always singular. (Most monosyllabic nouns require the syllabic plural affix -en, which would have interfered with the length manipulation.) All relative clauses consisted of a relative pronoun, a past participle, and a plural or singular version (both monosyllabic) of the auxiliary zijn be. 4. (slashes indicate segment boundaries) a. de carambulo's / van / de fup / die / verstritst / is b. de carambulo's / van / de fup / die / verstritst / zijn c. de fup / van / de carambulo's / die / verstritst / is d. de fup / van / de carambulo's / die / verstritst / zijn

long, short, NP2 long, short, NP1 short, long, NP1 short, long, NP2
The four versions of the 32 experimental items were divided across four lists, in such a way that each list contained all four conditions in equal numbers, but only one version of each experimental item. The experimental items were interspersed with 64 filler items, all of which had the same structure as the experimental items, but varied with regard to the lengths of the critical nouns, association of grammatical number with long and short nouns, and the identity of the auxiliary verb in the relative clause. Participants were randomly assigned to lists, and items within lists were presented in random order, different for each participant. Procedure. Participants were seated in front of a computer monitor in a dimly lit, soundattenuating room. Each trial started with a blank screen, on which appeared a small gray window containing the text push a button to proceed. Upon pressing a button on a button
box, the screen was refreshed, and a black window was presented slightly above the vertical midpoint of the screen, with a fixed height of 2 cm, and a variable width, dependent on the length of the item to be presented. The items were presented segment by segment, in nonincremental moving-window mode. Participants were instructed to press the right-hand button on the box if they thought the segment was an acceptable continuation of the preceding material, or the left-hand button if the segment was an unacceptable continuation. A left-hand button press ended the trial. The intervals between any two subsequent button presses were computed and stored for later analysis. After a trial ended, the screen was erased, and the window with the text push a button to proceed reappeared. Participants were instructed to read at a pace that allowed them to judge the acceptability of the items. It was stressed they should rely on their first impression, and not unnecessarily delay their response. They were allowed to take breaks between trials. Completion of the experiment took about 25 minutes. 4.2 Results and discussion Rejection rates. All items consisted of six segments, of which only the last one (containing the disambiguating auxiliary) is critical. No effects were observed in segments 1 to 5. In segment 6, the rejection rates were 37.5% in the LS (long NP1, short NP2) condition, and 38.4% in the SL condition (short NP1, long NP2). This difference is not statistically reliable (by participants F(1,54) =.22, by items: F(1,31) =.09). Across phonological length manipulations, the rejection rates for NP1 and NP2 disambiguation were 31.9% and 43.9%, respectively, which constitutes a significant difference (by participants: F(1,54) = 4.09, p =.048; by items: F(1,31) = 25.13, p <.001).

50 % rejection 0 NP1 Disambiguation NP2 LS SL
Figure 2. Experiment III: Mean percentage of rejection responses at the disambiguating segment 6 (auxiliary), as a function of the lengths of NP1 and NP2 (LS: long NP1, short NP2; SL: short NP1, long NP2) and disambiguation direction (NP1 or NP2). The critical question is whether RC-attachment disambiguation interacts with the manipulation of the lengths of the attachment sites. The relevant data are displayed in Figure 2, which shows that the difference in rejection scores between NP1 and NP2 disambiguations is slightly smaller in the LS (long NP1, short NP2) condition than in the SL (short NP1, long NP2) condition. It would seem, then, that the overall NP1 attachment preference is modulated by the difference in length between the two attachment sites. Unfortunately, this effect is not
statistically reliable (by participants: F(1,54) = 2.55, p =.116, by items: F(1,31) = 2.2, p =.148). When the results of the LS and SL conditions are analysed separately, we see a significant difference between NP1 and NP2 disambiguation in the SL condition (by participants: t = 2.59, df = 54, p =.012; by items: t = 4.13, df = 31, p <.001), whereas in the LS condition, the effect is marginal (by participants: t = 1.14, df = 54, p =.258, by items: t = 2.23, df = 31, p =.034). Admittedly, this is not exactly a strong result, but it suggests that it is slightly easier to accept an NP2-disambiguation when NP1 is long and NP2 short than when NP1 is shorter than NP2. Reaction times. No effects were found in segments 1 to 5. At the disambiguating auxiliary (segment 6), the average reaction times are 1670 ms (se = 140) in the LS condition and 1566 ms (se = 102) in the SL condition. The difference is not significant (by participants: F(1,43) = 1.92; by items: F(1,31) =.231). Across length manipulations, NP1 disambiguation yields an average reading time of 1708 ms (se = 144). The average reading time in the NP2 condition is 1554 ms (se = 123). This looks like a considerable difference, but it fails to reach significance (by participants: F(1,43) = 1.12, p =.295; by items: F(1,31) = 3.43, p =.073). Figure 3 shows the reaction times for the two disambiguation conditions, as a function of the NP length manipulations. In the LS condition, NP1 disambiguation leads to a longer reading time than NP2 disambiguation: 1933 ms (se = 216) vs. 1460 ms (se = 145). In the SL condition the effect is reversed: 1484 ms (se = 101) for NP1 disambiguation, and 1649 ms (se = 138) for NP2 disambiguation. This interaction is statistically significant (by participants: F(1,43) = 6.5, p =.014; by items: F(1,31) = 8.98, p =.005).

