Language Activity Monitoring (LAM)
Technical Guide for Developers
Language activity monitoring (LAM) is the automatic recording of the content and time of language events generated using AAC systems. LAM makes practical the collection of language samples, which can then be analyzed, for the support of AAC evidence-based practice. The primary LAM implementations are 1) as a built-in feature in modern AAC systems and 2) as software to allow a PC to act as a LAM (U-LAM). References on LAM are available elsewhere on the AAC Institute web site. This document sets forth the definitions needed for the implementation of the LAM function in an AAC system.
LAM data is recorded within the AAC system. After the collection of the language sample, the data is sent or otherwise moved to a computer for analysis. The format for the internal recording is open. However, when the data is sent or moved to the computer, it must comply with a standard format in order to be recognized and processed by the analysis software.
LAM data consists of two required fields and one optional field. For each language event, a time stamp and the content are required. Analysis of a language sample containing this information yields much information of clinical significance. The optional field is a three letter mnemonic indicating how the language event was generated in the AAC system. It has been found that the method used to generate language can have a dramatic impact on communication performance. This optional field provides information that addresses this issue.
The LAM file can start with a header of unspecified structure. It is recommended that the header include two components. First is the name and version of the AAC system. Second is a privacy notice. A sample is presented here:
### CAUTION ###
The following data represents personal communication.
Please respect privacy accordingly.Language Activity Monitor "device name"
Version 2.00 07/26/01
ACME AAC CompanyThe following example reflects the required format for LAM data reporting.
09:27:17 OWS "I "
09:27:19 OWS "am "
09:27:22 SMP "hungry "
09:27:24 OWS "and "
09:27:26 OWS "I "
09:27:29 SMP "want "
09:27:34 SPE "s"
09:27:36 SPE "o"
09:27:38 SPE "m"
09:27:43 WPR "some "
09:27:49 DWP "something "
09:27:51 OWS "to "
09:27:58 PAG "eat "
First Required Field
The first field in the language event reporting is the time stamp, reported with one second resolution. When the AAC system includes a real time clock, it is expected that the time stamp is derived from that clock and thus reports the actual time of day. If the AAC system does not have a real time clock, then the time stamp can be derived from an internally generated timer, starting at 00:00:00 when the system is powered up. If the clock does not reflect real time, this may be noted in the header.
The format is hh:mm:ss (hours, minutes, seconds) using a 24 hour number system.
If the AAC system includes date information, that can be included when the LAM function starts to record and when the date changes. The format for recording the date is *[YY-MM-DD=04-06-20]*, including the asterisks.
Second Required FieldThe second field is the content of the language event. This can be one or more letters or words that result from a single language generation action. The content is reported within quotation marks in order to make clear the presence of spaces.
Optional Source MnemonicWhen present, the optional three letter source mnemonic is used in the analysis of the language sample. This field is placed between the two required fields. It facilitates the identification of language representation methods used for generating the language events, an important factor in AAC clinical service delivery and research. Some of the summary measures in the AAC Performance Report presently are based on language representation methods. These are used in the analysis of utterances, or parts thereof, that are created using spontaneous novel utterance generation (SNUG). The three basic methods employed in all AAC systems are 1) single meaning pictures, 2) alphabet-based methods (spelling, word prediction, orthographic word selection, etc.), and 3) semantic compaction. Of these, five total variations are presently reported in the AAC Performance Report:
SMP: Single Meaning Pictures
SPE: SPElling
WPR: Word PRediction
OWS: Orthographic Word Selection
SEM: SEMantic compactionEach of these methods could be generated using different strategies. For example, a single meaning picture could be selected from a page or from an activity row. How it is selected may be of some interest and thus LAM allows reporting using other mnemonics. The PeRT (Performance Report Tool) analysis software used to generate the AAC Performance Report automatically assigns other mnemonics into one of the above five methods according to the following table.
Other mnemonics are used to designate other types of events.
PPU: Pre-Programmed Utterance (A PPU is any sentence or combination of words not commonly used in SNUG.)
CTL: ConTroL events, such as the date announcement covered above. Another example is the record of when recording is turned on or off. Such events should be inside the star/bracket combination as in the date example: *[YY-MM-DD=04-06-20]* If the logfile, does not include the mnemonic, PeRT automatically assigns CTL to events inside the star/bracket combination.
An AAC system developer or manufacturer wishing to use other unassigned mnemonics is encouraged to report such use to support@aacinstitute.org.
ACTivity row AUDio events recorded by U-LAM Derivative of Word Prediction (when the word prediction process uses more than one step) KEYboard events on a computer running the U-LAM program Orthographic Word Selection PAGe PreProgrammed Utterance SEMantic compaction SERial port event recorded by U-LAM SMP
Single Meaning Pictures SPElling Word PRediction It is possible that a strategy generally reserved for one language representation method might be used to implement another. For example, letters of the alphabet might be used on a page and thus the mnemonic will be wrong. When this is the case, the SLP or other user of the LAM file may open it in a word processor and use the REPLACE function to change them. Or they may be changed in the analysis program. PeRT automatically identifies individual letters and assigns them the SPE mnemonic, regardless of what mnemonic accompanied them in the original LAM file.
When a mnemonic is not present, a default language representation method may be assigned by the analysis software. This default method can then be overridden by exception.
Field Separation
The fields are separated by a single space.
Event Separation
The events are separated by a new line.
Controls
LAM controls need to allow:
1) the recording to be turned on and off,
2) the LAM file to be cleared, and
3) the LAM file to be sent or moved to a computer.These controls can be embedded in the AAC system with like controls for other functions. However, it is important that the system can be configured so that the person using it has convenient access to at least the function of turning it off.
Means of Transfer
A means of sending or moving the LAM file from the AAC system into a computer for analysis must be provided. This could be 1) a connection to the computer (serial, USB, IR, etc.) or 2) a means of saving the file and moving it intact (floppy disk, USB memory, Flash card, etc.).
When the transfer is implemented using a connection to the computer, LAMterm software may be used in the computer to receive the file. LAMterm is a free download from the AAC Institute web site: www.aacinstitute.org. LAMterm awaits serial port data and when the data string is interrupted it provides for storing the data that has been received. Therefore, if this method of transferring the data is used, be sure that the string is not interrupted.
When the transfer is implemented using mass storage media, the data should be saved as a text (.txt) file.
Acknowledgments
The LAM development and evaluation has been supported by two NIH (National Institutes of Health) SBIR (Small Business Innovation Research) grants awarded to Prentke Romich Company in collaboration with the University of Pittsburgh and Edinboro University of Pennsylvania. The research components of these grants were coordinated by Katya Hill, Ph.D., CCC-SLP.
NIH Grant No. 1 R43 DC04246-01. Augmentative Communication Language Activity Monitor. Awarded September 1999.
NIH Grant No. 1 R43 DC04472-01. Digitized Speech Language Activity Monitor (D-LAM). Awarded September 2000.
Comments and Questions
Please direct comments and questions on this Technical Guide for Developers to support@aacinstitute.org.
