Table of contents

What is the Moss Aphasia Psycholinguistic Project Database? ↑

The Moss Aphasia Psycholinguistic Project Database gives researchers access to experimental behavioral data from Aphasia patients. The web database represents years of data collection from patients who exhibited language impairments secondary to chronic left hemisphere stroke. Most were recruited to the research program at Moss Rehabilitation Research Institute (MRRI).

The core of the database is individual-trial performance data from picture naming. Picture naming is a primary test of lexical processing. The task taps a critical juncture in the language system because naming mediates between high-level conceptual and syntactic processing and low-level phonological processing. Difficulty in this task is present to varying degrees in nearly all aphasic individuals.

This site allows researchers to search through naming data from 178 patients. Searches can narrow in on data subsets based on patient characteristics (e.g. time since aphasia onset, clinical diagnosis), stimulus characteristics (e.g. semantic category, lexical frequency) and task performance (e.g. error type). The data available on this site can be used to test hypotheses about naming impairment and aphasic impairment generally. Once the basic analysis tools of the site are exhausted, users can export the raw data for further analysis and visualization.

What kinds of results can I get and how can I use them? ↑

Search results come out as trial-level data including information about task performance, trial stimuli and patient state. In addition to picture naming, other test data are included in the database for subsets of the patient population. In all there are 88 data fields for each entry in the database. The user decides which fields they want to display in the output using three sets of check boxes: patient info, stimulus info, and test performance info. A complete glossary is provided below.

Register to get access to the data ↑  

In compliance with ethical and regulatory standards, users of the site must register and be approved by a site administrator before getting access to the search form. Register here. Once you have registered, you will also have access to the Q&A section, where you can post questions on any topic for administrators or fellow users to answer.

One-click results with simple search ↑

Once registered, access the simple search page from the main search interface.

The simple search page gives you useful search results with a single mouse click. Before diving into  advanced searches, use the simple search to familiarize yourself with the format of results and the logic of designing searches.

Let's walk through the process using simple search #5:

• Click on load results. The results will load.
• Hover the mouse over different cells to see field descriptors; consider the significance of the results in relation to the search description. Each row shows a different type of formal error (i.e. an error of syntax or morphology) made with respect to the stimulus word ‘belt’. Each row gives multiple examples because the results collapse over multiple subjects.
• 
  
• Scroll up and review the search description. What are the elements of the search? Most searches will have just a few elements.
• 
  
• Scroll up further and review how the search is actually built in the form. You may need to expand some subsections of the search form by clicking on their headings. The search form will be populated with the values of the currently loaded search, so you can examine exactly what goes into making the search.

Repeat these steps for a few of the simple searches, and you will be ready to design your own search in a short time.

Getting started with advanced search ↑

The advanced search form gives users full and flexible access to the database. Setting up a search consists of defining constraints on data fields, selecting which fields should be displayed in the output, and making output settings that affect how the results are display (including powerful average by… and sort by… parameters). Once the form has been completed, press the submit button to get results. When the page reloads, the search form will contain all the same values that were entered, with the results table below.

• Section 1: Choose scope of search, either Philadelphia Naming Test (default) or Philadelphia Repetition Test.
• Sections 2-5: There are four sections for setting search constraints:
  
  • Constrain by patient characteristics, e.g. diagnosis, # months post onset.
  • Constrain by test items, e.g. semantic category, lexical frequency, imagibility
  • Constrain by performance outcomes, e.g. correct, semantic error, formal error, etc.
  • Constrain by data from specific publication(s)
• Section 6: Output settings determine which data fields get displayed in results (among the total of 88 fields), whether data should be averaged or sorted, and a few other options.

Explanation of output settings

The first part of output settings involves selecting which fields to display in the output. There are three sections: patient info, stimulus info, test performance info. Each section contains a group of check boxes. The user checks those fields that they want to have in the output. (See figure above.) In each section there is also three buttons: Select basic set, Select all, and Select none. Pressing these buttons automatically checks the corresponding checkboxes for each section. The 'basic set' for each section is a subset of checkboxes that are most commonly used. 

The second part of the output settings is picture here:

• 'Limit number of results' sets how many rows of results are displayed per page.
• 'Output type' sets whether results comprise trial level data, or should be averaged across one data field. The user selects which field the results should be averaged across from a drop-down menu. Two levels of averaging can be accomplished by selecting the 'Average by (second order)' option. 
• 'Display data as...' provides two options: Normal displays data in a formatted table, which contains tips that are displayed when the mouse is hovered over each field. The 'simple text' option displays the data as simple ascii text, that is more easily copied and pasted into an external application.
• 'Open new window' causes the search results to be sent to a new browser window. This option can be useful if a user wants to load a series of similar searches and view them in parallel in separate  browser windows.
• 'Filter out null values' excludes a results that include one of more null values. This option is useful for beginning users who are familiarizing themselves with the various data fields, and want an easy way to generate results for all fields, without worrying about null results. 
• 'Sort results by...' sorts the results by the selected data field. Option: ascending or descending.

Step-by-step directions for using the database to answer three basic questions ↑

1. How to compare the different errors made on PNT among aphasia subtypes
2. How to see if PNT performance improves across time
3. How to compare error type by semantic category and diagnosis
4. How to explore types of semantic errors

How to compare the different errors made on PNT among aphasia subtypes

1. Scope of Search, Constrain by patient characteristics, constrain by test items, constrain by performance outcomes and constrain by data from specific publications can remain as they are.
2. Click on Output settings.
   