2000 reaction time (ms) NP1 Disambiguation NP2 LS SL
Figure 3. Experiment III: Mean reaction time of accept responses at the disambiguating segment 6 (auxiliary), as a function of the lengths of NP1 and NP2 (LS: long NP1, short NP2; SL: short NP1, long NP2) and disambiguation direction (NP1 or NP2). Summarizing, the rejection rates indicate an overall preference for NP1 attachment, which agrees with previous findings on relative clause attachment in Dutch. The effect of the phonological length manipulation is weak: The apparent penalty on NP2 disambiguation is somewhat increased in cases where the two potential heads of the relative clause are short and long, respectively. The direction of this effect is in agreement with the hypothesis that a
long second NP will invite the insertion of a prosodic break, which blocks low attachment of the relative clause. It is somewhat surprising that the disambiguation contrast did not produce a reliable reaction time difference. There is a numerical trend, which, oddly enough, runs counter to the effect in the rejection rates. It would seem, however, that this trend is primarily due to the very long average reaction time in the case where NP1 disambiguation occurs in the LS condition. This long reaction time most likely also contributes considerably to the interaction of the phonological length manipulation and relative clause attachment disambiguation, which goes in the predicted direction: The LS condition is predicted to induce an NP2 attachment bias, and the data indicate that there is a strong penalty on forced NP1 attachment. The SL condition yields a mirror image effect, albeit less articulated. 5 General Discussion The primary aim of this study was to test the implicit-prosody hypothesis, the idea that readers generate a phonological representation on the basis of the written input. This representation contains prosodic boundaries that impact on syntactic parsing, notably with respect to structural-ambiguity resolution. The present study used a somewhat notorious parsing problem as a window on this matter, viz. the resolution of relative-clause attachment ambiguity in a structure with two potential heads. In order to suppress as much as possible the effects of lexico-semantic and pragmatic factors, the experiments reported here made use of Jabberwocky prose. The crucial manipulation concerned the phonological lengths of the NPs that are accessible as head of the relative clause. The rationale behind this was that if an effect of this phonological manipulation was to be observed (in attachment preference), there would be no other way to account for the data than by reference to an implicit, self-generated phonological representation, matching the phonological structure that would be most naturally generated in the spoken realisation of the string under scrutiny. The picture that emerges from the results is that in resolving relative-clause attachment ambiguity, in silent reading, the phonological lengths of the potential head nouns (Exp. I, III), as well as that of the relative clause itself (Exp. I vs. Exp. II), have an effect. The effects are not very strong; we see modulations of overall attachment preferences. The difference in outcome between experiments I and II agrees with the anti-gravity effect reported earlier. The results of experiments I and III are compatible with the idea that the relative lengths of conjoined NPs affects prosodic phrasing, which in turn modulates RC-attachment bias. The results of this study contradict those reported by Colonna & Pynte (2001), which suggests that adding an adjective to an NP (as Colonna & Pynte did) impacts differently on processing than the purely phonological manipulation applied in this study. Surprisingly, the overall attachment preference (NP1 vs. NP2) varies over tasks. Taking the two questionnaire studies together, we see quite a strong low-attachment preference (modulated by length of the relative clause), which is at odds with previous results on (real) Dutch. The continuous acceptability-judgment task, by contrast, yielded a bias toward NP1 attachment. I have no satisfactory explanation for this difference at present. Fodor (p.c.) suggests that processing Jabberwocky could be such a strain that subjects may resort to visual search in finding the attachment site for the relative clause. This would favor recency, i.e., NP2 attachment. Clearly, this strategy can only be applied when the stimuli are presented in full, not with a moving window presentation; hence the difference in bias between the offline and on-line experiments. It would be interesting to see what will happen when the task of experiments I and II is combined with the presentation mode of experiment III.

