JSONata mapper will not split any incoming arrays. If your integration flow logic requires splitting of the incoming array then the Splitter component must be used beforehand. If the Splitter component is not used the arrays are going to be passed as it is to the next step. 

To demonstrate the process let us assume an incoming array:

{
  "author" : "Twain",
  "book":"Adventures",
  "characters":[
    { 
      "fname":"Tom",
      "lname":"Sawyer",
      "town":"StPetersburg",
      "state":"Missouri"
    },
    { 
      "fname":"Huckleberry",
      "lname":"Finn",
      "town":"StPetersburg",
      "state":"Missouri"
    }
  ]
}

To split the array characters we add the Splitter component:

Since here we have only one array then we will see the following:

After selecting the array to be split we have the following interface:

Here we can select the data sample with which we want to proceed further and perform further testing. Since the incoming array had only 2 separate records we have only two data samples to choose from. This is the tree view of the incoming data sample and by clicking on JSON view you will be presented with the JSON structure:

JSONata mapper and the data aggregation.

If we have an incoming array structure we can use both the already split array values with the original incoming intact arrays. Here is how to do it. We select normally the mapping fields but now we scroll down to select data from the previous steps: 

In the message Body field, we have already used the split values from characters array. Now we want to select the intact character array.

And the resulting mapping would look like:

Here we clearly see that the initial array is left intact and passed to the next step.