   • Patient Info can remain as is.
   • Click on Stimulus Info and de-select Imageability and Lexical frequency.
   • Click on Test performance info and make sure the basic set is selected.
3. Under Output Type, click Average by (first order)
   
   • Select Diagnosis from the drop-down menu
4. Under Display data as, make sure that Normal is selected.
5. You may choose to display results in a new window by clicking the check box under open new window.
6. Constrain by data set can remain as is.
7. Click Submit. Your search results should then appear.
8. You should see a breakdown of patient response type ranging from correct to non-word errors.   Proportions of each error type are averaged by diagnosis. From here you will be able to see the varied performance among different groups.
9. Once the results are displayed, you can hide certain columns by clicking on Select fields to display above the results table. Then de-select the columns you do not want to view.

How to see if PNT performance improves across time

1. Scope of Search, Constrain by patient characteristics, constrain by test items, constrain by performance outcomes and constrain by data from specific publications can remain as they are.
2. Click on output settings
   
   • Within patient info, make sure that Test Iteration is checked in addition to the basic set
   • Within stimulus info, select none.
   • Test performance info can remain as is.
3. Under Output Type, click Average by (first order)
   
   • Select Anonymous Subject ID from the drop-down menu
4. Under Output Type, click Average by (second order)
   
   • Select Test Iteration
5. Under Display data as, make sure that Normal is selected.
6. You may choose to display results in a new window by clicking the check box under open new window.
7. Under Sort results by, choose diagnosis from the first drop-down menu.
8. Constrain by data set can remain as is.
9. Click Submit. You should then see your search results.
10. You should see results organized by individual patients; these will include patients who have only been tested one time. In order to see improvement across time, you can look at those who have multiple test iterations.
11. Once the results are displayed, you can hide certain columns by clicking on Select fields to display above the results table. Then de-select the columns you do not want to view.

How to compare error type by semantic category and diagnosis

1. Scope of Search, Constrain by patient characteristics, constrain by test items, constrain by performance outcomes and constrain by data from specific publications can remain as they are.
2. Click on output settings
   
   • Patient info can remain as is.
   • Within stimulus info, select Semantic category, in addition to the Basic set.
   • Test performance info can remain as is.
3. Under Output Type, click Average by (first order)
   • Select Diagnosis from the drop-down menu
4. Under Output Type, click Average by (second order)
5. Select Semantic category from the drop-down menu
6. Under Display data as, make sure that Normal is selected.
7. You may choose to display results in a new window by clicking the check box under open new window.
8. Under Sort results by, choose Semantic category from the first drop-down menu.
9. Constrain by data set can remain as is.
10. Click Submit. You should then see your search results.
11. You should see results averaged by patient diagnosis and broken down into semantic categories. These results allow for comparison among different diagnoses on a single semantic category or among semantic categories for a single diagnosis.
12. Once the results are displayed, you can hide certain columns by clicking on Select fields to display above the results table. Then de-select the columns you do not want to view.

How to explore types of semantic errors

1. Scope of Search, Constrain by patient characteristics and Constrain by test items can remain as they are.
2. Click on Constrain by performance outcomes
   • Select Semantic error from the second drop-down menu.
   • Select ‘is true’ from the third drop-down menu.
3. Constrain by data from specific publication(s) can remain as is.
   
   • Click on output settings
   • Patient Info and Stimulus info can remain as they are.
4. Within test performance info, Click Select none. Make sure that Semantic error, Phonetic Response and Regular Response are checked.
5. Constrain by data from specific publications can remain as is.
6. Under Output Type, make sure that Individual responses is selected.
7. Under Display data as, make sure that Normal is selected.
8. You may choose to display results in a new window by clicking the check box under open new window.
9. Constrain by data set can remain as is.
10. Click Submit. Your search results should then appear.
11. You should see an alphabetical listing of the PNT test words for which semantic errors occurred. The results are then further sorted by diagnosis. You can then compare the different subtypes of semantic errors among patient populations. These subtypes include superordinate, subordinate, category coordinate, taxonomic and associate errors. This further exploration needs to be performed according to your own error classifications.
12. Once the results are displayed, you can hide certain columns by clicking on Select fields to display above the results table. Then de-select the columns you do not want to view.

About the Philadelphia Naming Test ↑

The PNT is a 175-item picture naming test developed in the Language and Aphasia Lab of MRRI (Dir., Myrna Schwartz, Ph.D.) for the psycholinguistic exploration of lexical access in nonaphasic and aphasic speakers (Roach et al., 1996).

For more information about PNT and to download complete testing materials, follow this link http://www.ncrrn.org/assessment/pnt/

Glossary of search parameters ↑

(coming soon)

References ↑

• Dell, G. S., Schwartz, M. F., Martin, N., Saffran, E. M., & Gagnon, D. A. (1997). Lexical access in aphasic and nonaphasic speakers. Psychological Review, 104, 801-838. PMID 9337631
• Roach, A., Schwartz, M.F., Martin, N., Grewal, R.S., & Brecher, A. (1996). The Philadelphia Naming Test: Scoring and Rationale. Clinical Aphasiology, 24, 121-133.
• Schwartz, M.F., Dell, G.S., Martin, N., Gahl, S., & Sobel, P. (2006). A case-series test of the interactive two-step model of lexical access: Evidence from picture naming. Journal of Memory and Language, 54, 228-264.