The overall effects of the phonological manipulations are modest, and we may ask whether this is a result of the materials, the participants, or both. Conceivably, even in Jabberwocky prose, in which semantic and pragmatic influences on parsing decisions are believed to be virtually absent, other factors (beside syntax and prosody) can affect processing. It is also possible that generating implicit phonology is subject to much variation. Some readers may do it quite systematically, whereas others may never do it. Also, some stimuli may more often induce phonological encoding (in particular readers) than others. A replication of Experiment I using syntactic prose items yielded a similar pattern of results, but much less articulated. One explanation for this difference is that syntactic prose, which contains existing content words (in nonsensical combinations) allows lexical access, and therefore does not engage the phonological processor as much as Jabberwocky in which lexical access is ruled out. This is just to suggest that it might be worthwhile to further explore the conditions that stimulate or suppress phonological effects in visual language processing. In concluding this contribution, I am considering whether Sieb Nooteboom might be inclined to believe my account. Probably not. After all, the inferential chain linking the data to the underlying processes is long and, possibly, seriously convoluted. Nonetheless, I have the firm conviction that the strife of the carmactrolophy that was plimpered in this lasteract has a snuzzling porrifritch. References
Brysbaert, M., & Mitchell, D. G. (1996). Modifier attachment in sentence parsing: Evidence from Dutch. Quarterly Journal of Experimental Psychology, 49(A), 644-695. Colonna, S., & Pynte, J. (2001). Relative clause attachment in French: The role of Fodor's "same-size-sister" constraint. Paper presented at the Workshop 'Prosody in Processing', Utrecht. Cuetos, F., & Mitchell, D. G. (1988). Cross-linguistic differences in parsing: Restrictions on the use of the late closure strategy. Cognition, 30, 73-105. Cutler, A., Dahan, D., & Van Donselaar, W. (1997). Prosody in the comprehension of spoken language: A literature review. Language and Speech, 40, 141-201. Ehrlich, K., Fernndez, E., Fodor, J. D., Stenshoel, E., & Vinereanu, M. (1999). Low attachment of relative clauses: New data from Swedish, Norwegian and Romanian. Paper presented at the 12th Annual CUNY conference on Human Sentence Processing, New York. Fodor, J. D. (1998). Learning to parse? Journal of Psycholinguistic Research, 27, 285-317. Fodor, J. D. (2002). Prosodic disambiguation in silent reading. In M. Hirotani (Ed.), Proceedings of NELS 32 (pp. x-y). Amherst, MA: GLSA. Frazier, L., & Fodor, J. D. (1978). The sausage machine: A new two-stage parsing model. Cognition, 6, 291326. Hahne, A., & Jescheniak, J. D. (2001). What's left if the Jabberwock gets the semantics? An ERP investigation into semantic and syntactic processes during auditory sentence comprehension. Cognitive Brain Research, 11, 199-212. Kimball, J. (1973). Seven principles of surface structure parsing in a natural language. Cognition, 2, 15-47. Nagel, N. H., Shapiro, L. P., Tuller, B., & Nawy, R. (1996). Prosodic influences on the resolution of temporary ambiguity during on-line sentence processing. Journal of Psycholinguistic Research, 25, 319-343. Pynte, J., & Prieur, B. (1996). Prosodic breaks and attachment decisions in sentence parsing. Language and Cognitive Processes, 11, 165-191. Schafer, A., Carter, J., Clifton, C., & Frazier, L. (1996). Focus in relative clause construal. Language and Cognitive Processes, 11, 135-163. Steinhauer, K., Alter, K., & Friederici, A. D. (1999). Brain potentials indicate immediate use of prosodic cues in natural speech processing. Nature Neuroscience, 2, 191-196. Watt, S. M., & Murray, W. S. (1996). Prosodic form and parsing commitments. Journal of Psycholinguistic Research, 25, 291-318.